Professional Documents
Culture Documents
Dina D'Ayala, Enrico Fodde - Structural Analysis of Historic Construction_ Preserving Safety and Significance_ Proceedings of the VI International Conference on Structural Analysis ... SAHC08, 2-4 Jul
Dina D'Ayala, Enrico Fodde - Structural Analysis of Historic Construction_ Preserving Safety and Significance_ Proceedings of the VI International Conference on Structural Analysis ... SAHC08, 2-4 Jul
VOLUME 1
Editors
Dina D’Ayala & Enrico Fodde
Department of Architecture & Civil Engineering,
University of Bath, Bath, UK
CRC Press/Balkema is an imprint of the Taylor & Francis Group, an informa business
All rights reserved. No part of this publication or the information contained herein may be reproduced, stored
in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, by photocopying,
recording or otherwise, without written prior permission from the publishers.
Although all care is taken to ensure integrity and the quality of this publication and the information herein, no
responsibility is assumed by the publishers nor the author for any damage to the property or persons as a result
of operation or use of this publication and/or the information contained herein.
Table of Contents
Preface XV
Committees XVII
VOLUME 1
Invited papers
Preservation of workmanship or workmanship for preservation 3
K. Van Balen & R. Hendrickx
Experiences from the Northern Italy 2004 earthquake: Vulnerability assessment and
strengthening of historic churches 13
E. Giuriani & A. Marini
The structural behaviour of masonry vaults: Limit state analysis with finite friction 47
D.F. D’Ayala & E. Tomasoni
Cultural Heritage protection against flood – A European FP6 research project 109
Ch. Maierhofer, Ch. Köpp, S. Kruschwitz, M. Drdacky, Ch. Hennen, S. Lanza, M. Tomaževič,
D. Kolymbas, I. Herle, A. Virsta, L. Binda & P. Askew
V
Analysis of archaic fireproof floor systems 129
D. Friedman
Intervention limits in floor wood structure restoration of a historical building 137
R.T.M. Ribeiro & M. Hoirisch
The safety of historic load-bearing structures 143
B. Szabó & I. Kirizsán
Concrete: Too young for conservation? 151
H.A. Heinemann, R.P.J. van Hees, T.G. Nijland
Significance of a Mughal Mall and its role in achieving safety 161
S. Sanyal
Historical constructions – Authenticity and adaptation to the modern demands 169
I. Bucur-Horváth, I. Popa, M. Bulbuk & J. Virág
Champaner-Pavagadh – Managing conflicts – A conservation challenge 175
S.M. Modi
VI
Structure as a document 289
J. Jain
VII
Diagnosis and analysis of two king-post trusses 459
J.M. Branco, P.J.S. Cruz & M. Piazza
Non destructive and laboratory evaluation of strength of decayed wood members in a historic
construction located in Gorgan (North of Iran) 469
M. Madhoushi, J. Eimanian & M.P. Ansell
The structural behaviour of spires 473
E. Olosz & B. Szabó
In-situ and laboratory investigations on materials and structures for the static improvement:
“Torre Quadrata del Circo Romano” (Milan) 481
S. Bortolotto, E. Ciocchini, F. Zangheri & P.M. De Marchi
Dynamic multi-body behaviour of historic masonry buildings models 489
D. D’Ayala, Y. Shi & C. Stammers
Assessment and strengthening masonry arch bridges 497
C.L. Brookes
Field tests of a timber queen-post truss and numerical analysis 507
J.M. Branco, P.J.S. Cruz, M. Piazza & H. Varum
VIII
Earthquake based condition monitoring of the Katholikon of Dafni Monastery 621
H.P. Mouzakis, A. Miltiadou, N. Delinikolas, P. Touliatos & J. Dourakopoulos
Continuous and discontinuous modeling of the structures in Bayon temple, Angkor 629
T. Maeda, T. Yamamoto & T. Hirai
Safety assessment of temple E7 in Mỹ So,n, Viet Nam 635
P. Condoleo, A. Taliercio & L. Binda
Construction process, damage and structural analysis. Two case studies 643
R. González, F. Caballé, J. Domenge, M. Vendrell, P. Giráldez, P. Roca & J.L. González
Analysis and diagnosis of timber structures of Porto historical centre 653
T. Ilharco, J.M. Guedes, A. Arêde, E. Paupério & A.G. Costa
Baroque roof structures in Transylvania – Research and analyses 663
D. Makay & B. Szabó
The Athens Parthenon: Analysis and interpretation of the structural failures of the orthostate of
the northern wall 673
E.E. Toumbakari
Structural assessment of the Winter Garden of the Royal Glasshouses of Laeken, Belgium 683
L. Lauriks, I. Wouters, S. Adriaenssens, M. de Bouw & B. Verbeeck
Towards a systematic diagnosis of structural damage 689
I.A.E. de Vent, R.P.J. van Hees & G.J. Hobbelman
From geometrical and cracks survey to static analysis method: The case study of Santa Maria
del Quartiere dome, in Parma (Italy) 697
F. Ottoni
A benchmarking study of the analysis of non-reinforced structures applied to the structural
behavior of domes 705
P. Morer & R. Goñi
Structural analysis of the Caserta Royal Palace timber roof connections 715
G. Fabbrocino, G. Marcari, C. Laorenza & E. Cosenza
Real-time Probabilistic health monitoring of the Portogruaro Civic Tower 723
D. Zonta, M. Pozzi, P. Zanon, G.A. Anese & A. Busetto
VOLUME 2
Experimental studies
On the strengthening of three-leaf stone masonry walls 739
R.A. Silva, D.V. Oliveira & P.B. Lourenço
Long-term behaviour of historical masonry – A quantitative acquisition of the damage evolution 747
E. Verstrynge, L. Schueremans, D. Van Gemert, S. Ignoul & M. Wevers
Experimental assessment and modelling of in-plane behaviour of timber floors 755
M.R. Valluzzi, E. Garbin, M. Dalla Benetta & C. Modena
Bond behaviour of CFRP and GFRP laminates on brick masonry 763
M. Panizza, E. Garbin, M.R. Valluzzi & C. Modena
Experimental shear behavior of stone masonry joints 771
G. Vasconcelos, P.B. Lourenço & D. Oliveira
IX
Large scale testing of drystone retaining walls 781
C. Mundell, P. McCombie, A. Heath, J. Harkness & P. Walker
Seismic isolation, strengthening of walls with CFRP strips and heritage masonry buildings 789
M. Tomaževič, I. Klemenc & P. Weiss
Experimental survey on seismic response of masonry models 799
A. Baratta, I. Corbi, O. Corbi & D. Rinaldis
Bond performance of fiber reinforced grout on brickwork specimens 809
I. Carbone & G. de Felice
Experimental analysis of rectangular shaped sleeve connectors for composite
timber-steel-concrete floors: Bending tests 817
L. Calado, J.M. Proença, A. Panão, F.M. Mazzolani, B. Faggiano & A. Marzo
Shaking table testing of a typical Mexican colonial temple 825
M. Chávez & R. Meli
Capacity of a traditional timber mortise and tenon joint 833
A.O. Feio, P.B. Lourenço & J.S. Machado
Behaviour of refurbished timber floors characterized by different in-plane stiffness 843
M. Piazza, C. Baldessari, R. Tomasi & E. Acler
Stone masonry walls: Strengthening with TRM (I) 851
J.T. San-José, D. García, R. San-Mateos & J. Díez
The effect of friction joint and Gongpo (bracket set) as an energy dissipation in Korean
traditional wooden structure 861
J.K. Hwang, S.G. Hong, N.H. Kim, Y.W. Lee, S.J. Jeong & S.J. Joo
Optimization of cutting processes in archaeological sites 867
C. Cennamo, B.M. Chiaia, E. Masoero & S. Scaini
Experimental and numerical study on the application of the flat-jack tests to masonry walls 875
M. Acito, L. Binda, G. Cardani, J. Guedes & L. Miranda
The impact of the water/lime ratio on the structural characteristics of air lime mortars 885
R.M.H. Lawrence & P. Walker
Soil mortar’s evaluation characteristics for their use in conservation practice 891
G.E. Bei
Experimental and numerical analyses for static retrofitting intervention on the “Cappella
dei Principi” in Firenze 899
A. Antonelli, G. Bartoli & M. Betti
Comparison the effect of different repair method applying on masonry arch bridges 909
M. Miri & T.G. Hughes
X
Cross laminated timber panels to strengthen wood floors 949
A. Gubana
Lime mortar with natural pozzolana: Historical issues and mechanical behavior 957
E. Sala, I. Giustina & G.A. Plizzari
Blended lime-cement mortars for conservation purposes: Microstructure and strength
development 965
O. Cizer, K. Van Balen, D. Van Gemert & J. Elsen
Assessing workability of mortar by means of rheological parameters and desorptivity 973
R. Hendrickx, K. Van Balen & D. Van Gemert
Consolidation and reinforcement of stone walls using a reinforced repointing grid 981
A. Borri, M. Corradi, E. Speranzini & A. Giannantoni
Innovative and sustainable local material in traditional African architecture – Socio cultural
dimension 991
T.O. Odeyale & T.O. Adekunle
Imported construction materials and techniques in 19th century Ottoman architecture 999
U. Yergün & A. Çiftçi
Advance composites applications on historical structures in Italy: Case studies and future
developments 1007
P. Casadei & E. Agneloni
XI
Special structural solutions for adaptive use in grand hotel in Norwich, CT, USA 1107
J.F. Norden
The restoration of the foundation in the north side of the temple of Apollo Epikourios 1113
K. Papadopoulos
Structural consolidation methods for the Temple of Santa Maria della Consolazione in
Todi (Perugia, Italy), damaged by landslides and earthquakes 1123
M. Mariani
The diagnosis and arresting of settlement within Westminster Hall in the Houses of Parliament, London 1133
J.D. Miller
Istanbul – Fatih, Millet Library / Feyzullah Efendi Madrasah restoration 1141
R. Ozakin & A. Erdem
Application of non-destructive techniques at the Katholikon of Dafni Monastery for mapping
the mosaics substrata and grouting monitoring 1149
Ph. Côte, X. Dérobert, A. Miltiadou-Fezans, N. Delinikolas, O. Durand, J. Alexandre,
A. Kalagri, M. Savvidou, D. Chryssopoulos, L. Anamaterou & F. Georganis
The PROHITECH research project 1157
F.M. Mazzolani
The column of Marcian/Kıztaşı in Istanbul and an evaluation of its restorations 1167
A. Erdem & R. Ozakin
Restoration and strengthening strategies for 19th century iron pedestrian suspension bridges 1175
S. Adriaenssens, B. Verbeeck, I. Wouters & M. de Bouw
Reconstruction of the Sistani House at Bam Citadel after the collapse due to the earthquake 2003 1181
W. Jäger & C. Fuchs
XII
The effects of temperature on historical stone masonry structures 1271
C. Blasi & E. Coïsson
A GIS platform on main natural hazards for Valparaíso city (Chile) and vulnerability studies
for some historical constructions and urban sectors 1277
M. Indirli, F. Geremei, C. Puglisi, A. Screpanti, D. Blersch, L. Lanzoni, N. Lopez Izquerdo,
E. Milani, M. Miglioli, G. Simonini, M. Munari & F. Romanelli
The construction of a GIS database for the historical center of San Giuliano di Puglia 1287
S. Mazzanti, M. Miglioli, M. Nascosi, G. Simonini, M. Indirli, L. Lanzoni, S. Teston,
C. Alessandri & D. Pini
Vulnerability assessment of churches at Colima by 3D limit analysis models 1297
A. Orduña, A. Preciado, F. Galván & J.C. Araiza
Vulnerability screening by visual inspection of churches of North-East Anatolia 1303
A. Korkmaz, S. Toker & H. Hastemoglu
Seismic resistance assessment of heritage masonry buildings in public use in Ljubljana 1311
M. Lutman
Conservation practice
Relation between sub-soil and masonry structure of Angkor monument 1399
S. Yamada, M. Araya & T. Nagai
Beaufort Castle, Lebanon; conservation versus restoration project 1407
J. Yasmine
Structural faults in earthen archaeological sites in central Asia: Analysis and repair methods 1415
E. Fodde
XIII
Maria Birnbaum – Construction history, conservation history 1423
R. Barthel, H. Maus & C. Kayser
The structural strengthening of early and mid 20th century reinforced concrete diaphragms 1431
L. Sorrentino & C. Tocci
Strengthening design of Ganxi’s Former Residence 1441
C. Qing
The Trajan Markets and their Great Hall – The conservation problems and the structural
intervention for the improvement of the seismic safety 1445
G. Croci, A. Viskovic, A. Bozzetti, L.Ungaro & M. Vitti
Structural strengthening of the Dobrzyca Palace, Poland 1455
M.Y. Minch & J.P. Szołomicki
Structural analysis and restoration of the Guglia della Madonna di Bitonto 1461
T.M. Massarelli
Non conventional solutions for the consolidation of bell towers 1467
C. Mileto, F. Vegas López-Manzanares
Strengthening of a wooden covering built in the 18th century 1475
N. Augenti
Experiences of consolidation on archaeological UNESCO sites in the Sultanate of
Oman: The Fortress of Al Balid and the Citadel of Sumhuram 1481
M. Andreini, C. Cei, G. Mariani & M. Sassu
Conservation and restoration of a Ta Prohm temple 1491
T. Lakshmipriya
Restoration of a historical building for Cyprus Technical Chamber 1499
M. Pittas
The great hall of the Trajan Markets throughout the centuries 1509
L. Ungaro, M. Vitti & E. Speranza
The XIX century opera theatre of Catania: Conservation and seismic strengthening interventions 1519
C.F. Carocci & C. Tocci
House-towers in Campi Flegrei: History, conservation and re-use 1527
G. de Martino, R. de Martino, A. Pane & F. Delizia
Structural inspection and analysis of former British Consulate in Shanghai 1537
X.L. Gu, B. Peng, X. Li & D.F. Shang
XIV
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Preface
The successful preservation of an historic building, complex, or city, unless assigned to museum status, depends
on its continued use and the daily care and maintenance. The possibility of continued use depends on the
adaptation of the building to modern standards and practice of living, and this might require changes of its
constructional or structural features.
Conservation engineering can be defined as the process of understanding, interpreting and managing the
architectural heritage to safely deliver it to posterity. The term ‘safely’ embodies the assumption that the bodies
and individuals responsible for the care of historic buildings work towards ensuring maximum private or public
utility vis a vis minimum loss of fabric and value.
The conflict between safety standards and conservation philosophy usually stems from the fact that not
just the standards themselves, but the practice of achieving the standards, are based on, and refer to, modern
materials, techniques and processes. An historic building is considered of value not only because of its age,
but most importantly because of its uniqueness, its deviation from the norm and hence, to a certain extent,
from what is standard. This constitutes the building’s significance or authenticity, a quality vital to conserve.
With increasing global interest in conservation worldwide it is essential to open the debate on more inclusive
definitions of significance and on more articulated concepts for safety and for acceptable and reliable techniques,
in an attempt to reconcile and integrate further the activity of all the professions involved in conservation.
It is in this framework and with these aims that the Department of Architecture and Civil Engineering of the
University of Bath set out to organize and deliver the sixth international conference on Structural Analysis of
Historic Construction to be held in Bath’s Assembly Rooms between 2 and 4 July 2008.
A grade I listed building designed in 1769 by John Wood the Younger, the Assembly Rooms are located in
the heart of the World Heritage city. Their function was to host events and gatherings and this makes it the
perfect venue for the conference, which is the last of a series that is organized every three or two years. Previous
conferences were organized in New Delhi (2006), Padova (2004), Guimaraes (2001), and Barcelona (1998 and
1995).
The event is sponsored by CINTEC International Ltd, Gifford, English Heritage, ICOMOS, IStructE, and the
University of Bath. This sponsorship has made possible the participation to the conference of authors from other
parts of the world which could have not attended otherwise. Their contribution is critical to our understanding
of practical and theoretical issues in different contexts and to the widening and inclusivity of the debate.
The joint Department of Architecture and Civil Engineering of the University of Bath has a long standing
tradition of teaching conservation and sustainability within the built environment. The MSc in Conservation
of Historic Buildings provides education on the fundamental principles of conservation and is one of the most
successful in the country. Furthermore, the Department offers the only course on Conservation Engineering in
the UK, combining structural aspects with the study and conservation of the built heritage.
The main objective of the conference is to promote debate on an international scale, by the exchange of knowl-
edge and approaches among researchers and practitioners in the different professions involved in conservation and
from diverse cultural backgrounds. The conference is structured into eleven themes: ‘Definition of Significance
and Attribution of Value’, ‘History and Documentation’, ‘Monitoring, non Destructive Evaluation and Testing’,
‘Assessment and Analytical Techniques’, ‘Experimental Studies’, ‘Use of Traditional, Alternative and Innova-
tive Materials’, ‘Novel Conservation Engineering Techniques, Restoration and Strengthening’, ‘Vulnerability to
Natural Hazards, Preparedness and Retrofitting’, ‘Codes, Guidelines and Methods for Safeguarding Safety and
Significance’, and ‘Conservation Practice’.
A total of 180 contributions are collected in the two volumes of proceedings, of which 10 are invited papers
presented by speakers of international renown to the conservation audience. The reading of these proceedings
shows that the discipline of structural analysis of historic buildings is nowadays critical to the success of many
conservation projects and the skills required are increasingly complex and more articulated. The papers collected
all show the need for multidisciplinary approach in conservation research and practice and the continually
evolving need for more sophisticated solutions where both safety and significance play a critical role. An
increasing number of papers tackle issues not covered in previous conferences, such as the structural conservation
of modern materials. It is hoped that the papers presented here can form the basis for future research and the
XV
formulation of relevant issues that need to be advanced. This we hope will constitute the basis for a lively debate
during the 3 days of the conference and will inform and stimulate the directions for future conferences.
The editors are grateful to Marion Harney for managing the organization of the conference and to the depart-
mental office for the great support. The editors are also indebted to the Advisory Committee members for
providing the opportunity and the needed guidance to organize this conference, to the Organizing Commit-
tee members for the efforts and support given to make the conference possible, and finally but crucially to all
members of the Scientific Committee which have agreed to advise in the papers’ selection and reviewing process.
March 2008
Dina D’Ayala, Enrico Fodde
XVI
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Committees
Advisory Committee
Paulo Lourenço, University of Minho, Portugal
Claudio Modena, University of Padua, Italy
Pere Roca, Politecnico de Cataluña, Spain
Organising Committee
Dina D’Ayala, University of Bath, UK
Marion Harney, University of Bath, UK
Enrico Fodde, University of Bath, UK
Alex Copping, University of Bath, UK
Michael Forsyth, University of Bath, UK
Peter Walker, University of Bath, UK
Susan Denyer, Secretary ICOMOS-UK
Geoff Clifton, Chairman, Gifford, UK
Terry Girdler, Chief Engineer, English Heritage, UK
Scientific Committee
Takayoshi Aoki, Nagoya City University, Japan
Görün Arun, Yildiz Technical University, Istanbul, Turkey
Koenraad van Balen, Katholieke Universiteit Leuven, Belgium
Luigia Binda, Politecnico di Milano, Italy
Marcial Blondet, Catholic University of Peru, Peru
Miloz Drdácký, Institute of Theoretical and Applied Mechanics, Czech Republic
Peter Elliott, Elliott & Company, UK
Xianglin Ling Gu, Tongji University, China
Mehrdad Hejazi, University of Isfahan, Iran
Peter James, Cintec International Ltd, UK
Debra Leafer, University College Dublin, Ireland
Giorgio Macchi, Università di Pavia, Italy
Christiane Maierhofer, Fed Inst Material Research and Testing (BAM), Germany
Roberto Meli, Universidad Nacional Autónoma de Mexico, Mexico
John Ochsendorf, MIT, USA
Daniel Oliveira, University of Minho, Portugal
Maurizio Piazza, Universitá Degli Studi di Trento, Italy
Jan G. Rots, Technical University of Delft, The Netherlands
Marek Sklodowski, Institute of Fundamental Technological Research, Poland
Tom Swailes, University of Manchester, UK
Miha Tomazevic, Slovenian National Building and Civil Engineering Institute, Slovenia
Maria Rosa Valluzzi, Universita’ di Padova, Italy
Elizabeth Vintzileou, National Technical University of Athens, Greece
David Yeomans, ICOMOS, UK
XVII
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Conference sponsors
XIX
Invited papers
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The chain of preservation should be integrated into the authenticity debate of conservation. A
particular role can be given to workmanship. A study on repair of ancient masonry structures of a fortress and
its execution show that beside analysis, the exchange with craftsmen is essential in the final decision on the
materials used for repair. Ongoing laboratory investigation revealed that when masons working from various
geographic locations are requested to assess workability of mortar, they seem to agree better than standardized
tests would indicate. This finding together and the previously explained experience indicate that workmanship
is to be considered into the preservation process, as many other interdisciplinary issues. Evaluating the heritage
values of an architectural fabric, craftsmanship embedded in it should be considered, but also the continuation
of practice has heritage values and is part of the preservation process.
3
2. Determination of the composition of the original
mortar through analysis.
3. Determination of the properties of the original and
repair mortar.
4. Formulation of the repair mortar based on above
conclusions and interpretations.
5. Hiring skilled craftsmen and suitable technology,
in order to assist developing and
6. Applying repair mortar.
4
§18 Mortars – ordinary mortars will be composed evaluation of the thin sections which were made from
of one part of slaked lime (Fr.: chaux éteinte) and one some samples.
part of sand. From the analysis possible theoretical composition
of the original mortar could be estimated. Today’s
Information on mortar composition could be sum- reference materials were used for the calculation
marised as follows: taking profit of a program developed for that pur-
– Cinder track mortar will be made from two (2) parts pose. Based on the possible mixtures that fit with
of extinguished lime, one (1) part of sand and one the chemical analysis the following conclusions were
(1) part of cinder track; drawn:
– Hydraulic mortar 1 (for drowned works) will be
composed of one (1) part of trass of Andernach (or 1. There is a difference in mortar composition between
another hydraulic basis approved by the War Depart- the bedding and the pointing mortar;
ment); (this trass is a pozzolanic material exracted 2. The bedding mortar has approx. 50 weight% of sand
in Andernach, Germany) and 50 weight% of binder and most mortars have a
– Hydraulic mortar 2 (for vault capping) will be made moderately hydraulic binder (except M7).
of six (6) parts of slaked lime, one (1) part of sand 3. The pointing mortar is richer in binder and as will
and four (4) parts of Andernach trass (or another be explained later different sand with less glau-
hydraulic basis approved by the War department); conite was used. The pointing in an average has 67
weight% of binder and 33% of sand and the binder
Ambiguity and freedom of action is also included in is moderately hydraulic. The difference in sand may
the descriptions as: also be an indication that pointing was carried out
– In case exceptional circumstances require, the use in another construction phase.
of other type of mortar or the dosage will be deter-
In the next table a re-composition is given based
mined by the commander of the engineer corps.
on the combination of lime hydrate and trass. In
– If the mortar is prepared by labourers, it will have
the last column an alternative binder composition is
to be mixed and will be beaten during two days, it
given for which a higher fraction of the binder can
will be used the third day after it has been re-beaten
be recomposed. The compositions are theoretical as
(with as less as possible water) as to gain a degree
depending on the circumstances of hydration versus
of liquidity (workability?) for its use.
carbonation but also depending on the composition
– Mortar preparation can be accelerated using
of the raw materials that have been used; the final
machines; in that case the commander of the engi-
chemical composition may alter for a same mortar
neer corps will judge on the modifications that have
composition.
to be carried out on the previous stipulation.
From written historical sources we should con-
Those historical descriptions elucidate which mate- clude that normal mortar (including pointing?) should
rials were used, how they were mixed and in which be made from 1 volume of lime and 1 volume of
proportions. Aspects of workability could also be sand which in terms of weight proportions mean 30
found in conjunction with the preparation of the mor- weight% of dry lime hydrate and 70 weight% of sand.
tar. As well workability as other circumstances allowed This is different from the results found in the chem-
for adjustments on the prescribed methods. It is clear ical analysis that would correspond to 2.5 volumes
that other information available today is necessary to dry lime hydrate to 1 volume of sand. However the
decide on the composition of the repointing mortar. description makes reference to lime putty, made from
hydraulic lime. This means that the density of the lime
will be higher and that the water fraction of the putty
2.2.2 Chemical analysis of mortar is considered part of the binder.
Samples for further analysis were identified and were The preparation: slaking, beating of lime for 2 days
extracted from the building. includes the risk that if the lime contains a consider-
– M1 Bedding mortar from sampled core 1. able amount of hydraulic and reactive particles, they
– M1p Pointing mortar from sampled core 1. might hydrate and turn into granulates (they will not
– M2 Bedding mortar from sampled core 2. be able to contribute to the binding anymore) which
– M4 Bedding mortar from sampled core 4. changes the apparent proportions as identified from
– M5 Bedding mortar from sampled core 5. the analysis.
– M7 Bedding mortar from sampled core 7.
– M8 Bedding mortar from sampled core 8.
2.2.3 Thin sections of mortar including
– M8p Pointing mortar from sampled core 8.
mortar-brick interface
Results of the chemical analysis are given in Table 1. The thin sections have been prepared from the samples
Proper interpretation of those results also requires the taken from the masonry. As a general outcome of the
5
Table 1. Chemical and physical analysis, ratio of sand and binder, and recomposition of bedding and pointing mortar samples.
Analysis
W loss 540◦ C (%) 7,47 6,41 5,77 8,30 5,72 5,51 7,26 8,05
W loss 1050◦ C (%) 17,84 19,30 17,77 18,83 13,44 17,31 22,10 22,90
Insoluble residu (%) 47,38 49,32 52,04 45,44 54,34 53,74 33,82 31,70
Soluble SiO2 (%) 5,80 4,46 3,88 5,97 6,80 3,52 5,17 5,77
CaO (%) 22,70 23,07 20,02 21,97 20,78 21,18 29,76 31,17
Total (%) 93,72 96,15 93,71 92,21 95,36 95,75 90,85 91,54
CO2 (%) 14,13 16,37 15,46 16,42 8,82 13,59 19,98 19,89
App. Density 1,814 1,800 1,608 1,785 1,624 1,751 1,819 1,731
(kg/dm3 )
Porosity (vol %) 32,29 33,00 40,62 33,59 38,66 34,05 31,94 35,90
Ratio of sand and binder
Sand (mass %) 47,40 49,30 52,00 45,40 54,30 53,70 50.35 6,48 66 68 67
Binder (mass %) 52,60 50,70 48,00 54,60 45,70 46,30 49,65 6,57 34 32 33
Recomposition
Hydraulicity mod. mod. mod. mod. hyd. mod. mod. mod.
Lime hydrate (%) 70 71 69 66 71 72 69,77 2,76 76 72 74
Trass (%) 19 14 14 19 25 13 17,40 24,55 14 14 14
Alternative binder CEM NHL 2 NHL 2 NHL NHL 5 NHL 2
II + L 3,5
analysis of the thin sections the following conclusions possible. Properties of the mortar used for repair are
were drawn: critical when in contact with the original remaining
mortar parts that have to be conserved. Since matching
1. Difference in mortar composition between the bed-
the composition is no guarantee that the two mor-
ding mortar and the pointing mortar is clear:
tars will have compatible characteristics and future
pointing mortar has a higher binder content and
performance, it is essential to work toward a better
finer fraction while the sand has almost no green-
understanding of original and repair mortar properties.
ish (glauconite) grains that are found in the bedding
Therefore it is crucial that the craftsman is involved in
mortar.
formulating a repair mortar, as he has to do the job
2. Nodules of other materials, probably cinder track
with products or materials he may not be acquainted
are found in the mortar;
with.
3. Adhesion between mortar and bricks is sometimes
Therefore, an important issue during the restoration
very good and sometimes poor; this could be due
of the Fort IV was the repointing experiment of the
to a multitude of reasons, including the preparation
masonry walls as technical, historical and aesthetical
(polishing) of the thin section;
arguments were in play. However it may show how
4. It seems that between bedding and pointing mor-
important the intensive discussions with the crafts-
tar and in case of M8 between 2 layers of pointing
man before and during pointing were. In general we
mortar some “deposit” can be seen; this may be an
state that due to the lack of participation (bottom-
indication that the pointing mortar has been applied
up) within conservation, the craftsman is excluded
later, which could be consistent with the use of
(top-down) from the general thinking and decision
another sand.
making process. He’s familiar with compatibility and
reversibility concepts, yet he isn’t acquainted with
general conservation philosophies or interests.
2.3 Dialogue with the workman As argued by K. Van Balen (Van Balen 2003) it
Masonry conservation of historic structures poses sub- is clear that a uni-directional communication would
stantial dilemmas for architects, conservators, and not lead to an authentic conservation intervention. In
craft workers seeking to incorporate new mortars order to avoid poor craftsmanship due to un-addressed
with composition, performance, and characteristic know-how, it’s important that the different parties dis-
properties compatible to those of original mortars. cuss from the beginning the final technical, but also
In the late 20th century and early 21st century the philosophical context of the conservation.
conservation philosophy embraces the incorporation Architects could act as initiators of this debate in the
of original masonry materials and techniques where conservation team, as the present-day craftsmanship
6
training often focus on technical aspects and hardly
includes more theoretical and philosophical questions.
This early debate could help to assure the involvement
of proper craftsmanship.
In that respect the experiment in Fort IV was unique
as it tried to embrace the knowledge of a craftsman
in an early stage of the conservation process. So,
a dynamic interaction between pointer, experts and
architects arose as the pointer was asked to give pro-
posals (bottom-up) of possible mortar mixtures within
the guidelines formulated by the architect and the
guidelines resulting from the historical investigation
and the laboratory tests. In order to prevent a top-
down approach that obliges the pointer to work with
a certain mortar composition, that he doesn’t know or
isn’t familiar with, the strategy of trial and error was
Figure 2. Samples of repointing, pointing sample on top
followed. right is sample nr. 2 which eventually has been chosen. On
During the iterative process comments of the crafts- the left sample 1, at the bottom sample 3.
man were always collected after each evaluation of a
set of samples. Many comments were gathered during
conversation when the site works were ongoing.
mortar, which he preferentially applied in his projects
For the proper conservation of the decorated
as such mortars feel less “fatty”. As the demand was
masonry, two types of pointing would be needed, a
to use mortars without cement he proposed the same
dark and a light mortar. Further flush and cut-to shape
composition by omitting cement. He confirmed that
joints were requested.
this sample was quite difficult to work with, possibly
After some preliminary discussions with the crafts-
because the overall binder to sand ratio was quite low.
man, a first series of three samples was proposed with
In his opinion Sample 5 was too coarse to create cut-
the following compositions (in volume ratio):
to shape joints. Yet, the sample had good workability,
– Sample1: 1 part trass lime (German ready-mix prod- but he was afraid to use another type of mortar for the
uct with ≥55% trass) + 4 parts sand + ½ part “grey cut-to shape joints as colour differences would have
lime” (name and product proposed by the craftsman) been possible.
– Sample 2: 1 part trass lime + ½ part grey lime + 3 The evaluation of the samples demanded for a last
parts sand + 1 part yellow sand set that could help to determine the final choice:
– Sample 3: 1 part trass lime + 3 part sand + 1 part – Sample 8: 3 parts of Rhine sand + 1 part of trass
yellow sand + ½ part lime hydrate lime
According to the pointer the different samples had – Sample 9: 5 parts of sand + 1 part of Rhine sand +
more or less equal workability. An initial trial of sam- 1 part of trass lime + 1part of lime hydrate
ple 1 with 1 part of grey lime in stead of a ½ part – Sample 10: 6 parts of Sand + 1 part of trass lime +
resulted in a fatty lime, and was therefore according 1 part of cement
to the pointer unsuitable, as the binder-to-sand ratio Sample 8 was inspired by sample 4, yet with another
could have been too high. type of sand in order to obtain a colour that would
After the evaluation of these samples a more greyish match better. Nevertheless the pointer had to admit that
colour mortar was requested by the architect. There- this sample was too coarse to apply in the often narrow
fore a new set of samples was placed by the craftsman. joints. In order to make sample 5 less coarse he made
Meanwhile the craftsman was shown a video that some adjustments to obtain the mortar composition of
explained the use of lime in pointing mortars: sample 9. Although the workability improved, and it
– Sample 4: 3 parts sand + 1 part trass lime was easier to apply in the narrow joints, it was still too
– Sample 5: 3 parts sand + 3 parts of Rhine Sand + coarse to obtain a fair result.
1 part trass lime + 1 part of grey lime Sample 10 finally illustrated the usual compositions
– Sample 6: 6 parts sand + 1 part trass lime + 1 part with a binder to sand ratio of 1:3 and the use of cement
of cement (CEM II/A LL 32,5 R) and lime.
– Sample 7: 6 parts of Sand + 1 part of trass lime + Eventually sample 2 and sample 6 were chosen for
1 part of lime hydrate the light and dark mortar joints.
As in one of the buildings (the left “capponière”)
The pointer indicated that he normally would have the masonry showed a patchwork of light and dark
added some cement in sample 4 to create a bastard coloured bricks, the choice for the appropriate mortars
7
somewhat coarser than the Lommel-sand and there-
fore it is more difficult to make cut-to shape joints.
The craftsman was well aware of the problem, yet he
was afraid that another composition would differ too
much in colour from the approved samples.
The pointer experienced several problems when
brushing the joints, because in a prior phase the facades
were cleaned with a slightly abrasive sand-water mix-
ture. In his opinion several bricks were damaged at the
surface by this operation, the brushing of the joints pol-
luted the bricks with mortar remains that were difficult
to remove.
Some argue that many craftsmen are only inter-
ested in technical and aesthetical compatibility and
have few questions regarding the final conservation
philosophy or principles. Nevertheless, our craftsman
tried to identify peculiar problems, that in his opin-
ion could need another approach but were difficult to
judge as his training never included these more philo-
sophical issues. The most illustrative example that he
showed was the cut-to shape repointing of an area with
original brickwork. In this case the outer layer of the
bricks was damaged. The original joints had fallen out
and needed replacement. The craftsman struggled with
a technical-historical dilemma: would it be sensible to
create cut-to shape pointing as the outer layer of the
bricks had disappeared completely? His final judge-
Figure 3. Samples of repointing: lowest sample is sample ment was more practical by following the general rule
nr. 6 which eventually has been chosen.
that window arches were originally with cut-to-shape
pointing.
It shows that the input from the craftsman, a bottom-
was less straightforward. It was decided to use the
up strategy, from the beginning of the project can be
light coloured mortar (nr. 2) for the complete façade.
very useful to guide certain decisions.
The craftsman would have preferred the dark coloured
Eventually the craftsman was quite satisfied with
mortars, but he said this was just a matter of taste.
the overall pointing result and with the guidance and
After the works were finished, the pointer admit-
discussions during the project. Nevertheless he had to
ted that he was initially quite reluctant to use the trass
admit that he had lost a lot of time at the beginning
lime that was stipulated in the original job descrip-
when searching a correct pointing mixture. Further,
tions of 1859, as he wasn’t familiar with the product.
he found it quite difficult at the start to understand the
He even had problems to find the appropriate prod-
global framework of the different trials. He was very
uct as he only had found the first bags in a “garden
satisfied that he could work fast and effectively with
shop” close to Antwerp. Therefore it was important
the trass lime, a product he had never applied before.
that the pointer could “experiment” with the trass lime
It should be said that the craftsman had a very open
in order to obtain insight in the workability properties
mind towards this experiment and that therefore he
of the product, and that he could try several different
kept on trying to improve his samples to obtain the
mixtures.
desired result.
Eventually he was quite surprised that the trass lime
was easy to use and that a considerable strength was
obtained after some days. He mentioned that he has
been using trass lime since in other works after his 3 INVESTIGATION WORKABILITY AND THE
experience in Mortsel. So, one can argue the impor- ROLE OF WORKMANSHIP
tance of initiating craftsmen in ‘forgotten’ traditions,
as the use of a certain material is important and that In a very different project experiments were carried
the reluctance is often more related to habits than to out to connect measurements of workability according
bad will. to scientific laboratory procedures with a subjec-
The pointer applied in almost all the cases the tive judgement of workability by professionals. Some
so-called “velco” sand, a product that originates after of the experimental results are presented in another
washing the so-called Lommel sand. The product is contribution to this conference (Hendrickx 2008).
8
Workability of fresh mortar is the sum of the appli- Table 2. Description of binders.
cation properties which provide its suitability, whether
for masonry purposes, plastering or jointing. Sev- No Description
eral authors and normalising institutions published
lists of these properties in varying order of impor- 1 Calcic lime CL90S
2 Calcic lime CL90S (EN 459-1:2001) (75%) with
tance (ASTM 2004, Kampff 1961 and RILEM 1978).
hydraulic binder (15%) and pozzolana (10%)
Recent literature about the rheology of mortar has 3 Calcic lime CL90S (EN 459-1:2001) with air
focused mainly on concrete. A variety of techniques entraining agent
have previously been used in laboratory measure- 4 Calcic lime CL90S (EN 459-1:2001) (66.7%) and
ments: rotational rheometry, compressive rheometry ordinary Portland cement CEM I 42.5 R (33.3%)
(squeeze tests), slump tests and alternative methods 5 Natural hydraulic lime NHL 5 (EN 459-1:2001)
like the rotating ball method or a trowel-imitating 6 Ordinary Portland cement CEM I 42.5 R
method (Banfill 2006). Standardised methods are 7 Ordinary Portland cement CEM I 42.5 R with
available to measure consistence, plasticity, water admixtures: plasticizer and air entraining agent
retention, air content, etc.. Most of the consistence
tests provide a measure of spread or slump, or pen-
Table 3. Dimensions of masonry units.
etration depth of a falling object.
Their drawback when compared to the rheological Commercial name and producer L D H
tests is that they are not directly related to the funda- solid/perforated (cm) (cm) (cm)
mental physical properties of the material. The practi-
cal difficulties in designing suitable measuring meth- Mono 3009, Vandemoortel (B) 19 9 5
ods for masonry mortar are: poor reproducibility due (solid)
to variations in components and preparation, impor- Porotherm, Wienerberger (B) 29 14 19
tant sensitivity to deformation history (thixotropy), (perforated)
and slip and plug flow in prevalent test geometries. Poroton Plan-T 14-30,0; 37.3 30 24
Wienerberger (G) (perforated)
Assessing measuring methods is possible by com-
paring results to the findings of experienced practition-
ers. An experimental programme in the framework of
a RILEM committee TC13-MR on mortars involving
5 masons led to the conclusion that
1. all craftsmen were able to attain the desired consis-
tence independently of the (dry) mix composition,
2. each craftsman has an individual ideal consistency,
3. all known test methods are inferior to the crafts-
men’s judgement and some are not suited for
mortars made with specific binders (RILEM 1978). Figure 4. Overview of trowel shapes used in experiments.
The type on the left was used by the Dutch mason; the type on
It has been shown that the often used flow table test the right by the French mason; the others by Belgian masons.
yields contradictory results for cement mortars and
lime mortars (Van Balen and Van Gemert 1991).
work with his own tools, which lead to the use of a
variety of trowels (Fig. 4).
In the first section of the programme, with quarry
3.1 Materials and methods
sand, the binder to aggregate ratios (B:A) were fixed,
The selected binders include a wide variety of lime- and the water to binder ratios (W:B) were left free
based binders and an ordinary Portland cement for the masons to choose. All ratios are given in mass
(Table 2). The units are fired clay bricks of three dif- proportions. The B:A were derived from a practice-
ferent sizes, two of which are perforated. Binder 7 is based reference.
used in 2 subsets of mortars with a different dosage Batches of approximately 35 litres were mixed
of the air-entrainer: 0.4% of the binder weight in 7a in a floor-model Hobart mixer type M80. Approxi-
and 0.1% of the binder weight in 7b. The plasticizer mately 75% of the estimated needed water quantity
is added at 0.5% of the binder weight. The additives was poured in the bowl, then half of the aggregate,
used with binder 7 are Rheomix 359 and Micro-air 100 the binder, and the other half of the aggregate was
(BASF). added. Mixing was done at low speed, while water was
The masons in the test panel all have at least 5 years added by the mason, until the water content and homo-
of experience with mortars containing lime. The panel geneity were considered optimal. The amount of added
is composed of three nationalities: Dutch (1), French water is weighed and the consistence of each compo-
(1) and Belgian (4). Each workman was allowed to sition is measured immediately by plunger penetration
9
2.50
⫾ Standard deviation
1
2.50 2.00
W:B (kg/kg)
1.50
2.00
5
6 4
1.00 Non air entrained
1.50 2
Air entrained
W:B
7b
0.50 Linear (Air entrained)
0.00
Figure 6. Mean selected water to binder mass ratio (W:B)
1 2 3 4 5 6 7a 7b as a function of specific surface area (SSA) of the binder in
B type the mortar.
Figure 5. Mean selected water to binder mass ratio (W:B) mortars have lower values than non air entrained
for mortar compositions with different binder (B) types. mortars (Fig. 6).
The same trend in W:B is found in the second section
(EN 459-2:2001). Each batch is tested on all three of the programme, although with larger variation due
brick types: both types of perforated bricks are used to the varying B:A for each composition. The average
to add three bricks to a single-leaf wallet, and the fac- selected B:A are between 0.15 and 0.2 for binders 1, 2
ing brick is used to produce small columns of eight and 3; between 0.2 and 0.3 for binders 4, 5, 7a and 7b,
layers of three bricks in alternating bond. The masons and 0.4 for binder 6. The latter value is very high: it
are continuously interviewed and filmed during the appears that the test persons added a high quantity of
procedure. the ordinary Portland cement because a normal dosage
Reference compositions for lab experiments with gives an impression of being poor in binder.
the quarry sand are calculated by taking into account The lab experiments on reference mortars revealed
the average of the W:B chosen by the masons (Fig. 5). important differences in flow, yield stress, water reten-
They are tested for flow (prEN 1015-3:1998), plunger tion, etc., although all mixes were brought to opti-
penetration (prEN 1015-4:1998), bulk density and air mum water content. Different performance can be
content (EN 459-2:2001). The mixing procedure is not attributed to different behaviour of the binder particles
standardised, but attempts to approximate the proce- on micro-scale.
dure from the masons’ programme (see above), using What is important, is that the qualitative judge-
a table model Hobart mixer: 1.5 min mixing, 0.5 min ment of the different mortars is similar for all the
scraping and homogenising, 2 min mixing. All mixing masons, and this to a remarkable extent.The small vari-
is done at low speed. ation in ideal W:B ratio between the different masons
In the second section, with standardised sand, W:B for each of the mortars is remarkable. It shows how
and B:A are free to choose for the mason for the com- craftsmanship, knowledge and experience seem to be
position of small batches (1.2 liter). Lab experiments a very good discriminator for mix optimisation and for
with average values for W:B and B:A from these tests quality assessment.
sometimes yield visibly unworkable mortars. Hence it The applicability and use of mortar for different
was decided to discard them from the lab experimental types of bricks and masonry was also taken into
programme. account by the masons. For example the masons
For description of the scientific laboratory tests to revealed that Binder 2 had an important problem with
define workability aspects of mortar we refer to the the air entraining agent, which causes swelling and
other contribution (Hendrickx 2008). subsequent shrinkage of the mortar during mixing and
in the first 10 minutes after mixing. The same prob-
lem seemed present in binder 3 to a lesser extent. The
3.2 Experimental results and discussion
comments from the masons also depend on the type of
In the first section of the programme (fixed B:A), the brick used. For the small and large perforated bricks,
selected W:B values for mortars with each binder have it is judged important that the mortar adheres well to
a coefficient of variation of 3% to 10% between the the (higher) vertical sides, that it does not fall into the
different masons for one specific binder (Fig. 5). A perforations and that it allows a large unit to be manipu-
systematic difference between some of the masons lated in the mortar bed without too much force. For the
is observed. A correlation between W:B and the spe- small facing bricks, staining and floating of bricks are
cific surface area of the binder is found. Air entrained more important criteria. Floating must be understood
10
as instability due to lack of stiffening of the mortar, In the conservation work in Fort IV also the tangible
which should occur upon dewatering in contact with results of workmanship from people involved in the
the porous brick. construction in the nineteenth century were preserved.
An inquiry about the relative importance that the The origin of the research presented in the sec-
masons of the test panel attribute to different prop- ond part comes from the problem encountered when
erties of mortar (in fresh and hardened state), reveals comparing the so called workability of different types
three major concerns: workable time, adhesion to brick of mortar using standardized tests. Those tests have
and workability. Mechanical resistance and yield are started to live their own life and were considered
also considered important. Cost, thermal insulation, measurements of workability although it is noticed
frost resistance and environmental nuisance have low that they contradict the judgement of masons once
appreciation. leaving the (not explicit!) boundaries of the usual play-
The results of the scientific assessment of worka- ground for which the standards were made. Compari-
bility are given in (Hendrickx 2008). son of workability judgements of cement mortar versus
lime mortar with standardized tests is hardly possible
through the eyes of practitioners. Defining standards
3.3 Synthesis and conclusions today for “new” types of mortars (sometimes based on
The variance in chosen W:B between the different very old recipes that demonstrated durability) required
masons in practical tests is small. This indicates that a from the researchers to go back to the real appreciators
marked transition in the mortar’s behaviour takes place of workability: skilled workmen. The research project
around a well-defined water content: from a granular referred to, showed that a group of masons with differ-
material to a liquid material, from a frictional regime ent background could remarkably well agree on what
to a viscous regime. Small variations of water con- was considered a workable mortar. This from one hand
tent around this transition point, give a very different demonstrated that it is possible to define workabil-
mortar for the user. It seems more easy to identify ity and at the other hand it demonstrated that skilled
that transition point with masons than using laboratory people “inherited” the concept and understanding of
experiments, not to say with standardized tests. workability. It is the demonstration of the (intangible)
Laboratory research (Hendrickx 2008) indicates heritage based on practices, habits and tradition that is
that a combination of tests may be able to approach still worth to be transferred to future generations.
the masons judgement of workability. A combination As the preservation works on the walls at Fort
is proposed of yield stress measurement with a vane IV are the outcome of a dialogue strongly involving
apparatus, water retention tests with vacuum suction the craftsman acknowledging respectfully his skills,
or filter plates, and density and air content tests to knowledge and traditions the works themselves have
characterise the workability of a mortar in detail. contributed to the preservation of craftsmanship, to the
preservation (revival?) of a living tradition.
4 PRESERVATION OF WORKMANSHIP OR
WORKMANSHIP FOR PRESERVATION ACKNOWLEDGEMENT
The first case study illustrates that beside technical Town of Mortsel, EU Interreg IIIb project “Crossing
and scientific analysis workmanship is essential in the the Lines”, Lhoist R&D.
preservation of architectural heritage. Good workman-
ship influences the results of the appearance of the
historic masonry after preservation (“Workmanship REFERENCES
for preservation”). ASTM, 2004. ASTM C 270-04a Standard specification for
Historical analysis and a sensible way of dealing mortar for unit masonry, West Conshohocken: ASTM.
with conservation bring to light how workmanship is Banfill, P. 2005. The rheology of fresh mortar – a review,
embedded in heritage and how the preservation action In Cincotto, M.A., Silva, D.A., Oliveira, J., Roman, H.R
itself is “heritage” (“Preservation of workmanship”). (ed.), Proceedings of the 6th Brazilian and 1st Interna-
The historical sources referred to, were very infor- tional Symposium on Mortar Technology, Florianopolis,
mative on the materials and techniques used in the Brazil: 73–82.
original construction and they do reveal how concerns Cools, S. 2006. Crossing the lines: study and repair of
of workmanship were integrated in the construction historic masonry structures : case study: restoration of
Hoofdfrontgebouw and Caponnières at Fort 4 (2004–
process. Those historical sources show how some 2005), Unpublished Master thesis Raymond Lemaire
options were kept open to the judgement of the work- International Centre for Conservation, K.U. Leuven.
man. Eventually workmen had a very important impact Gils, R. 1997. Vesting Antwerpen Deel 1, Bakstenen
on the composition of the material mixture used as well Schoonheid 1830–1885, (België onder de wapens, 5),
as on the final state of the masonry when built. Erpe-Mere: De Krijger
11
Hendrickx R., Minet J., Van Balen K. & Van Gemert D. 2008. Van Balen, K. & Van Gemert, D. 1991. Kalk in metselmortel.
Workability of mortars with building lime: assessment Over het meten van verwerkbaarheid en hechting en over
by a panel of masons versus lab testing, 14th Interna- de invloed van luchtbelvormers en superplastifieerders.
tional brick and block masonry conference, Sydney, 16–20 Voorlopig rapport (unpublished report). Katholieke
February 2008. (accepted for publishing on CD-ROM) Universiteit Leuven.
Hendrickx R., Van Balen K. & Van Gemert D. (this volume) Van Balen, K. 2003. From conservation principles to materi-
Assessing workability of mortar by means of rheological alization (or the other way around: how is materialization
parameters and desorptivity, SAHC 2008, Bath, 2–4 July guided by principles?). In Van Gemert D. (Ed.), Consol-
2008. Rotterdam: Balkema. idation of Masonry. Advances in Materials Science and
Kampff, L. 1961. Workability of masonry mortars. In Mate- Restoration 1: 135–144. Freiburg: Aedificatio Publishers.
rials research and standards 1: 7–8. ASTM. Van Balen, K. & Van Gemert, D. 2005. Crossing the lines –
RILEM 13-MR committee on mortars and renderings 1978. restoration techniques Fort IV, Mortsel (B) (unpublished
Tentitative recommendation. Materials and structures 11: Laboratory report R/30299/05). Laboratorium Reyntjens,
207–216. Katholieke Universiteit Leuven.
12
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: This work is about the lesson learnt from the analysis of the damages caused to some churches
by the strong earthquake, which struck Northern Italy in the year 2004. Focus is paid to some mechanisms which
are not well known in the literature: the excessive and differential rocking of neighbouring transverse arches, the
differential deflection of single leave vault rings, the tie over-tension induced by the transverse arch rocking. In
the paper, the structural solutions appositely studied to repair and to reduce the structure seismic vulnerability
of these buildings are presented. Among these solutions: internal perimeter ties, lightweight wooden roof box
structure, lightweight spandrel ribs are illustrated.
13
In the churches surveyed following the 2004 earth- Table 1. Seismic vulnerability assessment of some churches
quake, other mechanisms, such as the excessive and hit by the 2004 earthquake with respect to some failure
differential rocking of the neighbouring diaphragm mechanisms.
arches and the differential deflection of the thin vaults,
were acknowledged as further causes of the severe C(*) = 2.5% (A)
C = 1.0% (B)
damages observed. C = 3.5% (C)
The seismic vulnerability of the churches was C = 5.0% (D)
assessed by reference to the limit analysis approach. C = 5.0% (E)
C = 5.2% (F) (A)
Table 1 shows the vulnerability coefficients esti- C = 3.3% (G)
Madonna della Rocca,
Sabbio Chiese, Brescia.
mated for some significant case studies. Façade overturning
C = 7.8% (H)
C = 10.7% (I)
C = 5.0 % (A)
2.1 Overturning of the main façade C = 4.0% (B)
C=- (C)
Overturning of the main façade was often observed C=- (D) (B)
during the survey campaign (Figure 1, Table 1). This C=- (E) San Lorenzo Church,
collapse mechanism was caused by the seismic action C = 6.0% (F) Clibbio, Brescia.
of the wall and by the possible lateral seismic thrust of C=- (G)
Lateral wall overturning C=- (H)
the roof beams. Provided that wooden roofs are usu- C = 12% (I)
ally unconstrained against lateral movements, the latter C = 3.5% (A) (C)
contribution to the toppling action was quite relevant C = 7.0% (B) San Pietro Church
in most of the analyzed churches. C = 6.3% (C) (Roè Volciano, Brescia)
The onset of the overturning mechanism is under- C = <0 (D)
C=0 (E)
lined by the development of cracks extending at the C=- (F)
wall edges and progressively reducing their width from C = <0 (G)
the top to the footing of the building. The cracks can be Diaphragm arch C = 7.4% (H)
excessive rocking C = 48% (I) (D)
vertical or angled, depending of the degree of anchor- Sant’Antonio,
age to the perimeter walls. However, it is worth noting remarkable
(Roè Volciano, Brescia)
that, the measured crack angle was always very small, crack pattern
thus the stabilizing contribution of the lateral walls was surveyed in A,
B, D, E, G, H
almost negligible in most cases. Collapse
Diaphragm arch surveyed in C
differential rocking (E)
2.2 Excessive rocking of the diaphragm and San Rocco,
(Roè Volciano, Brescia)
triumphal arches and existing tie over-tension remarkable
crack pattern
Excessive, unconstrained rocking of the diaphragm surveyed in A,
and triumphal arches was recognised as another source B, D, E, G, H
Collapse
of vulnerability of many churches in the Benaco area surveyed in C
(Table 1). Single leave vault differ- (F)
ential bending Santissima Trinità,
In rest condition, when only vertical loads are (Roè Volciano, Brescia)
applied to the structure, the arch lateral thrust can
be either largely or even entirely resisted by the but-
tress action developed by the abutments, depending on
their shape and massiveness (Figure 2a, Giuriani and
Gubana 1993 and 1995). As a result the possible ties (H) (I) (G)
are required to confine the arch thrust portion, which SS. Faustino e Giovita SS. Pietro e Paolo Santa Maria Assunta
(Montemaderno, Brescia) (Preseglie, Brescia) (Bione, Brescia)
exceeds the buttress action. The proportion of the con-
( )
fining contribution mainly depends on the structure * FS = C WTOT
where: F S is the seismic force triggering the mechanism; C is the
geometry and tie pretension. Conversely, in the case of load collapse multiplier (also regarded as vulnerability index); W TOT
rocking, the resisting mechanism significantly modi- is the structure total weight.
fies (Figure 2b, Giuriani et al. 2007 and 2008). The
crack pattern is characterized by two large horizon-
tal cracks developing at the abutment bases and two
cracks opening at the diaphragm arch springing (Fig- of full rocking, when the crack pattern is fully devel-
ures 2b, 3 and 4). The location of the cracks depends on oped, no abutment buttress action can be accounted
both the structure geometry and the local resistance. for, because the resisting ideal struts become parallel
The abutment buttress action dramatically dimin- to each other. As a result, the arch thrust has to be
ishes as the crack penetration progresses. In the case entirely confined by the existing ties.
14
Figure 1. Overturning of the façade induced by the roof
seismic thrust.
a)
b)
Figure 4. Detail of the horizontal crack pattern at the
Figure 2. Transverse arch resisting mechanisms under a) transverse arch abutment base induced by the seismic action.
vertical loads (rest condition) and b) horizontal loads (case
of full rocking, Giuriani et al. 2008).
between 0, 15 ÷ 0, 2, thus when shifting from rest con-
dition to rocking the arch thrust could be 30 ÷ 40%
Furthermore, following the cracking triggered by larger than the thrust at rest condition.
the rocking motion, the span of ideal arch significantly In the case of full rocking, the decrease of the abut-
increases and so does the arch lateral thrust (Figure 2b). ment buttress action and the increase in the ideal arch
The span of the ideal arch in rocking condition can be span cause the existing tie over-tension. When the tie
conservatively assumed to be equal to the span of the resistance is barely sufficient to withstand the traction
ideal arch in rest condition (L∗ ) plus the thickness of force in rest condition, no extra resources are available
the abutment (d) (Giuriani et al. 2007, 2008). For typ- in case of earthquake, and the collapse of the tie can
ical case studies the ratio d/L∗ was observed to range be expected.
15
Figure 5. Detail of the welded tie rod in San Antonio Church
(Manerba, Brescia, Italy). The tie rod collapsed during the
2004 earthquake.
16
contact
Fo D Fo
a b
A B
detachement
a)
vault support
at the springing
structural
discontinuity
17
Lx
t min
a) b)
L x t min 10
Lx
18
roof box structure
ideal constraint 2 facade panel nailed
Wc CWc Wc connections
eaves
chords
CWA CW
FVA 1
W W
plywood
roof box panels
ye structure head walls or
lateral 2
Fo 3 walls z triumphal arch
x y
ye'
ye h plywood panels
existing planks
nailed
steel straps
L
a) b)
19
g1 2
Ry c 13 f1y L x
艎12 p1 h1
f 1y = -r1y Ry = V1 =
2
1 f1y L2x
3 F13 Mmax =
8
a p3 A: FRAME B: BOX c 13
h3 BEHAVIOR STRUCTURE F13 Mmax Mmax f1y L2x
F13 = =
Lx 2 Ly 8Ly
Ly
z
Ly
x y
y r1z
g1* 2 Wg* head gable
h1 q1
3 fz
r1y f 1y = -r1y dy
q1 fz
4 2 Wg*
α zg head
h(y) q0 q0
∆W *g gable
wall
q0
A: FRAME fz −1
BEHAVIOR V1 L y V1
q1 = cos α =
z
B: nA 2 2 cos α Ly
5 1 x y BOX STRUCTURE
r1z q
q o = 1 cos α = q1
g1* zw W w* cos α
3 n A = (p1l12 h1 + p 3h 3h1 )L y − g1*l12 / 2 f z = 2q o tan α − w*g
r 1y
v A = p 3h 3 / 2
4 2 Figure 17. Simplified schemes for the evaluation of the
vA vA f1y = 2p1l12 + p 3h 3
nA nA force distribution in the box structure.
20
stud
connections
head gable chord
21
lightweight
spandrel ribs polystyrene ribs
cross section
clay plaster
plaster meshes
lightweight
spandrel ribs
Figure 22. San Pietro Church (Roè Volciano, Brescia): light
spandrel walls strengthening the single leave masonry dome
covering the presbitery.
4 CONCLUDING REMARKS
The inner lightweight core is made of polystyrene
elements. The analysis of the damages caused by the 2004 earth-
In San Pietro Church, spandrel ribs were placed quake to some churches, as well as the structural solu-
along the vault extrados as shown in Figure 20. The tions proposed to reduce their structure vulnerability
construction phases are briefly illustrated in Figure 21. were discussed in the paper.
22
The most significant results of the study can be Sicurezza e conservazione dei centri storici: Il caso
summarized as follows: Ortigia, pag. 189–205. Editrice Laterza, Bari.
Curti, E., Lemme, A., Podestà, S., Risemini, S. (2006).
• In the case of full rocking, the decrease of the abut- Criteri di verifica per la progettazione di interventi di
ment buttress action and the increase in the ideal miglioramento sismico di edifici monumentali. Ingegne-
arch span can cause over-tension of the existing ria Sismica, Anno XXIII, n. 1 gennaio-aprile.
ties. In the case of very weak ties, collapse of the D’Ayala, D. and Speranza, E. (2002). An integrated proce-
tie can be expected. In this case, the adoption of a dure for the assessment of seismic vulnerability of historic
roof box structure, limiting the rocking of the trans- buildings. In Proceedings of the 12th European Con-
ference on Earthquake Engineering, London, Elsevier
verse arches, together with the strengthening of the Science, paper n. 561 (CD-ROM).
existing tie system might significantly reduce the Doglioni, F., Moretti, A., Petrini, V. (1994). Le chiese e il
structure vulnerability. terremoto, Trieste, Edizioni Lint.
• Differential rocking of transverse arch pillars Gattesco N., Del Piccolo G. (1998). Shear transfer between
induces shear distortion within the vault ring, espe- concrete members and stone masonry wall through driver
cially in the first and last bay of the nave. Depending dowels. European Earthquake Engineering, Vol. 1.
on the magnitude of the differential rocking, large Giuffrè, A. (1993). Sicurezza e Conservazione dei centri
diagonal cracks my extend over the vault leading up storici: il caso Ortigia, Bari, Ed. Laterza.
to its collapse. In this scenario, the structure vul- Giuriani, E., Cubana, A. (1993). Recupero e consolida-
mento di volte in muratura. In Atti dei Colloqui Inter-
nerability can be reduced by adopting a roof box nazionali Castelli e Città fortificate,Storia Recupero,
structure and by limiting the shear distortion near Valorizzazione. Palmanova, 24–25 settembre.
the supports. Giuriani, E., Gubana, A., (1995). Extrados ties for structural
• In the case of a seismic event, pronounced differ- restoration of Vault. In Proceedings of the Fourth Inter-
ential deflection can be expected along the central national Conference on Structural Studies of Historical
part of thin vaults. Depending on the severity of Buildings Strema 95, Architectural Studies, Materials and
the earthquake, differential deflection can cause the Analysis, Vol. 1, Ed. C.A. Brebbia, Wessex Institute of
collapse of single leave vaults. Against this failure Technology, UK.
mechanism, lightweight spandrel ribs were pro- Giuriani E. (2004). L’organizzazione degli impalcati per
gli edifici storici. L’Edilizia. Speciale Legno strutturale,
posed. This solution allows to significantly enhance N. 134. (In Italian).
the stiffness and strength of the vaults. Giuriani E., Marini A., Porteri C., Preti M. (2007) “Don-
dolio degli archi diaframma e vulnerabilità sismica delle
chiese”. Technical Report n.7, Università degli Studi di
Brescia.
ACKNOWLEDGEMENTS Giuriani E., Marini A., (2008). Wooden roof box structure
for the anti-seismic strengthening of historic buildings.
The authors gratefully acknowledge Claudia Porteri Accepted for publication in the International Journal of
and Lorenzo Giuriani for their fundamental contribu- Architectural Heritage.
tion to the vulnerability assessment of the churches Giuriani E., Marini A., Porteri C., Preti M. (2008) “Vulnera-
and to the design work. bility of churches associated to transverse arch rocking”.
This work was developed within the research project Submitted to the International Journal of Architectural
DPC-ReLUIS 2005–2008, Research line n. 1: “Vul- Heritage for possible publication.
Griffith, M.C., Magenes, G., Melis, G., Picchi, L. (2003).
nerability assessment and anti-seismic strengthening
Evaluation of out-of-plane stability of unreinforced
of masonry buildings”. ReLuis and MURST financial masonry walls subjected to seismic excitation. Jour-
contributions are gratefully acknowledged. nal of Earthquake Engineering, Vol. 7, Special Issue 1,
pp. 141–169.
Housner, G. W. (1963). The Behaviour Of inverted pen-
REFERENCES dulum structures during earthquakes. Bulletin of the
seismological Society of America, 53 (2).
Abruzzese, D., Lanni, G. (1999). Some developments on the Magenes, G., Calvi, G.M. (1997). In-plane seismic response
lateral strength of historical reinforced vaulted buildings. of brick masonry walls. Earthquake engineering and
In Structural Studies, Repairs and Mainteinance of his- Structural Dynamics, Vol. 26, pp. 1091–1112.
torical buildings, VI International Conference, STREMA, Marini A. and Giuriani E. 2006. Transformation of
Brebbia and Jager Ed., Dresda. wooden roof pitches into antiseismic shear resistance
Bruhn E.F. (1973). Analysis and design of flight vehicle diaphragms. V International Conference on Structural
structures. Jacobs Publishing Inc. Analysis of Historical Constructions, November 6–8.
Como, M., Lanni, G., Sacco, E. (1991). Sul calcolo delle New Delhi. MacMillan Ed. ISBN. 10: 1403-93156-9.
catene di rinforzo negli edifici in muratura soggetti ad pp. 445–452.
azione sismica. VII convegno nazionale L’ingegneria Lagomarsino (1998). A new methodology for the post earth-
sismica in Italia, ANIDIS, Palermo. quake investigation of ancient churches. In Proceedings
De Benedectis, R., De Felice, G., Giuffrè, A., (1993). of 11th European Conference on Earthquake Engineering,
Restauro antisismico di un edificio. In: Giuffrè A., Balkema, Rotterdam, ISBN 90 5410 982 3.
23
Lagomarsino, S., Brun, S., Giovinazzi, S., Idri, C., Penna, A., of historical construction (C. Modena, P. B. Lourenço,
Podestà, S., Resemini, S., Rossi, B. (1999). Modelli di P. Roca Eds.), Proc. of IV Int. Seminar SAHC, Padova,
calcolo per il miglioramento sismico delle chiese. In Italy, A.A. Balkema, London (UK), Vol. 2, pp. 1091–1101.
Proceedings of 8th Italian Conference on Earthquake Tengattini C.G., Marini A., Giuriani E. (2006). Connessioni
Engineering. a taglio nelle murature. Tecnichal Report 3a.1-UR11-
Lagomarsino, S., Podestà, S., Risemini, S., Curti, E., 1 RELUIS – Progetto di ricerca N.1 – Vulnerability
Parodi, S. (2004). Mechanical models for the seismic vul- assessment and anti-seismic strengthening of masonry
nerability assessment of churches. In Structural analysis buildings.
24
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
S.J. Kelley
Wiss, Janney, Elstner Associates, Inc. Chicago, Illinois, USA
ABSTRACT: Central to the development of the American skyscraper was the concept of the curtain wall. This
development was rapid and was driven by the distinct cultural climate within American urban centers such as
Chicago and New York. Skyscrapers would not have been technically feasible without the lightweight curtain
wall, and it was through the skyscraper that the curtain wall achieved its greatest realization. The principal and
confrontational factors of its development were economics and the need for fireproof buildings. The historical
development of the curtain wall is presented and a comparison of deterioration mechanisms between masonry
and metal and glass curtain walls is discussed. Finally the issues of loss or redundancy and increase in complexity
of contemporary curtain walls are discussed and strategies for diagnosis and conservation are presented.
1 INTRODUCTION
25
2 DUELING FACTORS IN AMERICAN These structures needed to be fireproofed and
CONSTRUCTION: ECONOMY AND FIRE the urban fires revealed that iron work, though
SAFETY inflammable would yield and fail beneath the flames.
George H. Johnson, an English-educated archi-
Economy in construction of buildings is an axiom tect working with the Architectural Iron Works
of modern construction worldwide and the excep- of New York founded a business to manufacture the
tions that one may cite would only prove the rule. In fireproof terra cotta tile that he patented in 1871. On
America this axiom was very specific and tied to the trips Johnson made to Chicago to promote the sale of
development of the American West. his fireproof tile may have become a precedent for the
Consider the development of the balloon frame, a William leBaron Jenney’s 11-story Home Insurance
distinctly American framing technique that fulfilled Building (Chicago, 1885). The use of masonry, and
the requirements of a young and rapidly expanding lots of it, in skyscraper construction would prove to be
country. In a balloon frame structure, studs run con- difficult to move beyond in curtain wall development
tinuously from the sill to the second floor top plate due to its fire resistant qualities.
which supports the roof structure. The second floor Following these 19th Century building trends came
joists are nailed to the studs and rest on ledger boards the development of Modern Movement technologies
that are fit flush into the studs. This system relies on characterized by experimentation and innovation in
the external sheathing rather than triangular bracing construction materials and techniques, many of which
for lateral strength. had neither the resilience nor the longevity of tradi-
It was characterized by the use of plentiful wood tional construction. Materials such as cinder concrete,
from America’s once ubiquitous forests, rapid con- plywood, and particleboard; and systems including
struction, and the limited skill required by the workers. panelized construction, and thin curtain walls with lit-
The advent of cheap machine-made nails, along with tle redundancy came into use. This experimentation
water-powered sawmills in the early 19th century made with materials and systems remains unabated today.
balloon framing highly attractive. Standard wood ele-
ments such as studs were readily available. Although
3 THE DEVELOPMENT OF THE
lumber was plentiful in 19th Century America, skilled
SKYSCRAPER TYPOLOGY IN AMERICA
labor was not. Relatively unskilled workmen were able
to erect balloon frames quickly because the labor-
The curtain wall dichotomy can be traced back to
intensive mortise and tenon connection of the braced
numerous 19th and early 20th Century antecedents.
frame technique was supplanted by the use of cut
In the United States curtain wall development became
nails. Without balloon framing, the western boom-
intertwined with that of the skeleton frame. Skyscrap-
towns of America and Canada certainly could not have
ers would not have been technically feasible without
blossomed overnight.
the lightweight curtain wall. And it was through the
These important economic factors of construction –
skyscraper that the curtain wall achieved its greatest
namely rapid construction and limited skill required by
realization.
the worker – set a trend for American and then World
construction of large buildings in 20th century. 3.1 The skeleton frame
Another important but reactive factor was the need
for fireproof buildings as can be seen in Chicago. In the nineteenth century, engineers first utilized metal
Chicago’s building history recommenced after 1871, frame construction in bridges, factories, and ware-
when the Great Chicago fire devastated the City. houses, and cast and wrought iron were the major
Despite this catastrophe – and due to a strong local metals used in construction. The invention of the
economy, the high cost of land, and rapidly evolving Bessemer process in England in 1856 made it pos-
building technologies – Chicago quickly rose like the sible to produce large quantities of steel affordably.
Phoenix and afforded an opportunity for the realization In the United States, steel production on a large scale
of innovations in architecture, engineering, and con- was realized in the 1870s. The transition from iron
struction which established it as the birthplace of the to steel was gradual, with iron still used in building
skyscraper. The answer of who invented the skyscraper construction as late as the 1890s.
is debated to this day and depends entirely upon what Arguably, the first metal skeleton-framed skyscraper
are considered to be the defining characteristics. Com- was the Home Insurance Building designed by Jenney
parative characteristics are the separation of the wall in Chicago. In this structure, column loads were trans-
from the frame, the first use of the iron frame, appear- ferred to stone pier footings via the metal frame
ance of the beam-column moment connection, height without load-bearing masonry walls. Each level of
limits, use of the passenger elevator, and theories of the exterior wall was supported on a shelf angle fixed
frame stiffness. For this paper we use the wall and to the spandrel girder. At that time skyscrapers were
frame separation definition. designed without lateral bracing under the assumption
26
Figure 3. The New York Daily News building with a
masonry curtain wall that has been completely stripped of
Figure 2. The 20 story Masonic Temple in Chicago with a
detail as a cost measure (Hood and Howells, 1930).
laterally braced steel frame clad in a masonry curtain wall
skyscraper (Burnham and Root, 1892).
familiar construction techniques which had come into
widespread use at the end of the 19th century to facil-
that the heavy masonry cladding provided sufficient itate rapid construction. The construction site was lit
rigidity for the whole structure. so work could continue into the night. Work spaces
As skeleton framing came into common use, were enclosed and heated so the project could proceed
masonry bearing wall construction reached its prac- during the winter months. The structural frame, which
tical limit. Burnham and Root’s 16-story Monadnock began erection in mid July of 1894, was topped off on
Block (Chicago, 1891) utilized traditional load- August 1 and required only 15 days.
bearing masonry walls which at grade were almost 2 Steel and then concrete skeleton framing soon
meters thick. This building also utilized the first rigid became universally accepted for skyscrapers. There-
frame for lateral stiffness. At the same time, Burn- after, improvements of known design methods encour-
ham and Root developed a complete steel frame for aged the construction of increasingly taller buildings.
the Rand McNally Building (Chicago, 1890). They
also developed a steel frame laterally stiffened with
a diagonal bracing system in the 20-story Masonic 4 THE MASONRY CURTAIN WALL AND ITS
Temple (Figure 2). Within a three year period the DETERIORATION MECHANISMS
same firm had erected a bearing wall; steel frame; and
a diagonal-braced metal frame skyscraper revealing A good place to begin discussion of the early curtain
the experimental and quickly evolving nature of the wall is with the Reliance Building in Chicago, the first
typology. skyscraper to fully utilize terra cotta as a cladding. The
Several years later, the structural innovations of the terra cotta units of the curtain wall are connected to a
Chicago school were taken further by D.H. Burnham gridwork of cast-iron mullions, lintels, and sills which
in the Reliance Building (Chicago, 1895). The exte- span between levels. Unlike the Home Insurance and
rior bays were designed as rigid steel frames, and other similar buildings, the Reliance frame did not rely
two-story columns erected with staggered joints fur- upon the masonry curtain wall for lateral support.
ther increased frame rigidity. The construction of In New York City the steel frame and masonry cur-
the Reliance Building also illustrated all of the tain wall became established with the American Surety
27
Building in New York City (Price, 1894), and, once Another issue is the differential movement of the
adopted, skyscraper heights increased dramatically in curtain wall relative to the structural frame. By 1894,
that city. The once impressive twenty-story buildings lateral movement in curtain wall construction was
of Chicago were overshadowed by buildings 100, 200, actually being studied and analyzed. The masonry
and finally 238 meters with the Woolworth Building cladding is also exposed to temperature-related move-
(Cass Gilbert, 1913). The Woolworth building utilized ments, while the embedded frame is protected. Struc-
the latest developments in steel frame construction, but tural frames shorten under dead load and material
its curtain wall had not abandoned the use of masonry creep. Conversely, fired-clay masonry curtain walls
construction and ornament. expand due to the intake of moisture. Early curtain
The First World War administered the final blow walls were not built to accommodate these differential
to the arts and crafts movement in Europe, and the movements resulting in the introduction of unantic-
machine became the basis of a new architecture. Mod- ipated stresses into the curtain wall and frame. This
ern European architecture required that the labor of problem came to be understood as evidenced by dis-
producing the parts be performed in the factory rather placement measurements that were performed during
than by craftsmen on site. German intellectuals were construction of the Empire State building in 1931. The
in awe of the example of the American skyscraper, horizontal deflection of the top of the Empire State
a strong symbol of the new world for which they Building was monitored by the American Institute of
endeavored. Ludwig Mies van der Rohe, prior to relo- Steel Construction. Measurements were also made to
cating in Chicago, prepared a series of unrealized determine exactly how much lower the various floors
projects in which the most famous came to be known were from their theoretical position. These measure-
as the “Glass Skyscraper,” a highrise enveloped totally ments showed that the 85th floor was 16 cm. below
in glass. its theoretical elevation. (H.G. Balcom, “New York’s
Contemporary with the European modernists, Tallest Skyscraper,” Civil Engineering 1 no 6, March
America had entered a second skyscraper era. The post 1931).
World War I period brought a demand for increased To address this differential movement, a pressure-
speed in design and erection. Curtain wall con- relieving joint or “cowing” was developed by the
struction, however, continued to utilize the masonry 1930s. Cowing composed of corrugated lead was typ-
techniques that had been developed by the turn-of-the- ically laid horizontally at mid-span in selected levels
century. Though mass production and standardization of the skyscraper facade. This form of pressure relief
had begun to impact the details of architecture, the was utilized in masonry envelopes until the 1960s.
use of prefabrication and the module were not yet Besides serving the function of transmitting wind
extensively used. loads to the structural frame, the curtain wall also
H. R. Dowsell, one of the architects of the Empire must resist moisture infiltration. To enter the interior,
State Building (New York City, 1931) wrote of the the moisture must pass through the mass of the wall;
masonry curtain wall, “Tradition has clung to the her- however it may take a relatively long period of time
itage of thick masonry walls. We inherited masonry this infiltration to become apparent. The mass of the
walls and seem unable to outgrow our inheritance. wall may become saturated with slowly developing
The idea that masonry is the only form of permanent detrimental results.
construction was so deeply rooted that practically all
building codes made masonry walls mandatory . . .”
(Engineering News Record, 19 February 1931). 5 THE METAL AND GLASS CURTAIN WALL
The masonry trend continued into the Great Depres- AND ITS DETERIORATION MECHANISMS
sion while stripping away any costly ornamenta-
tion. Examples would include the New York Daily New technologies resulting from World War II had a
News (Figure 3), McGraw-Hill building (Hood and great influence on the acceptance of the machine-made
Fouilhoux, 1932), and the PSFS Building (Howe and metal and glass curtain wall. Given the abundant post-
Lescase, 1931). However, due to the Great Depression war supply, aluminum was reasonably priced. There
and the Second World War, construction grinded to a was experimentation with mild steel, stainless steel,
halt for almost 20 years. and bronze as well. Extruded components were suit-
Masonry curtain walls that we care for today con- able for standardization and could be prefabricated for
sist of facing material backed by several wythes of delivery to the site. This was important because labor
brick or clay tile, for a total wall thickness of as much had become a significant part of construction costs.
as 30 cm or more. Metal shelf angles to support the The new curtain wall technology further decreased
masonry infill are anchored to the structure at the floor building weight and construction cost, and increased
levels. Flashing was not commonly used and corrosion usable floor area. Prefabricated construction was less
of the shelf angles and other metal embedments is a limited by cold temperatures which prohibited erec-
significant cause of distress. tion of “wet” walls of brick and mortar. The invention
28
fire rating that code officials felt was not provided
by the curtain wall. The masonry wall could not be
immediately abandoned though it became hidden from
view.
The approach to curtain wall design that quickly
evolved was to make the joints as weathertight as pos-
sible, then provide positive means for conducting any
water leakage out of the wall. Thus an interior drainage
system was provided to collect water that leaks through
the cladding and direct it back to the exterior.
The development of new curtain wall materials
occurred in the post-war years: thin stone veneers,
precast concrete, brick veneers, and structural silicone
glazed facades. To bring things full circle, the aesthetic
development of “Post-Modernism” led to a return to
earlier architectonic forms, but not a return to earlier
methods of construction.
Metal and glass curtain walls that we care for today
consist of factory-fabricated and preassembled metal
units that are connected to the structural frame. Glass
has gone through radical technical developments and
is typically no longer a monolithic material. In a cur-
tain wall it can appear as an insulated glass unit or
a sandwich of materials developed to strengthen it. It
may have clear, colored, or reflective coatings installed
on one or more of its surfaces or may have transpar-
Figure 4. The Lever House in New York where the prefabri- ent or translucent colorants integral with the glass.
cated metal and glass curtain wall realized its potential (SOM, The assemblies may also include panels of aluminum,
1952). ceramics, precast concrete, or stone.
Though not the vision of 1950s designers, metal and
and development of float glass by Alistair Pilkington glass curtain walls are wholly reliant on sealants to per-
in the 1950s would make large panes of glass afford- form adequately. High performance sealants include
ably available. American architectural philosophers of newly developed families of elastomeric sealants such
the day lauded the fact that craftsmanship had been as polysulfides, solvent acrylics, urethanes, and sili-
transplanted from the site to the factory. cones. Numerous sealant products are now available
One of the first post-war buildings to be constructed and have been developed to be used in either curtain
with a glass curtain wall was the Equitable Building wall glazing, installation, or repair.
(Pietro Belluschi, 1948) in Portland, Oregon. Belluschi Modern structural frames are more flexible because
was able to take advantage of leftover aluminum stock- they are designed to tighter limits with less mate-
piled for World War II by smelters and to utilize rial and are more exposed to temperature extremes
assembly techniques derived from West Coast airplane than the frames of a masonry-clad building. Prefab-
plants. The 860–880 Lake Shore Drive buildings in ricated curtain wall units are detailed to accommodate
Chicago (Figure 1) were among the first residential these increased movements. However, lateral move-
buildings in the United States to be sheathed entirely ments of the frame and differential movement between
in glass, and were the realization of Mies’ 1920 pro- the frame and the cladding can lead to distress in the
posal for a glass skyscraper. The steel, aluminum, and glass and metal curtain wall.
glass skin was assembled on the buildings’ roofs in
two story high units, and then lowered into place on
the facade. 6 LOSS OF REDUNDANCY AND INCREASE
At the Lever House (Figure 4), the curtain wall has IN COMPLEXITY
an interior frame of mild steel clad with stainless steel.
At the United Nations Secretariat Building (Harrison In a general sense, redundancy in design can prevent
and Abramovitz, 1950), curtain walls were conceived failures of a building system. Structural redundancy,
as an assembly of aluminum windows held in place allows for loads to follow an alternate path if the
with a grid of reinforced mullions. At both buildings, primary supports fail. Redundancy of design for cur-
the lower portion of the curtain wall at each level was tain walls can apply not only to structural but to
backed up by a concrete masonry wall to provide the performance issues such as weathertightness.
29
The mass of the masonry curtain wall provided a and regulated by the US Government. Consequently
redundant system with empirically developed strength there are numerous instances where curtain walls are
capacities that extend far beyond that required. While discarded and replaced rather than conserved.
lack of maintenance could lead to water infiltration Another issue is obsolescence. Curtain walls are the
problems or deterioration of the cladding system, these embodiment of pure hard economics on the construc-
problems do not become immediately apparent but tion industry. Early metal and glass curtain walls, being
become manifest only over a period of time. avant garde, were not designed with a discernible lifes-
In the metal and glass curtain wall strength capac- pan.The aluminum was cheap and easy to fabricate and
ities are engineered to meet specific standards and not finished in a way that would preserve its luster.
redundancy is reduced. An engineered system of per- Some glass products were manufactured using tech-
formance redundancy has been introduced that pro- nically sophisticated and obsolete processes, and it
vides a weathertight barrier while providing internal is no longer feasible to authentically recreate them.
drainage as well. However, a breakdown of these pro- Original sealants or sealants used for repair are not
tective systems in the metal-and-glass curtain wall always compatible with sealants that are presently
would lead to immediate water leakage on the interior. available.
High-tech materials have been adopted for use on A third issue is defining significance of the metal
these curtain walls that offer unique maintenance chal- and glass curtain wall. Is significance defined by the
lenges. Tempered glass may fail due to nickel-sulfide material itself or by the transparency that it provides?
inclusions. Thin-stone veneers may become distressed Perhaps a key to all of the issues cited can be found
from material weakness, loss of strength hysteresis, or in the Nara Document on Authenticity, “authenticity
deterioration due to anchorage. Incompatible materials judgments may be linked to the worth of a great vari-
such as certain sealants and building stone or dissimi- ety of sources of information. Aspects of the sources
lar metals can cause staining which can pose long-term may include form and design, materials and substance,
maintenance problems. Little is known regarding the use and function, traditions and techniques, location
long term performance of sealants that are now being and setting, and spirit and feeling, and other external
used to hold glass panels in place thus eliminating the factors.” This statement infers that the historic fabric
metal grid. does not necessarily need to be conserved to maintain
integrity.
Perhaps our approaches to energy inefficient build-
7 CONCLUSIONS
ing claddings will need to be reassessed as we enter
into a new energy paradigm. Be that as it may, as
Due to the sophisticated engineered systems that have
our post-World War II buildings become historic land-
been introduced, and the decreased redundancy, metal
marks, the challenges that we face will be complex
and glass curtain wall systems pose unique and com-
as well and will require greater degrees of scientific
plex problems. Diagnosis of this unique typology has
inquiry and the active participation of the engineer.
been made much easier by following the guidelines
provided in the ICOMOS Charter: Principles for the
Analysis, Conservation and Structural Restoration of
Architectural Heritage (ISCARSAH Principles) that REFERENCES
was ratified by the ICOMOS 14th General Assembly
in Zimbabwe in 2003. The ISCARSAH Principles pro- Kelley, S., “The Glass and Metal Curtain Wall: History, Diag-
vide a comprehensive and well considered philosophy nostics, and Treatment,” Preserving Post-War Heritage,
Edited by Susan MacDonald, London: Donhead, 2001.
for building diagnostics which proves just as valu-
Kelley, S., “Conflicts and Challenges in Preserving Curtain
able (if not more so) for engineered systems as they Walls,” APT Bulletin, Volume 32, Number 1, 2001.
would for empirically designed medieval structures Kelley, S., “Office Buildings of the Chicago School: The
composed of masonry and wood. Restoration of the Reliance Building,” Konservierung
However there is a wide chasm that presently exists Der Moderne, ICOMOS, Hefte des Deutschen National
between professionals in the preservation community Komitees XXIV, Munich 1998.
and environmental professionals. The obvious debate Kelley, S., and Dennis Johnson, “The Glass and Metal Cur-
is whether it is a sustainable strategy to maintain tainwall: The History and Diagnostics,” Modern Move-
an aesthetic and keep original material or whether ment Heritage, London: E & FN Spon, 1998.
Kelley, S., and B. Kaskel. “Curtain Walls in the USA: Fail-
it is more important to achieve better water shed-
ures, Investigation, and Repair,” Curtain Wall Refurbish-
ding and thermal performance by compromising the ment, A Challenge to Manage, Eindhoven, Netherlands:
original material. Windows are sources of heat loss DOCOMOMO International, 1997.
and gain and curtain walls even more so. This was Kelley, S., “The History of the Glass and Metal Curtain Wall:
not a great concern to designers of large structures From the Reliance Building to the Lever House,” Window
enveloped in glass prior to the Energy Crisis of 1973. Rehabilitation Guide for Historic Buildings, Washington
However energy use is of primary importance today DC: Historic Preservation Education Foundation, 1997.
30
Kelley, S., “Aluminum: History and Conservation.”Twentieth R. Shiffer, editors. Preserving the Recent Past. Washing-
Century Building Materials.Thomas C. Jester, editor. New ton, D.C.: Historic Preservation Education Foundation,
York, New York: McGraw-Hill, 1995. 1995.
Kelley, S., “The History of the Curtain Wall: From Crafts- Kelley, S., C. Paulson, and D. Slaton. “Assessment Tech-
manship to Machine Made.” Preserving the Recent Past. niques Utilized with Historic American Highrises,” in
D. Slaton and R. Shiffer, editors. Washington, D.C.: Structural Repair and Maintenance of Historical Build-
Historic Preservation Education Foundation, 1995. ings. Southampton, England: Computational Mechanics
Hunderman, H., J. Koerber, and S. Kelley. “Curtain Wall Publications, 1989.
Development: The Loss of Redundancy.” D. Slaton and
31
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M.R. Valluzzi
DAUR, University of Padua, Italy
ABSTRACT: Modern techniques and innovative materials are often rapidly proposed and allowed in the current
practice, even for restoration of historical constructions, where fundamental preservation criteria have to be taken
into account.The large variability and complexity of masonry structures and typologies make particularly difficult
the preliminary choices for proper structural models and interventions, that should be based of suitable knowledge
of both existing and new materials, and of their interaction under environmental and loading conditions. Despite
the increasing number of specific studies of FRP reinforcement on masonry structures, still limited codes and
recommendations are available so far. Harmonization of test procedures and methods should be pursued, in order
to compare results and calibrate analytical and numerical models for design and assessment rules.
33
modification of mechanisms at collapse, which can
involve further resisting phenomena.
What essentially emerges from the analysis of a
number of works available in literature aimed at inves-
tigating the mechanical performances of strength-
ened structures and components, and from the code
proposals, are still needs of:
Figure 1. Textile sheets and laminates (a), and various
– definition and putting into practice specific cau- features of FRP bars (c).
tious criteria for possible application of composite
materials in the historical construction preservation
field;
– clarification of critical aspects of application tech-
nologies (e.g., bond and anchorage of textiles and
bars);
– standardization of methods and experimental pro-
cedures for the characterization of the mechanical
performance of strengthened components to define
proper design and assessment criteria;
– definition and validation of investigation proce-
dures for the evaluation of the effectiveness and Figure 2. Wet lay-up system (a), structural repointing
durability of the intervention. (b) and specific anchorage devices for bars (c).
34
Figure 3. Modern residential (a) and masonry historical
(b) buildings in severe conditions.
Figure 5. Loading test on masonry assemblages: vaults
(a) and pillars (b), in-plane (c) and out-of plane (d) testing on
walls.
35
where M is the maximum moment at the section under
consideration, V is the corresponding shear force,
d the distance from the extreme compression fibre to
the centroid of tension reinforcement, fm is the com-
pressive strength of masonry, An is the compressed
area of masonry, Af is the reinforcement area, P is
the axial load, ffe is the design strength of FRP, and
s in the spacing of horizontal reinforcement in the
vertical direction. For CNR DT/200, the analogy with
the design formulas proposed for reinforced masonry
beam, as in the Eurocode 6, is evident:
36
Figure 7. Debonding of CFRP strips due to peeling (a), FRP
rupture beyond bonding limit (b).
37
Figure 11. Double-lap shear test executed on CFRP sheets:
experimental set-up (a), fingerprint on brick after test (b),
Figure 10. Effectiveness factor for shear strength for differ- peeling of brick surface on FRP (c) (Panizza et al. 2008).
ent amount and configuration of FRP reinforcement.
where γFk = c1 · f · fmtm (N mm) is the characteristic
value of the fracture energy, Ef is theYoung modulus of
FRP in the direction of the applied force, tf is the FRP
thickness, fmtm = 0.1fmk is the mean tensile strength
of masonry (considered coincident with the strength Figure 12. Scheme of wall adopted for FE simulation: unre-
of the blocks), γM and γf ,D are partial safety fac- inforced wall (a), horizontal FRP strip application in lintels
tors, varying from 1.1–1.25 and 1.2–1.5, respectively, (b), vertical strips on main vertical walls to reinforce flexural
behavior (c), addition of horizontal strips on main vertical
depending on the certification of the entire bonding
sects to reinforce shear behavior (d).
system on the support, or of only the single materials.
c1 is a coefficient to identify on experimental basis or
to adopt equal to 0.015 (0.03 is proposed for concrete connection between reinforcement and masonry. The
with the same equation). comparison with the application of equations (5) and
The non homogeneity of masonry due to the pres- the elaboration of experimental data able to define effi-
ence of mortar joints, and the consequent influence in ciency laws for debonding as in (Panizza et al. 2008),
the bonding phenomenon of the different mechanical allowed to calibrate a FE model (with DIANA), in
properties and of the geometrical discontinuity, are not order to simulate the performance of shear and flexural
considered in the model; moreover, a unique signifi- strengthening with CFRP sheets applied in a masonry
cant value for the fracture energy along the connection wall including openings. The scheme of the model is
is assumed. depicted in Figure 12, and the comparison among the
A fundamental contribution to clarify these aspect main parameters at the interface between reinforce-
is done by several research groups (Briccoli Bati et al. ment and masonry, to be used for the model, are given
2001, Aiello et al. 2003 and 2006, Casareto et al. in Table 2. Experimental elaboration obtained by DL
2003, Basilio et al. 2005, Panizza et al. 2008), but shear tests have been considered, taking into account
different test procedures are adopted, thus comparison a reduction of 30% to obtain reasonable characteristic
of results is often unreliable. In particular, the most values, to compare to the ones computed according to
suitable arrangements could be the double-lap (DL) the CNR standard.
(Figure 11) and the single-lap shear test (SL), the latter The main characteristics of materials were derived
being the most effective, due to the problem of repro- from available experimental tests, or computed on the
ducing actual symmetry of load distribution in the DL basis of the national standards. For clay bricks, a char-
configuration, but not simple to realize, in comparison acteristic compressive strength of 41.2 MPa, a mean
with the former one. tensile strength of 2.4 MPa and an elastic modulus of
The identification of ffdd is crucial, as it represents 16 GPa, were assumed; a mortar M2.5 (MPa, com-
the parameter strictly related to the efficiency of the pression) was considered, whereas computed global
38
Table 2. Mechanical properties of materials.
39
Figure 15. Comparison among plain masonry and three
strengthening configuration by using CNR DT-200 for-
mulation and elastic-brittle or elasto-plastic law for FRP
strips.
40
Figure 17. Repair intervention on masonry vaults of S.ta
Corona Church in Vicenza: view of church and detail of
cracked cross vault (a), scheme of intervention at extrados
and view of FRP application combined with injections (b).
41
more and more increasing. Nevertheless, many aspects
still are under investigation, and specific experimental
procedures and models need to be homogenized in
standards and recommendations.
Therefore, from more and more modern concep-
tions, today more than in the past, we need to resume
traditional values, in order to not forget our learning
from history and to respect constructive specificity
and functions of the original structures. Innovative
solutions can be very useful even in the Cultural
Heritage context, provided that we are able to recog-
nize their limits, and to pursue the clarification of all
aspects (both positive and negative) that are involved
in the delicate question of the preservation of historical
constructions.
ACKNOWLEDGEMENTS
Figure 20. Intervention on ‘Arca of Cansignorio’ (Verona): The author wish to acknowledge C. Bettio, E. Garbin,
view of deteriorated hoof and first phase of intervention (a), G. Guidi and M. Panizza for their contribution in the
protecting lime putty and final view of the statue after researches and applications and for data processing
intervention (b). and modelling.
42
FRP: Experimental Analysis. British Masonry Society, Corradi, M., Grazini, A., Borri, A. 2007. Confinement of
proc. of the 6th Int. Conf. brick masonry columns with CFRP materials. Composites
Basilio, I., Oliveira, D., Lourenço, P. 2004. Optimal FRP Science and Technology, 67 (9): 1772–1783.
strengthening of masonry arches. Brick and Block Dai, J.G., Ueda, T., Sato,Y. 2005. Development of the Nonlin-
Masonry Conference, proc. of the 13th Int. Conf., Ams- ear Bond Stress-Slip Model of Fiber Reinforced Plastics
terdam, Netherlands. Sheet-Concrete Interfaces with a Simple Method. ASCE J.
Basilio I., Oliveira D., Lourenço P. 2005. Experimental Char- of Composites for Construction, 9 (1): 52–62.
acterization of FRP Masonry interface Behaviour. Proc. De Lorenzis, L., D., Tinazzi, A., Nanni, A., 2000. Near
of the 5th Int. Conf. on Analytical Models and New Con- Surface Mounted FRP Rods for Masonry Strengthening:
cepts in Concrete and Masonry Structures AMCM-2005, Bond and Flexural Testing. Meccanica delle Strutture in
Gliwice-Ustron (Poland). Muratura Rinforzate con FRP Materials, proc. of the 2nd
Binda L., Cardani G., Saisi A., Valluzzi M.R. 2006. Vulnera- Nat. Symp., Venice, Italy.
bility analysis of the historical buildings in seismic area by De Lorenzis, L., Dimitri, R., La Tegola, A. 2005. Strength-
a multilevel approach. Asian Journal of Civil Engineering, ening of masonry edge vaults with FRP composites.
7(4): 343–357. Composites in Construction – CCC2005, proc. of the 3rd
Borri, A., Casadei, P., Castori, G., Ebaugh, S. 2007. Research Int. Conf., Lyon, France.
on composite strengthening of masonry arches. Fiber De Lorenzis, L., Dimitri, R., La Tegola, A. 2007. Reduction
Reinforced Polymer Reinforcement for Concrete Struc- of the lateral thrust of masonry arches and vaults with FRP
tures – FRPRCS-8, proc. of the 8th Int. Symp., Patras, composites. Construction and Building Materials, 21 (7):
Greece. 1415–1430.
Borri, A., Corradi, M., Vignoli, A. 2002. New materials for De Lorenzis, L., Galati, N., Ombres, L. 2004. In-plane shear
strengthening and seismic upgrading interventions. Ari- strengthening of natural masonry walls with NSM CFRP
adne 10, Arcchip, proc. of the Int. Workshop, Prague, strips and FRCM overlay. Structural Analysis of historical
Czech Republic. constructions SAHC 2001, proc. of 4th Int. Sem., Padova,
Brencich, A., Gambarotta, L. 2005. Mechanical response of Italy, 843–855.
solid clay brickwork under eccentric loading. Part I and De Lorenzis, L., Miller, B., Nanni, A. 2001. Bond of FRP
II. Materials and Structures, 38. laminates to concrete. ACI Materials Journal, 98 (3):
Briccoli Bati, S., Rotunno, T. 2001. Environmental durabil- 256–264.
ity of the bond between the CFRP composite materials Ehsani M.R., Saadatmanesh H., Al-Saidy, A. 1997. Shear
and masonry structures. Structural Analysis of historical behavior of URM retrofitted with FRP overlays. ASCE J.
constructions – SAHC 2001, proc. of the 3rd Int. Conf., of Composites for Construction, 1 (1): 17–25
Guimaraes, Portugal, 1039–1046. El-Dakhakhni, W.W., Hamid, A.A., Hakam, Z.H.R.,
Briccoli Bati, S., Rovero, L. 2000. Consolidation of masonry Elgaaly, M. 2006. Hazard mitigation and strengthening of
arches with carbon-fiber reinforced plastics. 12th Int. unreinforced masonry walls using composites. Composite
BRICK/BLOCK Conference, proc., Madrid, Spain. Structures, 73 (4): 458–477.
Briccoli Bati, S., Rovero, L. 2001. Experimental validation El-Gawady, M.A., Lestuzzi, P., Badoux, M. 2005. Aseis-
of a proposed numerical model for the FRP consolida- mic retrofitting of unreinforced masonry walls using FRP.
tion of masonry arches. Structural Analysis of historical Composites Part B, 37 (2): 148–162.
construction – SAHC 2001, proc. of the 3rd Int. Conf., EN 1996. Eurocode 6. Design of masonry structures.
Guimaraes, Portugal, 1057–1066. Foraboschi, P. 2001. Strength Assessment of Masonry Arch
Casareto M., Oliveri A., Romelli A., Lagomarsino S. 2003. Retrofitted Using Composite Reinforcements. Masonry
Bond behavior of FRP laminates adhered to masonry. International, 15 (1): 17–25.
Advancing with Composites – Plast-2003, proc. of the Int. Foraboschi, P. 2004. Strengthening of Masonry Arches with
Conf., Milan, Italy. Fiber-Reinforced Polymer Strips. ASCE J. of Composites
Cecchi, A., Milani, G., Tralli, A. 2004. In-plane loaded for Construction, 8 (3): 191–202.
CFRP reinforced masonry walls: mechanical character- Foster, P., Gergely, J., Young, D., McGinley, M. 2005.
istics by homogenisation procedures. Composites Science Strengthening masonry buildings with FRP composites.
and Technology, 64: 2097–2112. Structural Faults & Repair SF&R2006, proc. of Int. Conf.,
Charter of Krakow (2000). Principles for conservation and Edinburgh, UK.
restoration of built heritage. ICOMOS, Paris. Gabor, A., Ferrier, E., Jacquelin, E., Hamelin, P. 2006. Anal-
Charter of Venice (1964). International charter for the ysis and modelling of the in-plane shear behaviour of
conservation and restoration of monuments and sites. hollow brick masonry panels, Construction and Building
Decision and resolutions. ICOMOS, Paris Materials, 20: 308–321.
Chen, J.F., Teng, J.G. 2001. Anchorage Strength Models for Galati, N., Micelli, F., Tumialan, J.G., La Tegola, A., Nanni, A.
FRP and Steel Plates Bonded to Concrete. ASCE Journal 2004. Comparison between FRP strengthening techniques
of Structural Engineering, 127 (7): 784–791. on the Out-of-Plane behaviour of URM masonry walls.
CNR-DT 200/2004. Guide for the Design and Construc- Innovative Materials and Technologies for Construction
tion of Externally Bonded FRP Systems for Strengthening and Restoration – IMTCR04, proc. of the 1st Int. Conf.,
Existing Structures. National Research Council, Italy. Lecce, Italy, 440–457.
Corradi, M., Borri, A., Vignoli, A. 2002. Strengthening Garbin, E., Valluzzi, M.R., Modena, C., Galati, N., Nanni, A.
techniques tested on masonry structures struck by the 2006. In-plane design for masonry walls strengthened by
Umbria-Marche earthquake of 1997–1998. Construction FRP materials, Structural Faults & Repair – 2006, June
and Building Materials, 16 (4): 229–239. 13–15 2006, Edinburgh, UK (12 pp. on CD-ROM).
43
Gaudini, G., Modena, C., Casarin, F., Bettio, C., Lucchin F. Micelli, F., De Lorenzis, L., La Tegola, A. 2004. FRP-
(2008), Monitoring and strengthening interventions on confined masonry columns under axial loads: experimen-
the stone tomb of Cansignorio della Scala, Verona, Italy, tal results and analytical model. Masonry International,
6th Int. Seminar on Structural Analysis of Historical 17 (3): 95–108
Constructions, 2–4 July 2008, Bath, UK. Modena, C. 1997. Criteria for cautious repair of his-
Gilstrap, J.M., Dolan, C.W. 1998. Out-of-plane bending of toric building. A valuation and strengthening of exist-
FRP-reinforced masonry walls. Composites Science and ing masonry structures, Binda L. and Modena C., Ed:
Technology, 58 (8): 1277–1284. RILEM.
Giuffrè A. 1993. Sicurezza e conservazione dei centri storici: Morbin, A. 2003. Strengthening of Masonry Elements
il caso di Ortigia, Laterza: Bari, Italy. with FRP Composites, Report No. CIES 02–23, Cen-
Grando, S., Valluzzi, M.R., Tumialan, J.G., Nanni A. ter for Infrastructure Engineering Studies, University of
2003. Shear Strengthening of UMR Clay Walls whit Missouri-Rolla.
FRP Systems, 6th International Symposium on Fiber- Mosallam, A.S. 2007. Out-of-plane flexural behavior of
Reinforced Polymer (FRP) Reinforced for Concrete Struc- unreinforced red brick walls strengthened with FRP
tures (FRPRCS-6), Singapore, 8–10 July 2003, 1229– composites. Composites: Part B, 38: 559–574.
1238. Nanni, A., Tumialan, G. 2003. Fiber-Reinforced Compos-
Hamid, A.A., El-Dakhakhni, W.W., Hakam, Z.H.R., ites for the Strengthening of Masonry Structures. Struc-
Elgaaly, M. 2005. Behavior of Composite Unreinforced tural Engineering International, 13(4), 1 November 2003,
Masonry – Fiber-Reinforced Polymer Wall Assemblages pp. 271–278
Under In-Plane Loading. ASCE J. of Composites for Nurchi, A., Valdes, M. 2005. Strengthening of stone masonry
Construction, 9 (1): 73–83. columns by means of cement-based composite wrapping.
Hamilton III, H.R., Dolan, C.W. 2001. Flexural Capacity of Composites in Construction – CCC2005, proc. of the 3rd
Glass FRP Strengthened Concrete Masonry Walls. ASCE Int. Conf., Lyon, France.
J. of Composites for Construction, 5 (3): 170–178. Oliveira, D.V., Basilìo, I., Lourenço, P.B. 2006. FRP strength-
Hamoush, S., McGinley, M., Mlakar, P., Scott, D., Murray, K. ening of masonry arches towards an enhanced behaviour.
2001. Out-of-plane Strengthening of Masonry Walls with Bridge Maintenance, Safety and Management – IAB-
Reinforced Composites. ASCE J. of Composites for Con- MAS’06, proc., Porto, Portugal.
struction, 5 (3): 139–145. OPCM 3431/2005. Ordinanza 3431/2005 del 3 Maggio
Haroun, M., Mosallam, A.S., Allam, K.H. 2003. Cyclic In- 2005: Norme tecniche per il progetto, la valutazione
Plane Shear of Concrete Masonry Walls Strengthened by e l’adeguamento sismico degli edifici. P.C.M., Italy.
FRP Laminates. Journal of Engineering Mechanics, 128 Panizza, M., Garbin, E., Valluzzi, M.R., Modena, C. 2008.
(5): 562–573. Bond Behaviour of CFRP and GFRP Laminates on Brick
Kiss, R.M., Kollàr, L.P., Jai, J., Krawinkler, H. 2002. Masonry Masonry. Structural Analysis of Historical Construction –
Strengthened with FRP Subjected to Combined Bend- SAHC08, proc. of the 6th Int. Seminar, Bath (UK) .
ing and Compression, Part II: Test Results and Model Papanicolaou, C.G., Triantafillou, T.C., Karlos, K., Pap-
Predictions. Journal of Composite Materials, 36 (9): athanasiou, M. 2007. Textile-reinforced mortar (TRM)
1049–1063. versus FRP as strengthening material of URM walls:
Krevaikas, T.D., Triantafillou, T.C. 2005. Masonry con- in-plane cyclic loading. Materials and Structures, 40:
finement with Fiber-Reinforced polymers. ASCE J. of 1081–1097.
Composites for Construction, 3–4: 128–35. Prota, A., Marcari, G., Fabbrocino, G., Manfredi, G.,
Kuzik, M.D., Elwi, A.E., Roger Cheng, J.J. 2003. Cyclic Flex- Aldea, C. 2006. Experimental In-Plane Behavior of Tuff
ure Tests of Masonry Walls Reinforced with Glass Fiber Masonry Strengthened with Cementitious Matrix-Grid
Reinforced Polymer Sheets. ASCE J. of Composites for Composites. ASCE Journal of Com-posites for Construc-
Construction, 7 (1), 20–30. tion, 10 (3): 223–233.
Li, T., Galati, N., Nanni, A. 2004. Research on FRP strength- Saadatmanesh, H. 1997. Extending service life of concrete
ening of URM walls with opening. Mechanics of Masonry and masonry structures with fiber composites. Construc-
Structures Strengthened with FRP-materials: Modeling, tion and Building Materials, 11 (5): 327–335.
Testing, Design, Control, proc. of the Int. Symp., Venice, Savoia, M., Ferracuti, B., Mazzotti, C. 2003. Non linear bond-
Italy. slip law for FRP-concrete interface. Fibre-Reinforced
Liu, Y., Dawe, J., McInerney, J. 2005. Behaviour of GFRP Polymer Reinforcement for Concrete Structures FRPRCS-
sheets bonded to masonry walls. Bond Behaviour of FRP 6; proc. int. symp., Singapore.
in Structures – BBFS 2005, proc. of the Int. Symp., Hong Schwegler, G. 1994. Masonry construction strengthened
Kong, China, 473–480. with fiber composites in seismically endangered zones.
Lourenço, P., Poças Martins, J.P. 2001. Strengthening of the Earthquake Engineering, proc. of the 10th Europ. Conf.,
architectural heritage with composite materials. Compos- Rotterdam, The Netherlands, 454–458.
ites in Construction – CCC2001, proc. of the 1st Int. Conf., Shrive, N.G. 2006. The use of fibre reinforced polymers to
Porto, Portugal. improve seismic resistance of masonry. Construction and
Luciano, R., Marfia, S., Sacco, E. 2001. Reinforcement of Building Materials, 20 (4): 269–277.
masonry arches by FRP materials: experimental tests and Tan, K.H., Patoary, M.K.H., Roger, C.S.K. 2003. Anchor-
numerical investigations. Composites in Construction – age Systems for Masonry Walls Strengthened with FRP
CCC2001, proc. of the 1st Int. Conf., Porto, Portugal. Composite Laminates. J. of Reinforced Plastics and Com-
Luciano, R., Sacco, E. 1998. Damage of masonry panels rein- posites, 22 (15): 1353–1371.
forced by FRP sheets. International Journal of Solids and Tomaževič, M. 1999. Earth quake resistano design of
Structures, 35 (15): 1723–1741. masonry buildings. Imperial College Press: London, UK.
44
Triantafillou, T.C. 1998. Strengthening of masonry struc- Valluzzi, M.R., Tinazzi, D., Garbin, E., Modena, C. 2003.
tures using epoxy-bonded FRP laminates. ASCE J. of FEM modelling of CFRP strips bond behaviour for bed
Composites for Construction, 2(2): 96–104. joints reinforcement techniques. Computer Methods in
Triantafillou, T.C., Fardis, M.N. 1997. Strengthening of Structural Masonry; proc. 6th int. conf., Rome, Italy.
historic masonry structures with composite materials. Valluzzi, M.R., Valdemarca, M., Modena, C. 2001. Behavior
Materials and Structures, 30: 486–496. of brick masonry vaults strengthened by FRP laminates.
Turco, V., Secondin, S., Morbin, A., Valluzzi, M.R., Mod- ASCE Journal of Composites for Construction, August
ena, C. 2006. Flexural and shear strengthening of un- 2001, vol. 5, n. 3, pp. 163–169.
reinforced masonry with FRP bars, Composites Science Velazquez-Dimas, J.I., Ehsani, M.R., Saadatmanesh, H. 2000.
and Technology, 66(2):289–296. Out-of-Plane Behavior of Brick Masonry Walls Strength-
Valluzzi, M.R., da Porto, F., Modena, C. 2004. Strengthen- ened with Fiber Composites. ACI Structural Journal, 97
ing of a masonry vault with GFRP laminates: the case (3): 377–387.
of the church of S. Fermo in Verona. Innovative Mate- Yu, P., Silva, P.F., Nanni, A. 2004. Application of Bond
rials and Technologies for Construction and Restoration Polyurea in Structural Strengthening of RC Beams and
-IMTCR04, Lecce (Italy), 6–9 June 2004, pp. 430–439. UMR Walls, Final Report, Report No. CIES 01-49, Cen-
Valluzzi, M.R., Tinazzi, D., Modena, C. 2002. Shear behav- ter for Infrastructure Engineering Studies, University of
ior of masonry panels strengthened by FRP laminates. Missouri-Rolla.
Construction and Building Materials, Special Issue,
16(7): 409-416.
45
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
D.F. D’Ayala
Department of Architecture and Civil Engineering, University of Bath, UK
E. Tomasoni
University of Brescia, Italy
ABSTRACT: The increasing interest in historic architectural heritage and the need for preservation of historical
structures has led to the continuous development in the past 20 years of a growing number of methods for the
analysis of masonry vaults. This notwithstanding some type of vaults, for instance pavilion vaults and fan vaults,
despite their broad use in past centuries, have not been thoroughly studied, mainly due to the difficulty of applying
simplified theories to their complex shapes. The major simplification that is usually carried out is to reduce the
vault to a series of adjacent arches, without transversal connection. Even though simple and accurate, the arch
model does not take into account the interaction between the arches and it is not able to properly simulate the
three-dimensional effects in the vaults. Although the results are conservative for uniform load distributions, the
model limits substantially the set of loading conditions that can be analysed and hence a thorough assessment
of vaults performance. This limitation and the need for a flexible method to study the different types of vaults
could be solved by using the concept of the surface of thrust within the framework of limit state analysis. The
main object of the present paper is the development of a computational procedure which allows to define the
3D structural behaviour of masonry vaults: using limit state analysis with finite friction, the proposed analytical
method, based on lower bound approach, allows to obtain, for a generic type of vault, the actual crack pattern,
the stress field and the horizontal thrust at the supports for both gravitational and localized loads. In the present
paper the limit state analysis with finite friction has been applied to pavilion vaults. The results obtained are
compared with non linear F.E. analysis simulations.
47
of the existing pre-stresses, a structure may or may not reduction to an ideal system of many arches. While
collapse. Sometimes, it is nevertheless possible to find this oversimplification could be admissible for barrel
non-null safe limits. vaults and spherical domes under simple gravita-
The existence of a safe domain within the friction tional loads, cannot do justice of the three-dimensional
domain is extended further by Josselin de Jong (1964, effects developing in other types of vaults, ribbed or
1973) by including the positive influence of normal with discontinuity lines. Such effects may be essential
forces on stability. If through the analysis of the entire to explain stability.
structure the value of the minimal normal force acting For Heyman (1966), the actual stress state cannot be
on the interfaces can be determined, the lower bound found. He abandons the behaviour’s ‘quest’ to concen-
safe domain can be extended. Palmer technique can trate on safety. In the spirit of ‘standard’ limit analysis,
be used to construct it. The limit of applicability of all methods able to give an admissible stress state can
this procedure to masonry structures has been stud- be used to give a lower bound of the collapse load.
ied by Smars (2000). A prove of uniqueness of the Geometry and specific masses are the basic input data.
safety factor for vaulted and domed structures under In this approach, the possibility of sliding is ruled out
symmetric loading condition, once the magnitude of a priori. However cracking is considered and resis-
the normal force is known is given by D’Ayala and tant mechanisms able to assure the equilibrium after
Casapulla (2001). cracking are identified (Heyman, 1977).
Since Heyman fisrt formulation of the plasticity The Heyman’s study presents a useful and intu-
theorems for masonry (Heyman 1966), limit analysis itive approach to understand the behaviour of masonry
has been successfully applied to masonry in the ‘stan- arches and vaults and provides the value of minimum
dard’ format under the assumption that the friction thickness to span ratio as a safe solution under the
coefficient would be high enough to prevent sliding assumptions of infinite compressive strength and fric-
in the range of problems considered. This powerful tion resistance and zero tensile strength. Furthermore,
tool was specifically applied to the safety analysis of this work suggests that, for complex vaults, the junc-
masonry arch bridges and later extended to the analy- tion between two shell surfaces leads to a large stress
sis of vaulted structures; in the search for lower bounds concentration.
different methods have been used to find admis- In real vaults however, sliding does occur, especially
sible state of stresses: funicular methods (Heyman if accompanied by a loss of shape. In that case, the
1966, Harvey 1988, Huerta 2001), membrane analysis ‘lower bound’ estimates can be unsafe. Stereotomy of
(Heyman 1966), network of forces (O’Dwyer 1999), the vault, coefficient of friction and resistance to ten-
minimum of complementary energy (Maier 1990) or sion must then be considered. D’Ayala (1993, 1994,
FE with incremental failure analysis. 2001) uses the knowledge of extreme meridian nor-
Livesley (1978, 1992), by adopting a static mal forces in a dome to demonstrate its stability, even
approach, was the first to developed a formal linear in the case of potential sliding. Smars (1993, 2000)
programming procedure to discuss the existence of showed that locally stronger structures are not neces-
safe load factor of two-dimensional vaulted structures. sarily safer, proposing a technique to build a lower
Within this approach research has developed substan- bound domain for structures having possible local
tially in the past decade (D’Ayala 1993, Boothby 1994, tension resistance.
Baggio & Trovalusci 1998, Ferris 2001). Friction is Nowadays and to our knowledge, no general-
often taken into account but the question of applica- purpose software exist permitting 3D limit analysis
bility of the general theorems of plasticity is not much of vaults. Existing tools are not tailored to their actual
discussed. If it is usually recognised that the maxi- behaviour and complex geometry.
mum safe load factor can be overestimated, it does not In the last years the developments of constitutive
seem to be clear that it is not unique. Only Mauldon laws for masonry structure are been made and most
et al. (1997) make references to stable, instable and finite element analysis of the masonry vaults are been
potentially stable domains. carried out (Lourenco, 2006).
Among masonry structures, arches, vaults and but- Nevertheless, the analysis of historical masonry
tresses represent the principal structural component of constructions is a complex task and only few propos-
historic masonry buildings. They epitomise masonry als specifically oriented to the non-linear analysis of
features and problems. They are mainly subjected to masonry domes and vaults exist.
compressions, are shaped accordingly and commonly Oppenheim et al. (1989) propose a limit state anal-
present cracks. Techniques developed for vaults can ysis of masonry dome, but does not take into account
help to understand and define techniques for other the interaction between the slices.
masonry structures. Nart (2003) describes the mechanical behaviour of
However the interpretation of masonry vaults’ masonry domes in relation to the applied loads and
behaviour is difficult, especially when the vaulted sur- geometry. The author’s analysis, based on the geomet-
face is not smooth and continuous, leading often to a rical parameters, consider a great variety of shapes and
48
loads, but assume zero hoop stresses and neglect the the binding materials, the original friction coefficient
sliding mechanisms. could be substantially reduced. Therefore the shear
Interesting is the work by Block (Block et al. 2006), strength at blocks interfaces is not infinite, but deter-
that propose a structural analysis tools based on the mined by the cohesion and the internal friction angle,
limit state analysis for vaulted masonry buildings. having assumed the Mohr–Coulomb criterion as a
This study extends the graphical method for limit good representation of the real behaviour. Although
analysis using the line of thrust. Nevertheless it does very little testing has been performed on historic
not provide a complete analysis of three-dimensional masonry to quantify both characteristic shear strength
behaviour and mechanisms of vaulted structures. and friction coefficient, this assumption is supported
Only few studies explain the tree-dimensional by work conducted by Hendry et al. (1986) on new
effects. Modelling the principal stresses in a masonry masonry assemblies, providing value of friction coef-
vault as a discrete network of forces, O’Dwyer (1999) ficient lower than 0.4.
develops the limit analysis for vaults, able to take Modelling a vault of any shape as a three-
into account redistribution effects. However this work dimensional discrete system of rigid blocks, along
assume an initial value for the horizontal component meridians and parallels, it is possible to determine, by
of the resultant of stresses under the condition that the use of the general shell theory, the meridian stresses,
friction between the voussoirs is sufficient to prevent the hoop stresses and the shear stresses (Fig. 1).
failure due to sliding. Careful observation of historical vaults and domes
An iterative procedure for the analysis of the highlights that, unless they are corbelled, and indepen-
masonry structure, that take into account the non- dently of the bricks bond (for instance brick courses
associative frictional joints, is proposed by Gilbert parallel to the spring, perpendicular to the spring or
et al. (2006) that provide a tool for analysing the sta- herringbone brickwork), the joints are always placed
bility of masonry gravity structures. This method, easy in the arch’s normal plane, i.e. towards the centre of
for masonry walls and for rib arches, could be very curvature. Hence, as the weaker elements in the fabric
difficult to apply for three-dimensional structure. are the joints the analysis can be most usefully car-
Both the three-dimensional effects and the slid- ried out at the blocks’ interface, where the stress field
ing mechanism are studied by D’Ayala e Casapulla is related to the frictional (plastic) limits, and where
(2001).This study analyzes hemispherical domes by failure by shear or tension will first occur.
a new analysis tools based on the limit state analysis While it is evident that the discretization scale could
with finite friction and provides a simple proof of the not be representative of each single brick or block
unique solution. constituting the fabric, each portion identified by the
The same procedure, based on membrane theory intersection of 2 parallels with two adjacent meridians
with limit state analysis, could be applied at different can be considered as a macroelement of homogenous
type of vaults. masonry material, with infinite internal compression
Hence, ascertained the value of finite friction in and tensile and shear strength at the interfaces between
the masonry analysis and the importance of three- portions defined by a Coulomb criterion.
dimensional effects for the study to the complex vaults, In the case of double curvature vaulted structures,
the paper present a limit state analysis with finite fric- for gravity load distributions, if the material has tensile
tion applied to a masonry domes, pavilion vaults and strength, then membrane theory might be assumed to
fan vaults, in order to evaluate their actual structural apply with resultants of meridian and parallel stresses
behaviour and the minimum thickness to define the tangent to the middle surface of the vault at any point.
stability condition, very important element with regard As masonry by hypothesis is generally not able to resist
to structural rehabilitation. tensile stresses for most shapes of vault the gravity load
distribution will cause tensile stresses near the spring
level that the material is not able to absorb and hence
2 STRUCTURAL ASSUMPTIONS FOR cracks will form. This means that the stress field is
MASONRY VAULTS modified with respect to the membrane theory, the
meridian stress resultant is no more tangent to the
The structural behaviour of masonry vaults and their meridian geometric curve and generally not centred,
collapse mechanisms depends on the material prop- so that both resultant bending moment and shear stress
erty, so the present theory is founded on the assumption resultant are present at the block interface. The curva-
of infinite compressive strength for the blocks, no ture of the thrust surface is hence not known a-priori
tension transmitted across the joints and finite friction. and it is generally not constant.
The last hypothesis, often neglected by previous As a consequence of the accruing of such stress
studies, reflects a more realistic masonry mechanical resultants, the equations of equilibrium for classic
behaviour because, especially for historical structures, membrane theory are not sufficient to solve the prob-
after the deterioration of the contact surfaces or of lem, which has ∞2 possible solutions. Among those,
49
Figure 2. Oversimplification of forces Nj and Tj, compo-
nents of Sj normal and parallel to the interface J respectively,
and Coulomb’s cone projection.
50
complex 3-dimensinal state of stress that develops in
presence of uniform gravity loading does not lend
itself easily to simplified approaches.
These vaults have continuous support along walls;
have finite curvature in the direction of the meridi-
ans, but infinite curvature along horizontal plane. This
means that there can be no contribution to recentre the
line of thrust from hoop stresses and hence membrane
theory is inherently non applicable. This means that
such type of vaults is often affected by cracks along the
diagonals, where the geometry of the vault is farthest
from the geometry of the thrust surface generated from
the gravity load distribution. Cracks can also appear
toward the centre of the web of each portion so, in
the past, they are been often modeled as independent
arches. Figure 3. Geometric characteristics of the vault’s generatrix.
However the simplified arch model, does not
account for the capacity of the vault to transfer load
associated with shear and for the arch effect that can
be develop within the horizontal strips due to their
non negligible thickness, similarly to the arch effect
accounted for in walls or slabs.
On the basis of this considerations and suitably sim-
ulating the boundary conditions and the condition of
equilibrium that arise along the diagonal due to the
singularity of the surfaces along the diagonals, limit
state analysis can be applied to this family of vaults to
determine their state of stress at failure.
The main objective of the present section is to under-
stand, through the limit state analysis approach, the
structural behaviour of pavilion vaults, in order to eval-
uate with more accuracy the actual stress field that can
cause cracks.
Moreover this method allows finding the thrust line
position and the admissible thrust surfaces, leading to
the calculation of the minimum thickness which sat-
isfies at the same time equilibrium and compatibility.
This is of great importance with regard to strength-
ening interventions because it allows determining the Figure 4. Geometric characteristic of the diagonal and of the
vaults’ safety factor, which can be expressed, as pro- vault’s generatrix, used for the centring work (Levi, 1932).
posed by Heyman as the ratio between the geometric
thickness over the minimum required thickness, but the equation of an ellipsis (Curioni, 1870; Breymann,
also the accurate position of the hinges at failure and 1885; Levi, 1932).
hence the appropriate positioning of ties or other thrust Assuming the n slices are made up of m blocks,
contrasting devices. each block is identified, in the global system, by the
The simplest case to analyse is a pavilion vault over a coordinates of its centre of mass:
plane square area and of semicircular curvature, which
is also the most frequent occurrence according to the
technical literature (Scamozzi, 1615), subject to self
weight loading.
Figure 3 shows the geometric characteristic of the
vault’s generatrix, where R is its radius, f is the vault’s
rise, l is the span. In the numerical procedure the slices
that make up half of the web between to successive for i = 1 to m and k = 1 to n
ribs are considered, these being present only along were αk is the horizontal angle between the x axis
diagonal. The ribs according to historic the technical in the global system and the horizontal projection of
literature, as shown in figure 3, can be described by x-axis in the local system as shown in Figure 5 and θi
51
Hence, the eccentricity of the thrust surface at each
point is:
52
Figure 7. Cracked masonry showing the shear force Txθ and
Figure 6. (a) view from the top of the pavilion vault, with the force Sj transmitted between the blocks.
the forces for a generic element, (b) axonometric showing the
angles γi and α’k . Hence the horizontal force Hp along the diagonal is:
The shear force at the interface between two slices,
shown in Figure 6a, can be quantified using translation
equilibrium:
Starting from the diagonal slice, for horizontal
equilibrium, the parallel force Hp−1 is:
53
Imposing the frictional constraint:
54
Figure 10. Meridian force S before cracking, after cracking Figure 12. S*, normal force N and shear force T at the
and after the optimization for the slice near to diagonal. interface for the slice near to diagonal.
Figure 11. S*, normal force N and shear force T at the Figure 13. Thrust surface’s eccentricity for the
interface for the central slice. non-optimized solution and for the optimum solution
(slice near to diagonal).
and imposing the material constrains. This force
presents the same trend of the S* before optimiza- 4.3 Thrust surface’s eccentricity
tion, but the optimum S* diverges earlier from the
The thrust surface’s eccentricity is shown in Figure 13.
membrane solution, indicating that cracks will extend
It should be noted that the non-optimized solution dif-
further in along the meridian but that the value of the
fers everywhere from the optimum solution. The non
resultant is smaller. This change in value id directly
optimized solution, indeed, is equal to zero along the
related to the distribution of eccentricity, especially for
uncracked area and it has an exponential trend from
the slice closer to the diagonal as it will be seen in 4.3.
the haunches to the support, because the absence of
the hoop stresses requires an infinite thickness.
It is well known that, with a lower bound approach,
4.2 Normal force N and shear force T at the
between the ∞ possible solutions, the actual solution
interface
is the one that maximise the load-bearing capacity
The resultant meridian normal force N and shear force of the vault while the thrust surface is constrained
T at the interface, components of S* normal and paral- within the vault’s thickness: imposing the condition
lel to the interface respectively, are shown in Figures 11 of minimum eccentricity and the condition that the
and 12. As it can be seen, the force N shows the same friction between the voussoir is sufficient to prevent
trend of S* while the shear force T is equal to zero failure due to sliding, the actual solution can be iden-
up to the cracking point, where the thrust line is per- tified. Hence, as shown in Figure 13, the eccentricity
pendicular to the block interface, and it increases in in the optimum solution reduces drastically and the
the cracking area, where the funicular breaks away presence of relative peaks indicate the possible for-
from the middle surface because of the absence of axial mation of plastic hinges at the extrados, at x equal
forces along the parallel. Besides, it can be observed to 1.5 m (θ j = 30◦ ), and at the intrados, about at x
that, from the centre of the web to the diagonal, the equal to 2.78 m (θ j = 65◦ ). The eccentricity presents
shear force T increase. Indeed the angle γj , repre- its absolute maximum at the support.
senting the inclination of the thrust line, increasingly The chart in Figure 14 shows the corresponding
differs from the geometric angle θj moving from the bending moment for the four slices, calculated as
centre of the web to the diagonal. the product of the meridian normal force N and the
55
Figure 14. Bending moment for all the slices belonging to Figure 15. Horizontal component of S* along the meridian
half web. in the web centre.
56
Figure 19. Comparison between the meridian forces S
obtained by the limit state analysis and the Finite Elements
Analysis along the slice near the diagonal.
57
Figure 23. Possible vault’s collapse mechanism for any
slices.
58
Figure 27. Comparison between the eccentricities obtained
Figure 25. Comparison between the meridian forces S by the limit state analysis for the rise-span ratio equal to
obtained by the limit state analysis after meridian cracking 1/3 (rise = 2 m), 1,5 (rise = 1,5 m) and 1,2 (rise = 1,2 m)
and before meridian cracking for the rise-span ratio equal respectively (central slice).
to 1/3 (rise = 2 m), 1,5 (rise = 1,5 m) and 1,2 (rise = 1,2 m)
respectively (central slice).
59
The results obtained by the limit state analysis with
finite friction show that this schematisation, especially
for complex vaults, do not provide information about
the actual stress field. In fact, in the case of pavilion
vaults, it is emerged the importance of the interac-
tion between the arches for the evaluation of the crack
pattern and the horizontal thrust.
Therefore, the present paper provide a new anal-
ysis tool for a computer procedure able to value
both the rotational and sliding mechanisms, the three-
dimensional effects in the vaults, very important
especially for complex vaults and the vaults’ safety
factor, given by the minimum thickness over the actual
thickness ratio.
60
D’Ayala D. & Casapulla C. 2001 Limit state analysis of hemi- Levi C. 1932. Trattato teorico-pratico di costruzioni civili,
spherical domes with finite friction. In Lourenço P.B., rurali, stradali ed idrauliche. Milano: Hoepli.
Roca P. (Eds.). Structural Analysis of Historical Construc- Livesley R. K. 1978. Limit Analysis of structures formed
tions, Proc. intern. seminar, Guimaraes 7–9 November. from rigid blocks. International Journal for Numerical
Guimaraes: Balkema. Methods in Engineering 12(11): 1853–1871.
Drucker D. C. 1954. Coulomb friction, plasticity and limit Livesley R. K. 1992. A computational model for the limit
loads. Journal of Applied Mechanics 21(1): 71–74. analysis of threedimensional masonry structures. Mecca-
Ferris M. C. 2001. Limit analysis of frictional block assem- nica 27(3): 161–172.
blies as a mathematical program with complementarity Lourenço P. B. 1996. Computational strategies for masonry
constraints. International Journal of Mechanical Sciences structures. Thesis Delft University of Technology. Delft
43(1): 209–224. University Press.
Gilbert M. & Melbourne C. 1994. Rigid-block analysis of Lourenço P. B. 2001. Analysis of historical constructions:
masonry structures. The structural engineering 72(18): From thrust-lines to advanced simulations Historical Con-
356–361. structions. In Lourenço P.B., Roca P. (Eds.), Structural
Gilbert M. et al. 2006. Limit analysis of masonry block Analysis of Historical Constructions, Proc. intern. semi-
structures with non-associative frictional joints using lin- nar, Guimaraes 7–9 November. Guimaraes: Balkema.
ear programming. Computers and Structures 84 (13–14): Lourenco P. B. 2006. Recommendations for restoration of
873–887. ancient buildings and the survival of a masonry chimney.
Giuriani E. et al. 2001, Structural rehabilitation of masonry Construction and Building Materials 20(4): 239–251.
vaults. In UNESCO-ICOMOS millennium congress; Nart M. 2003. Limit analysis of masonry domes. Masonry
Proc. intern. symp., Paris, 10–12 Septembre 2001. international 16(1): 12–20.
Giuriani E. et al. 2002. Studio del comportamento strutturale O’Dowyer D. 1999. Funicular analysis of masonry vaults.
delle volte a padiglione in muratura. Technical Report of Computer and Structures 73(1–5): 187–197.
University of Brescia 17: 1–327. Oppenheim, I. J et al. 1989. Limit state analysis of masonry
Guarini G. 1968. Architettura civile (1737). Milano: Il domes. Journal of Structural Engineering 115(4):
polifilo. 868–882.
Harvey W. J. 1988. Application of the mechanism analysis to Palladio A. 1980. I quattro libri dell’architettura (1570).
masonry arches. The structural Engineer 66(5), 77–84. Milano: Il polifilo.
Heyman J. 1977. Equilibrium of shell structures, Oxford: Scamozzi V. 1964. L’idea dell’architettura universale (1615),
Clarendon Press. Ridgewood: Gregg.
Heyman J. 1996. Arches, Vaults and Buttresses, Norfolk: Smars P. 2000. Etudes sur la stabilité des arcs et voûtes,
Variorum. Ph.D. thesis, Department of Civil Engineering, Katholieke
Huerta S. 2001. Mechanics of masonry vaults: The equi- universiteit Leuven.
librium approach. In Lourenço P.B., Roca P. (Eds.), Tomasoni E. 2008. Le volte in muratura negli edifici storici:
Structural Analysis of Historical Constructions, Proc. tecniche costruttive e comportamento strutturale. Ph.D.
intern. seminar, Guimaraes 7–9 November. Guimaraes: Thesis, DICATA, Università degli Studi di Brescia.
Balkema.
61
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
D.T. Yeomans
ICOMOS, UK
ABSTRACT: The conservation of timber structures has improved in recent years with a growing appreciation
of their historical significance. However there are still difficulties presented because of the limited number of
engineers with an understanding of timber structures and the failure of design codes to consider the kind of
detailing often used in historic structures. This paper will point out a number of areas where research on the
behaviour of ‘traditional’ carpentry would be valuable.
1 INTRODUCTION day (Charles & Horn, 1973: Horn & Charles, 1966,
1984). At much the same time, Stuart Rigold (1966)
Since the eighteenth century timber in Britain has been did work on the many timber barns in Kent, and Cecil
a rather Cinderella material, essential to construction Hewett (1967), having looked at the barns of Cressing
but not highly regarded compared with masonry as Temple, went on to draw, examples of major medieval
an architectural material. It did enjoy something of a carpentry from all over the country, his work culmi-
revival during the nineteenth century for roof struc- nating in two large collections (Hewett, 1980 & 1985).
tures of the Gothic revival but, apart from that, it His work did a great deal to draw people’s attention to
has been a poor relation to other materials for struc- these structures but, because of the inaccuracy of his
tural purposes. That is to some extent because Britain, drawings, regrettably little to advance scholarship in
not having large supplies of structural species has the area. (One could even argue that it set back scholar-
imported most of its building timbers, but also a legacy ship since many must have assumed that this work had
of the Fire of London and the switch to brick. The sub- already been done.) Timber buildings were also a com-
sequent ‘Georgianising’ of timber-framed buildings mon interest of students of vernacular architecture, so
across the country was probably as much a fashion that by the end of the 1970s there was an extensive cov-
statement as a concern for fire protection. One could erage of medieval carpentry. In contrast little had been
argue that it is not having a continuing tradition of tim- done at that time on later carpentry work and for that
ber structures that has led to a lack of concern for the reason I chose to look at seventeenth and eighteenth
history and historic value of such structures, except century carpentry to provide an outline of develop-
for exposed half-timbering. Be that as it may, the his- ment during that period that I hope others might follow
torical significance of timber structures has only been up on.
recognized relatively recently, so that when we con- This still leaves the nineteenth century, both in terms
sider the approaches to conserving timber structures of the development of industrial structures, the tran-
we have to consider the development of ideas about the sition to iron and architects’ use of exposed timber
significance of these structures as well as that of the structures in both Gothic revival buildings and their
technical solutions; an understanding of their history later treatment of the material. Only Booth has done
as much as of their structural behaviour. It is true that work on nineteenth century laminated timber. There
it has for some time been a material of interest to a few are also significant 20th century timber structures that
enthusiasts, people like the late Freddie Charles, and its we are now loosing. Oxford Road Station, Manchester,
importance was recognized in those buildings where a group of conoids, was recently imitated in steel.
it was a prominent architectural material: the roofs of The histories of early carpentry provided an account
open halls and East Anglian churches for example. But of developments in the overall form of timber struc-
otherwise it was taken little account of. tures, their structural design if you will, but carpentry
In the nineteen sixties there was a series of studies is a craft skill and we also need to consider its details.
of the major monastic barns, some undertaken through Again Hewett provided the general reader with draw-
the collaboration of Freddie Charles and the American ings of joints but that is something requiring careful
scholar Walter Horn that recognized the significance observation and measurement. In this the work that I
of what were the greatest timber structures of their know best is that of Richard Harris, now curator of the
63
Weald and Downland Museum, and Peter McCurdy, Much the same might be said of the structures of the
who I would describe as a scholar craftsman. These are nineteenth century and even the early 20th century, and
people who have worked closely with the timbers and the danger is that we will loose historically significant
who have come to understand and bring to our atten- structures before their significance is recognized.
tion the significance of setting out marks that one finds
within the timbers, and have a sound understanding of
the process from conversion of the log to the erection
of the building. Although they have published rela- 2 ANALYSIS OF THE STRUCTURE
tively little, in their teaching they have shown us that
carpenters marks are more than just the numbering cut Recognizing the value of structures is the first step to
into the timbers. For imported timbers there has been their preservation, but the second is their proper struc-
a little interest in merchant marks that indicate their tural analysis. Here it is unfortunate to have to report
sources, significant because of the changing patterns that their behaviour is not always properly understood,
of trade, particularly during the late eighteenth and with the result that unnecessarily heavy-handed repairs
early nineteenth centuries, although this work is still are sometimes specified leading to the loss of historic
in its infancy. material and hence the loss of historic character. On
What historical significance these clues have I leave more than one occasion I have been contacted by those
to others, but the significance of this work is that concerned when their engineers have specified the use
aspects of the surviving timbers are now being rec- of steel connectors for timber structures that had been
ognized that were simply not seen a few decades ago. performing in a perfectly satisfactory manner. This can
Moreover, as many timbers have been reused, it has be either because of an unwillingness to treat the struc-
sometimes been possible to reconstruct the design of ture on its own terms, or simply the application of
earlier structures that the timbers were a part of. For computer software better suited to steel or concrete
example, the earlier roof of Lincoln Cathedral has been structures.
reconstructed from timbers that now form part of the In some cases it is clear from the places where the
present roof (Foot et al., 1986), but which were reused steel has been specified that the structural form has
from the previous. An example of more recent reuse been analysed as if it were a steel structure, result-
was that timbers from the original gallery structure at ing in assumed tension forces that the timber joints
Christ Church, Spitalfields were reused for the floor cannot take. However, if these members are removed
of later pews. from the analysis what remains might be a perfectly
This means that surviving timbers are now being satisfactory structure. In spite of what seems obvious
recognized as conveying evidence of the development foolishness, there have even been a number of papers
of structural design, as evidence of craft practices and published where finite element analysis has been inap-
as evidence of the earlier form of the buildings in propriately applied to historic timber structures. In one
which they are found. This is a change from the scant case that I was sent to referee (I will not give a refer-
value that they were given only a few decades ago. ence to protect the foolish.) the author came out with
I recall visiting the roof of Norwich Cathedral to look the startling discovery that in a king post structure, it
at the timbers in the roof shortly before they were was the post rather than the tie beam that would be
taken away. By that time the transepts had already been carrying the majority of the load. A moment’s thought
re-roofed using reinforced concrete A frames, involv- would have told him that the principal rafter/king post
ing the total loss of the original timbers. In today’s cli- assembly must be stiffer than the tie beam. I have a
mate an alternative and rather different solution would feeling of unreality when reading such work. I have
surely be sought in similar circumstances, perhaps to assume that such studies are carried out to impress
using a supplementary structure rather than a complete academic colleagues rather than as contributions to the
replacement. We are perhaps beginning to recognize practice of either history or conservation.
that structural design is a cultural activity whose his- The message here is that ‘work is no substitute for
tory is to be valued as much as carving or wall painting. thought’, but it remains a problem when the work is
However, medieval structures still seem to be the main done by a computer while it is the engineer who is
focus of attention with far less understanding of the required to do the thinking. It is likely to remain a
history of early modern structures. problem because the modelling of structures using
This is disappointing because without the develop- computer programs has become a relatively simple
ment of new carpentry structures in the seventeenth task. Entering the data into the computer is a fairly
and eighteenth centuries the forms taken by the great mechanical operation that can be performed without
public buildings, churches and country houses would much thought. It is a sad reflection that there seems
simply not have been possible. But the carpentry to be a large number of people for whom thought is
structures that made this architecture possible are a painful activity to be avoided at all costs. So much
largely invisible and so little attention is paid to them. easier to just put the data into the machine and let it
64
do the work. Of course, computer analysis can be used paramount importance. Between these two there may
intelligently if one realizes that large tension forces at be a choice of options that the engineer should be able
joints are unlikely to occur and the structure needs to to assist in deciding between. This requires the to engi-
be modified accordingly. Sometimes one even finds neer have a greater involvement in the process than
that the result is a statically determinate structure and simply providing the technical fix after the strategic
with a little thought one could have started from there. decisions have already been taken.
One also needs to recognize that the nature of the In considering the overall strategy the choices are
workmanship can affect the support conditions for between:
members that in turn will affect the overall structural
– repairing and restoring the as-found structure and
scheme. Also, the section properties that one can insert
– providing supplementary structure
as data for an historic structure can often be little more
than a guess. This is equally true of the behaviour of And if the first option is chosen one needs to
connectors, and any analysis needs to be based on a consider the choice between
thorough qualitative grasp of the structure following
– Repairing in-situ and
close observation of its construction and the way that
– Dismantling for repair, either completely or in part.
it has behaved. Often the uncertainties are such that
more than one structural scheme has to be considered. There is often resistance from conservation offi-
Of course, timber is not unique in this regard but per- cers (and possibly clients) to the latter, even thought
haps the linear nature of its elements encourages a false there are some cases where this might be the cheapest
idea of simplicity. and safest approach. It is an option for timber struc-
tures because of their prefabricated nature, and there
is almost certainly a failure to understand that where a
structure has distorted over time, for whatever reasons,
3 REPAIR STRATEGY to repair it in its as-found condition is to preserve a
cripple. It will often have locked in secondary stresses
This brings us to the methods of repair, and here we that have arisen through the distortion that has taken
should recognize that different approaches might be place and which now have to be accommodated within
appropriate for different kinds and periods of structure. the repairs, a complicating factor.
A simple contrast can be drawn, for example, between I have argued (Yeomans, 2007) that in some build-
the decorative roofs of the middle-ages and the purely ings, particularly barns, the structural design is what
utilitarian structures of nineteenth century industrial we should be seeking to preserve. In contrast, where
buildings. In spite of both being equally visible, the the timber has been treated decoratively this will
former are clearly more valuable architecturally. As for take precedence and we may have to compromise the
all historic structures, we should begin by considering integrity of the structure in order to preserve the sur-
the possible historical value and hence the strategy to face. Perhaps there are those occasions where the
be adopted rather than focusing immediately on the choice is that simple, but this dichotomy means that
tactics of repair. This issue of conservation strategy is some judgement must often be made. In such circum-
one in which the engineering historian has something stances the engineer responsible must be prepared to
to contribute and about which the engineering consul- engage with other conservation professionals and to
tant might be better informed than other members of explain his point of view. It may be necessary to offer
the conservation team. alternative structural strategies that would have dif-
This means making some strategic assessment that ferent effects on the historic fabric. One is balancing
should take into account: structural safety, the heritage value of the structure, and
perhaps a choice of future use. The client and other
– The historical value of the fabric.
conservation professionals will be involved in such
– The overall condition of the structure and hence
decisions, and the engineer must be prepared to explain
– The scale of the repairs required, but also
the engineering issues in terms they can understand.
– The options for future use.
Although I have emphasised the need for con-
Consider the extremes here. We might try to retain servation engineers to work with other conservation
the timbers in a condition as close as possible to the professionals in determining the overall strategy, it is
original structure, i.e. as a repaired structure carrying also advantageous to work with the carpenters. I regard
out its original function of supporting the building. the simple model of professionals producing a design
At the other extreme is provision of new structure for contractors to price as inappropriate for the conser-
to support the original timbers that simply remain vation of timber structures. We are not dealing with an
as an historic artefact reminding one of what was empty site on which a building is to be assembled by
there originally. The latter might well be the preferred reasonably tried and tested methods. We are dealing
option where the decorative nature of the timbers is of with unique situations where it is necessary to ensure
65
the continued stability of an existing structure that is
being changed while it is being repaired. This is a situ-
ation in which the methods used may affect the details
adopted, or in which the practicalities of a particular
detail might affect the scope of the work to be car-
ried out. In many cases an experienced carpenter may
make an invaluable contribution at the design stage
and, where possible, it is worth employing one as part
of the design team.
I raise all these issues before discussing repair meth-
ods in detail because both the repair recommendations
of ICOMOS UK’s Wood Committee and the Recom-
mendations of ISCARSAH, while informed by such
considerations, do not draw the conclusions as clearly
as they might. My view is that more consideration
needs to be given to this stage of the process in the
Figure 1. Repairs using steel carried out to the heel joint of
advice given to professionals. a Wren roof truss at Hampton Court.
4 REPAIR DETAILS
66
In the 1970s the use of epoxy-resins as either a con- sensible to use it – but this is an argument normally
solidant, or to replace lost timber appeared to offer a applied to non-structural elements.
solution for those who wanted the minimum loss of
historic fabric, but this ran into technical difficulties.
5 FASTENERS AND DESIGN CODES
Early research in the US on this method, where they
were interested in it for repairing softwood trusses in
Once the forces in a structure have been determined,
the roofs of aircraft hangars, showed the difficulty of
and unless there is particular concern over deflections,
controlling the run of what was a liquid and the obvi-
our engineering problems become a series of con-
ous change in the structural characteristics of joints.
nections with members in between; member stresses
However, it is questionable whether or not this work
are generally low. Unfortunately the jointing methods
was noticed by those in Britain who were enthusiastic
used in new structures are not always appropriate for
for the use of the material and there are some unfor-
historic structures. Moreover, historic structures use
tunate results of its early and inexpert use. Baguely
details that are not used in modern carpentry and there-
Hall, on the outskirts of Manchester, is an example of
fore for which there is no guidance within design codes
the kind of thing that can go wrong. Epoxy-resin was
framed for new buildings, using modern materials and
used to repair the main fames of this fine medieval
modern methods. This was never so clear as during
hall but was allowed to run over the surface of timbers
the lacuna when the newly introduced British Standard
and is now a permanent disfigurement. Many early
(BS5268) in 1984 failed to include oak as a structural
repairs using this material were carried out by a firm
material and conservation engineers had to continue
who appeared to have little technical knowledge and
to use the obsolete code (CP112). While that issue has
even less interest in the proper training of their opera-
been resolved, there are still questions over the per-
tives. The result was a number of failures, sometimes
formance of traditional joints and the values that one
because of the use of the technique in unsuitable loca-
might use for metal fasteners.
tions where exposure to the weather seems to have
The British code of practice is even inadequate in
resulted in accelerated decay of timber adjacent to the
the information given for steel fasteners. One example
epoxy-resin repair.
is that safe loads for screws in shear are only given up
The difficulty then was that with considerable anec-
to 10 mm diameter when carpenters may well wish to
dotal evidence about failures following the use of
use 12 mm diameter screws. (Even that is an improve-
epoxy-resin, and with little scientific evidence, there
ment on the earlier editions of the code where loads
was deep concern among some over the long-term per-
were only given up to 8 mm diameter.) An engineer
formance of such repairs. This was especially so where
who is unfamiliar with timber may be unwilling to go
they might be used in timbers with high or fluctuat-
beyond the limits of the table, particularly as the for-
ing moisture content. If there were to be deterioration
mulae for determining the loads from first principles
of the adjacent timber it would have an effect on the
are rather daunting. Bolts can be seen used in circum-
strength of the joint. One might also be suspicious of
stances where screws would have been better but where
its behaviour when used as an adhesive to join steel
M10s were presumably inadequate. Moreover it might
reinforcing plates to timber and the effects of temper-
not be clear to those unfamiliar with the code that the
ature changes. As the steel must move more than the
spacings and edge and end distances for bolts rather
timber what is happening at the interface between the
than screws apply to these larger sized screws.
materials? Here we have no information and can only
There is the serious question about how we should
wait and see.
view the allowable loads on bolts and screws used in
Opinions on the use of the material seem to still
green timbers that will dry in service. The problem for
be divided between those with long memories and an
conservators is complicated by the fact that one side
instinctive distrust of these techniques and those for
of the joint will be dry timber and the other side green.
whom it is a valuable tool. TRADA (2001) have pub-
I have no idea what reasoning or experimental results
lished a book on its use and at the Whitbread Brewery,
lie behind rather draconian reduction factor required
London, Hockley and Dawson used it to effect in the
by the code. If it is to allow for the possibility of split-
repair what may well be the largest surviving eigh-
ting as the timber dries one would have thought that
teenth century trusses. More recently the Weald and
such an event might well reduce the capacity of the
Downland Open Air Museum has been developing
joint to zero.
skills in the use of epoxy resins for repairing struc-
tures that have been dismantled. (Their techniques are
not suitable for in-situ work.) Richard Harris points 6 TRADITIONAL JOINTS
out that this material allows one to make repairs that
retain a lot more of the historic fabric. He has also The terminology is a little loose here because there
argued that if something is going to be lost through are traditional joints, either surviving in existing struc-
decay and epoxy-resin will prolong its life, then it is tures or used in repair work that are no different from
67
Figure 3. In loading to determine shear capacity of timber
the forces are applied at each end.
joints contemporary with the original construction. boundary without the kind of rotation that occurs in
At the same time there are those joints that have a bolts.
superficial resemblance to traditional joints but which Unfortunately what work has been done so far has
rely on modern fasteners, the most common being the produced conflicting results. Drawing on the work of
scarf joint. The difficulty is that any tests to deter- Jonathan Shanks carried out at Bath (Shanks & Walker,
mine working loads for such joints must be affected 2005), Ross et al (2007) have given possible work-
by the workmanship in the carpentry. One can envisage ing loads for pegs in oak, whereas a formula derived
the problem of relating test results to the performance from quite different tests by Schmidt (2004) at the
of historic joints, both with very uncertain standards University of Wyoming suggests a working loads of
of workmanship and the effects of time and chance – less than half their figures. Given such a wide dis-
or perhaps that should be wind and weather. crepancy, prudence suggests one should use the lower
One wonders whether the standard test for shear figure for design, but science surely requires some
resistance is a suitable basis for the design of tradi- explanation for this large difference. The Bath tests
tional carpentry joints relying upon shear along the were carried out on complete joints while those in
grain? The test loads the timber at either end so that Wyoming in more ‘traditional’ laboratory experiment
one might expect a sensibly uniform shear stress along that allowed the effects of varying density in both pegs
the plane of failure – the dotted line in Figure 3. and member timbers to be explored. What surprises me
The joint where timber is loaded in shear parallel is that the Bath tests that used carpenter-made joints,
to the grain is a typical heel joint where the normal and presumably subject to variations of workmanship,
condition in practice is for restraint to be provided by nevertheless resulted in the higher figure.
the tie beam behind a notch (Fig. 4) or a mortice and The behaviour of dovetail joints is another problem-
tenon joint. In such configurations one would expect atic area as these rely on timber loaded across the grain
the shear stress along the plane of failure to fall off (Fig. 6). When there is drying shrinkage in the timbers
towards the unloaded end; but in what way? A com- there is a change in the angles of the two parts of the
plicating factor in some buildings is that this timber joint resulting in a reduced contact area and a corre-
may be rather exposed and have a higher than normal spondingly higher compressive stress.The stress might
moisture content. be well above the elastic limit with collapse of some
Pegged joints is one of those areas where there is an of the cell walls. However there is no readily available
overlap between new-framed construction and conser- data from which one can estimate the movement that
vation, the question being whether such joints can be might result.
relied upon to transmit tensile forces. The difficulty is A serious problem occurs when existing structures
that pegs do not behave in the same way as steel bolts or requiring repair incorporate details that are ‘prohib-
dowels so that Johansen’s equations cannot be applied ited’ by the present code. An example occurred where
to find the allowable loads. Experimental work shows rafters had birdsmouth cuts over the supporting plate
that the pegs fail in shear close to the tenon/mortice with the depth of the cuts exceeding half the depth of
68
Figure 7. Birdsmouthed rafters. Figure 9. A more traditional scarf joint with screws in
withdrawal at the tension end of the joint.
69
characteristics of this material in their analysis of its Hewett, C. 1980, English Historic Carpentry, Philimore,
behaviour.At the detailed design stage, there have been London.
some developments as a result of the present fashion Hewett, C. 1985, English Cathedral and Monastic carpentry,
for green-oak structures. However, work in this area is Philimore, London.
Horn, W. & F.W.B. Charles 1966, ‘The cruck-built barn
still patchy and rather inconclusive. If this seems a little of Middle Littleton, Worcestershire, J. Soc. Architectural
pessimistic, what I have not discussed is the quality of Historians.
the pool of carpentry skills that has been growing over Horn, W. & F.W.B. Charles 1984, ‘The cruck-built barn
the last few decades that facilitate good standards of of Frocester in Gloucestershire’, J. Soc. Architectural
conservation Historians.
Rigold, S.E. 1966, ‘Some major Kentish barns’,Archeaologia
Cantiana, 81, pp. 1–30.
REFERENCES Ross, P., C. Mettam & A. Holloway, 2007, Green Oak in
Construction, TRADA, High Wycombe, p. 163.
Baines, F. 1914, Westminster Hall, report to the first commis- Schmidt, R.J. 2004, ‘Considerations for mortise and tenon
sioner of H.M. Works, etc., on the condition of the roof joint design’, Wood Design Focus, 14, No. 3, pp. 14–17.
timbers of Wstminster Hall, with suggestions for main- Shanks, J.D. & P. Walker 2005, ‘Experimental performance
taining the stability of the roof, London, HMSO. (Cd. of mortice and tenon connections in green oak’, The
7436). Structural Rngineer, (6 September), pp. 40–45.
Charls, F.W.B. & W. Horn 1973, ‘The cruck-built barn TRADA 2001, Resin-bonded Repair Systems for Structural
of Leigh Court, Worcestershire’, J. Soc. Architectural Timber, Timber Research and Development Association,
Historians. High Wycombe.
Foot, N.J.D., C.D. Litton & W.G. Simpson 1986, ‘The high Yeomans, D.T. 1992, The Trussed Roof: its history and
roofs of the east end of Lincoln Cathedral’ in Medieval development, Scolar: Aldershot.
Art and Architecture at Lincoln Cathedral (B.A.A. Conf. Yeomans, D.T. 2007 ‘Appropriate technologies for conserva-
Trans., 8), pp. 47–74. tion’, Journal of Architectural Conservation, 13 No. 3,
Hewett, C. 1967, ‘The barns at Cressing Temple, Essex’, pp. 9–20.
J. Soc. Architectural Historians, 26.
70
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
A. Miltiadou-Fezans
Directorate for Technical Research on Restoration, Hellenic Ministry of Culture
ABSTRACT: The Katholikon of the Byzantine monastery of Dafni (world heritage list of UNESCO), is one
of the most important monuments of middle Byzantine period, famous worldwide for its mural mosaics. Being
constructed in a seismic area, the Katholikon was affected throughout its history by a large number of earthquakes
that caused structural problems and damages to its mosaics. This paper focuses first on the historical and recent
pathology studies of the monument and the emergency measures undertaken by the Hellenic Ministry of Culture
after the earthquake of 1999 in Athens. Then, a synthesis of the results of research programs, investigations
and structural restoration studies is presented. Furthermore, a brief presentation is made of the first phase of
interventions executed for the consolidation of the masonry and the control of their efficiency using novel
methodologies, non destructive techniques and monitoring systems.
1 INTRODUCTION
71
of the HMC. On the basis of a synthesis of the
obtained results, adequately completed with detailed
in situ observations and archive researches, the mul-
tidisciplinary group of the HMC has elaborated the
structural restoration study (Delinikolas et al., 2003,
Miltiadou-Fezans et al. 2003b). Following the propos-
als of this study, approved by the Scientific Committee
and the Central Archaeological Council, it was decided
to implement the structural restoration design and
interventions in two phases.
Given the importance of the monument, a step
by step multidisciplinary approach was adopted, both
concerning the design and the implementation of the
structural restoration interventions. Such an approach
gives the possibility to start the execution of a series Figure 2. N-E view of Katholikon.
of interventions, and in the same time to perform the
in situ and laboratory investigations that are necessary
for the design of the next step. Furthermore, the assess-
ment of the effect of the first step interventions to the
structural behaviour of the monument can be carried
out and taken into account for the design of the works
of the second step.The first phase of interventions (that
is completed) comprises measures taken to repair and
strengthen masonry elements. The interventions of the
second phase aim to improve the overall behaviour of
the entire structure; their design is still in progress.
Moreover, within the HMC, it was judged as abso-
lutely necessary to support the two phases of works
with further research and investigations, comprising
the design of adequate mortars and grouts, the exper-
imental estimation of mechanical characteristics of
masonry before and after grouting, the seismic moni-
toring of the structure, the application of NDT’s for Figure 3. The central dome and its drum, curried by eight
investigating invisible parts of the monument and pendentives and eight arches.
controlling the grouting effect and efficiency.
In this paper, a brief synthesis of all the aforemen-
cross-shaped in plan, the hemispherical dome rising
tioned investigations, studies, emergency and struc-
over its square core. The dome is 8,2 m in diameter
tural restoration works will be presented.
and 16,4 m high, and rests on an almost cylindrical
drum with 16 piers and 16 vaulted windows. The
2 BRIEF DESCRIPTION OF THE dome and its drum are carried by eight pendentives
KATHOLIKON OF DAFNI MONASTERY and eight arches (four semicircular and four embodied
in the squinches of the corners), forming an octagon
The Byzantine monastery of Dafni comprises various and achieving in this way the transition from circle
buildings (laid out in a square plan), constructed over to square. Thus, twelve piers (laying out in a square
a long period of time, starting on the 11th century AD plan), provide support to the dome together with the
(Delinikolas et al. 2003). Currently, most of the build- groin vaulted arms of the cross, situated in a higher
ings are in ruins, with the exception of the Katholikon, level (Figs 3 & 4). All the other parts of the monument
part of the internal range of cells, the cistern and the are covered with byzantine groin vaults.
northern fortification walls. As one can clearly distinguish in Figures 1 and 3, the
The Katholikon belongs to the octagonal type and exterior face of the vertical perimeter walls are built
preserves large part of the original mural mosaics. It according to the enclosed brick system (stones with
comprises the main church, the sanctuary, the narthex bricks around them), following two different masonry
and four chapels, which complete its orthogonal plan. construction types. In fact, there is a lower zone, built
In the western part, only the perimeter walls of an with large dimension stones (often, reused material),
exonarthex or portico and those of a spiral stair- their length placed horizontally or vertically, in order to
way tower leading to the upper floor have survived form crosses, and un upper zone constructed with ash-
(Figs 1, 2, 5). The central part of the main church is lar masonry, using smaller and perfectly cut orthogonal
72
stones (Delinikolas et al. 2003). Bricks are placed in all
horizontal joints, and in most of the vertical ones. The
thickness of mortar joints is of 3 cm approximately.
Plain brick masonry was used in the construction of
all the windows and doors (Fig. 2). In the internal face
of the perimeter walls, as well as in internal masonry
elements in general, the enclosed brick system is not
followed systematically, whereas in various locations,
a mixture of cut, semi-cut and rubble stones are used
together with bricks. Thus, the thickness of masonry
joints is varying. Finally the entire vaulted roof was
constructed with plain brick masonry.
Both the lower and upper parts of the perimeter
walls are constructed following the three-leaf masonry
type, with varying widths of the leaves, as will be more Figure 4. Plan of the Katholikon and schematic presentation
analytically presented below (Section 9). of the arrangement of the raking shores and vertical props.
73
Figure 8. West and south façade: 1st construction phase,
drawings of Benouville as modified by Delinikolas (Millet
1899, Delinikolas et al. 2003).
74
Figure 11. West façade of the exonarthex before and after
Stikas restoration interventions (photos Stikas, 1962).
75
(a) East-West section. View to South
76
arches just below the squinches presented severe dis-
location near their crown, followed by out of plane
deformations of the squinches themselves. Cracks
appeared also in all the groin vaults of the church.
The structural condition of the dome (reconstructed
at 1891 and damaged soon after its reconstruction at
1894), was assessed as extremely critical immediately
after the earthquake (Delinikolas et al. 1999, Miltiadou
et al. 2003a, 2003b). Horizontal cracks have appeared
along the perimeter of the drum (both at its base and
top). In the piers of the drum that are situated perpen-
dicular to the East-West direction, horizontal cracks Figure 16. N-S transversal section in the narthex. The south
(due to out-of-plane bending) have opened at their top wall, although reconstructed in 1896 (see photo), presents
and bottom. In the piers that are situated parallel to the today a 90 mm out-of-plane deformation.
E-W axis, diagonal or bi-diagonal (shear) cracks have
appeared. In the intermediate piers, mixed type of (less
should have played an important role towards this
severe) cracks was observed. This pathological image
pathology.
does seem to confirm seismological data regarding the
Moreover, the increase of crack openings with
predominant direction of the 1999 earthquake. Thanks
height was found more pronounced along the transver-
to the external upper metal tie-rod, confining the hemi-
sal axis (N–S), than along the longitudinal one (E–W).
spherical dome near the springing level, the occurrence
Such a difference in behaviour is usual in churches
of severe cracking of the dome’s shell was prevented.
with an orthogonal plan and could be attributed to
the larger number and sections of vertical elements
6 QUALITATIVE INTERPRETATION OF available along the E–W axis in the main church (see
DAMAGES Figure 12). This behaviour was also noticed in the past,
whereas previous interventions were applied with the
As reported in Miltiadou-Fezans et al. 2004, it was aim to alleviate this problem (external stone buttresses
observed that the number and the opening of crack in in the north, metallic trusses and confinement of piers
the vertical elements of the Katholikon increase from in the south), as described in Section 4 (Figs 13 &14).
the base to the top of the monument. It was there- Although those corrective measures were in the right
fore obvious that the monument exhibits the tendency direction (allowing the church to withstand the 20th
to “open” from the base to the top along both main century earthquakes without local collapse), they were
directions. This deformed shape of the church, verified proven to be insufficient to prevent extensive cracking
by the photogrammetric survey (Georgopoulos et al. of the monument.
2003), is confirmed also by the history of the monu- The damages observed in the drum of the cupola
ment. The South façade of the narthex reached in 1894 may, therefore, be attributed to the (increasing with
a total out-of-plane deformation larger than 200 mm height) tendency for out-of-plane deformations of the
and it was reconstructed (Fig. 13,14). The photo in church. It should be reminded here that the damages
Figure 16 shows the region between the original and that made imperative the demolition and reconstruc-
the reconstructed part of that façade, where one may tion of the cupola at the end of the 19th century were of
clearly see the permanent out-of-plane deformation of the same nature, as those observed now; this is proved
the original part. by the missing parts of the mosaics. As described
However, even this reconstructed part of the mon- in Section 3, most of the piers in the drum exhib-
ument presents today a total out-of plane deformation ited out-of-plane deformations. Since the substructure
of 90 mm (drawing in Fig. 16). This is another element on which the system of the cupola rests is deforming
proving that the feature we observe now in the mon- out-of-plane and the cupola itself (being very stiff) is
ument is an inherent characteristic (due to its initial practically non deforming, the piers of the drum (being
construction scheme and the extended alterations/ rather flexible out-of-plane) are called to follow the
interventions undertaken during its lifetime). deformations of the substructure.
Thus, out of plane deformations were reported both Some final comments should be made here, regard-
for the perimeter walls and for all the main arches ing the foundation and foundation soil of the mon-
bearing the dome in the central area of the monument, ument and their effect on its behaviour. The recent
followed by a geometrical deformation (and loss of pathology does not give signs of major differential set-
initial shape) of the arches themselves. The lack of tlements that might have contributed to the damages of
wooden or metallic ties (typical structural elements the monument. This is also confirmed by geotechnical
for the Byzantine architecture) or other horizontal ele- investigations conducted immediately after the 1999
ments connecting the vertical walls, pillars and piers earthquake (O.T.M., 2000). On the other hand during
77
the application of urgent measures local investigations
revealed that rather shallow foundations (0.5 m deep
in the middle of South perimeter wall, 1.25 m deep in
the N–E chapel) are provided to the perimeter walls).
Thus, in order to collect qualitative information on
foundation elements and on eventual empty under-
ground spaces or local variations of the ground
material characteristics, geophysical investigation was
also undertaken, using mainly 3D seismic tomog-
raphy (Polymenacos et al. 2005). This investigation
suggested that, in the north-northwest part of the
church and exonarthex, inferior ground material qual-
ity has to be expected, as low velocities were recorded.
Moreover, archaeological excavation in the region of Figure 17. The model prepared for parameter analyses.
the N–W chapel revealed that Troump had probably
noticed that the foundation were relatively shallow and
has dictated their strengthening during the extended
reconstruction of the upper part of the north wall in
1896. For this purpose, a 2.0 m deep wall was con-
structed under the NW perimeter wall of the church. To
this end extensive excavations had to be undertaken in
that area, thus disturbing the foundation soil. This fact
could probably explain the lower velocities found with
the seismic tomography. Nevertheless, all the above
information and eventual further investigations con-
cerning the foundation conditions of the church and
exonatrhex have to be further examined during the
design of the interventions of the second phase.
7 NUMERICAL VERIFICATION OF THE Figure 18. Inner face of shell elements. Tensile stresses due
PATHOLOGICAL IMAGE to self-weight of the structure.
One of the major steps of the assessment of an existing of several arches, in the groin vaults of the narthex, in
structure is the analytical reproduction of its patholog- the base of the cupola and that of its drum, as well as in
ical image. In the case of the Katholikon, preliminary the four squinches. The obtained results are conform
linear parameter analyses were performed (Miltiadou- with the qualitative interpretation of the pathology of
Fezans et al. 2004), as a means for selecting adequate the monument and they prove the inherent vulnerabil-
emergency interventions. Those analyses provided a ity of the structural system, in which (a) a stiff central
quite satisfactory numerical verification of the patho- cupola is resting (through the drum) on four major
logical image of the monument. For the preliminary arches parallel to the two main axes of the church
analytical study, using the computer code ACORD, (see Figures 3, 4), as well as on four arches oblique in
the structure was modeled by shell elements (Fig. 17), respect with the longitudinal and the transverse axis,
whereas the mechanical properties of elements belong- (b) the vertical, as well as the horizontal component
ing to various parts of the structures were assumed of the self weight of the whole system of (intersect-
on the basis of the available data for the construc- ing) groin vaults arranged around the central dome,
tion materials. Linear elastic analyses were performed are transferred to rather flexible stone masonry piers,
for various combinations of actions (self weight alone and (c) there are not currently ties in the arches and
or combined with seismic action). Both static and vaults, or other elements to confine critical structural
dynamic analyses were performed. elements or link the various parts among them. There-
Figure 18 shows the calculated stresses for the inner fore, the tendency of the structural system to deform
face of shell elements, due to vertical loads. Irrespec- laterally in its upper region is expected to be apparent
tively of the accuracy of numerical values of stresses, even under self-weight alone.
one may clearly distinguish the vulnerability of the This working hypothesis seems to be confirmed by
region of arches and domes (especially in the west the deformed shape of the structure in-plan, shown
part), even for the self-weight of the monument alone. in Figure 19. One can even observe the more pro-
As expected, tensile stresses are developed in the apex nounced lateral deformation along the N–S axis of the
78
Although this preliminary analytical work allowed
for a better understanding of the structural behaviour
of this important monument, it was judged that a
more accurate assessment of its seismic behaviour was
of primordial importance for the design of optimum
interventions.
79
9 DESIGN OF GROUTS AND INVESTIGATION Table 1. Mechanical and injectability characteristics of the
OF MASONRY BEHAVIOUR BEFORE AND grouts selected to be injected in the wallettes.
AFTER GROUTING
Grout NHL5-based
9.1 Design of alternative grout compositions and properties Ternary (1) grout (2)
tests on cylinders simulating the infill material
T36 (sec) 19 22.5
The design of high injectability grouts was carried out Sand column
on the basis of the methodology proposed by Miltiadou 1.25/2.50 mm
& Tassios 2006. First, were taken into account the (Wnom ∼.0.2 mm)
Bleeding 2% 3%
performance requirements deriving from the struc-
App.viscosity 20.5 22
tural restoration study (Miltiadou et al. 2003b), and td=4.7 (sec)
then the following target values were set for the basic
mechanical properties of the grouted masonry: ten- Age (days)
sile strength approximately double than that one of the
masonry before grouting, and compressive strength Compressive and 28 90 180
approximately equal to 3.0 MPa. flexural strength
On the basis of the available literature (Vintzileou (MPa) fgc fgt fgc fgt fgc fgt
et al., 1995 and Tassios, 2004), it was estimated that
the compressive strength of the grout at the age of Ternary 4.08 2.11 8.16 2.29 10.6 3.13
six months should lie between 6 MPa and 10 MPa; a NHL5-based 2.82 2.47 4.50 2.52 6.36 3.87
grout
grout flexural strength of the order of 2 to 3 MPa was
required. In addition, the physical-chemical properties (1) Ternary: 30% white Danish cement, 25% Lime (pow-
of the raw materials should be selected in a way that der), 45% natural pozzolan Petr. (<45 µ m), superplas-
the durability of the structure and its precious mosaics ticer 1%, water 80% of the solid phase of the grout
would not be jeopardized. Finally, the grouts should (2) NHL5-based grout: 100% NHL5 (St Astier), superplas-
have high injectability capacity, so that, under low ticer 1%, water 80% of the solid phase of the grout
pressure (∼ 0.075 MPa), they enter and fill fine voids
and cracks, with a nominal minimum width (Wnom )
equal to two tenths of millimeter. Two main categories strength and durability requirements (Table 1). There-
of grouts could satisfy injectability, strength and dura- fore, they were selected to be applied to six three-leaf
bility requirements: (i) ternary grouts composed of stone wallettes, simulating the masonry of the upper
lime, pozzolan and a low cement content (30%) and parts of the monument. It has to be noted however
(ii) natural hydraulic lime – based grouts. that the hydraulic lime based grout presented a better
Thus, various grout mixtures, belonging to the sulphate resistance than the ternary one (Miltiadou –
above two categories, were designed and tested (in Fezans et al. 2007).
order to assess their physical, chemical and mechan-
ical properties) at the laboratory of the Directorate
9.2 Construction and testing of wallettes
for Technical Research on Restoration of the Hellenic
Ministry of Culture (DTRR/HMC). The main results In order to estimate in the best possible way the
of the research are reported in Kalagri et al. (2007) and mechanical characteristics of the masonry before and
Miltiadou et al. (2007). after grouting, considerable effort was devoted to the
In order to determine injectability characteristics, identification of the construction type of masonry.
the penetrability, fluidity and stability of the sus- For this purpose, radar and endoscopy were applied
pensions were fully examined in various water/solids in a systematic way. The in-depth geometry of the
ratios, with or without superplastisizer. The com- perimeter stone masonry was rather accurately identi-
positions presenting satisfactory injectability capac- fied (Vintzileou et al., 2004). As anticipated, both the
ity were further tested to evaluate their behavior to lower and upper parts of the perimeter walls belong
salt decay and estimate their mechanical characteris- to the three-leaf masonry type, with varying widths of
tics (compressive and flexural strength). Furthermore, the leaves. Although both regions were examined, the
six alternative grout formulations presenting similar investigation was concentrated mainly on the upper
injectability were injected into twenty eight cylindrical (more vulnerable) zone of masonry. In Figure 21 char-
specimens, simulating the infill material of three-leaf acteristic examples of vertical sections of the lower
stone masonry. The cylinders were then subjected to and upper zone of perimeter masonry are shown.
compression in different hardening ages (Kalagri et al, The materials characteristics (stones, bricks and
2007). After comparative evaluation of the results, two mortars) being determined in the Laboratory on sam-
grout compositions (a ternary grout and a NHL5-based ples taken from the monument and the type of masonry
grout) fulfilled simultaneously the injectability, the being determined in-situ, an attempt was made to
80
Table 2. Basic mechanical properties of Wallettes before
and after grouting.
(MPa) (GPa)
fws / εv0 εvs Es /
W fw0 fws fw0 (‰) (‰) E0 Es E0
∗
1 1.82 3.00 1.65 −1.76 1.00 1.20 1.20
2 1.74 3.75 2.16 −1.6 −2.50 1.44 1.55 1.08
3 2.26 3.73 1.65 −2.25 −3.39 1.50 1.30 0.87
ft,0 ft,s ft,s /ft,0
4 0.1 0.21 2.10
5 0.1 0.33 3.30
6 0.1 0.22 2.20
Figure 21. Vertical sections of lower (a) and upper zone (b). W2 and W5 injected with the ternary grout; W1, W3, W4,
W6 injected with the hydraulic lime based grout.
81
Table 3. Optimum grout composition and injectability char-
acteristics measured in the laboratory and in situ at the first
pilot preparation.
Grout composition
NHL5 (St Astier) 90%
Pozzolan Petrotechniki 10%
Superplasticizer (1), (2) 1%
Water (1) 80%
Grout properties In lab In situ
T36 (sec) – Sand column 19–22
1.25/2.50 mm
(Wnom ∼ .0.2 mm)
Bleeding <1% 1%
Apparent viscosity − td=4.7 (sec) In lab In situ
0 min after mixing 21 22
60 min after mixing (agitated) 23 25
Apparent density (gr/cm3 ) In lab In situ Figure 24. The groin vaulted structure after grouting.
0 min after mixing 1.5050 1.4978
60 min after mixing (agitated) 1.4986 1.4870
It has to be noted that the groin vault was built with-
(1) % of the solid phase of the grout. out any formwork, following the traditional way of
(2) superplasticizer based on polycarboxylic ether. byzantine masons, as reported by Delinikolas et al.
2003. The total dimensions of the model are in plan
2.70 m × 2.60 m; the thickness and the height of the
walls are 0.45 m and 2.60 m respectively, whereas the
thickness of the vault in its centre is 0.20 m. The total
height of the model is 2.85 m. These dimensions were
selected taking into account the limitations imposed
by the capacity of the seismic simulator.
The model was tested after nine months from its
construction by imposing seismic loads gradually
increasing, until rupture. Then local application of
grouts and installation of ties to the arches was per-
formed and after a suitable period of time the tests
were repeated, until rupture. The model was then
injected with the final grout composition (Table 3) to
Figure 23. The groin vaulted structure on the seismic homogenize the whole structure, following the same
simulator. methodology with that used for the injection of the
wallettes and the injection of the monument itself
(Miltiadou-Fezans et al. 2006, 2008). Tests were again
9.4 Construction and testing of a large scale repeated until rupture. In Figure 24 the model after
structure covered with a byzantine groin vault the grouting application is shown. As expected, it was
As aforementioned in section 2, byzantine groin-vaults observed that the dynamic characteristics of the model
are used for covering the most of the parts of the were changed after the application of grouting to the
Katholikon of Dafni Monastery, including the arms of whole structure, and the model could suffer stronger
the cross, where severe damages have been occurred. base motions. The results of this experimentation are
Thus, important effort has been undertaken for achiev- still under elaboration and are going to be presented in
ing a better knowledge of the dynamic behaviour of a separate paper. An attempt will be made to correlate
such structures. An experimental research was carried these results with those of the wallettes and the in situ
out by DTRR/HMC in collaboration with the Labo- seismic monitoring.
ratory of Earthquake Engineering of NTUA. A model
of a byzantine groin-vaulted structure bearing locally
mural mosaics was constructed. The materials and 10 BRIEF PRESENTATION OF THE WORKS
the construction type used for its masonry walls were OF THE FIRST PHASE
exactly the same with those used for the construction
of the wallettes, whereas bricks and mortar were used As reported in the relevant study, (Mitiadou-Fezans
for the construction of the arches and the groin vault et al. 2003b), the main objective of this first phase of
(Fig. 23). structural restoration interventions was (i) to improve
82
Figure 25. Old mortars of first and second construction
phase and cracked mosaics that have to be preserved in situ.
Figure 27. Typical examples of in situ mural mosaics con-
servation works (including installation of fine tubes for
grouting).
83
the protection and in situ conservation of old pointing 0.45
0.40
mortars, on the external facades, the upper central area 0.35 032 039
Period, T (sec)
just below the dome, and the groin vaults of the sanc- 0.30
0.25
027 030 026 026
020
tuary, the deep re-pointing works have been executed 0.20
022
by experienced Conservators. 0.15
0.10
As anticipated special attention has been given to 0.05
84
Figure 30. Tomographical reconstruction in a grouted and
in a non grouted area.
In 2005, an exhaustive high frequency ground- The high values of the Katholikon of Dafni Monastery
penetrating radar survey has been realized on the fifty and, hence, the need for accurate information to serve
main mosaics of the Katholikon of Dafni Monastery, the design of optimum structural interventions led
with the aim to investigate the possibility of GPR to the adoption of a step by step multidisciplinary
application to locate doubtful zones (delaminations, approach, both concerning the design and the imple-
changes of substrata mortar or other buried het- mentation of the structural restoration interventions.
erogeneities, etc), related to the mosaics very near This approach proved to be very efficient, as it gives the
bearing-structure. As presented in Côte et al. 2004 & possibility to perform the in situ and laboratory inves-
2008, a methodology has been developed for the tigations that are necessary for the design of the next
execution of specific parallel radar profiles, the pro- step, as well as for the evaluation of the previous ones.
cessing of data and the construction of gray-color scale In this framework, a series of quite novel investiga-
maps from chosen trenches of the structure, related tions were undertaken; the most important results are
to the amplitude of the echoes in the selected trench. summarized below.
These gray-scale maps were correlated to a level of
detachment or other heterogeneities by comparison 1. The holistic design of hydraulic grouts, carried out
with manual-sonic maps and pathology ones realized on the basis of rational criteria, laboratory and in
by the Conservators of the mosaics (see Fig. 30). It has situ pilot tests, led to the improved knowledge on
been shown that the GPR maps, giving a more detailed the hydraulic lime based grouts and, hence, to the
and less subjective qualitative evaluation of mosaics selection of appropriate mixes.
substrata, can be a very useful tool for the Conserva- 2. The experimental assessment of the mechani-
tors in order to locate doubtful zones in a more refined cal characteristics of three-leaf stone masonry
way. Furthermore, the possibility of using GPR proce- (before and after grouting) contributed to better
dures to monitor the movement of the grout behind the understanding of the behaviour of this type of
mosaics, in real time, during injection was also inves- structures and proved that their repair with ade-
tigated; detailed information is given in Côte et al. quately designed highly injectable hydraulic grouts
2008. (hydraulic lime based ones or ternary grouts) can
Investigations have been also realized to control the be very efficient.
grouting effect into the masonry walls, using seismic 3. A hydraulic lime based grout was considered as
techniques, including some sonic 2D travel time tomo- optimum composition to be applied in situ, on
graphies. The objective herein was to propose a simple the basis of injectability, durability, strength and
survey methodology which gives information for the deformability characteristics.
mechanical characterization of the materials inside the 4. Testing a scaled model of a byzantine groin-vaulted
structure, before and after grouting. The processing structure on an earthquake simulator (before
of the results after the completion of grouting is still and after grouting) provided valuable information
85
regarding the structural behaviour of this com- final design of the optimum interventions will be car-
plex type of structures, as well as regarding the ried out and the respective works will be implemented
strengthening capacity of injections. during the second and final phase of restoration works.
5. The developed methodology for grouting appli- Last but not least, it is strongly believed that the
cation gives the possibility for a more rational experience and knowledge acquired by all levels of
and fully controlled implementation of injections, involved stuff in the framework of the structural
not only to ordinary structures, but also to impor- restoration of this important monument of World Her-
tant monuments bearing mosaics, frescoes and old itage constitute a solid base for similar investigations,
mortars, that have to be preserved in situ. studies and structural restoration interventions to other
6. The entire procedure for the design and applica- historical structures and monuments.
tion of adequate lime-pozolan mortars enhanced
our knowledge on this kind of mortars, traditionally
used in the past. ACKNOWLEDGEMENTS
7. The monitoring system, installed to record the seis-
mic behaviour of the monument, provides valuable The contribution of Ass. Prof. E. Vintzileou for the
data on the effect of interventions, as they are critical reading of the text is gratefully acknowledged.
applied in steps to the monument. Thus, the effects The close collaboration othe Scientific Committee
of interventions to various characteristics of the with the personnel of the 1st Ephorate of Byzantine
monument, as well as their efficiency can be con- Antiquities, the Directorate for Restoration of Byzan-
tinuously assessed. It should also be noted that, a tine and Post Byzantine Monuments, the Directorate
similar monitoring system having been installed in for Technical Research on Restoration and the Direc-
another important monument (the Katholikon of torate for the Conservation of Ancient and Modern
Osios Loucas Monastery), the comparative eval- Monuments was the base for this multidisciplinary
uation of data obtained in the two churches will approach. The entire project was included in the Oper-
allow for a better understanding of the structural ational Program “CULTURE”. It was co-funded by the
seismic response of middle-Byzantine churches of European Regional Development Fund (ERDF-75%)
octagonal type. and by National Funds (25%).
8. The potential of NDT was further enhanced to serve
the needs of investigating important monuments.
High frequency GPR combined with endoscopy REFERENCES
proved to be efficient for the identification of the Bouras, Ch. 1998. The Dafni Monastic Complex
masonry construction type. High frequency GPR Reconsidered, AETOS Studies in honor of Cyril Mango,
contributed also to locate unsafe regions of mosaics B. G. Teubner Stuttgart und Leipzig.
substrata. Furthermore, sonic techniques, under- Chryssopoulos D., Anamaterou L., Georganis F. 2003. Docu-
taken before and after grouting, were proven very mentation study for the mosaics of the Katholikon of Dafni
efficient in checking the effect and efficiency of Monastery after the 1999 earthquake. Hellenic Ministry
grouting intervention, whereas geophysical inves- of Culture (in Greek).
tigation by means of 3D seismic tomography pro- Côte Ph., Dérobert X., Miltiadou-Fezans A., Delinikolas N.
vided valuable information about the foundation 2004. Mosaic-grouting monitoring by ground-penetrating
radar. Proc.IV Int. Sem. Struct. Anal. of Hist. Const.,
elements. Padova, Vol. 1, pp. 401–406.
The entire plan of investigations has supported the Côte Ph., Dérobert X., Miltiadou-Fezans A., Delinikolas N.
rational design and implementation of the first phase Durand O., Alexandre J. Kalagri A., Savvidou M,
of works, as well as the evaluation of their effect and Chryssopoylos D., Anamaterou L, Georganis, F. 2008.
efficiency. Application of non-destructive techniques at the Katho-
likon of Daphni Monastery for mapping the mosaics
The collected results are now being used for the substrata and grouting monitoring. Sub. to the 6t intern.
design of the second phase interventions. A synthe- conf. on Structural Analysis of Historical Constructions,
sis of all data provided by the situ surveys during the Bath.
first phase of interventions, the results of the monitor- Delinikolas, N., Miltiadou-Fezans, A., Tsofopoulou, E.,
ing system and those obtained by testing the wallettes Minos, N., Chrissopoulos, D. 1999. Initial Inspections,
and the groin vaulted structure is under elaboration. Assessment and Proposals for Protecting the Monastery
Analytical work will be extended, with further devel- of Dafni, Internal report. Hellenic Ministry of Culture (in
opment and calibration of detailed numerical mod- Greek).
els. This/these calibrated model/models will then be Delinikolas N., Miltiadou-Fezans A., Chorafa E., Zaroyianni
E. 2003. Study on restoration of the Katholikon of Dafni
used to check the efficiency of various strengthening Monastery: Phase A – Architectural and historical Survey.
intervention techniques (installation of ties, metallic Hellenic Ministry of Culture (in Greek).
confinements, diaphragmatic structures in the extra- Georgopoulos A. 2003. Novel survey and digital pho-
dos of the vaults and the exonarthex, etc). Thus, the togrammetry methods for geometrical and architectural
86
digital survey of Byzantine monuments. The case Dafni Miltiadou-Fezans A., Kalagri A., Kakkinou S., Ziagou A.,
Monastery. Research Report, NTUA and Hellenic Min- Delinikolas N., Zarogianni E., Chorafa E. 2008. Method-
istry of Culture. ology for in situ application of hydraulic grouts on historic
Kalagri A., Miltiadou–Fezans A., Vintzileou E. 2007. Design masonry structures. The case of the Katholikon of Dafni
and evaluation of hydraulic lime grouts for the strengthen- Monastery. Submitted to the 6th intern. conf. on Structural
ing of stone masonry historic structures. Proc. int. symp. Analysis of Historical Constructions, Bath, UK.
on Studies on Historical Heritage, Antalya, Turkey, 2007. Mouzakis Ch., Miltiadou-FezansA.,Touliatios P., Delinikolas
Mariolakos, I., Fountoulis, I., Andreadakis, E., 2000. N., Dourakopoulos J. 2008. Earthquake based condition
Engineering geological problems caused by human inter- monitoring of the Katholikon of Dafni Monastery. Sub-
ference on monuments that influence their seismic mitted to the 6th intern. conf. on Structural Analysis of
behavior. Compatible materials recommendations for the Historical Constructions, Bath, UK.
preservation of European cultural heritage. Ac. Civil O.T.M. 2000. Geothechnical Investigation of Dafni Monastery.
Engineers of Greece. Internal Report, Hellenic Ministry if Culture (in Greek).
Millet, G., 1899. Le Monastère de Daphni. Histoire, Archi- Papagianni. I. 2002. Investigation of physicochemical charac-
tecture, Mosaiques, Monuments de l’Art Byzantine I, teristics of the historic mortars of the Katholikon of Dafni
Paris. Monastery. Internal Report, AUTH and Hellenic Ministry
Miltiadou-Fezans A., Tassios T.P., Delinicolas N., Chorafa E., of Culture (in Greek)
Zarogianni E., Chandrinos I. 2003a. Earthquake struc- Polymenakos L., Papamarinopoulos S., Miltiadou A.,
tural problems and urgent measures undertaken to support Charkiolakis N (2005). Investigation of the foundations
the Katholikon of Dafni Monastery in Athens, Greece. of a Byzantine church by three-dimensional seismic
Proc.8th inter. conf. STREMAH 2003, Halkidiki, Greece, tomography. Elsevier Journal of applied geophysics 57,
7–9 May. pp. 81–93.
Miltiadou-Fezans A., Delinikolas N., Chorafa E., Stikas, E. 1962. Restoration and consolidation of the
Zaroyianni E. 2003b. Study on restoration of the Katho- exonarthex of Dafni’s Katholikon, DChAE ser. 4, 3,
likon of Dafni Monastery, Phase A – Structural Survey, 1962–1963.
Analysis and remedial measures. Hellenic Ministry of Tassios, T.P. 2004, Rehabilitation of three-leaf masonry.
Culture (in Greek). In Evoluzione nella sperimentazione per le costruzioni,
Miltiadou-Fezans A., Vintzileou E., Delinikolas N., Seminario Internazionale 26 Settembre- 3Ottobre, CIAS.
Zaroyianni E., Chorafa E. 2004. Pathology of the Dafni Troump, E., Quelques vielles églises byzantines de la Grèce
Monastery: survey, monitoring of cracks, interpretation moderne, Marseille, 1896, 5–11, 11–15.
and numerical verification. Proc. 4th inter. Sem. on Vintzileou, E., Tassios, T.P.1995. Three leaf stone masonry
Structural Analysis of Historical Constructions, Vol. 2, strengthened by injecting cement grouts. Journal of Struc-
Padova. tural Engineering, ASCE, Vol. 121, No 5.
Miltiadou-Fezans A., Vintzileou, E., Papadopoulou, E., Vintzileou E 2002. Methods for documenation of pathology
Kalagri, A. 2006. Mechanical properties of three-leaf and assessment of strengthening measures in Byzan-
stone masonry after grouting. Proc. 5th inter. Sem. on tine churches. Application in Dafni Monastery. Research
Structural Analysis of Historical Constructions, New Report, NTUA and Hellenic Ministry of Culture (in
Delhi, India. Greek).
Miltiadou-Fezans A., Tassios T.P. 2006. New rational crite- Vintzileou E., Miltiadou-Fezans A., Palieraki V. Delinikolas
ria for the holistic design of hydraulic grouts. 1st Conf. N. 2004. The use of radar techniques and endoscopy in
of the Society for Research and Promotion of Scientific investigating old masonry: the case of Dafni Monastery.
Restoration of Monuments (in Greek). Proc. of 4th intern. sem. on Structural Analysis of Histor-
Miltiadou-Fezans A., Kalagri A., Delinikolas N. 2007. ical Constructions, Padova Vol. 2, pp. 351–360.
Design of hydraulic grout and application methodol- Vintzileou E., Miltiadou-Fezans A., Vrouva A,
ogy for stone masonry structures bearing mosaics and Anagnostopoulou S. 2006. Mechanical properties of
mural paintings: the case of the Katholikon of Dafni three-leaf stone masonry. Proc.5th intern. Conf. on Struc-
Monastery. Proc. int. symp. on Studies on Historical tural Analysis of Historical Constructions, New Delhi,
Heritage, Antalya, Turkey, 16–21 September, 2007. India.
87
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
G.A. Clifton
Director Historic Buildings – Gifford Ltd
ABSTRACT: The paper is an overview of the structural interventions made to English Cathedrals. It reviews
some of the most common problems and disasters and looks at how they have been overcome with reference to
particular events.
1 OVERVIEW
89
1170 Gloucester South tower to the west front
collapsed
1175 Worcester Western tower fell into the river Severn
1185 Lincoln The whole building collapsed during an
earthquake
1210 Chichester West tower collapses
1238 Lincoln Central tower fell
1248 Wells Central spire fell in an earthquake
1272 Norwich Spire burnt down during rioting
1322 Ely Central tower removed before it fell
1362 Norwich Replacement timber spire blown down
in a gale
1407 York Central tower and Spire collapsed
1463 Norwich Second replacement spire destroyed by
lightning The stone replacement from 1480 still stands!
1548 Lincoln Timber spire (524 ft high) destroyed by
lightening
1561 Old St. Paul’s Timber spire destroyed by lightning
1615 Ripon Central spire collapses
1635 Chichester West Tower collapses again
1640’s Durham West Tower demolished by Cromwell’s
army
1642 Lichfield Central spire demolished by Cromwell’s
army Figure 3. Chichester spire collapse.
1704 Canterbury Central spire removed due to storm
damage
from flying buttresses. On visual inspection one is
1786 Hereford Central spire removed after collapse of
immediately worried that the vault is in tension and
west front
that the slender columns are undergoing excessive
1807 Lincoln West front spires demolished as unsafe
bending.
1832 St. Albans Spire removed as unsafe
I have investigated these carefully and can show
1861 Chichester Central tower and spire collapse
that they have a factor of safety greater than 1.0 which
1880’s Peterborough Tower replaced to prevent a
is just as well as they have been standing unchanged
collapse.
for 900 years. On searching back through the archives
This is by no means an exhaustive list as it not I find that every Cathedral Architect, Engineer, and
does not include many of the major fires that have before that the master masons, were all expressing con-
caused untold loss to the Cathedrals, but does show cern about the retro choir and all setting up their own
the determination to recover from disaster. monitoring systems – just as I have done! Indeed the
Victorians had a series of specially fabricated props
which were kept close to the retro choir for instant
3 LONG TERM WORRIES installation if any cracks appeared. My own advice is
that if cracks appear everyone is to run, as there will
Over this time the buildings were subjected to much probably be localised and fast compression failures! It
repair, renovation and alteration and we know, as did gives one a profound sense of historical continuity to
our forefathers, of the value of constant maintenance find that we share the same worries as our forbears.
and the dire consequences of its omission. We have Another fascinating case is that of the Deans Eye
records of many of these campaigns of repair and window at Lincoln Cathedral.This is one of the top five
alteration and it is fascinating to find that many of rose windows in Europe with fabulous 1220 stained
our current concerns are merely the continuation of glass and a very ambitious form.
worries of the past generations. The companion window, the Bishop Eye collapsed
I have personal experience of two very clear exam- within 50 years and the archives show a long history
ples of such ongoing concerns, one which, so far, has of repairs to the Deans Eye, some successful and some
required no intervention, but a second which resulted a failure. A constant theme in Clerk of Works records
in significant works. over the centuries is how to keep the window stable.
The retro choir at Wells Cathedral is a vaulted This was initially by adding cramps and lead filling
area between the high altar and the Lady Chapel, the to the joints as they moved, then by bracing across
supporting columns are very slender and support an the central opening and latterly by adding major cross
unbalanced vault onto which are imposed large loads bracing behind the window. Until by the year 2000 no
90
Figure 4. Wells Cathedral retrochoir.
91
Figure 7. Fire at York Minster.
92
Figure 10. Ely Cathedral – strengthening of the West Tower.
7 SALISBURY CATHEDRAL
Figure 9. Ely Cathedral – strengthening of the West Tower.
A different approach was undertaken at Salisbury
Cathedral where there were serious concerns about the
strengthened at its top by a new reinforced concrete stability of the spire in the 1970’s. Salisbury spire has
ring beam, with stainless steel cables inserted and a series of bands of open stonework at several levels in
pre-tensioned at a lower level. its height and the lowest level of stonework was decay-
This is an excellent example of the progression of ing such that there were serious fears as to its ability
repairs to Towers in English Cathedrals. As we have to continue to carry the wind loads and weight from
seen there were numerous collapses of towers rela- the spire. It was not possible to individually replace
tively soon after their construction or alteration and the stones as it was highly likely that the spire would
there was little that could be done with the technology collapse with the removal of some of the stones. Vari-
available at the time to save them. They were often ous options were considered including the dismantling
significantly under designed. and rebuilding of the spire – though needless to say that
As the centuries passed by the continuing settle- was thought of as a last resort.
ment of overloaded ground below the central piers The successful solution was to introduce a stainless
caused ongoing cracking of the major towers and this steel frame inside the spire which effectively transmit-
resulted in loss of stiffness which caused further redis- ted the load from the upper part of the spire into the
tributions of load. This often increased the load on the frame and bypassed the open section. The frame was
piers, with an ongoing racheting increase in cracking then jacked against the upper spire, to ensure that it
and movements. Sometimes this was added to other was taking some of the load from the open section.
interventions such as the removal of the choir screen
at Chichester which ultimately caused the failure of 8 REPAIR AND MAINTENANCE
the spire. In other cases the inclusion of wrought iron
ties and strapping alongside the repair of crumbling Alongside these major interventions there is an
stonework was sufficient to maintain stability. ongoing need for constant repair and renewals. All
93
Figure 11. Salisbury Cathedral.
94
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
F. Peña
Instituto de Ingeniería, UNAM, Mexico
ABSTRACT: Anyone involved in the conservation of cultural heritage buildings is aware of the enormous
amount of information generated by the different specialists, which is generally not handled adequately by owners
and/or authorities involved in the process. The consequence is that valuable information is lost in a complex
process of reaching a decision that involves many different experts and information tends to get forgotten
or misplaced in the course of time. A case study with extensive damage, which recently suffered significant
conservation works, is described in detail, so that the adopted remedial measures are justified. Then, the case
study is also used to develop an application for the efficient management and visualization of the information
related to interventions of cultural heritage buildings.
Monastery of Santa Maria de Salzedas (Portugal) The Monastery and Church of Salzedas are located
recently suffered major works in one of the cloisters. in Salzedas, Tarouca and the church was recently
This case study is addressed here from a compre- classified as a National Monument. The church is
hensive perspective, including aspects of the inter- essentially set in an urban environment, whereas the
vention project and aspects related to information monastery is set in a more rural environment (Fig. 1a).
management. The plan dimensions are very large, 75 × 101 m2 . The
The first part of the paper describes in detail the monastery and church possess a longitudinal irregu-
Monastery of Santa Maria de Salzedas and the con- lar plan with different volumes, typical of a Cistercian
servation works in the cloister. This includes survey Abbey (Fig. 1b). The conservation works addressed
and characterization of the damage, together with in here focus in the cloister dated from the 17th century
situ and laboratory testing, which allowed to gather the (Main Cloister, in the picture).
information necessary to establish the need of an inter-
vention and to bound this intervention. The obtained
information allowed obtaining a computer simulation
2.1 Previous documented works in the
of the building, which resulted in clear information on
main cloister
its structural behavior. The details on the consolidation
project and the execution works are also provided. The main cloister is regular and substitutes part of
The second part of the paper addresses a database the primitive cloisters (Fig. 2a). It possesses cross
created for the management of the information gener- vaults in the 1st level and barrel vaults in the 2nd
ated during the intervention process and the historical level. The walls, brackets and ribs are made of gran-
information collected, which has been produced along ite and the vaults are made of brick masonry with
the time and has been selected by different areas of clay or masonry filling. After repeated statements of
study. The information is accessible via intranet for the pre-collapse status of the cloister, Leitão (1963)
practioners and is planned to be accessible on the and Cocheril (1978), the former General Directorate
monastery for visitors, in way that a non-specialist and for National Buildings and Monuments (DGEMN)
a team specialist can obtain information with different carried out remedial works in 1980/1981 and 1983
contents. The paper discusses the underlying princi- (Figs 2b-d), including: (a) Demolishing and replac-
ples in which the application is based on and details ing the vault of the 2nd level of the West wing
the main characteristics of the application. by a reinforced concrete vault; (b) Dismounting and
95
Church
Sacristy
Figure 3. Plan of Clairvaux, Cocheril (1978).
Main Chapter
Little
Cloister Cloister Room reassembling the wall separating the large and small
Inn cloisters, between the 1st and the 2nd levels.
The works do not comply with modern theories
of intervention in historical structures and would be,
today, very debatable.
(b)
3 VIRTUAL RECONSTRUCTION
Figure 1. Monastery and Church of Salzedas: (a) aerial
view; (b) plan and spatial units. The majority of the monastery disappeared after the
extinction of the religious orders in the Age of Enlight-
enment without letting any vestige. A 3D CAD model
of the virtual reconstruction of the medieval monastery
of Santa Maria de Salzedas was made in order to
allow visitors to better understand the monument and
to assist the conservation works.
96
The church was oriented in the East-West direction,
with the apse to the East. The plan was a Latin cross,
with three naves, being the aisles lower than the cen-
tral nave. Usually the body of the naves was preceded
by a narthex and the transept was formed by a single
nave. The naves were divided by square pillars with (a)
engaged or truncated columns and were covered by
barrel vaults, oriented perpendicularly.
Sometimes, on the side of the transept, engaged to
the exterior wall, a bell tower could exist, built in stone
or wood. The construction of towers similar to tradi-
tional secular churches was forbidden. The church’s
facade was simple and divided in three parts.A detailed (b)
summary description on the arrangement of Cistercian
abbeys can be found in Amado et al. (2006).
97
Figure 5. Reference model mesh of the medieval church.
98
Figure 9. Crack survey: (a) 1st level; (b) 2nd level;
Figure 8. Internal views of the medieval model. (c) South wing, 2nd level; (d) South-West corner, 2nd level;
(e) West wing, 1st level; (f) South-West corner, 1st level; (g)
wall, West wing.
The out-of-plumb displacement of the internal walls
reaches values of 0.18 m, 0.14 m, 0.09 m and 0.07 m freeze-thaw cycles, effloresce and biological coloniza-
in the wings West, South, East and North, respectively. tion (Fig. 12). A detailed survey on stone deterioration
The brackets supporting the crossed vaults of the and salt effloresce in the cloister are given in Alves &
first level show signs of compression/shear damage, Pamplona (2006, 2007), see also Figure 13.
particularly in the West wing (Fig. 11a). This can be
explained by the tilting movement of the walls. The
absence of connection between the infill of the crossed 5 IN SITU TESTING AND LABORATORY
vaults and the walls resulted in a much localized area to INVESTIGATION
transfer the load, i.e. only the brackets. Also, a signifi-
cant number of bricks show deterioration, particularly In order to better characterize the materials, to justify
around the cracked areas (Fig. 11b). This occurs at the observed damage and to define corrective mea-
bothflorr levels and can be explained by frost-thaw sures, an experimental in-situ and laboratory testing
cycles and water infiltration, as the amount of rainfall program was carried out.
per year in the region is high and the temperatures in
the winter are excellent for ice formation (daily cycles
5.1 Soil and foundation survey
with ± 0◦ ).
Other perturbing signs, less relevant from the struc- This survey consisted of seven boring holes and
tural point of view, include damage of the stone due to three pits to define the mechanical and physical
99
Figure 10. Aspects of the wall movements: (a) separation
between wall and vault (intrados of 1st level); (b) detail of
previous separation; (c) separation between wall and vault
(extrados of 2nd level) and (d) horizontal movement inwards
to the court.
Figure 12. Material deterioration: degradation of stone
carving and biological colonization.
100
Figure 16. Flat-jack testing: (a) aspect of one test;
(b) stress-relative displacements results.
Figure 14. Soil characterization: (a) identified layers; As a result of the inspection with the rigid endo-
(b) inspection pit 2; (c) inspection pit 3. scope, it was concluded that the granite walls of the
cloister are adequate and there is no danger of collapse
due to desegregation under vertical loading. Coring
and other techniques to estimate the strength of the
walls were considered not necessary and it was decided
to carry out two simple flat-jack tests.
101
Figure 17. Samples for visual identification and mechanical
testing.
(Fig. 18b, c, d). The mortar from the brick masonry was
also characterized with uniaxial compression tests.
The representative samples were extracted from the
construction or the cored samples.
The tests indicated the composition of the plaster
and mortar (1:3 in volume) and the composition of the
vault infill. The bricks are of low quality and non-
durable, with an absorption in cold water around 20%
and a volume mass of 1560 kN/m3 .
The uniaxial compression tests were carried out
in samples of 45 × 45 × 45 mm3 , tested with greased
Teflon layers to avoid the plate confining effect. The
obtained Young’s modulus and strength for the bricks
were Eb = 7300 MPa and fb = 5.2 MPa, respectively.
These values are quite low and confirm the poor
quality of the bricks. The obtained Young’s modulus
and strength for the mortar were Em = 8600 MPa and
fm = 3.8 MPa, respectively. This strength value can be
considered normal for the mortar composition.
With these results it is possible to estimate the
strength of the brick masonry as:
6 SAFETY ASSESSMENT
Figure 18. Examples of chemical, physical and mechani-
The objective of the structural analysis carried was the cal testing: (a) X-ray diffraction test on vault infill material;
safety assessment of the cloister and the definition of (b) absorption test on brick; (c) samples tested in uniax-
remedial measures. For this purpose, two models of ial compression; (d) compression test results on brick (two
samples).
102
Figure 20. Adopted 2D model for the 2nd level: (a)
thickness in different parts; (b) deformation and boundary
conditions.
103
experience of the stone master in charge of the works
(Humberto Reis de Sousa), as a joint decision by the
authors and the technicians in charge of the monu-
ment (ArchitectsAngela Melo and Jorge da Costa from
the Cultural Property Service, Porto). The option not
to intervene in the foundations was made from the
beginning, as: (a) the intervention would need to be
very invasive; (b) it would lead to the destruction of
the buried remainings; (c) the authors believe an inter-
vention in the superstructure is sufficient to stabilise
the structure. In the modern spirit of a step-by-step
minimal intervention, the owner was alerted to the
fact that a (possible, but unlikely) intervention in the
foundations might be required in the future.
The viability of the proposed works depended on the
possibility of cracked masonry to accommodate move-
ments and the technical capability of the contractor, as
the structures would be moved but not dismounted.
The operation entails some risk due to the precari-
ous stability, significant weight and non-monolithic
behaviour of part of the structure. The operation was
made possible only by the careful execution of the
stone mason Humberto Reis de Sousa, which knew
how to straighten and move structures walls using
hydraulic jacks, cable tensioning tools and adjustable
props.
Figures 22 and 23 illustrate the proposed struc-
tural remedial measures. All metallic elements are in
stainless steel AISI 316, which provides the highest
corrosion resistance. The ties are applied only in the
wings that exhibit larger damage, possibly due to the
lack of external transverse walls: South and East, and
South and West. The ties are placed in the vault infill,
meaning that the horizontal thrust from the vault is not
aligned with the tie. This non-alignment produces a
Figure 21. Results from non-linear analysis: (a) load- bending moment, which is balanced by vertical stitch-
displacement diagram; (b) separation between vault and ing (or reinforcement) for the 2nd level and an uneven
wall; (c) cracking of vault. vertical distribution of stresses for the 1st level. This
uneven distribution of stresses, which does not pro-
voke any tensile stresses, is possible due to the weight
displacements of magnitude comparable to the values of the upper structure.
observed in the structure, it is necessary to consider the It is noted that other possible solutions were ini-
soil-structure interaction. It seems therefore that the tially considered for the 2nd level, and a final solution
foundations play a key-role in the observed damage; remained to be decided during execution and clarifi-
(d) the large movements recorded in the construction cation of the composition of the vaults. It was found
and the deterioration of the brick vaults indicate that that the masonry infill of the vaults from the 2nd level
the safety level of the structure is not compatible with was extremely hard and impossible to remove with-
any use and immediate intervention was necessary. out significant loss of fabric and possible damage to
the structure. This contributed to the conception of the
adopted solution for the 2nd level.
7 REMEDIAL WORKS All ties possess a coupling element capable of
adjusting the tie. In the 2nd level, the ties are connected
The cloister required consolidations works and the to a vertical bar inserted in a fabric sock, capable
proposed solution included repositioning the walls in of containing the injected grout. As the masonry in
plumb, elevation/re-centring vaults and arches, and the internal walls is made of large ashlars, additional
hidden tying of the walls as an additional strength- anchorage is provided with a transverse element. For
ening. This strategy resulted from the inspection, the internal walls, sufficient bond occurs in the con-
diagnosis and safety assessment, and from the previous tact with the irregular masonry. In the 1st level, vertical
104
Figure 23. Consolidation works for 1st level: (a) plan with
location of ties; (b) transverse section view.
Figure 22. Consolidation works for 2nd level: (a) plan with
location of ties; (b) transverse section view; (c) details.
bars are not needed, and the ties are directly anchored
to the walls at a 30◦ angle.
Figure 24 shows images of the preparatory works
for protection and for propping and lifting the vaults.
Details on the installation of the ties are shown in Figure 24. Preparatory works: (a) protection of floor and
Figure 25 and several complementary works are shown walls; (b) adjustable propping for 2nd level; (c) adjustable
in Figure 26. propping for 1st level.
105
Figure 25. Execution of consolidation: (a, b) removal of
infill from wings in 2nd and 1st level; (c) openings to remove
debris and lift barrel vaults; (d) coring for vertical reinforce-
ment and tie anchorage in 2nd and 1st level; (e, f) installing
ties, repairing brackets and final view in 2nd and 1st level
ceiling.
106
Figure 27. The database can be browsed through: (a) menus; (b) photographs, 3D models, panoramic views; (c) hotspots.
8 INFORMATION MANAGEMENT TOOL place; (c) updated records can be easily incorporate;
(d) the format of the records can be standardized and
Anyone involved in the conservation of cultural her- indexed; (e) information is accessible 24 hours a day,
itage buildings is aware of the enormous amount year round, from any point of the world; (f) information
of information generated by the different specialists, can be published, disseminated and be easily accessi-
which is generally not handled adequately by owners ble and/or restricted depending on the classification of
and/or authorities involved in the process. The conse- the information and the user; (g) the information can
quence is that valuable information is lost in a complex be easily presented in different languages in real time
process of reaching a decision that involves many dif- allowing its dissemination around the world.
ferent experts and information tends to get forgotten
or misplaced in the course of time. 8.2 Structure of the database
Upon the recognition of this evidence, University
of Minho (UMinho) in partnership with Centre of The information about the Monastery is presented
Computer Graphics (CCG) and the former Portuguese using virtual representation (VR), i.e. through mul-
Institute of Architectural Heritage (IPPAR) developed timedia tools, the users will have access to all infor-
an application for the efficient management and visu- mation stored in the database.
alization of the information related to interventions The application, and the database, is structured in
of cultural heritage buildings. The development was three main groups, which define the way in which
based in a case study with extensive damage and that the information can be accessed, namely spatial units,
recently suffered major works in a cloister, namely level of access and thematic areas. The spatial units
the Monastery of Santa Maria de Salzedas (Portugal). should be understood as the different areas in which
A database was created for the management of all the the Monastery has been divided. The level of access
information generated during the intervention process defines the privileges of the user. Thematic areas are
of 2005 and the historical information collected which the options for searching the information contained in
has been produced along the time and has been selected the database. The database can be browsed through
by different areas of study. The information is acces- menus or hotspots in photographs, 360◦ panoramic
sible via intranet for practitioners and, planned soon, views or 3D model (Fig. 27). The thematic areas
to be available on the monastery for visitors, in way depend on the level of access. A tree-like structure
that a non-specialist and a team specialist can obtain was thought for the thematic areas (menus and sub-
information with different contents. menus), in which each one can be subdivided in three
different levels.
The main menus or thematic areas are shown in
8.1 Internet as a tool for information
a fixed top bar, while the submenus are displayed at
management
the top-left of the screen. The bottom-left area dis-
Internet is an excellent tool that fulfils the ICOMOS plays the image with hotspots. The hotspots link with
(1996) principles for the management, dissemination some particular information or display files stored in
and sharing of records, namely: (a) the original elec- the database. The information is displayed in the main
tronic version can be safely archived; (b) backups are window. The information is presented as image ele-
easily made and can be stored in a different and safe ments (pictures, drawings, movies, panoramic views,
107
ACKNOWLEDGEMENTS
108
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M. Drdacky
Institute of Theoretical and Applied Mechanics ASCR v.v.i, Prague, Czech Republic
Ch. Hennen
Institut für Diagnostik und Konservierung an Denkmalen in Sachsen und Sachsen-Anhalt e.V., Dresden, Germany
S. Lanza
University of Genova, Italy
M. Tomaževič
Slovenia National Building & Civil Engineering Institute, Ljubljana, Slovenia
D. Kolymbas
University of Innsbruck, Austria
I. Herle
Technische Universität Dresden, Germany
A. Virsta
UNESCO Chair ECOLAND, University of Bucharest, Romania
L. Binda
Politecnico di Milano, Italy
P. Askew
Industrial Microbiological Services Ltd., UK
ABSTRACT: For the protection of Cultural Heritage from environmental damage and in particular against
flood, a research project has been proposed under the 6th European Framework Program to support policy-
oriented research. The main aim is the development of preventive conservation strategies and the identification
of appropriate measures. Clear recommendations for mitigating or removing damage from Cultural Heritage
objects will be provided through the project results. CHEF proposes the integration of multidisciplinary research
as scientific support to European policies. In this frame the necessary technological basis and cost efficient and
effective tools for the development of new and innovative strategies will be provided.
1 INTRODUCTION (1997), Rhine (1993 and 1995) and many more. Floods
can not be prevented. Europe will have to face further
The European Community has suffered from disas- flood catastrophes due to the change of climate and due
trous floods in recent years, which caused enormous to further building activities in flood-prone regions [1]
damages (compare Fig. 1) and left hundreds of peo- [2] [3].
ples dead in several European countries. The flooding Although very comprehensive, projects, strate-
occurred across regional borders and involved neigh- gies and further administrative measures, which
bouring states along the rivers Elbe (2002), Odra started after the recent flood events, do not
109
Figure 1. Historic power station in Prague (CZ) during flood
in 2002. Photo: V. Herle.
110
c. Analysis of exposure by means of adequate
survey
d. Risk assessment of Cultural Heritage based on
quantified hazards, exposure maps and GIS.
2 Analysis and classification of damage processes
in different materials (soil, brick, stone, wood,
metal, textile, paper, etc), structures (buildings,
bridges) and sites (cities, gardens, archaeological
sites) induced by flood and flood related environ-
mental impacts in close cooperation with museums
and further stakeholders. This includes:
a. Set-up of a damage catalogue
b. Investigation of the influence of water, salt and
biodegradation on material properties
c. Comparison to results of modelling.
Figure 3. Investigation of the moisture distribution at Vel-
3 Validation and evaluation of methods and sen- trusy Castle (Czech Republic) after the Elbe 2002 flood with
sors for non-destructive testing and monitoring of radar, detecting damage at a balcony.
material and structural parameters, environmental
technologies will have to be provided.
4 Definition of threshold levels for exposure and dam- which will be practical and adapted to the needs of
age before restoration, also in relation to the use of a broad array of institutions, operational services and
the objects/buildings. specialists in the public sector dealing with civil pro-
5 Analysis of preventive measures before flood and tection, land development, regional planning and risk
temporary (emergency) measures during and soon management.
after flood for movable and immovable Cultural Her-
itage. The influence of these measures on material
and structure will be adressed. 3 WORK PLAN
6 Assessment of restoration and repair techniques
after flood. The scientific and technological approach aims at
7 Assessment of running and new case studies on dif- identifying characteristic parameters for flood and
ferent objects concerning infrastructure, buildings, flood related hazards, which have an impact on mov-
parks, cities. able and immovable Cultural Heritage depending on
8 Generation of practical recommendations by defini- their vulnerability (compare Fig. 3 and 4). The under-
tion of strategies for the protection, conservation and standing of damage processes in materials and struc-
repair of movable and immovable Cultural Heritage. tures by using environmental technologies will provide
The potential impact of the project on the conser- the basis for evaluation and further developing strate-
vation of Cultural Heritage is direct and transparent: gies for preventing, mitigating and repairing damages
The strategies to be developed in the project can help at Cultural Heritage related to the objectives and poli-
to make protection and repair of Cultural Heritage cies of the European Community. It is planned to
against flood and related hazards far more effective implement a project with a clear multidisciplinary
and allow carrying out these measures on a large scale. approach as an obvious reaction upon the requirements
With a reasonable management of Cultural Heritage described in task 4 of the above-mentioned call.
sites, whole neighbourhoods can be prevented from The research will be carried out through eight work
deterioration. Another aspect is the understanding of packages. These work packages and their interaction
damage processes not only induced by flood but also will provide a clear progress beyond the current state
by other aggressive environmental conditions, such as of the art:
air pollution, microclimate, earthquakes, traffic vibra- • WP1: Identification of typical environmental haz-
tions, etc. This knowledge will help to take precautions ards related to flood and being decisive to Cultural
against it and will thus reduce the impact of environ- Heritage. Here, flood mechanisms and hydrology
mental damage on historic structures. But also social will be one important aspect. Movable and immov-
problems like unemployment, equal opportunities and able Cultural Heritage will be classified according
health and safety of citizens will be positively influ- to its sensitivity. This is going to be investigated in
enced by CHEF. The project results will be dissem- the first of the work packages and serves mainly
inated by publications, project reports, contacts with as an input for the measures, strategies and meth-
end-users and mainly by publishing recommendations ods, which are worked on in detail in the subsequent
111
of material (brick, stone, metal, wood, paper, textile,
etc), on the material parameters (porosity, grain size,
density, etc) and the structure and conditions of the
objects. Therefore, the damages are manifold and
gained knowledge from recent floods and a damage
classification is urgently required.
The work of WP 2 will be carried out by the
institutions BAM, TUD and ISML.
• WP3: Classification of damage on historic struc-
tures and sites including buildings, infrastructure
(e.g. bridges) and cities for understanding the mech-
anisms of damage. Static and dynamic loading,
moisture and salt transport, contamination and ero-
sion problems will be considered as well as the
vulnerability of whole structures and sites. Inves-
tigations will be based on previous experience, new
and existing data recorded from on-site investiga-
tions, monitoring and laboratory tests and numerical
simulations. Existing databases and past experience
Figure 4. Determination of the moisture distribution at Vel- will be analysed and damages as they belong to
trusy Castle (Czech Republic) with radar and microwave types of buildings or structures and to sites will be
borehole method. summarized related to different materials. A damage
catalogue containing exposure data and applicabil-
work packages. This part of the project will also ity tips for remedial works will be created. Damage
provide an overview about existing plans for protec- as well as protective measures will be analysed in
tion of Cultural Heritage objects and will examine order to identify the most urgent problems for further
the experiences (historic and recent) made during more detailed studies. Previously defined vulnera-
floods especially in Europe. The threat posed at bility indicators will be validated (element-at-risk,
Cultural Heritage by flooding is quantified in this susceptibility and exposure). A cooperation with the
subtask, based on the analysis of existing litera- User Group will be mobilised and exploited.
ture and archived documents. The selection of the The work of WP 3 will be carried out by the
best-suited technical investigation tools is also institutions ITAM, POLIMI and IDK.
addressed, with reference to the hydraulic and hydro- • WP4: According to the long term experience the best
logic models available and their interpretation in preventive measure for movable heritage objects
view of Cultural Heritage vulnerability assessment. is their replacement from endangered localities to
The work of WP 1 will be carried out by the safe storage. Such a simple procedure failed in
institutions ITAM, CRUIE and IDK1 . many recent cases and this sub-task is suggested
• WP2: Damage analysis of different materials and to analyse the reasons and to prepare more robust
movable Cultural Heritage, their properties and their and reliable approaches. Therefore, its main focus
interaction with moisture, salt, pollution and other will be not only on administrative measures but
phenomena related to flooding catastrophes. In this it will involve also technical measures as well as
context a survey of possible flood-related damage, a critical evaluation of methods and possible needed
validation of non-destructive and minor-destructive improvements. Preventive measures for immovable
methods for damage detection and quantification, a heritage are designed according to the classifica-
classification of damage and the definition of dam- tion of individual objects at risk. Conventionally,
age threshold will be given. The direct contact of both “structural” and “non-structural” interventions
water and contaminants occurring during the flood are addressed in flood protection related issues; the
as well as the modified environmental conditions latter ones include both macro-structural interven-
after the flood leads to various damages and/or con- tions, when involving interventions along the river
taminations of soil, building materials and movable network, and micro-structural ones, when limited
heritage objects. These damages and modifications to the passive protection of the resources located
strongly depend on the flood and environmental on the flood prone territory. The management of
parameters (duration of flood, height of water, flow objects and sites is usually ensured by different sub-
rate, contaminations, temperature, etc), on the type jects and also the measures and their impacts may
differ quite substantially. Therefore, the objects are
1
The abbreviation and full names of the involved institutions treated separately from site to site. Temporary emer-
will be explained in the next section. gency measures prepared before the flood and their
112
effectiveness will be studied considering technical • WP7:The work in the project has to be framed within
measures. This will encompass temporary barriers a strict project management. This work package
as well as construction measures for stabilization includes the administrative and the technical coor-
of the whole building or strengthening of structures dination of the project. The coordinator is supported
based on knowledge of building condition and prob- by the WP-Leaders, who take on the responsibility
able exposure parameters. For restoration and repair, for the technical supervision in their work packages.
post-flooding short term techniques (drying and The work of WP 7 will be carried out by BAM.
dehumidification, reducing temperature, after flood • WP8: The planning of dissemination activities will
care etc.) as well as long term techniques (pollution be carried out during the project under the leader-
removal, salt removal, desalination etc.) for repair of ship of the Exploitation Manager with involvement
damage of historic building materials, which require of WP-Leaders and User Group members. This
a very distinct and careful treatment (paintings, task includes the dissemination of non confidential
frescoes, original building materials) shall be col- information about the project objectives and work
lected and critically analyzed. To correctly perform plan, coming events (seminars), published papers,
a restoration activity after catastrophic events, it is demonstration actions, training courses, lectures at
highly recommended to execute a careful diagnosis universities and especially the dissemination of the
on the damaged works of art. final practical recommendations.
The work of WP 4 will be carried out by the The work of WP 8 will be carried out by the
institutions ITAM and ZAG. institutions ITAM, BAM and IDK
• WP5: Assessment of running and new case stud-
ies on different objects concerning infrastructure,
buildings, parks, cities. Two to three cases will be 4 PARTNERS AND USER GROUP
studied in detail concerning a multitude of aspects
like a small castle surrounded by a park and con- The consortium includes members from 7 different
taining a museum. Full-scale models in the shape European countries:
of small buildings will be also used as intermediate
case between laboratory and real buildings. It will be 1. Bundesanstalt für Materialforschung und – prü-
studied how historic floods occurred and have geo- fung, Berlin, Germany (BAM)
graphically spread. Former concepts and methods 2. Institute of Theoretical and Applied Mechan-
used to restore the affected case studies are going to ics ITAM ASCR v.v.i., Prague, Czech Republic
be evaluated. An analysis of previous and running (ITAM)
case studies carried out in similar projects will be 3. Institut für Diagnostik und Konservierung an
part of this work package in order to get a wider Denkmalen in Sachsen und Sachsen-Anhalt e.V.,
base for the development of recommendations and Dresden, Germany (IDK)
guidelines. 4. University of Genova, Italy (CRUIE)
The work of WP 5 will be carried out by the insti- 5. Slovenia National Building & Civil Engineering
tutions ECOLA ND, ZAG, ITAM, IDK, CRUIE and Institute, Ljubljana, Slovenia (ZAG)
POLIMI. 6. University of Innsbruck, Austria (UIBK)
• WP6: Development of guidelines and recommen- 7. Technische Universität Dresden, Germany (TUD)
dations for the strategies concerning the assessment 8. UNESCO Chair ECOLAND, University of
of vulnerability of Cultural Heritage against flood, Bucharest, Romania (ECOLAND)
the prevention and mitigation of damage, the emer- 9. Department of Structural Engineering (DIS),
gency and the medium and the establishment of long Politecnico di Milano, Italy (POLIMI)
term post-flooding action plans is the main sub- 10. Industrial Microbiological Services Ltd., United
ject of this work package. The collected damages Kingdom (IMSL)
(damage catalogue) as well as the case studies will These institutes provide a lot of experience on dif-
be included in existing data bases. A development ferent skills related to flood protection of Cultural
of strategies for the protection and conservation of Heritage:
buildings, historic areas, movable Cultural Heritage
objects which bear the risk of flooding is taking • Experience with problems related to civil and struc-
place in close collaboration with the end-users. tural engineering with a strong focus on historic
Meetings during the project will serve as knowl- materials and structures
edge exchange between consortium members and • Development of methods for testing and monitoring
end-users. Towards the end of the project, an open of different parameters of materials, structures and
workshop will be organised. objects with only minor invasion
The work of WP 6 will be carried out by the • Measuring climate and climate control of historic
institutions ITAM, CRUIE and IDK. buildings
113
It is envisaged that another major opportunity to
promote the results obtained from the proposed project
will be in the form of advice to local authorities,
historic monument protection trusts and associations,
national governments and the EU in the field of spa-
tial development, specifically sustainable protection
of historic monuments.
This in turn should serve to put future EU policies
on protection of historic monuments onto an even more
widely respected scientific footing both in Europe and
world-wide.
The transfer of the research results will be in the
form of reports, scientific papers in journals, presen-
Figure 5. CHEF Consortium members during the Kick-Off
Meeting on 26th of February 2007 in Brussels. tations at international conferences, case study fact
sheets, patents on novel technologies developed within
the proposed project and an Internet Website. Other
• Characterization of historic building material and possibilities will be hands-on demonstrations and
successful implementation of modern material in courses and the inclusion of the results in regulations
historic structures and systems and guidelines.
• Numerical modelling, laboratory determination and In the universities presented by the partners, the
experimental verification of transient hygrothermal results of the research will be exploited through
transport processes in materials and building com- workshops and continuing education courses for the
ponents; physico-chemical modelling (influence students.
of salt transport and crystallisation) and physico- Several dissemination activities are targeting at a
biological modelling (conditions for mould and broad audience and are meant to demonstrate the
algae growth) connected with deterioration and transformation of scientific research into a practical
aging effects of the building substance context.
• Investigation of continuous drying out processes All these actions have the following objectives:
• Structural assessment of historic building
• Structural modelling • Exploitation of project results
• Town and urban planning and engineering of the • Information of possible users
territory • Information of public
• Safeguard of the historic-architectonic heritage of • Integration of related projects
city centres against the hydraulic and hydrologi- • Involvement of other organisations
cal hazard, vulnerability assessment and mitigation • Transnational networking
initiatives • Future education of staff.
• Water and soil protection engineering, hydrology Dissemination activities will be organised by the
and water resources management Exploitation Manager in close co-operation with the
• Mechanical behaviour of soil, theoretical modelling coordinator and those project partners with close ties
(theory of hypoplasticity) and experimental deter- to end-user, managers or government or local officials
mination of material parameters for soil models. responsible for operating and maintaining historic
buildings. In that context the User Group plays an
The balance between old EU member countries and
important role for the integration of organisations and
new EU members together with associated states is
other projects related to the maintenance and care of
especially adapted to the experience gained with floods
Cultural Heritage buildings.
from rivers and costs in neighbouring regions: Czech
The consortium of CHEF has already contacted sev-
Republic, Slovenia, Romania, Austria, Italy, United
eral national trusts, authorities, companies, museums,
Kingdom and Germany. These countries were heavily
universities in order to establish a large User Group
affected by flood hazards during the last 10 years.
for dissemination activities. Many institutions all over
To involve the public and to raise awareness it is
Europe have expressed their interest in the project and
important to plan dissemination actions, which target
their willingness to support the activities.
at many different groups. It is necessary to present
the project and its objectives, results and approach in
an easily understandable and straightforward way. To
keep the balance between scientific approach and pub- 5 FIRST RESULTS
lic acceptance is a difficult task, but the experienced
consortium in cooperation with the User Group will Since the official project started in February 2007,
be able to manage this. some results are already available.
114
5.1 Report on experiences from floods and includes short comments on their content and main
results.
As an overview about the whole complex topic, it was
The European Commission has launched several
necessary to perform a state of the art survey and to
international research projects on natural hazards
find out, which measures have been taken in the past
affecting Cultural Heritage. These projects are valu-
to prevent or reduce the impact of natural flooding
able for the CHEF project because most of them
disasters and to survey possible repair interventions
European relevance and provide highly objective infor-
that have proven to be suitable for the restoration of
mation. To take advantage of the synergies of similar
valuable buildings after the flood.
research projects it is necessary to use the contacts that
Several factors are influencing Cultural Heritage
have been established within many networks, espe-
and the losses due to flood and it is important to
cially with end-users, who are the most important
develop all-embracing strategies taking into account
addressees for the project results.
not only technical measures against flood but to con-
It is a major issue that during the next project years
sider also the “human factor” and other aspects that
the listed contacts with the leaders and partners of the
bare a threat to vulnerable buildings.
related projects will be activated in order to establish
An evaluation of countermeasures against flood
a common scheme for sharing experience and to col-
damage has to start with an analysis of former inci-
laborate in filling the gaps presently existing in our
dents. In many cases data are not available but still it
knowledge concerning protection of Cultural Heritage
is possible to draw valuable conclusions from reports
from flood effects.
and recordings of earlier catastrophies. Even the com-
Examples of very prominent international projects
parison of old pictures or photographs that show, how
related to damage of flooding events are NOAH’S
a Cultural Heritage building is located in its natural
ARK [6], RIMAX [7], and ELLA [8]. The objectives
surrounding allows to identify some of the relevant
of NOAH’S ARK (Global Climate Change Impact
characteristic parameters of flooding.
on Built Heritage and Cultural Landscapes, running
The “Report on experiences from previous and his-
from 2004–2007) were to determine the meteorologi-
toric floods, summary about the characteristic param-
cal and climate change parameters that are most critical
eters of these events” was worked out by the University
to the built cultural heritage. Moreover to research
of Genoa, Italy under the leadership of Prof. Luca
and address future expected damages and dissemi-
Lanza and with the contributions of other project part-
nate optimum adaption strategies. The RIMAX (Risk
ners. It provides a summary of the experiences gained
Management of Extreme Flood Events) project aims at
from major floods occurred in the past and their syn-
developing and implementing improved instruments
thetic descriptive parameters, which may help in future
of flood risk management by the integration of dif-
assessment of the severity of floods in relation to Cul-
ferent disciplines and several participants. It focuses
tural Heritage. The main types of floods are presented,
on extreme flood events in river basins which means
and typical problems are illustrated by means of recent
events with a return period of more than a 100 years and
and past examples related to the impact on Cultural
a highly destructive potential. ELLA (Preventive flood
Heritage.
management measures by transnational spatial plan-
The report is one of the agreed deliverables of the
ning, running 2004-2007) focusses on damage related
project and will be published after the next project
to flood events of the river Elbe in Germany. It is
meeting in April 2008.
worked towards trans-national strategies and regional
planning for flood control. This includes international
knowledge, experience and data exchange.
5.2 Interfaces to running Flood and Cultural
Heritage projects
Another deliverable of the project compiles a list 5.3 Literature compilation
of other projects, both national and international,
A large number of articles, publications, reports and
related to mitigation of damages from floods. This
websites have been collected and analyzed in order to
report was made by the Institute of Theoretical and
get a broad overview about the topic. Especially for
Applied Mechanics ITAM ASCR v.v.i., Prague, Czech
the recent floods in Germany and the Czech Repub-
Republic, led by Miloš Drdácký.
lic in 2002 a lot of valuable material is available. For
It provides basic information about other related
an effective analysis the material has been classified
running projects dealing with flooding, and on their
according to the following topics:
impact on Cultural Heritage. An analysis is made also
on other natural hazards, e.g. earthquake, landslides, • Recommendations on floods
subsidence etc. and their threat for Cultural Heritage. • Reports about floods – lessons learned
The report “Interfaces to running Flood & Cultural • Risk management of floods
Heritage projects” provides a list of identified projects • Damage of materials.
115
This compilation allows the comparison of counter-
measures that have been taken in different European
countries to reduce the impact of flooding. The reports
about various regional flooding events show that the
visible and unvisible damages and cosequences and
may vary significantly even if flooding occurs in the
same region at different times.
116
have the possibility to establish contingency plans. The proposed project CHEF already deals with top-
Therefore it is very important to run the project with ics, which are important for the framework described
participation of end-users and to address the public above. It can be a very good starting point for further
as often as possible. It is planned to invite end-users initiatives in this field. The large consortium and the
to most of the project meetings and to discuss the involvement of the User Group offer many possibili-
usefulness and impact of the project results. ties for networking. It is very likely that the consortium
The Kick-Off Meeting of the CHEF Consortium or groups within this partnership will react upon future
members took place on 26th of February 2007 in Brus- calls for projects in FP7. The work plan of the project
sels. The work plan to achieve the ambitious goals has described here can be further extended by including
been presented to the representatives of the European more case studies or by concentrating on different
Commission. aspects of Cultural Heritage protection. There are sev-
The 2nd Periodic Meeting of the project was orga- eral starting points for new initiatives or for extended
nized as a two days event. The first day was scheduled co-operations with other groups or institutions dealing
as a regular working meeting to exchange first results with Cultural Heritage. This project will fulfil the pro-
and to plan further dissemination activities. On the sec- posed work plan and provide the agreed deliverables
ond day invited end-users were present and together but should also be regarded as some kind of “think
with the group an exchange of experiences, gained tank” for future research.
during previous floods took place.
Presentations about flooding as a global phe-
nomenon, followed by impressive reports about recent ACKNOWLEDGEMENTS
flooding events in Germany, Slovenia, Czech Republic
and the United Kingdom highlighted the importance of The project CHEF is funded by the European Com-
the topic. The introduction of methods for assessment mission in the 6th Framework Program. The presented
and diagnosis of affected buildings and sites as well results have been achieved by all Partners of the
as the presentation of treatments for damaged historic project, which are listed in Chapter 4.
materials contributed to the exchange of knowledge
between consortium and end-users. This was extended
by lectures about preventive flood management mea-
sures, flood management in historic cities, urban REFERENCES
development, open space planning, monument preser-
vation and strategies for risk minimization. [1] An EU policy on flood risk management, http://
Another meeting, the 3rd Periodic Meeting is sched- www.europa.eu.int/comm/environment/water/
flood_risk/index.htm
uled for the 14th and 15th of April 2008 and will be [2] Becker, A. and Grünewald, U. (2003) Flood risk in
held in Genoa, Italy. Again, besides a regular project Central Europe. Science Vol. 300, 16.5.2003, p. 1099.
meeting, one day will be dedicated for the involvement [3] Dombrowsky, W.R. and Ohlendiek, L. (1998) Flood
of end-users and interested public. Management in Germany. In Rosenthal, U., ‘t Hart, P.
(publisher): Flood Response and Crisis Management
in Western Europe. A Comparative Analysis, Berlin,
7 CONCLUSIONS AND OUTLOOK New York a.o. Springer, pp. 153–187.
[4] Drdácký, M. and Slížková, Z. Damages and Failures
The risk of flooding or other natural hazards has of Historic Materials and Structures from Natural Haz-
ards, in: Proceedings on 2nd H & mH, International
increased significantly in recent decades due to cli- Conference on Vulnerability of 20th Century Cultural
matic change and building activities in risk-prone eritage to Hazards and Prevention Measures, Island of
areas. The European Community is aware of this situ- Kos, Greece, 3–5 October 2005.
ation and has raised programmes to support research [5] Lanza, S. (2003). Flood HazardThreat on Cultural Her-
initiatives, which aim at prevention of natural disas- itage in the Town of Genoa. J. Cultural Heritage, 4(3),
ters and minimizing the impact on Cultural Heritage. pp. 159–167.
For the 7th Framework Programme of the European [6] http://noahsark.isac.cnr.it
Commission a Tentative Roadmap has been set up for [7] http://rimax-hochwasser.de/index.php.?id=430&L=1
the area Environment including Cultural Heritage. The [8] http://ella-interreg.org/kontakt.htnl?&L=1
[9] http://www.onsiteformasonry.bam.de
research need covers the following topics: [10] European Commission – Community Research,
• Managing environmental changes, hazards, dam- Project Report (2006) Onsiteformasonry project: On-
age, mitigation and preventive conservation site investigation techniques for the structural evalu-
• Protection of archaeological sites and cultural land- ation of historic masonry buildings. EUR 21696 EN.
ISBN 92-894-9601-0.
scapes
• Fostering the integration of Cultural Heritage in the
rural/urban setting.
117
Definition of significance and attribution of value
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
C. Thelin
Tyréns AB, Växjö, Sweden
Chalmers University of Technology, Göteborg, Sweden
K. Linscott
Göteborg University, Göteborg, Sweden
ABSTRACT: Few roof structures remain unscathed from the 12th and early 13th centuries. The relatively
large amount of well-preserved early roof structures in churches in Sweden is used as a point of departure to
structurally define medieval Romanesque roof structures and to make comparisons within North-Western Europe.
Investigations of actual roof structures along with study of written sources provide the basis for the study. The
structures are defined and compared according to load-carrying system, included parts, centre distances, slope of
roof, connection with wall, joints, outer roofing and decorations. We found many structural similarities and the
structures seemingly belong to the same building historical context while also showing regional differences. The
early medieval roof structures of North-Western Europe can generally and concisely be described as common
rafter roofs with tie beams, without longitudinal bracing systems and having strut beams and collar beams in
different combinations to support the rafters.
121
structures (Sjömar, 1990, 1992, 1995, Thelin 2005) the 14th century, from raising Romanesque roof struc-
and newer, unpublished, recordings and surveys of tures to building roof structures without tie beams
about 40 church roofs. The information was put into adapted to high Gothic vaults. The two different sys-
a database together with dendrochronological dating tems were used in parallel for at least 150 years. In the
(www.timmerdraget.org/php_takkons4) and the report middle of this period, the second half of 14th century,
also refers to searches in this database. Roof structures it seems as if very little was built at all, probably as a
of the same type were raised during several hundred result of the Black Death. The use of the special hewing
years in Sweden. The oldest known roof structure is in technique, “sprättäljning”, that is found in all the 12th-
Herrestad church 1112+5d (Bartholin 1990). Inves- and 13th-century structures terminates completely in
tigation and dating so far show that as many as 24 this period.
medieval churches have roof structures that most likely
were raised in the 12th century. 3 DEFINITION OF STRUCTURAL TYPE
Several varying types of early medieval roof struc-
tures are found in Sweden. The different types can be In order to find and to define the characteristics of
divided into two major groups, which are found in two the structural types used in the early Middle Ages in
geographical regions. In southern Sweden the builders Sweden 70 roof structures have been used.The name of
raised roof trusses with tie beams, struts and later collar the church, which part of the church (nave or chancel)
beams. In northern Sweden roof trusses with passing the roof structures are standing on, their geographi-
braces and collar beams were raised instead (Sjömar & cal location and the type of structure are presented in
Storsletten 1993). Only four roofs of this type remain Table 1 and in Figure 1. Of the 70 roof structures 46
in Sweden. They have not been included in the defini- have been dated according to the database described
tion of structural type below since they are a different above. Unfortunately, many of these datings still have
structural type even though they are contemporary. At uncertainties. The uncertainties are due to a part of the
the same time, the roof structures from the 12th and the roof structure being dated but not enough investiga-
13th centuries show many similarities. The tools that tions of it having been carried out in order to certainly
were used are the same, axe and draw-knife “skave” ascribe the dating to the entire roof structure. Eigh-
and the wood is prepared in the same way. The crafts- teen have parts dated to before 1150, fifteen to before
men used a special hewing technique, “sprättäljning”, 1200, nine to before 1250, one to before 1300 and
whereby the axe cut back and forth along the side of three to before 1350. Of the 70 roof structures 21 have
the log, which makes a very characteristic herringbone been examined and documented by the author. Thir-
pattern. Some surfaces are also very finely planed with teen roof structures have not been examined by the
the draw-knife. The different parts in a roof truss are author in person but there is documentation in draw-
also put together in the same way, with straight lap ings and descriptions. For the final 36 roof structures
joints. The builders, in both northern and southern the information consists solely of written documenta-
Sweden, had the same basic idea of how the wood tion. This data cannot be considered as certain as in
should be treated and how a roof should be put together the other cases.
and raised. They seem to have belonged to the same To identify possible common characteristics of the
building tradition. early Swedish medieval roof structures several criteria
The very oldest remaining roof structures are tech- have been used: load-carrying system, elements and
nically fully developed and the craftsmanship is excel- shape, centre distance, slope of roof, connection with
lent. The builders, in both northern and southern wall, joints, outer roofing system and decorations.
Sweden, that raised the roof of a remote parish church All of the 70 roof structures carry the vertical (grav-
in the first half of 12th century mastered the materials itational) load cross-wise to the longitudinal direction
and the construction completely. The highly developed of the building from wall to wall, with trusses of the
skill of craftsmanship shown in the structures, the per- same design and size which are generally repeated con-
manence of the building tradition over several hundred secutively at a close distance, see Figure 2. This system
years, and the spread of the same building tradition in of carpentry is called common rafter roof or uniform
such a large geographical area indicate the existence scantling.
of an older (wood) building tradition in Sweden that The roof structures have tie beams that connect the
had existed for a long time before the 12th century. bases of the roof trusses and provide support for the
Dendrochronology was developed in the 1980s in horizontal thrust. They do not have a separate struc-
Sweden (Bartholin, 1994). There are references to pine ture for longitudinal stability. Longitudinal stability
and oak. From the dendrochronological dating we can is achieved by boarding (Fig. 10) below the outer
get a picture of how the roof structures developed over roofing, which possibly provides a stabilizing plate
time in Sweden, even if very few of the younger roof together with roofing material and in many cases con-
structures, connected to Gothic architecture, have been nects the roof trusses with stone gables. In the early
investigated. In principle, the builders changed during roof structures the tie beam is commonly embedded
122
Table 1. Churches with roof structures that have been part
of the study and their type.
∗
Province: ÖG Östergötland, VG Västergötland, G Gotland,
S Skåne, SÖ Södermanland, SM Småland, N Närke, B
Blekinge.
123
Figure 5. The roof structure at Garda church, Gotland.
It has one pair of strut beams between the tie beam and the
Figure 3. The tie beam and the rafter foot are completely rafters. The roof structure was once open and is decoratively
embedded into the masonry on top of the wall at the Kaga carved. Today there is a ceiling nailed to the tie beams.
church, Östergötland.
124
Figure 8. The image to the left shows the ridge beam in the
Garda church, Gotland, from above and the image to the right
show the steering plate in the Old Jät church, Småland.
Garda church, Gotland, having a slope of 59 and 53 Figure 11. Decorations in the nave roof structure of
degrees respectively, see Figures 5 and 7. The oldest Gökhem church, Västergötland.
dated roof structure has the flattest roof, see Figure
4. The timber churches generally have a steeper pitch
between 55 and 60 degrees. Different types of outer roof were used. In most
The tie beams of the early medieval Swedish roof cases they have been changed or replaced several
structures usually lie more or less embedded into the times. Common types today are shingle, wood, tiles,
masonry of the top of the walls, see Figure 3. There slate and tin. Below the outer roofing material there
is usually one wall plate situated at the outside of the is boarding. If the original boarding is preserved it is
wall, to which the tie beam is connected by a notched usually of cleaved and cut timber, see Figure 10. Later
joint, see Figure 9. The close relation between the wall types are sawn, and if they are replaced in modern time
and the tie beam leads to the assumption that the roof planed boarding is normally used.
structure has a vertical support along the width of the Studies of roof structures from the 12th century
wall which provides a shortening of the span since show that the church interior originally was open to the
the walls usually have a thickness of about one metre. roof (Sjömar 1992). An investigation of Hagebyhöga
The connection with the wall plate as well as with church, Östergötland, raised in the 12th century with
the masonry gives good possibilities for transferring walls of stone, shows that the interior both in the nave
horizontal forces by friction between the roof trusses and in the chancel originally had no ceiling; the roof
and the wall. This works as proved by examples where structure was visible. The wooden roof construction
the tie beams have been cut off for various reasons. probably gave the room a character like that in the
The situation is different for timber churches where Norwegian stave churches today. Several roof struc-
the tie beam lies notched over the top timber of the tures from the 12th century are given sculptural forms
wall, and it is probable that the tie beams also act to (see Figures 5 and 11). Some parts were probably
hold the timber walls together. painted. Other signs of open roofs are plaster that
125
remains on the walls above later vaults or ceilings, Liège, and Saint Pierre-de-Montmartre, Paris (Corvol-
see Figure 3. This means that the wood craft was an Dessert et al. 2002). Randerup church in Denmark
important means to create character in the interior, (Madsen 2007) and St Martin in Sindelfingen (dated
also in the stone churches. Later a flat ceiling made to 1132) in Germany (Binding 1991) are of the type
of planed boards was put up in Hagebyhöga church with only one pair of strut beams, like for example
before, finally, some hundred years later, stone vaults Garda church in Sweden, see Figure 5. There are also
were built. Traces of such wooden ceilings are also examples of roof structures with crossed strut beams
found in other churches and some still exist today. In in France and Denmark. The church of St Christophe
many Swedish parish churches vaults were built in the in Chabris, France, has a roof structure with six cross-
14th and 15th centuries. ing strut beams (Epaud 2003), similar to the old Jät
church (Fig. 7). Vester Nebel church in Denmark
has a roof structure with four crossing strut beams
4 COMPARISONS WITHIN A (Madsen 2007), similar to Asby church (Fig. 6). Both
NORTH-WESTERN EUROPEAN CONTEXT the Danish examples mentioned have in later times
been reinforced with collar beams.
When comparing the early medieval Swedish roofs The use of ridge beams (or ridge pieces) in some of
with other North-Western European roofs that were the Swedish roof structures is interesting, see Figure
raised about the same time, it is clear that they are 8. Courtenay (1993) claims that ridge beams cannot be
all part of the same context. Published surveys and found in England or France from 1050–1300 when dis-
studies, in Denmark (Aaman Sørensen 1995, Madsen cussing the late Norwegian stave churches that contain
2007), Norway (Storsletten, 2002), England (Smith, both ridge beams and purlins. Maybe the ridge beams
1975, 1981, 2006, Hewett, 1981, 1985, Walker, 1999, on the early medieval roof structures demonstrate a
Courtenay 1993), France (Corvol-Dessert et al. 2002, connection between the different Scandinavian roof
Epaud, 2003, Courtenay 1993) and Germany (Binding structure types.
1991, Lohrum 2006) show that the constructions in Courtenay (1993) thinks that lack of longitudinal
all these countries and Sweden have similarities. Also bracing is characteristic of the early medieval roof
eastern parts such as Romania seem to have some sim- structures. This is also described by Corvol-Dessert
ilarities (Szabó 2005). The studies show, just as in et al. (2002) and Binding (1991). Examples of diagonal
Sweden, that the roofs were raised by skilled crafts- elements on the inside of the rafters to provide longi-
men and that the crosswise load-carrying system of tudinal bracing are described by Fletcher & Spokes
the common rafter roof with tie beams on every truss (1964), Corvol-Dessert et al. (2002) and Madsen
seems to have been used throughout Europe for the (2007). This type of longitudinal bracing is used in
early medieval Romanesque roof structures. later medieval times in Swedish roof structures but
Most sources refer to the early medieval generally not in the early medieval types.
(Romanesque) roof structures as having collar beams The centre distance between the roof trusses, nor-
often in combination with strut beams (Fletcher & mally being less than a metre, as well as the slope
Haslop 1970, Corvol-Dessert et al. 2002, Courtenay of the roofs, normally between 40 and 50 degrees,
1993, Szabó 2005), but there are exceptions in France, seem to be generally similar across Europe for the
Belgium, Denmark and Germany of early medieval early medieval roof structures. Joints are generally
roof structures without collar beams similar to those described as being of the simple lap type (like the
in Sweden. Binding (1991) describes a similar devel- Swedish ones) or being notched or having dovetailed
opment to that in Sweden where the earlier roof profiles (Courtenay 1993, Fletcher & Spokes 1964,
structures lack collar beams. The collar beams seem Fletcher & Haslop 1970, Szabó 2005, Madsen 2007)
to appear in Germany in the second half of the 12th Courtenay (1993) describes the earliest
century. Courtenay (1993) and Epaud (2003) describe Romanesque roofs in Northern Europe as being open
the roof structure of St Gertrude at Nivelles, con- and visible from the inner space or intended to sup-
structed in 1046, which no longer exists, and Binding port a panelled ceiling. In contrast to the Swedish
(1991) describes the monastery church in Billigheim ones, these types of roof structures are rare and only
constructed in 1180/1190. Both St Gertrude and fragmentary preserved. Evidence of decorated roof
Billigheim are very similar to the roof structure at structures from early medieval times outside Sweden
Herrestad church in Sweden dated to 1112. It has is difficult to find. Binding (1991) says that the ques-
four angled struts connecting the rafters with the tie tion of open roofs in Germany is debated and many
beam, see Figure 4. The roof structure of St Denis roofs of churches without vaults have or show evidence
in Liège is also similar but has one extra pair of of earlier ceilings. There are later examples of open
struts (Corvol-Dessert et al. 2002). In France there roofs from the 13th century but they are of a different
are several examples of roof structures with vertical structural type. In England the tradition of open and
struts, as in the roof structure of Saint-Barthélemy, visible roof structures was developed in later medieval
126
times (Smith 2006). Smith also describes the use of a
“locking-plate” in Lincoln cathedral from around 1200
that seems to be similar to the Swedish steering plates.
The connection between the roof trusses and the
wall normally consists of wall plates in order to dis-
tribute loadings more uniformly to the top of the walls.
Examples of the use of masonry-embedded tie beams
and rafter feet exist in Germany, Denmark and France
(Binding 1991). One German example very similar
to the Kaga church (Fig 3) is St Ägidius in Mittel-
heim, which was lost in the 20th century. There is one
English example in the nave roof of Kempley, dated to
1120–1150 (Smith 2006). There are also examples of
roof structures totally devoid of wall plates (Courtenay
1993) although most seem to have double wall plates.
5 STRUCTURAL BEHAVIOUR
127
the strut beams act as hanging elements supporting the Revolution: The Art and Structure of Large-Scale Build-
tie beam instead of loading it, since the support of the ings. Massachusetts: The MIT Press.
rafters has been taken care of by the stiffer load path Epaud, F. 2003. Evolution des techniques et des structures de
provided by the collar beam, see Table 2. The use of charpenterie du XIe au XIIIe siècle en Normandie: Une
approche des charpentes par l’archéologie du bâti. Paris.
collar beams also gives an increasing horizontal thrust Fletcher, J.M. and Haslop, F.W.O. 1970. The West Range at
at the base of the roof that has to be transferred by the Ely and its Romanesque Roof. Archaeological Journal
joint between the rafters and the tie beam, see Table (126): 171–176.
2. The collar beam does not provide any wind bracing Fletcher, J.M. and Spokes, P.S. 1964. The Origin and Devel-
more than making the rafters act together. The strut opment of Crown-Post Roofs. Medieval Archaeology (8):
beams, on the other hand, can act as wind braces in the 152–183.
roof structures. Hewett, C. A. 1981. Church Carpentry, a Study Based on
Essex Examples. London: Phillimore.
Hewett, C. A. 1985. English Cathedral and Monastic Car-
pentry. Chichester: Phillimore
6 CONCLUSIONS Linscott, K. 2007. Medeltida tak, bevarade takkonstruk-
tioner i svenska medeltidskyrkor, Del 1: Rapport om
It seems that surprisingly many Romanesque roof kunskapsläget 2006. Göteborg: Göteborg University.
structures remain in Sweden. Of the very few North- Lohrum, B. 2006. Early Variations of the “Sparrendach”
Western European standing roofs that were raised in the “Pfettendach”-Region of South-Western Germany.
in the first half of 12th century, at least 10 still In Drdácký M. (ed.), European Research on Cultural
exist in Sweden. These, together with the rest of Heritage: State of the Art Studies Volume 4. Prague: Insti-
the well-preserved medieval roof structures, are an tute of Theoretical and Applied Mechanics Academy of
important and valuable cultural heritage in a European Sciences of the Czech Republic.
Madsen, P. K. 2007. Middelalderlige kirketagværker i
perspective. Sydvest- og Sønderjylland – eksempler på alder og typer.
The remaining roof structures from early medieval Aabøger for nordisk Oldkyndighed og Historie 2003.
times in North-Western Europe clearly show that they København: Det Kgl. Nordiske Oldskriftselskab.
are part of the same context and have many structural Sjömar, P. 1990, Medeltida takkonstruktioner i kyrkor, Stock-
similarities but also some differences. They are com- holm: Sveriges Kyrkor, Riksantikvarieämbetet.
mon rafter roofs that carry their load crosswise to the Sjömar, P. 1992. Romanska takkonstruktioner: ett värde-
longitudinal direction of the building; they have tie fullt och outforskat källmaterial. In Ullén M. (ed.), Från
beams on every roof truss and in many cases con- romanik till nygotik: studier i kyrklig konst och arkitek-
sist of similar inner elements. The sources indicate tur tillägnade Evald Gustafsson. Stockholm: Riksantik-
varieämbetet.
that the chronology of early roof structures without Sjömar, P. 1995. Romanskt och gotiskt – takkonstruktioner i
collar beams is true also outside Sweden. The collar svenska medeltidskyrkor. Hikuin(22): 207–230.
beam is shown to be an innovation that improves the Sjömar, P. and Storsletten O. 1993. Där Norge och Sverige
load-carrying system for vertical loads. mötas. Kirkearkeologi og kirkekunst: studier tilegnet
Additional studies are needed to complete the pic- Sigrid og Håkon Christie. Øvre Ervik: Alvheim & Eide.
ture of early medieval roof structures, not least in Smith, J. T. 1975. Norwegian Stave-church Roofs from an
Sweden. More systematic investigation and surveys of English Standpoint. Universitetets Oldsaksamling Arbok
the medieval roof structures in combination with dat- 1975/76/77: 123–136.
ing methods could provide a means to achieve a more Smith, J. T. 1981. Mittelalterliche Dachkonstruktionen
in Nordwesteuropa. Frühe Holzkirchen im nördlichen
comprehensive overview that would shed further light Europa: 379–390.
on this significant heritage of carpentry. Smith, J. T. 2006. English Medieval Roofs. In Drdácký M.
(ed.), European Research on Cultural Heritage: State of
the Art Studies Volume 4. Prague: Institute of Theoretical
REFERENCES andApplied MechanicsAcademy of Sciences of the Czech
Republic.
Aaman Sørensen, M. 1995. Middelalderlige tagværker, Storsletten, O. 2002.Takene taler: Norske takstoler 100–1350
særlig på Sjælland. Hikuin (22): 41–46. klassifisering og oprinnelse. Oslo: Arkitekthøgskolen.
Bartholin, T. 1994, Träets År. Kulturmiljövård 1994(1): 38. Szabó, B. 2005. Illustrated Dictionary of Historic Load-
Binding, G. 1991. Das Dachwerk auf Kirchen im deutschen bearing Structures. Cluj-Napoca: Kriterion Utilitas.
Sprachraum: vom Mittelalter bis zum 18. Jahrhundert. Thelin, C. 2005. Några medeltida takstolar: En delrapport
München: Dt. Kunstverl. i ett forskningsprojekt kring takfotens historiska utveck-
Corvol-Dessert, A., et al. 2002. Les charpentes du XIe au ling och konstruktiva verkningssätt. Göteborg: Chalmers
XIXe siècle: Typologie et évolution en France du Nord University of Technology.
et en Belgique. Paris: Centre des monuments nationaux/ Walker, J. 1999. Late 12th & Early 13th Century Aisled
Monum, Éditions du patrimoine. Buildings: A Comparison http://www.ccurrie.me.uk/vag/
Courtenay, L. T. 1993. Timber Roofs and Spires. In Mark, walker/aisled1.htm (2007).
R. (ed.), Architectural Technology up to the Scientific
128
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
D. Friedman
Old Structures Engineering, New York, US
ABSTRACT: As engineers and builders developed modern steel framing in the late nineteenth century, the
existing options for floors to span between steel beams were forms of masonry vaults. Many possible alternate
floors were developed in the United States, but few had rational bases for design. Testing programs put in place by
building officials in NewYork City promoted the use of certain systems in NewYork and, by providing a rationale
for those systems, nationwide. Ten systems are described and analyzed.
Conservation and analysis of early modern structure by thousands of years; it is far more interesting from
often begins with the main structural material: we the modern perspective that catenary systems were
have one set of techniques for steel-framed build- popular before beam systems.
ings, another for reinforced concrete, and another for There was no mechanism for the creation of national
those buildings which are architecturally modern but standards in the United States during 1880s and 90s
are constructed using traditional wood and masonry. because there were no national building codes or
This focus is inevitable but tends to obscure secondary authorities. Various floor systems were promoted by
structural materials and systems, which include the manufacturers and contractors, and reviewed by local
framing for facade ornament, structural adaptations building officials who often had no technical train-
for mechanical systems, and floor systems. This paper ing. Empirical load and fire-resistance analysis was
will address the original design and current-day analy- used by some insurance companies, local building
sis of floor systems used in the United States between departments, and engineering schools. After 1896, the
1890 and 1930 to fill in between iron and steel beams. NewYork City Department of Buildings and Columbia
Since iron-framed floors were first used in the University’s School of Mines together conducted hun-
late eighteenth century, floor systems have devel- dreds of tests on samples representing dozens of floor
oped in parallel with the iron-framing development. systems, beginning with systems already in use and
A wood-joist floor is most often associated with wood inviting floor-system manufacturers to submit their
plank flooring, and masonry vault structures provide products. Given the lack of national standards and
masonry floors, but iron or steel plates were not the rigorous testing regime, the results of these tests
used spanning between iron and steel beams. Instead, were influential far beyond New York and long after
combinations of other materials have been employed the tests ended. For example, standard E119 of the
between the beams. Because the designers and builders American Society for Testing and Materials, currently
who used metal-framed floors were usually interested used in the United States to determine fire-rating of
in increasing their buildings’ resistance to fire, floor floor assemblies, is closely related to the New York
systems have typically been made of masonry and testing standard and can be used to corroborate some
concrete. of the older results.
Floor systems can be divided into three broad This paper will review the floor systems that dom-
groups: arches, catenaries, and beams. These cate- inated “fireproof” construction in the Untied States
gories are, of course, the three basic methods of between 1890 and 1930 as well as selected less-
supporting vertical load across a horizontal span, but common systems. For each system, analysis using
unlike complex structures such as bridges where two current methods will be compared with available anal-
(or, rarely, all three) basic methods may be combined, yses from the era of construction and historic test
floor systems tend to show only one method. That data. Discussion will include implications for contin-
the earliest floor systems were arches is unsurpris- ued use and alteration in terms of design load capacity,
ing since masonry vaults predate structural iron beams detailing, and fire resistance.
129
1 HISTORICAL CONTEXT having a depth of not less than one and one-quarter
inches to each foot of span, a variable distance being
Structural steel technology and skeleton framing of allowed of not over six inches in the span between the
buildings matured in the United States between 1890 beams.” Some version of this was in the New York
and 1910. At the beginning of this period, reinforced code from 1891 to 1916. (Birkmire 1898, no author
concrete was just beginning to be used and was still, to 1899, no author 1901)
American engineers and builders, partly experimental. The first clause allows for traditional brick and
There was not a general consensus on reinforced con- stone vaulting, which was rarely used because of the
crete until the late 1910s. Given the late development weight of the materials and the intensive labor required
of concrete relative to steel, designers and builders had for construction. The second clause allows some
already addressed the need for fireproof floors before segmental terra-cotta tile arches. The third clause,
concrete became common. The use of bar-reinforced centered on the vague phrase “some equally good
concrete slabs never became the dominant solution for fire-proof material,” and led to a proliferation of pro-
floors; after the invention of composite metal deck in prietary systems from manufacturers looking for a
the 1950s, designers and builders rarely used any other portion of the lucrative New York construction mar-
system. As a result, from 1890 to 1960, floor systems ket. After several years of politicized review of various
were used which are now unfamiliar to most engineers systems, the city Superintendent of Buildings, Steven-
and builders. son Constable, arranged for a series of tests, beginning
As large steel-framed buildings were constructed in 1896 and performed by the Bureau of Buildings in
in the 1890s, designers and builders became aware concert with Columbia University. (Freitag 1899)
of an acute need for inexpensive, lightweight and
“fireproof ” structural floors. This need was not new:
1.1 Testing
the Chicago fire of 1871 and the Boston fire of
1872 – large-scale conflagrations that had between The building code enacted in 1898 in New York
them destroyed 18,000 buildings including much of allowed for brick and segmental tile vaulted floors.
both commercial downtown areas – had emphasized Other floors were addressed: “. . . or between the said
the need for buildings better able to resist fire. By the beams may be placed solid or hollow burnt-clay, stone,
1880s, various proprietary forms of terra-cotta tile arch brick, or concrete slabs in flat and curved shapes, con-
floors had been patented and were in use, although all crete or other fireproof composition, and any of said
were functionally similar. materials may be used in combination with wire cloth,
The word “fireproof,” as used at that time, meant expanded metal, wire strands, or wrought-iron or steel
what is now called “fire resisting:” a building or a bars; but in any such construction and as a prece-
structural element that could withstand some degree dent condition to the same being used, test shall be
of exposure to fire without failure. In this paper, “fire- made . . .” (no author 1899)The test defined in the code
proof ” will be used in that sense, not in its literal consisted of three stages of loading on a completed
meaning of withstanding any exposure to fire. panel of the floor in question: first, with the 7.2 kPa
As new ideas for fireproof floors were devel- load in place, the floor was heated so that an average
oped, building officials and insurance companies were temperature of 927 C was kept for four hours; second,
forced to consider how floors built with new, often pro- with the load in place, the floor was cooled by pres-
prietary and patented technology, could be reviewed surized streams of water meant to imitate fire-fighters’
for approval. In the end, the various parties, includ- hoses; and third, the floor was reloaded to 29 kPa. The
ing engineers, companies selling proprietary systems, physical integrity of the floor was observed at all times
and insurance companies, cooperated with officials to and deflection measurements taken during the third
create standardized tests. (Hill 1895) Many of these loading stage. (no author, 1897)
tests were performed in the late 1890s and early 1900s Because the New York code was interpreted lit-
by various agents under the supervision of the New erally, each variation on a system had to be tested
York City Bureau of Buildings, in accordance with the separately. A floor system had to be tested for each
requirements of the New York City Building Code at possible load and span combination for which it was
that time. (no author 1905) sold, leading to over 190 tests between 1895 and 1915.
The portion of the code that defined floor systems (Perrine & Strehan 1915)
read: “All brick or stone arches, placed between iron Similar tests were used in other cities, by insurance
floor beams, shall be at least four inches thick and companies in setting fire rates for buildings using new
have a rise of at least one and a quarter inches to each floor systems, and by builders and inventors attempt-
foot of span between the beams. Arches over five feet ing to prove the reliability of their systems to owners
span shall be properly increased in thickness . . . or the and local building officials. The fireproofing aspect
space may be filled in with sectional hollow brick of of the tests is obvious, but in many cases there was no
burnt clay or some equally good fire-proof material, structural design for the floors beyond the empirical
130
Figure 2. End-construction Flat Tile Arch (Birkmire 1898).
131
Figure 7. Rapp Floor: wood sleepers to attach finish floor-
ing, fill (“concrete”), brick, and inverted Ts. (Birkmire
Figure 4. Roebling Arch: wood sleepers for finish flooring, 1898).
concrete fill, wire-mesh arch, hung ceiling. (Birkmire 1898).
132
anchored at support.” (no author 2004, no author
2007a)
The load analysis for the different floor types that
follows is based on current codes except as noted.
Modern code analysis is somewhat simpler with
regard to the fireproofing capacity of the old floors.
Under currentAmerican practice, fire ratings for struc-
tural assemblies are determined using tests defined
in the standard specification ASTM E119. The stan-
dard has been published and periodically revised by
Figure 10. Roebling Flat Slab Floor, showing strap rein- the ASTM since 1917 and carries no authority of its
forcing twisted at the beam top. (Freitag 1899).
own, but is regularly adopted into building codes and
other statutes. (no author 2007b)
The Roebling Flat Slab Floor was basically similar The provisions of E119 are similar in many respects
to the Columbian floor, except that the reinforcing was to that of the New York City tests. The floors
rectangular bars set vertically and twisted horizontally are to be loaded “to simulate a maximum load
to rest on the floor beams. (Fig. 10) (Freitag 1899, Hool condition . . . under nationally recognized structural
1913) design criteria,” the heating is to follow an upward
curve from 538 C at five minutes to 1093 C at 4 hours
3 RE-USE AND ANALYSIS (and up to 1260 C at eight hours if a rating past four
hours is sought), the floor is to be cooled by a pres-
Since the original use of these floors depended only surized hose stream, and the test report is to include
on empirical testing, analysis came after the fact. For information on damage and deflection. Since (a)
the arch floors, there was rarely analysis at all until the the maximum design load on most floors is less than
floors fell out of use, while justifications for the beam the 7.2 kPa used in the New York tests and (b) there
and catenary floors have changed over time. are no standard uniform loads of 29 kPa as was used in
the New York reloading test, the current loading crite-
3.1 Current codes
rion is less stringent than the New York tests. The total
During the course of the twentieth century, numerical amount of heat to be absorbed in the early stages (as
analysis replaced the empirical tests remaining in the measured by the area of the time/temperature curve)
structural design. Steel beam design had been ratio- is less in the ASTM tests than in the New York tests,
nalized in the nineteenth century, and concrete beam the New York tests are more conservative for systems
design after 1900. At the time of the floor testing, all with low fire-ratings. At four hours, the total energy
American building codes were local, created by either is nearly identical between the two tests, so that a sys-
municipalities or states. Regional and national codes tem that passed the New York test would be assigned
were only established later, and have only allowed a four-hour rating using the ASTM test. Only when
analysis-based structural design. In the case of New the tests are continued past four hours does the total
York, the 1916 building code removed the testing energy in the ASTM test exceed that of the New York
requirement in favor of analytic design. This code was test. As few floor systems are now required to have rat-
revised but kept in use until 1968, when an entirely ings greater than three hours, it can be generally stated
rewritten code was put into effect. All references to that any floor that passed the New York test has effec-
tile arches or other masonry floors were removed, but tively passed the ASTM test and can be considered to
draped-mesh slabs remained as “Short-span concrete have a four-hour fire rating if it has been maintained
floor and roof construction supported on steel beams.” in its original state.
The formulas for minimum allowable slab thickness The most significant difference between the New
and wire stress were the same as the old code, and York tests and the ASTM test is one of perception: the
values were given for stone concrete and unspecified- older tests were used as proof of both fire-resistance
material “lightweight aggregate concrete.” This code and structural load capacity, while the current tests
remained in effect, with revisions, until July of 2007, are used only to determine fire ratings for assemblies
when it was superseded by a local version of the 2003 assumes to have been already numerically designed
International Building Code, with a one-year overlap for structural loads. This gap is less of a problem for
until the summer of 2008.The new code retains one last existing buildings than it might appear, since the cur-
vestige of the old floors: a statement that wire mesh rent building codes allow existing structure to be load
reinforcing “is permitted to be curved from a point tested to prove capacity, in a manner reminiscent of
near the top of the slab over the support to a point the historic tests.
near the bottom of the slab at midspan, provided such The use of historic test data, common among engi-
reinforcement is either continuous over, or securely neers, has been given a government imprimatur in the
133
form of a guide to fire ratings on historic structures A modern simplified formula for arch thrust used by
compiled by the US National Institute of Building the Brick Industry Association is H = 3 WS/8d, where
Sciences and published by the federal Department of H is the thrust, W is the total uniform load, S is the span,
Housing and Urban Development. (no author 2000) and d is the arch depth. (no author 1986) Again, the
This guide is a compilation of historic and modern analysis is for a unit width of arch. In this case, the main
tests on various types of building elements, including simplifying assumption, represented by the “3” in the
floor systems. Gypsum slabs, terra-cotta tile arches, formula, is that the thrust line is within the middle
and tile and concrete rib systems are all included and third of the arch height. This is more conservative than
assigned fire ratings of up to four hours. While the the older formula in that it increases the thrust force
guide does not comprehensively address all of the by confining its path. Using the previous example, the
floor systems of interest, it does provide an official thrust is 34 N/mm and the net compression on the webs
opinion on some and it does show that the reuse of his- is 0.93 MPa, which is still within the capacity of clay
toric test data can be an officially-accepted method of masonry.
research. The formulas for segmental arches do not require
the complicated assumptions regarding the location of
the thrust curve: it is taken within the curve of the vault.
3.2 Arch analysis Since the floor vaults based on segmental curves typi-
cally support fill above, a thrust line that extends past
Arch analysis is one of the oldest problems in structural
the vault material on the top side and passes through
design. In ordinary masonry practice, symmetrical and
the fill may still be viable.
uniformly loaded arches are rarely dependent on the
Using the Guastavino vault as an example of a
compressive strength of the material, but the terra cotta
segmental-arch floor, with a span of 1830 mm and
tile used in the floors was often weaker than stone or
a rise of 150 mm, the average total load is 7.9 kPa,
ordinary brick, and was used in thin webs vulnerable
with the extra dead load resulting the deeper fill over
to local concentrations in stress.
the arch ends. The basic thrust formula, H = WS/8d,
Among the various formulas developed for masonry
gives a thrust of 22 N/mm. Unlike terra cotta vaults, the
arches exists a class of simplified formulas for minor
stress in a Guastavino vault is evenly distributed along
arches – those with short spans and low rises. In such
the solid masonry section, so with a vault thickness of
cases, the formulas also include the assumptions for
75 mm, the compressive stress is 0.29 MPa.
a flat arch of uniform loading and a specific location
Finally, the Roebling arch is the simplest of the type.
for the thrust line.
The minimum arch depth is the distance from the floor
A simplified formula suggested while terra-cotta
to the top of the arch center, and the material is stone
floors were still in use is H = WS / 8D, where H is
concrete, with a compressive strength in the range of
the thrust, W is the total uniform load, S is the span,
14 MPa, so an analysis similar to the Guastavino floor
and D is the effective rise of the arch. This formula
shows a gross over-capacity.
assumes that the thrust line curve was 61 mm less than
the height of the terra cotta blocks, as D is the total
arch depth minus 61 mm. (no author 1919) Note that
the load and analysis are for a unit width of arch. In this 3.3 Catenary analysis
case, the main simplifying assumption, represented by Analyses for catenaries with fixed supports were avail-
the use of D rather than the arch depth is that the thrust able in the late nineteenth century, both as general
line curve is proportional to the block depth excluding formulas based on statics and simplified formulas that
the lower portion of the blocks which extend below the became part of building codes and manufacturers’ rec-
steel beams. For example, a typical application span- ommendations during the first half of the twentieth
ning 1830 mm between beams and consisting of wood century.
flooring, 125 mm inches of cinder fill, 254 mm deep One of the simplified catenary formulas that was
terra cotta blocks making up the flat arch, 12 mm of publicized by the wire-products division of United
plaster on the block soffits as a ceiling, and an office States Steel was also incorporated in the New York
live load of 2.9 kPa, had a total weight, including steel City Building Code in 1916 (no author 1944) This for-
framing, of 6.8 kPa. This translates to an arch thrust of mula is W = 3CAs / L2 where W is the total allowable
15 N/mm to be carried by tie rods from beam to beam. floor load, L is the beam-to-beam centerline spac-
A typical terra cotta block had 19 mm thick webs ing , As is the unit wire cross-sectional area, and
at 100 mm spacing. Since the horizontal webs are ori- C is a constant representing the maximum allowable
ented incorrectly to carry the arch thrust, the thrust wire stress and equal to 138 MPa. The referenced ver-
must lie entirely within the vertical webs. With an sions of this formula refer to stone or cinder-concrete
effective web area of 254–61 or 193 mm by 19 mm, the slabs; however, the results also agree closely with the
compression in the terra cotta is .41 MPa, well within stated capacities of the Metropolitan Floor’s gypsum
the allowable range for clay masonry. slab. It should be noted that the maximum allowable
134
tension in the wire of 138 MPa is very low for steel moment (the distance from the centroid of the tri-
wire produced after 1900. The minimum yield stress angular elastic compression block to the centroid of
expected in historic wire systems has been conserva- the reinforcing) is jd, where j = 1 − k/3. The resisting
tively established as 345 MPa. (no author 1981) Given moment is the lesser of the two calculated moments
that current reinforced-concrete codes have combined Ms = As fs jd and Mc = fc kjbd2 /2, where Ms is the max-
safety factors in the range of 1.5 to 1.8, it is evident imum resisting moment if the slab is under-reinforced
that there is excess capacity in the wire in the original and steel yield controls, As is the reinforcing area in
designs. square inches per foot, fs is the allowable tensile stress
For example, in a 100 mm-thick draped-mesh slab in the steel, Mc is the maximum resisting moment if
with a 1830 mm span, the total load based on wood the slab is over-reinforced and concrete crushing con-
flooring, the cinder-concrete slab, 12 mm of plaster on trols, and fc is the allowable compressive stress in the
the soffit as a finish ceiling, and an office live load concrete.
of 2.9 kPa was 5.1 kPa. With mesh reinforcing of 8- For example, the code-enforced maximum stress
gage (4.2 mm diameter) wire at 75 mm spacing, the values would show a 100 mm stone-concrete slab
maximum allowable load is 41 KPa. with 6 mm diameter bars at 150 mm on center, and
A more basic formula derived from statics by spanning 2440 mm, to have a maximum allowable
approximating the catenary curve as a parabola (a rea- moment of 4.9 kN-m/m, or a maximum total load of
sonable assumption since the actual wire curve is not 6.5 kPa, compared to the 7.2 kPa allowed from testing
particularly accurate) is T = (wsL2 / 8 h) + wsh where the same floor. This is not a hypothetical compari-
T is the tension in a wire, w is the combined dead and son: the Monier Floor System, which was essentially
live load, s is the wire spacing, L is the span, and h is a numerically-designed, modern reinforced concrete
the wire sag. By setting T equal to the allowable ten- slab as used in the in Europe, was tested for use in
sile stress (Ta ) multiplied by the wire cross-sectional New York by 1900. If the current American Concrete
area (As ), the maximum floor load can be deter- Institute code and ultimate-stress theory is used on the
mined as W = 8 hTaAs / (sL2 + 8 sh2 ). In the previous same slab, assuming the 14 Mpa concrete often used
example, using the 138 MPa maximum stress in the in that era and a steel yield stress of 138 MPa based on
wire and assuming a wire sag of 75 mm for the 100 mm the weakest available reinforcing steel bars, the maxi-
slab thickness, the allowable load per square foot mum allowable moment is 8.5 kN-m/m, and, assuming
is 25 kPa. that the dead load to live load ratio is approximately 1,
In a Metropolitan floor example using the simpli- a maximum unfactored total load of 8.0 kPa.
fied formula, with pairs of 12-gage (2.8 mm diameter) The Columbia and Roebling Flat Slab floors can
wire at 25 mm spacing and a 1830 mm span, the total be analyzed like any other reinforced slab as long as
load capacity is 36 kPa. Using the basic formula, the certain restrictions are observed: the bars must be con-
total load capacity is 22 kPa. verted to a cross-sectional area per unit width of slab,
In short, the catenary floors show the excess capac- the slab strength must be adjusted to match the mate-
ity required to pass the load test. rial, and the lack of deformations must be accounted
for by a check on the development length of the bars.
The Expanded Metal Floor can be analyzed using
3.4 Beam analysis ordinary reinforced-concrete formulas and converting
the cross-sectional area of the expanded-metal sheet
Beam theory is generally more complex than arch or
into the bar equivalent.
catenary theory as long as fixed supports are assumed.
The analysis of the Rapp Floor, a beam-type floor
In addition, the design of reinforced concrete in the
that did not rely on reinforced concrete, is simpler. The
United States at the time of the testing was ham-
bricks spanning 200 mm and the light-gage Ts span-
pered by over-conservative allowable stresses. When
ning the beam-to-beam spacing can be checked for
the New York code was amended in 1911 to allow use
maximum stress using a simple beam formula.
of reinforced-concrete floors proved by analysis rather
than by testing, the maximum compression in the con-
crete was limited to 4,5 MPa, the maximum tension 4 CONCLUSIONS
in the reinforcing was limited to 110 MPa, and n, the
ratio of the elastic moduli, was fixed at 15. (Waite The simplest result of the analysis is the conclusion
1914) Only elastic concrete theory was in common that these floors can be demonstrated to pass mod-
use at the time, and using the notation common for ern requirements for fire resistance and load capacity.
that theory, the location of the neutral axis from the In the course of renovation projects, many engineers
compression face of the slab is kd, where d is the depth and contractors prefer the removal of unfamiliar and
from the compression face of the slab to the centroid of archaic structural elements rather than their reuse, even
the reinforcing and the ratio k = (2 pn + p2 n2 )0.5 − pn, if no damage is visible. The explanation for this behav-
where p is the reinforcing ratio.The arm of the resisting ior is often “better to be safe than sorry,” as if the reuse
135
of existing structure that has functioned properly for no author. 1919. Pocket Companion for Engineers, Architects
decades is somehow unsafe. It must be emphasized and Builders Containing Useful Information and Tables
that unknown structural capacity and fireproofing are Appertaining to the Use of Steel. Pittsburgh: Carnegie
not excuses for the wholesale demolition of historic Steel Company.
no author. 1944. American Welded Wire Fabric for Concrete
fabric that often takes place, leaving a historic facade Reinforcement. Pittsburgh: United States Steel Corpora-
covering an essentially new building. It is incumbent tion.
upon the engineers involved to investigate unfamil- no author. 1981. Evaluation of Reinforcing Steel Systems in
iar structures and find methods by which they can be Old Reinforced Concrete Structures. Chicago: Concrete
analyzed. Reinforcing Steel Institute.
This is not to say that there are no difficulties in no author. 1986. “Technical Note 31a – Structural Design of
reuse once an archaic floor is analyzed. There may be Brick Masonry Arches.” Reston, Virginia: Brick Industry
damage to the floors, particularly with the more frag- Association.
ile systems such as tile arches, that reduce their load no author. 2000. Fire Ratings of Archaic Materials and
Assemblies: Guideline on Fire Ratings of Archaic Mate-
capacity or fire protective abilities. Modern seismic rials and Assemblies. Washington, D.C.: U.S. Department
analysis of a building frame may depend on diaphragm of Housing and Urban Development, Office of Policy
action of the floor, which may or may not be available Development and Research.
depending on the details of original construction. The no author. 2004. Building Code of the City of New York. New
need for a diaphragm implies a second investigation York: New York City Department of Citywide Adminis-
into details and a second analysis, not automatic dis- trative Services.
qualification. Finally, alterations accompanying reuse no author. 2007a. New York City Building Code. New York:
may require special details, such as providing new New York City Department of Citywide Administrative
anchorage for catenary floors when new openings are Services.
no author. 2007b. “Standard Test Methods for Fire Tests of
cut for circulation or HVAC shafts. Some of the floors Building Construction and Materials,” ASTM E119–07a.
are particularly vulnerable to damage during alter- West Conshohocken, Pennsylvania: American Society for
ation: theTs in a Rapp floor can shift when an adjoining Testing and Materials.
portion of the floor is removed. Birkmire, W. 1898. The Planning and Construction of High
In a broader sense, the continued viability of many Office Buildings. New York: John Wiley & Sons.
of these floors emphasizes that the products of empir- Buel, A.W. & Hill, C.S. 1906. Reinforced Concrete, 2nd
ical design can still be used today. The rise of “scien- edition. New York: The Engineering News Publishing Co.
tific” numerical analysis in the twentieth century can Collins, G.R. 1982. “Guastavino y Moreno, Rafael and Guas-
obscure the value of older forms of engineering design, tavino y Esposito, Rafael,” Macmillan Encyclopedia of
Architects. New York: Macmillan Publishing Co.
even as we still rely on some of the methods and data Guastavino, R. Jr. 1892. “Hollow Cohesive Arch,” United
produced by those older forms. States Patent 471,173. Washington DC: United States
Patent and Trademark Office.
Freitag, J.K. 1921. Fire Prevention and Fire Protection, 2nd
REFERENCES edition. New York: John Wiley & Sons.
Freitag, J.K. 1899. The Fireproofing of Steel Buildings. New
no author. 1889. “A New System of Fireproof Floor Construc- York: John Wiley & Sons.
tion.” Engineering News, November 9, 1889: 434–35. Hill, G. 1895. “Tests of Fire-Proof Flooring Material,” Trans-
no author. 1895. “The Metropolitan Concrete and Wire actions of the American Society of Civil Engineers 34:
Floor.” Engineering News, November 14, 1895: 333. 542–568.
no author. 1897. “Comparative Standard Fireproof Floor Hool, G. A. 1913. Reinforced Concrete Construction,
Tests of the New York Building Department.” Engineer- Volume II: Retaining Walls and Buildings. New York:
ing Record, September 18, 1897: 337–40; September 25, McGraw-Hill.
1897: 359–63; October 2, 1897: 382–87; October 9, 1897: Perrine, H. & Strehan, G. 1915. “Cinder Concrete Floor
402–405. Construction Between Steel Beams.” Transactions of the
no author. 1899. Building Code of the City of New York. New American Society of Civil Engineers. 79: 523–621.
York: New York City Bureau of Buildings. Waite, G.B. 1914. “Cinder Concrete Floors.” Transactions of
no author. 1901. Building Code of The City of New York. New the American Society of Civil Engineers. 77: 1773–1823.
York: Department of Buildings.
no author. 1905. Report of the Bureau of Buildings of the City
of NewYork for the Borough of Manhattan for the Quarter
and Year Ending December 31, 1904. New York: Martin
B. Brown Co., 1905.
136
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: In 1995, a routine inspection carried out by the Restoration Department of Federal University
of Rio de Janeiro disclosed instability of wood floors in a room located in a Brazilian 19th century building.
This paper presents a concise report of this structural degradation. It focuses mostly on the project of structural
restoration of areas that brought about the practice works to be implemented in a historical building listed by
the Brazilian Historical Artistic Heritage Institute. Not limiting this study to a description of the strengthening
work stages, this research analyses the actions adopted to restore the damage caused by termites in wood beams,
trying to evaluate the adequacy in this specific work of the guidance of international charters of architectural
heritage. The main goal of this paper is to analyse the criteria that oriented the feasible restoration and the limits
of this intervention.
137
that restoration does not falsify the artistic or his- of the production methods, but otherwise. Hereupon,
toric evidence”. In addition to that, in the article 10: he points out two situations in which an intervention
“Where traditional techniques prove inadequate, the could be considered a counterfeit:
consolidation of a monument can be achieved by the
use of any modem technique for conservation and (1) Production of a purpose with the specific inten-
construction, the efficacy of which has been shown tion of deceiving either the period, the material
by scientific data and proved by experience” (In: coherence or the author, or
http://www.icomos.org). (2) Diffusion of an object, not necessarily intended
The Italian Charter of Restoration of 1972 to deceive as if it were an authentic work, from
approaches a prevailing architectonic restoration stan- that period; by non-authentic authors and produc-
dard: “the replacement and possible integration of ers different from the genuine one, and made out
ornaments [. . .] ought to differ from the existing ele- of distinct materials. In other words, it has been
ments, implying the material and surfaces to stand out, created with no counterfeit intention, but meant to
making them distinguishable from the originals”. It be divulged as the original.
also mentions the interventions must be implemented
for fear no doubt is left and by the most discrete manner According to Brandi, to get the counterfeit judge-
(In: http://www.icomos.org). ment, it is necessary to prove the swindle.
The General Conference of UNESCO at Nairobi This issue has been developing, oftentimes being
in 1976 recommends researches and systematic sur- debated in international congresses and seminars. In
veys on the use of modern techniques in conserva- Japan, November 1994, the Nara Conference, based
tion works, as well as studies about craftsmanship upon the Venice Charter, has discussed the authen-
techniques, important to the building safeguard. ticity issue linked to the intrinsic heritage value and
the cultural identity of the region in which it is
placed.
3 THE MATTER OF AUTHENTICITY IN Following the international stream, Brazil has
INTERVENTION hosted the V Regional Meeting of ICOMOS in 1995.
A regional document on authenticity of the southern
The legitimacy of the restorative intervention in his- cone countries was the result of this assembly. Brasilia
torical buildings was debated since the very beginning Charter established that authenticity of values is man-
of restoration theories. Authors like Boito stated as a ifest, supported and preserved in the truthfulness of
fundamental concept the distinction of the new parts the heritage that we receive and transmit. Interven-
from the existing ones, as well as emphasizing the pru- tion must rescue building or site character, underlining
dence of limiting intervention actions to the minimum its authenticity without transforming its essence and
required. All material to be added or replaced should balance, avoiding extraordinary actions and enhanc-
have a different feature, in the type of material to be ing its values. The Charter avows the quality of the
used, or should bear a plaque, indicating the successive treatments as fundamental and all new elements intro-
architectural alterations. The new parts to be inserted duced must be both reversible and harmonious with the
should exhibit simple lines, being perfectly balanced whole.
with the ancient building. The issue of authenticity is rooted in the values
Later on, the Venice Charter has established that the attributed to the heritage. The judgement of value
new elements ought to differ from the existing parts, was broadly discussed in “The modern cult of mon-
and the same criterion was sustained by the Italian uments” by Riegl in 1903. This matter was the main
Charter of Restoration. point of Nara Conference, in 1994, related to scientific
This differentiation theory was strengthened in later studies on cultural heritage, conservation and restora-
congresses, aiming at the visibility of the interven- tion planning and procedures for the World Heritage
tions to be carried out, refusing the idea of utilisation Convention.
material and techniques similar to the existing ones, The Nara Document claims that the authenticity
preventing historical fraud. emerges as the chief criterion when it comes to values
In his book Theory of Restoration written in 1963, attribution. “The understanding of authenticity plays a
Brandi (2004) mentions the judgement of counterfeit. fundamental role in all scientific studies of the cultural
The author refers to the essential attributes the subject heritage, in conservation and restoration planning”.
should possess, but is not able to earn: the falseness is (http://www.international.icomos.org).
in the judgement and not in the purpose. One of two The interventions carried out in the University
identical objects may be considered a fake, according Palace of UFRJ – Federal University of Rio de Janeiro
to its intention. will be herein analyzed based on these concepts, more
Pursuant to him, the distinction between copy, imi- specifically the floor restoration of three rooms located
tation and counterfeit is not in the specific diversity in this building.
138
Figure 2. Assessment discloses a drastically reduced section
Figure 1. University Palace. Photo: Soares, B., 2007. of some beams. Photo: Hoirisch, M., 1995.
139
Figure 4. Fixation of floor beams in Science and Culture
Forum coordinator cabinet. Photo: Hoirisch, M., 1996. Figure 6. Golden room after restoration. Photo: Soares, B.,
2007.
140
The necessary strengthening in some of the tim- The key point about the restoration proceedings def-
bers was accomplished by the “sandwich” technique, inition and the materials to be utilized depends upon
putting two new pieces parallel to the pre-existing the limits of the intervention; in other words, up to
ones. what extent is it possible to intervene in a heritage?
The remaining doubt refers to the highly perishable This intervention ought to be limited by the preser-
feature of the wood, mainly when utilized in humid vation of a historical building, its values, authenticity
tropical weather (Rio de Janeiro relative humidity is and identity of the local cultural heritage where the
above 70% all year round). If this high humidity level monument belongs. One must dedicate special care
foments the termite infestation why had not the tim- to the values to be preserved and what would be the
bers been replaced by a more long lasting material, essence of restoration practice.
like steel? Regarding our heritage actions, this is a concept
The option for metallic beams could pinpoint the and paradigm changing era. The target should not only
exact intervention period, and the advices of the her- be the material safeguarding, but mainly the historical
itage charters would be abided. A future survey could building image and authenticity, as a cultural identity
allow the identification of steel utilization as dating offspring.
from our times. An extra benefit would consist of the
prolongation of the inserted beam durability, when
compared to the pre-existing one. By the way, one REFERENCES
of the main goals in restorative interventions is to
hinder new interventions on a short span. The replace- Boito, C. 2003. Os restauradores. São Paulo: Ateliê Editorial.
Brandi, C. 2004. Teoria da restauração. Trad.: Beatriz M.
ment of wood beams by metallic ones could be threat
Kuhl. São Paulo: Ateliê Editorial.
a future, for the utilization of modern materials and Hoirisch, M. 2005. Restauração das Estruturas de madeira
techniques would make a “patchwork” out of the em prédio Histórico da Universidade Federal do Rio de
heritage building. Losing the constructive original fea- Janeiro. In: 4th International Conference on the behavior
tures dated from its erection period would mean losing of damaged structures. Anais do DAMSTRUC 2005, João
its authenticity as well. Pessoa, p. 1159 a 1178.
Concerning its physical aspect, a primary mapping Prudêncio, W.J. & Ribeiro, R.T.M., 1998. As bases éticas
of the pre-existing wood pieces had been performed, da restauração do patrimônio cultural. In: Del Rio, V.
before the removal of the floor wood clapboards, (org.). Arquitetura: Pesquisa & Projeto. Rio de Janeiro:
Pró-Editores.
allowing its fixation to the previous position. The new
Ribeiro, R.T.M., Pimentel, V.L., Coelho, C.M.T.,
inserted wood pieces followed the original pattern, the Andrade, I.E. & Melo, C.M.S. 2005. Olhares sobre o
usual proceeding in these cases. patrimônio edificado: o conceito de valor. In: XXIII Sim-
pósio Nacional de História: Guerra e Paz / Associação
Nacional de História – ANPUH. Anais. . . XXIII Simpósio
6 CONCLUSION Nacional de História: Guerra e Paz. Londrina: Editora
Mídia, 2005.
The current theories tell the insertions from the whole, Riegl, A. 1999. El Culto Moderno a los Monumentos.
and were applied to this restoration work. The mate- Madri: Visor Dis, S.A.
http://www.icomos.org/
rial had not been changed, likewise preserving the
http://www.nardinirestauro.it/download/pdf/carte/Carta_
authenticity of the building. Moreover, the durability restauro%20_1972.pdf
of the restoration practice carried out may disagree http://www.international.icomos.org/naradoc_eng.htm
with future conservation decisions.
141
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
B. Szabó
Technical University of Cluj-N., Romania
I. Kirizsán
Built Heritage Research and Design Centre UTILITAS, Cluj-N., Romania
ABSTRACT: The safety of the historic load-bearing structures – being a part of the general issues regarding the
load-bearing structural safety – is a very topical problem for the sustainable management of the built environment.
Being to a great degree related to the objective (the characteristics of the historic load-bearing structures, as well
as of the historic and/ or current structural environment) and subjective premises of providing safety (social-
political conditions, the level of technical knowledge, research possibilities, quality control system, available
materials, aptitude etc.), the historic load-bearing structures need interventions that improve their safety and
preserve their intrinsic heritage values as completely as possible. If the “updating” of their safety requires a
technical-economic optimization, then this optimization is greatly influenced by the objective and subjective
premises of providing safety and by the condition of a minimal loss from the heritage values.
143
Figure 1. The structural sensitivity of the masonry to Figure 2. The structural sensitivity of the timber structure
moisture, freezing cycles. to biological decay.
3.1.1 The characteristics of the historic structural environment differ from those of the historic
load-bearing structures environment:
These characteristics are well-known, including their
share in the safety of historic buildings; for example: – the artificial corrosion factors changed essentially
both in quality and quantity – the historic build-
(a) the masonry structural sub-units: ings being extremely vulnerable to some of these
– are rigid to gravitational actions, if these lead to actions;
compression and if due to these no differential – the level of the ground water was essentially modi-
settlements are produced; fied in the 19th century, influencing the foundation
– have minimum reserves for taking over the tension, conditions and the historic foundations to a great
thus they are very sensitive to dynamic actions; degree;
– are sensitive to climatic actions, moisture, freezing – the impermeable flooring and pavement basi-
cycles etc.; cally changed the hydrostatic balance of the load-
– are sensitive to natural or artificial corrosion bearing structural environment surrounding the
factors; buildings, increasing the humidity (and dissolved
– the fatigue, the seismic energy-absorption of the salts) absorbed by the load-bearing structures due
masonry mortars and the consequences due to to the lack of horizontal waterproofing;
corrosion can be tested in the mass of the mason- – the vibrations due to traffic are very damaging,
ries with difficulty, the evaluation of their state especially for the masonry mortars – operating
parameters being possible only in quality. almost constantly, making also the efficiency of the
(b) the timber structural sub-units: interventions questionable;
– are load-bearing structures, which work in an elas- – the vibrations due to industrial or household appli-
tic limit state, being suitable for both static and ances have an effect similar to the one of the
dynamic actions; vibrations due to traffic.
– are subject to biological decay, some of the dam-
ages – like the ones caused by the moisture of
the supporting structure of the floors – causing 3.1.3 Special historic conditions relating to the
an immediate, unforeseeable collapse. historic load-bearing structures
In the course of history the ecclesiastic constructions
(c) being hundreds of years old, the durability of the
(especially the churches) have been built according
historic load-bearing structures is a certainty.
to prototypes. These «typical projects» are made for
(d) the state of the historic load-bearing structures,
exactly formulated functional requirements, but are
the safety reserves are continuously changing, the
elaborated also (more or less consciously) for technical
periodic re-evaluation of safety being compulsory.
requirements, demanded by the climatic, seismic, cor-
3.1.2 The historic and/or current characteristics rosive historic conditions etc. of the place, where the
of the historic load-bearing structural prototype was created. The prototype being generally
environment propagated – for example by the Cistercian monks –,
These are of great significance, influencing to a both the functional and technical characteristics are
great degree the safety of load-bearing structures. entirely transmitted, without being adapted to the local
The current characteristics of the historic load-bearing technical climatic or soil conditions (shrinkable clays,
144
Figure 4. Historical skill in England.
145
(d) skill:
– the traditional technologies are developed for
the processing of the local materials (there are
situations, in which the traditional processing
offers a higher quality);
– the manual operations present a development
in time, going into decline only after the great
industrial revolution from the 19th century.
146
Figure 6. The use of the traditional technologies, interven-
tion on a fractured masonry vault.
Figure 7. The use of the traditional technologies, interven-
tion on a historic timber roof construction.
load-bearing structures; we have to choose
between preservation, restoration, retrofitting only with ”virgin lands”), the behaviour of the
or re-construction – only that specialist can historic timber – rheological aspects etc.];
choose, who knows the terminology; – the use of the research results of the different
– the doctrinal attitudes have to be adopted care- fields in the interventions carried out for the
fully: for example the Mediterranean fashion safety of the load-bearing structures: art history
of natural surfaces of the 1950s and ‘60s (the building of a construction in phases – the
has always had negative results in continental source of damages in the load-bearing struc-
Europe, as the number of freezing cycles is tures), archaeology (the former foundations –
incomparable in these places. damaging factors), building physics (moisture
(d) available materials: containing dissolved salts – causing damages)
– we have to take into consideration that the cur- etc.;
rent materials do not reach the quality level – providing laboratories for those industrial
of the historic ones (the solid brick with the branches that carry out the interventions on
old format, the lime or the timber from the historic buildings, where the quality of the load-
planted forests are only a few examples for the bearing structural materials would be studied,
poor quality of the recently produced traditional creating conditions for the routine experiments:
materials); e.g. stone is generally used without a quality
– the contemporary building materials are pro- check;
duced so as to be compatible with other recently – in-situ testing of the building materials and a
produced materials – their compatibility with qualified personnel.
the historic building materials has to be stud- (g) the quality control system:
ied in each case; laboratories, specialists, the – the historic load-bearing structures have to be
awareness of the issue, as well as the will to levelled with the current safety standards – this
take measures are thus needed. is impossible without a quality control system,
(e) skill: taking into consideration the aforementioned
– one of the tasks is the studying of the traditional deficiencies;
technologies, avoiding the technological incom- – the personnel designated for the quality control
patibilities between the traditional technologies has to have thorough knowledge of the historic
and the intervention; load-bearing structures;
– we need to study the technological conditions – the quality control laws have to deal with
for the reinforcement to begin immediately to the characteristics of the historic load-bearing
work (reinstating the original geometry with the structures in detail;
introduction of counter-bending etc.). – uniform measures are recommended regarding
(f) research possibilities: the safety of the built environment – taking of
– the basic research programs are necessary [the course into consideration the characteristics of
slow transformation of the slopes, the mechan- the historic buildings (it is not efficacious, if a
ics of the earth banks and of the poor-quality group of institutions takes care of the contempo-
foundation soils that have been loaded for rary buildings and another group watches over
centuries (the soil mechanics regulations deal the historic ones).
147
4 ENSURING THE SAFETY OF HISTORIC focusing their attention on the safety aspects: (i) the
LOAD-BEARING STRUCTURES [6], [7], elimination of the causes of insufficiencies in the
[8], [9] load-bearing structure, (ii) ensuring the performance
expectations at the load-bearing structure and (iii) pre-
The acceptable behaviour of the historic load-bearing serving the performance expectations in the case of the
structures in the resistance and stability limit state – load-bearing structure – ensuring the durability of the
safety measurement of load-bearing structures – is intervention.
anticipated through: (i) investigations for the deter- From the principles adopted for the safety interven-
mination of the extant safety level, (ii) interventions tions the following are to be stressed:
needed for the correlation of the extant level with the
contemporary safety requirements, (iii) the mainte- – the interventions are in connection with the load-
nance of the acceptable behaviour in the resistance bearing structure concept: the correct concepts
and stability limit state during use. allow conservation or restoration works, in the
case of the wrong ones retrofitting or possibly
reconstruction is needed; safety is provided in the
4.1 Investigations for the determination of the case of both permanent and temporary, as well as
extant safety level extraordinary (special) actions;
– reversible (that can be dismantled) interventions are
The investigations for the determination of the extant
preferred, especially when the compatibility of the
safety level of the historic load-bearing structures are
intervention with the heritage values is uncertain;
related to technical, as well as to heritage issues.
– the efficiency of the interventions has to be
The technical investigations take place during [9]
checked; in this respect it is recommended to test
the diagnosis phases carried out on the historic load-
the dynamic characteristics of the load-bearing
bearing structure (characterized by the approach of
structure before and after the interventions: accel-
the structural issues from the point of view of build-
eration, speed etc.;
ing mechanics, physics and biology, as well as of
– the consequences of the interventions regarding the
soil mechanics and building services engineering),
improvement of safety at all performance expec-
focusing their attention on the safety aspects: (i) the
tations: in connection with building physics and
identification (survey and presentation) of the historic
biology, as well as soil mechanics;
load-bearing structure; (ii) finding the insufficiencies
– the reconditioning of the quality of historic materi-
in the historic load-bearing structure; (iii) the formu-
als is also recommended, with a view to regaining
lation and testing of the performance expectations at
their initial safety level (for example re-freshening
the historic load-bearing structure; (iv) the determi-
the mortar from the vaulted slab masonries through
nation of the causes of insufficiencies at the historic
its moistening with lime milk [8]).
load-bearing structure.
The heritage values have to be recognized – the The interventions carried out on the heritage values
investigations are linked in this respect with the artis- should be minimal. The problem consists in, whether
tic elements or architectural details, taken over from the intrinsic heritage values of the load-bearing struc-
archaeology or art history studies, as well as from the ture are recognized or not: these values have to be
expert opinion on the artistic elements or the non- protected or not, or the load-bearing structure has only
bearing structures. The intrinsic heritage values of the the role to support, to carry the heritage values of the
load-bearing structure are determined by experts in the non-bearing structures (artistic elements and architec-
field of historic load-bearing structures. tural details). If the intrinsic heritage values need to
be protected, then the diagnosis and prediction of the
historic load-bearing structures have to focus on the
4.2 Interventions needed for the correlation of the
identification and protection of the intrinsic heritage
extant level with the contemporary safety
values, going through each phase of the diagnosis and
requirements
prediction, mentioning what happens to the intrinsic
The interventions needed for the correlation of the and worn heritage values every time, thus: (i) survey
extant level with the contemporary safety require- – the heritage values are surveyed; (ii) insufficiencies:
ments are also approached from a technical point of it will be described to what degree the insufficiencies
view, as well as with a view to the heritage values. affect the heritage values; (iii) testing the way in which
The technical interventions take place during [9] the performance expectations are provided: whether
the diagnosis phases carried out on the historic load- the heritage values have got deficiencies in this respect
bearing structure (characterized by the approach of or not, (iv) the determination of the causes of insuffi-
the structural problems from the point of view of ciencies: whether the causes dispose of heritage values
building mechanics, physics and biology, as well as or not; (v) the elimination of the causes of insufficien-
of soil mechanics and building services engineering), cies: whether they are destroyed through elimination
148
in the case of heritage values as well, (vi) providing the (b) The load-bearing structural concept has to be ade-
performance expectations, (vii) preserving the perfor- quate and has to meet the contemporary safety
mance expectations – whether the heritage values are requirements. If it does not meet them, what will
affected through the suggested interventions or not. happen to the historic building? Will it be demol-
ished? If an intervention is carried out for the
correction of the load-bearing structural concept,
4.3 The maintenance of the acceptable behaviour what are the acceptable limits of the intervention?
of load-bearing structures during use What is the quantity of heritage values that can be
The maintenance of the acceptable behaviour of sacrificed in order to provide safety?
load-bearing structures during use requires – among (c) The empirical-intuitive concept can be correct, but
others – the following: the historic materials – of which the concept is
made – do not comply with the pretensions of the
– the functional and technical (conceptual, mate- concept to meet the contemporary safety require-
rial and technological) compatibility of the inter- ments: Do we replace the materials? Do we change
ventions has to be provided (shear thresholds, the concept?
compressions, counter-bending etc.); (d) When are the safety requirements satisfied through
– the incompatibilities in connection with building the conservation of the load-bearing structure?
mechanics, physics, biology and soil mechanics What about restoration? When is retrofitting or
have to be avoided; reconstruction applied? The reconstruction is car-
– specifying certain measures for the perpetuation of ried out with the same material, or only the
the effects of the interventions carried out on the geometry is re-built by using the materials that
load-bearing and/ or non-bearing structure (avoid- provide the necessary safety (for example concrete
ing heat bridges that bring about vapour, ventilation instead of masonry, possibly a concrete framework
conditions, anticorrosive treatments etc.); embedded in the masonry)?
– the maintenance of the interventions, with a view to (e) Temporary and definitive interventions are
obtaining the response of the load-bearing structure accepted in order to ensure the safety of the historic
to extraordinary actions at any time (for example load-bearing structures. Is the emergency inter-
earthquakes); vention to provide safety an accepted version or
– reducing the aggressiveness of the environment. not: in the first phase the emergency intervention
is carried out to provide safety that temporarily
guarantees the safety of the load-bearing struc-
5 CONCLUSIONS ture, after which the final intervention follows –
after a certain period of time: days, months, years,
The safety of the (historic) constructions is defined decades? Do these interventions have to respond to
by the relation between the qualities (possibilities) of all the possible actions, or only to some of them? Is
the (historic) load-bearing structures and the demand there a maximum duration, for which a temporary
of the environment. Being built on the basis of intervention can efficiently function?
an empirical-intuitive concept, of historic materials
and with traditional technologies, the historic load- 5.2 There is a series of questions
bearing structures display their qualities (possibilities) (a) The safety of the historic load-bearing structures
arranged according to these conditions of existence. is influenced by the general factors that affect
The pretentiousness of the environment is always in all the constructions from the built environment
connection with the place, in space and time. The and by factors specific for historic constructions.
spatial coordinates mark the natural and artificial cor- Both categories of factors deal with a significant
rosive, technical and social conditions in which the objective, as well as a subjective part.
load-bearing structure exists. The time coordinates (b) The characteristics of the historic load-bearing
present the evolution of these conditions. structures, as well as of the historic and/ or
current load-bearing structural environment are
considered to be the objective premises.
5.1 A series of questions have to be asked,
(c) The social-political conditions, the level of tech-
which need to be answered
nical knowledge, the research possibilities, the
(a) Beside utility and economical nature, safety plays quality control system, the available materials,
an important role in the sustainable management aptitude and others are considered to be subjective
of the built environment. The question is stated: in premises. These premises operate at the formation
the case of historic buildings, is the preservation of of the constructions or during the subsequent inter-
the heritage value a criterion of equal importance ventions (the «historic» ones), being present in our
with that of safety and economical nature? time («contemporary» ones) as well.
149
(d) Local historic materials, technologies developed concepts (Byzantine-eastern and western building
in connection with the local materials and the cultures) and others.
local conditions relating to the development of the (m) The accidents due to incapacity and frauds
technologies, the local climatic conditions (load cannot be avoided through technical regulations –
resulting from snow, wind, the number of the however perfect these might be. The measures nec-
repeated annual freezing cycles etc.), the inten- essary for the reduction of the causes of these
sity of the seismic activity, all these have led to the accidents do not influence the conditions regarding
local empirical-intuitive concept of historic load- the safety of the constructions.
bearing structures (postulating also a huge her-
itage value). The investigations will compulsorily
identify all these. REFERENCES
(e) The investigations will pay special attention to the
time coordinate (for example to the depreciation [1] Ferry Borges J. & Castanheta M. 1971. Structural Safety.
of the quality of the materials due to explosions, Lisbon: LNEC
floods, fires, earthquakes taking place over time). [2] ∗ ∗ ∗ 2003. ICOMOS Charter – Principles for the Analy-
(f) The safety of the historic load-bearing structures sis, conservation and structural restoration of architec-
tural heritage. Victoria Falls, 14th General Assembly of
is sufficient, if the foundation conditions are ade- ICOMOS
quate, the corrosive agents controlled, taking the [3] Szabó B. Gy. & Kirizsán I. 2003. Built Heritage in Crisis
following into consideration in connection with Situations – Disasters and Historic Load-bearing Struc-
the superstructure: (i) the safety of the historic tures. In:Imola Kirizsán, Bálint Gy. Szabó (ed.), Historic
masonry load-bearing structural sub-units is max- structures and disasters; Proc. 7th Scientific Conference
imal, if the dynamic loads and differential settle- Cluj-N., 6–8 November 2003, Cluj-N.: Utilitas
ment can be avoided; (ii) the safety of the historic [4] Szabó B. Gy. 2004. Illustrated dictionary of historic
wooden load-bearing structural sub-units is max- load-bearing structures. Cluj-N.: Kriterion & Utilitas.
imal, if they are protected against the biological ISBN 973-9377-40-8
[5] Szabó B. Gy. & Kirizsán I. 2004. Conservation of His-
attacks. toric Load-bearing Structures and the Protection of their
(g) The state of the historic load-bearing structures, Intrinsic Heritage Values – Legacy of Historic Load-
the safety reserves are continuously changing, the bearing Structures. In: Imola Kirizsán, Bálint Gy. Szabó
periodic re-evaluation of safety being compulsory. (ed.), Heritage values of historic structures; Proc. 8th
(h) The reconditioning of the quality of historic mate- Scientific Conference Cluj-N., 28–30 October 2004,
rials is also recommended, with a view to regaining Cluj-N.: Utilitas. ISBN 973-9377-38-6
their initial safety level (for example re-freshening [6] Szabó B. Gy. & Kirizsán I. 2006. The Safety of the His-
the mortar from the vaulted slab masonries through toric Load-Bearing Structures In: Imola Kirizsán, Bálint
with lime milk). Gy. Szabó (ed.), Safety of Historic Load-Bearing Struc-
tures; Proc. 10th Scientific Conference Cluj-N., 26–28
(i) Reversible (that can be dismantled) interventions October 2006, Cluj-N.: Utilitas. ISBN 973-9377-45-9
are preferred – for their reconsideration together [7] Roca P. 2006. The Study and Restoration of Histori-
with the development of intervention techniques-, cal Structures. In: P. Lourenco, P. Roca, C. Modena,
especially when the heritage values are affected. S. Agrawal (ed) Structural Analysis of Historical Con-
(j) Issues regarding safety and heritage values are structions; Proceedings of the 5th International Con-
required to be treated simultaneously. Theoreti- ference, New Delhi 6–8 November 2006. New Delhi,
cally we apply measures that are compatible with Macmillan India Ltd. ISBN 978-1403-93155-9
the intrinsic heritage values of the historic load- [8] Szabó B. Gy., Kirizsán I., Kirizsán Il. & Vass L. 2006.
bearing structures, wishing to destroy as few of The Improvement of the Safety of the Vaulted Slabs
with Lime Treatment In: Imola Kirizsán, Bálint Gy.
these values as possible. Szabó (ed.), Safety of Historic Load-Bearing Struc-
(k) Uniform measures are recommended regarding tures; Proc. 10th Scientific Conference Cluj-N., 26–28
the safety of the built environment – taking of October 2006, Cluj-N.: Utilitas. ISBN 973-9377-45-9
course into consideration the characteristics of the [9] Szabó B. Gy. 2007. Metoda complexa de investigare
historic load-bearing structures, as it is not effica- a starii de degradare la structuri portante istorice.
cious, if a group of institutions takes care of the In: Raport tehnic UTILITAS_07009, Cluj-N.: Utilitas.
engineered structures and another group watches ISSN: 1842–9866
over the historic ones.
(l) Regional (even local) instructions are needed for
the right handling of the issues regarding the safety
of the historic load-bearing structures; the delim-
itation of the regions can be made according to
seismic zones, slope stability, empirical-intuitive
150
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
H.A. Heinemann
Chair of Building Conservation, Faculty of Architecture, Technical University of Delft, The Netherlands
T.G. Nijland
TNO Built Environment & Geosciences, Delft, The Netherlands
ABSTRACT: The 20th century built heritage is one of the new conservation challenges, due to its architectural
differences from the traditional heritage and new materials. One major new material is concrete; its quantity and
importance for the new heritage requires a tailored conservation approach. Until now, there is a dependence on a
repair approach. The dangers of such an approach for the cultural-historical values of the original concrete will
be shown by the means of a case study, Fort Bezuiden Spaarndam (1901), part of the UNESCO World Heritage
Site Defence Line of Amsterdam. An underlying problem is that concrete is too young for conservation experts,
causing a dependence on concrete repair expertises. The used repair techniques are developed for structural aims,
not integrating conservation aims. Fundamental research on the history of concrete and suitable conservation
approaches and techniques is therefore needed.
151
During the research, the most important involved
parties were interviewed (architect, surveyor, and con-
crete repair company) to understand their background,
knowledge, and experience and to see where the
dependence on concrete repair experts starts. Liter-
ature, documents, and surveys related to the fort, its
construction and conservation were consulted to inves-
tigate the historical background, conservation aims,
and conservation process. During a visual inspection
in November 2006, the performance and impact of the
conservation was evaluated.
The study focussed on concrete which was used
for exterior elements, stressing the complexity to bal-
ance technical, aesthetical and conservation demands,
meaning a minimal loss of existing fabric and a maxi- Figure 1. Site plan of Fort Bezuiden Spaarndam.
mal preservation of values. Additionally, petrographic
analyses of samples were made to determine the con-
crete composition and possible damage. To evaluate 4 FORT BEZUIDEN SPAARNDAM
the suitability of standard concrete surveys for con-
servation tasks, literature and guidelines related to 4.1 Historical background of the Defence Line of
concrete assessment were studied. Amsterdam
The Defence Line of Amsterdam, the Netherlands
3 BACKGROUNDS OF REPAIR AND (1880–1920), is a ring of fortifications (45 forts)
CONSERVATION around Amsterdam, intended to defend the city by sys-
tematic inundation in case of an attack. This ingenious
To understand why following a repair approach in a defence technique was one justification for the listing
conservational context can endanger values, it is nec- as a UNESCO World Heritage Site in 1996. In con-
essary to understand why and how the main aims nection with the listing, attention was drawn to the
of repair and conservation diverge radically. Repair need for conservation and re-use, as the condition of
methods are usually developed to preserve a structural many forts declined after the military function of the
function for a limited lifetime, lately also influenced by forts ceased in the 1960s. Fort Bezuiden Spaarndam
sustainability aspects. Surveys focus on physical and (1901), a provincial monument in the province North
mechanical properties and the criteria for the choice Holland, was constructed as part of the Defence Line
of a repair approach are, beside the structural perfor- of Amsterdam.
mance, the service life, economical issues, and the
possibilities of maintenance (ENV 1504-09 1997). 4.2 Historical background of Fort Bezuiden
Conservation, however, is much more than ful- Spaarndam
filling technical and financial requirements. Besides
considering structural performance and safety, the Fort Bezuiden Spaarndam is a non-reinforced concrete
preservation of the authenticity of and values attached fort built in 1901. It was planned as a standard model
to the site, building and material must be considered. of the Defence Line of Amsterdam on a fort island with
The value and authenticity of the cultural heritage two cupolas with disappearing turrets (Fig. 1). The fort
is understood in ‘artistic, historic, social, and sci- itself did not suffer from major damages during the
entific dimensions’ and the conservation principles Second World War, though the cupolas were blown up
‘respect. . . the existing fabric’ and the cultural heri- by the German occupation army. The remains of the
tage should be preserved for ‘present and future cupolas still lay spread around the grounds. Since the
generations’ (ICOMOS 1999). cease of the military function in 1960, the fort was
Yet, due to the youth of the field of concrete con- used for agricultural storage.
servation, concrete repair experts are entrusted with
concrete conservation projects.Although these experts
4.3 Historical background of the concrete of
have considerable knowledge of the technical aspects,
Fort Bezuiden Spaarndam
they are not trained in specific conservation aspects,
as the complexity and variety of monumental values To recognise the historical value of the original con-
and authenticity or the properties of historic con- crete, it is necessary to understand the construction and
crete. Because conservation experts often do not have technology history of the late 19th century. Knowl-
sufficient knowledge on the material, an evaluation of edge of concrete and its construction was mainly
the suitability of such a process is difficult. based on empiric studies; standard design codes only
152
appeared at the beginning of the 20th century. In assess the technical state of the concrete and to deter-
this context, the knowledge gained from the con- mine the causes of the encountered leaching, cracking,
struction of the Defence Line of Amsterdam was and leakages. Several cores were taken, not only to
important for the Netherlands, where early concrete determine the composition and damage causes, but
technology was dominated by the army corps of also to gain information on the wall construction.
engineers (e.g. Scharroo 1910). The first Dutch rein- It was stated that the walls were massive concrete
forced concrete guidelines Eerste Gewapend-Beton- and the concrete was composed of cement with ground
Voorschriften from 1912 were influenced by the expe- granulated blast furnace slag (GGBS) and crushed
rience gained during the construction of the Defence rocks (granite, porphyry) as aggregates. The following
Line of Amsterdam (Noord-Holland 1996). damages were assessed in the survey:
Therefore, the forts of the Defence Line of Ams-
– shrinkage cracks
terdam are the tangible artefacts of this pioneer phase
– cracks along casting segments caused by missing
of concrete technology. Initially, the forts were built
expansion joints
as traditional brick constructions; yet the invention
– water bearing cracks with leaching
of brisance grenades demanded new and stronger
– map cracking caused by the formation of thaumasite
construction methods. Hence from 1897 onwards,
or ettringite.
the forts were built as non-reinforced concrete con-
structions, later as reinforced concrete constructions The advice was to reconstruct the missing earth
(ICOMOS 1996). This transition from brick to rein- shelter of the roof and to seal the roof to prevent
forced concrete and the still preserved high degree further infiltrations. Cracks that influenced the struc-
of authenticity was part of the justification for the tural performance should be injected; the injection of
Netherlands to nominate the Defence Line of Ams- other cracks was optional. In general, the condition
terdam as a UNESCO World Heritage Site (ICOMOS of the concrete, considering the construction time and
1996). The value of the concrete as a preservable method, was stated as good and with adequate repairs,
cultural-historical artefact was stated prior in the the building could last another 100 years.
description of the Fort Bezuiden Spaarndam in con-
nection with the listing as a provincial monument in
the early 1990s (Provincie Noord-Holland 1998). 5.2 Repair of the concrete shell
The main goal of the intervention was to repair the
5 CONSERVATION 1996–1999 outer shell of the fort in a way that would solve the
problems ‘for once and for all’, repairing the roof
Fort Bezuiden Spaarndam was one of the first forts and treating all cracks. As a detailed advice for repair
to be restored after the listing of the Defence Line techniques was not commissioned, the choice of an
of Amsterdam as a World Heritage Site in 1996. It adequate repair technique was left to the executing
was supposed to be a role-model conservation project concrete repair company. In 1997, the repair campaign
with extra subsidies and attention, as further forts of the fort started and included for the exterior concrete
needed restoration in the near future. The first aim walls following intervention steps:
was to find a new locally orientated function for the – high water pressure cleaning of the façade
fort, which should be compatible with the surrounding – filling of cracks with standard repair mortars
nature recreation area. In 1996, it was decided to use – injection of cracks with micro-cement
the northern wing as a day nursery, the postern as an – creating expansion joints along the cracked vertical
art gallery, and the southern wing as storage. segment joints.
The task of conservation was handed over to a
restoration architect, who had experience with the No specific requirements concerning durability and
conservation of a brick fort, but no experience with compatibility were prescribed. The concrete repair
the material concrete. This is not unusual for the works were not supervised by external experts. Main-
late 1990s, as little knowledge and experience with tenance and regular inspections were not planned,
the conservation of concrete existed. The heritage since it was assumed that the repairs would be long
care authorities of the province North Holland had no lasting.
specialist for the material concrete either, and an exter-
nal concrete expert was entrusted with the survey of
the fort. 6 EVALUATION OF THE CONSERVATION
PROCESS
5.1 The survey of the concrete
Although intended to be a role-model conservation
The survey did not differ from standard, non- project, the approach of Fort Bezuiden Spaarndam
conservation concrete surveys. The assignment was to is an example of a concrete repair approach. This
153
case shows the problems resulting from an unsuit- interior of the fort (day nursery and art gallery), the
able approach and reflects an underlying problem climatic conditions were satisfactory.
of concrete conservation: unawareness of the values
of historic concrete. Examples from the survey and
6.2 Awareness of values
execution illustrate the consequences and show how
valuable information could have been gained or values The visual impact of the conservation on the façade
preserved. raised the question why the aesthetical values of the
First, the crucial phase of assessment and decision- monument were not respected. A possible aesthetical
making was not guided by adequate experts who value of a building is usually widely accepted and the
understood both the material, including its history, visual impact of a repair technique can be evaluated
and conservation principles. Consequently, values without large-scale research. However, standard repair
attached to the concrete were neither determined nor techniques do not consider aesthetical compatibility.
the impact of the repair techniques on the values Therefore, the question arose how the conservation
attached to the concrete evaluated. of concrete was realised and if values were consid-
However, it is possible to gain information on ered in the process. During our investigation, it became
possible values during a survey (Heinemann 2007). apparent that the concrete was only perceived as the
A review of the conservation process show how values structural material of the shell. This reflects the under-
could have been detected and thus incorporated in the lying problem of concrete conservation, that concrete
conservation process; it also reinforces that knowl- is often not seen as a valuable historical material.
edge on the material, its history, conservation aims and In the case of Fort Bezuiden Spaarndam, it was offi-
repair techniques is needed, and that trained experts for cially stated that the original concrete was valuable in
these tasks are still missing. the justification for the listing as a UNESCO World
Heritage Site (ICOMOS 1996) and as a provincial
monument in 1992 (Noord-Holland 1998). The origi-
6.1 Present condition 2006
nal concrete is important because of its uniqueness and
A visual inspection of the performance and impact remains of this phase of the technical development of
of the interventions was made during a site visit of concrete in Europe are rare. However, an implemen-
Fort Bezuiden Spaarndam in 2006. From far distance, tation of the preservation of and awareness for these
the impact of the repaired cracks was already visible, values was missing in the conservation process.
changing the appearance from a monolithic structure This unawareness led to the commission of a stan-
to a patched structure (Fig. 2). On closer examination, dard (i.e. purely technical) survey of the concrete
it appeared that the repaired cracks next to the newly which was not adapted to the needs of conservation.
created expansion joints re-cracked and the concrete Without sufficient knowledge of historic concrete and
of the casemates showed map-cracking with leaching possible values, the weaknesses of a purely technical
and local disintegration. On several places on the main survey in this context were not seen. Consequently, the
façade, traces of camouflage paint were found that correlation between values, the material, and the state
were not documented in former surveys.The interior of of conservation was not determined. Therefore, the
the fort showed no signs of infiltrations or re-cracking, values were not further integrated in the conservation
and according to the tenants, who appreciated the process.
154
6.3 Survey of the surface coarse aggregate was crushed porphyry from Quenast,
Belgium, and the fine aggregate river sand (Fig. 3).
Aesthetics are criteria for the value assessment of
From the study of historical documents, we found
buildings and for the compatibility of conservation
out that crushed rock was assumed to be superior to
techniques. Therefore, it is important to describe the
gravel because of its rough and angular surface which
surface of a monument during a survey. For techni-
was supposed to improve the bond to the cement paste.
cal surveys, as it was commissioned at Fort Bezuiden
The use of crushed rock and in particular porphyry of
Spaarndam, a description of the surface is not included
Quenast as an aggregate was advised in contempo-
as it usually does not influence the mechanical or
rary books (Kloes 1908). In the construction plans of
physical properties of the concrete.
the fort from 1900, it can be seen that a differentia-
A description of the properties of the surface such
tion between the used aggregates and function of the
as colour, texture, deposits, weathering, or traces of
concrete element was made. For important elements,
former events can be used to evaluate the impact of an
the use of porphyry was designed, while secondary
intervention (e.g. material loss due to cleaning) or help
elements where designed with cheaper gravel or even
to adapt the structure and colour of a repair mortar to
with rubble.
the surrounding concrete.
The high amount of voids and the average to high
Other essential information which should have
capillary porosity of the concrete mentioned in the
been documented is the presence of camouflage paint
survey can be explained with knowledge of the com-
traces and the original surface finish. Further colour-
position and construction method. They are in this
historical research should have been carried out to
case typical, because the rough shape of the aggre-
evaluate the options for reconstruction of the paint and
gate complicated the manual compaction. To obtain
to broaden the historical knowledge of the forts.As this
the workability, the use of a higher water-cement ratio
case was supposed to be a role-model, opportunities
was common. Both the compaction and the high water-
to study the forts in connection with the survey should
cement ratio influence the porosity and durability of
have been taken.
concrete.
How far the cleaning of the façade with high-water
Another result from the composition is the occur-
pressure caused losses of the original surface, and
rence of alkali-silica reaction (ASR) (Fig. 4). The used
with it historical information, cannot be determined
aggregate, porphyry of Quenast, is now known to
now, since there is no detailed documentation of the
be alkali-silica reactive (Nijland et al. 2003). When
façade to refer to. Given experience with façade clean-
used in combination with Ordinary Portland Cement,
ing, however, such a loss is highly likely. It also shows
which, in contrast to the GGBS cement that was
how easily standard approaches are followed, and con-
wrongly considered to be present in the original survey,
sequences for the monument and its values are not
ASR is not inhibited. With knowledge of the construc-
considered.
tion history of concrete certain damage types can be
explained, here the assumed superior combination of
6.4 Survey of the composition and construction porphyry and Ordinary Portland Cement causingASR.
An analysis of the composition, structure, and cause
of damage of concrete, as for example by petrographic
examination, is not standard, as it might not always be
necessary to diagnose the cause of damage. However,
if samples are taken, they can be used to gain insight
into the composition and construction of historic con-
crete. Besides giving relevant technical information,
it can reveal important historical information, and
therefore indicate value.
Historic concrete differs from modern concrete. In
the case of Fort Spaarndam Bezuiden, it is known that
it was constructed in a time when concrete construc-
tions were not standardised. Also the statement that the
concrete is a rare remainder of concrete development
should have led to an investigation of the samples from
a historical point of view as well.
Our petrographic analysis of samples of the south-
ern cupola and of the main building show that the Figure 3. Microphotograph showing detail of the cement
concrete is composed of a coarse-grained Portland paste, with relics of unhydrated Portland clinker (indicated
cement and not ground granulated blast furnace slag as calcium silicates C2 S and C3 S), and no blast furnace slag
(GGBS) cement as described in the survey. The at all (view 0.7 × 0.45 mm, plane polarized light).
155
preservable.Yet it is likely that the high pressure clean-
ing of the façade affected this layer of the surface. As
no references were found on the building anymore nor
could be found in the documentation of the project,
the amount of loss of original fabric cannot be said.
With historical and technical knowledge, a petro-
graphic analysis can help to determine the authenticity
of the original concrete, which is important when deci-
sions have to be made of how much material loss is
acceptable. The results from our petrographic analy-
sis show that the composition of the concrete of Fort
Bezuiden Spaarndam is of high quality relative to the
state of the art of 1900. The given historical informa-
tion confirms the value of the authentic concrete and
any intervention should respect that value.
Figure 4. ASR-gel (A) in voids (V) and porphyry (P) 6.5 Advices and decision-making
and sand (S) as aggregates in the cement paste (C) (view
1.8 × 1.12 mm, plane polarized light). As the assignment of the survey was pure technical,
additional historical aspects of the concrete were not
considered. As a result, no special treatment respecting
the historical values (significance of original concrete
and the visual appearance) was advised. The choice of
the repair techniques was done by the restoration archi-
tect and the concrete repair company. Neither of them
was familiar with the properties of historic concrete
and the complexity and consequences of repair choices
were not seen. One example of the consequences is
the repair approach of the cracks caused by thermal
movements.
A common problem of the forts of the Defence Line
of Amsterdam is the lack of expansion joints. Expan-
sion joints were not common at that time, in the 1920s
there was still discussion if expansion joints are needed
(Kleinlogel 1927). Also here the missing insight into
the properties of concrete caused damage, cracking,
Figure 5. Microphotograph of the section through the which becomes a historical evidence of the trials and
surface, showing three lime-based layers (plane polarized errors of the technological development.
light). The concrete forts of the Defence Line of Amster-
dam are approximately 150 m long, most of them not
It has to be kept in mind that the perception of reinforced. The stresses caused by thermal movement
cement types differed from our current perception. At caused regular linear cracks along the weakest points,
the time, the use of Ordinary Portland Cement was pre- usually the casting segments. Cases of severe struc-
ferred to the use of blast furnace slag cement. Latter tural damage are not known, as each casting segment
was assumed to be of poor quality, mainly because it is equivalent to one vault and the cracks appeared at
was a cheap waste product. It became, due to economic the points of support in the middle of the cross walls.
reasons, only widely used in the Netherlands from the Yet, the ingress of aggressive agents can be facil-
1920s onwards. itated by the cracks. As there was no reinforcement,
A petrographic analysis of the concrete can also corrosion was not possible but the composition of the
tell stories about the history of a building. A section concrete was ASR sensitive, a reaction which needs
through the surface (Fig. 5) shows three layers of a water and can lead to severe damage. However, no
lime based layers on the surface. The core was taken signs for increased deterioration were visible next to
before the surface was cleaned with high-water pres- the cracks.
sure. Therefore it cannot be determined if they are During the restoration campaign of Fort Bezuiden
remains of a finish (Noord-Holland 1996) or deposits. Spaarndam is was decided to repair the outer shell
If the indications from the thin section would have been ‘properly’, which included a repair of the cracks
used prior to the intervention, research could have been caused by thermal stresses. The idea was to compen-
carried out in situ and decisions made if the surface is sate the thermal stresses with expansion joints, which
156
Figure 7. Close up of an expansion joint with a reappearance
of the former cracks; the colour of the used patching mortar
was not adapted.
157
Similar to a repair process, the survey focused kept in mind. It is the task of the conservation expert
on physical properties of the concrete and did not to explain the diverging requirements for conserva-
integrate historical relevant aspects as the unique tion towards repair experts. This requires sufficient
composition of the concrete or the traces of camou- knowledge of the material concrete from the conser-
flage paint. Consequently, the amount of lost original vation expert; otherwise the deficiency of a repair
concrete and related values was not considered. approach cannot be seen and may bring to risk the
Support and guidance for the involved parties values which were the initial reason for listing and
by concrete conservation experts was not possible, conservation.
as the field of concrete conservation barely existed As long as concrete conservation projects are still
and the complexity of concrete conservation was not isolated cases and no clear guidelines or references for
understood. Therefore, possible conservation aims, the concrete conservation exist, general guidelines such
cultural-historical values, the significance of the state as the Venice Charter (ICOMOS 1964) should be
of the building and their relationship could not be consulted to evaluate possible approaches. Finally, dis-
discussed openly before starting the conservation cam- cussions of possible values of concrete are needed to
paign. Such discussion should not be limited to pos- see that it can be more than a greyish bearing mate-
sible re-use or economical issues, but should include rial. As concrete was initially mainly understood as a
tangible and intangible values of the building/material structural material and only later the architectural val-
as well. ues were discovered, we have to make now a similar
A critical review of the repair approaches was not evolution: from structural repairs to conservation.
taken into account, as it was assumed that repair
experts have sufficient knowledge of the material.
There was no awareness for the fundamental differ-
ences between a repair and conservation, therefore the ACKNOWLEDGEMENTS
influence of a repair approach on the values was not
considered. Figures 1 (based on the construction drawings from
However, this case is not isolated and it is likely that 1900), 2, 6, and 7 by H. A. Heinemann, figures 3, 4
the results would have been similar if the conservation and 5 by T.G. Nijland.
would have been carried out by other people. The fun-
damental problem of concrete conservation is that the
conservation process is often not guided by conserva- REFERENCES
tion experts which understand both the material and Construction plans Fort Bezuiden Spaarndam. 1900. Nation-
conservation principles. aal Archief, The Hague, OPV Oorlog Plans van Vestingen,
17e–20e eeuw, access number 4.OPV, inventory number
A268–A26.
7.2 Concrete conservation Heinemann, H.A. 2007. Bedreigde waarden: De invloed van
de technische benadering op de cultuurhistorische waar-
Values and their preservation are key issues of heritage den van betonnen monumenten.Praktijkreeks Cultureel
care. The value of age, that, as Riegl (Riegl 1903) says, Erfgoed. 1(2):1–24.
even ‘appeals the masses’ is often not seen since to ICOMOS. 1964. International charter for the conservation
the ‘youth’ of concrete buildings and the more sophis- and restoration of monuments and sites. The Venice Char-
ticated historical values need art-historical knowledge ter. http://www.international. icomos.org/e_venice.htm.
to be recognised. Until now, this art-historical knowl- Accessed 08-05-2007.
edge hardly exists and this reflects itself in the (value) ICOMOS. 1996. Advisory Body Evaluation no. 759
assessment of historic concrete buildings. For the http://whc.unesco.org/archive/advisory_body_evaluation/
759.pdf. Accessed 19-11-2007.
majority of restoration stakeholders, concrete build- ICOMOS. 1999. The Burra Charter. The Australia ICO-
ings are too young, as their working field is mainly MOS charter for the conservation of places of cultural
buildings prior to 1900. significance. http://www.icomos.org/australia/. Accessed
It is necessary to overcome the knowledge gap con- 26-11-2007.
cerning 20th century buildings in the working field Kloes, J. A. van der. 1908. Onze Bouwmaterialen – Mortels
and education of conservation experts. Parallel, it is en Beton. Maassluis, the Netherlands: J. van der Endt &
necessary to develop a tailored conservation approach Zoon.
to overcome the dependence on a pure repair approach. Kleinlogel, A. 1927. Bewegungsfugen im Beton- und Eisen-
When applying repair techniques, one should not rely betonbau. Berlin: Wilhelm Ernst & Sohn.
Nederlands Normalisatie Instituut 2006. ENV 1504-09:1997
on the fact that concrete repair has been carried out Products and systems for the protection and repair of
for decades, but assessed whether they are applicable, concrete structures – Definitions, requirements, quality
since they were developed for different aims. control and evaluation of conformity – Part 7: Reinforce-
As long as there is a dependence on a repair ment corrosion protection. Delft: Nederlands Normal-
approach, the background of repair experts has to be isatie Instituut.
158
Nijland, T.G., Larbi, J.A. & Siemes, A.J.M. 2003. Experi- Riegl, A. 1903. The Modern Cult of Monuments: Its essence
ence of ASR in the Netherlands. Proceedings European and its development. Published in The Getty Conserva-
marine sand and gravel – shaping the future, EMSAGG tion Institute. 1996. Historical and Philosophical Issues in
Conference. Delft, the Netherlands. the Conservation of Cultural Heritage. Los Angeles: The
Provincie Noord-Holland. 1996. Onderzoek naar de Getty Conservation Institute (originally published 1903
bouwkundige staat van de forten van de stelling van as Der moderne Denkmalstatus: Sein Wesen und seine
Amsterdam. Haarlem, the Netherlands: Provincie Noord- Entstehung Vienna: W. Braunmüller).
Holland. Scharroo, P.W. 1910. Inleiding tot de studie van het gewapend
Provincie Noord-Holland. 1998. Stelling van Amster- beton. Amsterdam: van Mantgem & de Does.
dam. Positie bij Spaarndam.(2nd ed.) Haarlem, the
Netherlands: Provincie Noord-Holland.
159
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Saptarshi Sanyal
School of Planning and Architecture, New Delhi, India
ABSTRACT: The fascinating and unique covered market, Chatta Chowk Bazaar, was designed for the fortress-
palace, Redfort, the magnum opus of 17th century Mughal Emperor Shahjahan’s new capital in Shahjahanabad,
Delhi. This building was the subject of documentation and comprehensive qualitative studies, undertaken by
architectural conservation students of School of Planning and Architecture in 2006–07. The direction and
approach given by the conservation studio’s faculty, one of knowledge systems, was significant in guiding
the studies selected and the way the structure was analysed, resulting in an appreciation of the many dimensions
of the historic building. These revealed architect Shahjahan’s immense knowledge of architecture, qualified
through primary studies of Mughal architecture in Delhi including the palace-complex and secondary studies
of his other projects. This paper attempts to illustrate, through this case, how superior structural know-how
and stability, are integral to a significant historic building, which is a knowledge system, and how an essential
consequence of such significance, is safety.
161
Figure 1. The Chatta Chowk Bazaar in Shahjahanabad, its
articulation of the processional imperial axis from the city to
the palace. Figure 3. Chatta Chowk Bazaar, 1911–12, octagon and
arcade; photo: Archaeological Survey of India (ASI) Annual
Report.
162
3 STRUCTURAL VALUES IN THE BUILDING
163
Figure 7. Pendentives used in the cells, photo of corner.
164
Figure 10. Sunken roof above “Darwazah-i-Chatta-
Chowk”.
Figure 9. “Safe” structural design of Lahore Gate and the pendentives has been employed; this also min-
arcade interface highlighting the critical components in the imizes threat to structural safety through equitable
diagram. distribution of stresses.
Examination of the intersection with the arcade sys-
tem reveals the use of a combination of a segmental
space, owing to planar design of the elements, are the
dome, which is a three-dimensional composite ele-
structurally the most fallible.
ment and dual piers at each end (Figure 9,A) of spans
of 4.8 m (gate octagon) and 8.3 m (arcade). The dimen-
3.5 Gate and arcade module interface sions of the pier sections designed in the octagon are
roughly 1.1 m square, qualifying the extra load taken
The longer ends of Chatta Chowk Bazaar are defined
by them from the arcade arch. The horizontal thrust of
by two distinct gateways. These are the Lahore
the arch-rib of the arcade is directed vertically over the
Gate and the “Darwaza-i-Chatta-Chowk”, the Bazaar’s
curved surface of the segmental dome that links the two
entrance and exit points respectively as one walks into
differently sized arches.. This intersection thus demon-
the palace-citadel.Though different in scale and mildly
strates an elimination of unbalanced thrusts, arising in
different in design, the structural behaviour of these
two very different structural systems.
two spatial zones is quite similar. The more elaborate
Apart from the Lahore Gate, it is also noteworthy
and larger of the two, the Lahore Gate and its junction
that the opposite gateway to the east the “Darwaza-
with the arcade system, shall be discussed here.
i-Chatta-Chowk”, responds to safety in massing. This
Being a complex structural system in itself, this
is done by sinking the floor above the dome to make
needs to be described briefly prior to the examina-
the roof light (Figure 10), a balance between smaller
tion of the intersection with the arcade. Spatially, the
span, lower rise of its dome and dead-load of roof.
internal clear area of this gate is an irregular octagon
(Figure 9, C) that is spanned by a dome (Figure 9,
3.6 Salient aspects in construction
B) touching its longer edges over a distance of 9.3 m.
The vertices of the octagon are defined by eight piers Formal aspects in structural values in Chatta Chowk
that rise to form, two vaults laterally and two arch-ribs Bazaar are reinforced by certain very significant built-
longitudinally, along the longer edges of 4.8 m each. in values in construction. These are the links to the
Four pendentives are formed along the shorter edges building’s stability, durability and timelessness.
of 2.7 m each. These support the large dome overhead Along with secondary technical sources, the pri-
while along the transverse direction, smaller stone mary information on construction was extracted from
columns support a mezzanine spanning the length of exposed parts of the building’s fabric itself. As men-
the vault. The significant transverse depth of the gate, tioned earlier, the hunting lodge at Jaunti, a technolog-
largely of solid masonry, accommodates any lateral ical contemporary to Chatta Chowk Bazaar, was used
thrust. as a resource to understand construction through its
The presence of the pendentives in this system, exposed details.
illustrates a magnified, finer resolved version of the Though the discussion on this aspect is limited
sub-system described earlier for the cells. Here, all by sources of information, logical conjectures were
walls in the cells are replaced by piers, and their drawn to answer questions of longevity and dura-
corners, by the shorter edges of the octagon. This bility of the building’s fabric. As mentioned earlier,
is a visible sign of structural respect warranted by the Bazaar is, with rare exceptions, a compressive
the threefold span. Again, structural reliability of brick masonry structure. Two observations described
165
Figure 11. Wall construction, Jaunti; and possibly Chatta
Chowk.
Figure 13. Built interventions and consequences: profile of
collapsed arch at the octagon; built cubicles on the first floor.
166
Chatta Chowk Bazaar. This building is an irreplaceable
and timeless resource in this regard and will have to
be treated with its due respect. The knowledge sys-
tems approach has helped acknowledge the value of
this resource comprehensively and direct an attitude
for best practices in the future.
In the debate between safety and significance, cases
like Chatta Chowk illustrate that they are actually sides
of the same coin. The most effective route to achieve
safety in such outstanding knowledge systems, is thus
through authentic structural conservation, informed by
Figure 14. Synthetic and natural threats. a systematic method to generate knowledge, given the
significance of these structures.
external historic walls (Figure 14, A). Since 1992,
foliated arches that form the shop openings were also
gradually walled for signboards. ACKNOWLEDGEMENTS
Electrical interventions have been the most marked
where ceiling fans and other such fixtures have been The author immensely thanks Prof. Nalini Thakur,
hung from tubular steel rods that puncture the penden- School of Planning and Architecture (SPA), for crit-
tives in the cells on the first floor. In some cases, the ical guidance, encouragement and contribution in the
buttressing wall has also been punctured to link two or work on which this paper is based, especially on using
more cells from the inside. the Knowledge Systems approach that greatly helped
Apart from the synthetic interventions mentioned the studies. Sessions with co-faculty of studio, Ar.
above, the southeast corner of the building is also Sanjay Bharadwaj, Morphogenesis, Delhi and on site
threatened by roots of a large tree that has grown into resource person, Ar. Rachna Vishwanathan were also
the wall (Figure 14, B). instrumental in guiding the documentation and analy-
ses. The Archaeological Survey of India (Delhi Circle)
5 CONCLUSIONS and the Chatta Chowk Bazaar Traders’Welfare Associ-
ation were very helpful and cooperative during the site
The Chatta Chowk Bazaar today is an expression of a work. Further, the author thanks all his colleagues who
unique technological memory. Historically conceived worked on this project, batch of 2006–08, Masters in
and built with cognizance of processes, techniques and Architectural Conservation, SPA: Allahyar Allahyari,
systems evolved and perfected over time, it is also sig- Bhavesh Patel, Moinak Bose, Muzakkir Bijli, Sidharth
nificant in its geographical (here, Indian) context by Roy & Sonal Chitnis.
being a lone example of a building type. This typolog-
ical significance was achieved through the architect’s
knowledge of the structural and construction systems, REFERENCES
amongst other things, that could make this architecture
not just possible, but also durable and timeless. Asher, Catherine B. 1992. The Architecture of Mughal India.
New Delhi: Cambridge
For substantive knowledge generation as explained
Croci, Giorgio. Structural Preservation of the Architectural
above, the Chatta Chowk Bazaar and its context on Heritage. Structural Aspects in Restoring Monuments,
the ground were used as resources to firstly inform its symp. Rome 1993.70: 15–18
typological significance and then extend it to include Koch, Ebba. 2002. Mughal Architecture ,New Delhi: Oxford
structural values. It substantiates how the building’s Thakur, Nalini. 1988. Shahjahan and his relevance to the
values are linked to the contemporary need for its future of Shahjahanabad. Indian Institute of Architects
safety and the safety of its users. The issues also indi- Conf.., Goa, 1988: 8–15
cate that the practices that undermine the Bazaar’s Thakur, Nalini. 2006. Technological Knowledge Systems
typological significance as a historic mall also under- Approach - from little insights to a paradigm shift in struc-
tural knowledge and application. Structural Analysis of
mine its safety because of this significance embodying
Historic Constructions Vol. 1: 25–30
numerous structural values. Thakur, Nalini. 2002. The “Architectural Knowledge Sys-
The above definition of significance notwithstand- tems” Approach: as a Solution for the regeneration and
ing, the scope for further enquiry is immense and will Conservation of Indian Heritage. Architexturez: South
contribute further to the body of knowledge about Asia: 1–10
167
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The study deals with the possibilities of rehabilitation of historical buildings while preserving
their authenticity, as much as possible. It starts with a comprehensive definition of the historical constructions
and emphasizes the features of their authenticity. The study shows that the issue of historical constructions is
included in the overall problem of rehabilitation, but with specific requirements regarding authenticity. It presents
the basic principles of the historical buildings’ rehabilitation. Case studies regarding the structural strengthening
and the possibilities of preserving the authenticity of historical buildings during their rehabilitation are presented.
They are ordered increasingly, according to the degree of the intervention needed.
1 INTRODUCTION
169
action (ground sliding, severe earthquake, environ- at the same time, the rehabilitation decision on the
mental catastrophe etc.). The rehabilitation is also structural interventions could be a compromise.
needed when one wants to keep the building func- Sometimes, the necessary functional improvement
tional, beyond its normal service life.This is the overall of an old building implies adding new blocks, preserv-
case of the historical constructions. ing the structural and architectural authenticity of the
original one.
As a general demand, it is to be mentioned that the
added structures and structural elements should have
2 REHABILITATION OF HISTORICAL
an appropriate degree of reliability with those of the
CONSTRUCTIONS PRESERVING
original structure.
AUTHENTICITY
170
Table 1. Strengthening, preserving the original structure.
171
Table 2. Strengthening, adding compatible structural elements.
172
Table 3. Indirect strengthening providing additional bearing structure.
173
Figure 3. New staircase – exterior view.
Figure 4. New staircase – interior view.
original structure shall be repaired or restored. From
that point on, it has to carry only its own weight. solution should respect some basic principles, but
Another idea is that of using additional helping sys- every case represents a new challenge. The specialist
tem in order to correct displacements or deflections has to establish the priorities in preserving authentic-
that affect valuable elements of authenticity of the ity, without neglecting the functional and structural
historical building. reliability of the building.
Sometimes, the necessary functional improvement
of an old building could imply adding new blocks to the
building, preserving the structural and architectural REFERENCES
authenticity of the original one.
For instance, in the case of the Palace of Jus- Bucur-Horváth, I. et al. Study on old masonry structures in
tice in Odorheiu Secuiesc, Romania, built in 1835, brick vaults. In Dr. GörünArun & Dr. Nadide Seckin (eds),
belonging to the architectural heritage, an improve- Studies in ancient structures; Proc. 2nd intern. congress,
ment of the circulation flux was necessary, responding Istanbul, 9–13 July 2001. Istanbul: Yildiz Technical Uni-
to the current requirements concerning emergency versity, 2001.
Bucur-Horváth, I. et al. Theoretical and experimental stud-
exits (Bucur-Horváth 2007). A new staircase made of
ies for strengthening Bohemian brick vaults. In Clau-
reinforced concrete and glass was added to the ini- dio Modena, Paulo B. Lourenço & Pere Roca (eds),
tial building (Figs 3–4). Placed in the inner courtyard, Structural Analysis of Historical Constructions; Proc.4th
this very effective modern solution emphasizes the itnern. seminar; Padova,10–13 November 2004. Leiden:
historical character of the old building. Balkema, 2004.
Bucur-Horváth, I. et al. Researches on historical construc-
tions built in several stages. In Paulo B. Lourenço,
4 CONCLUSIONS Pere Roca, Claudio Modena & Shailesh Agrawal (eds),
Structural Analysis of Historical Constructions; Proc.5th
intern. conference; New Delhi, 6–8 November 2006.
Any intervention on a historical building has to be
Delhi: Macmillan India Ltd., 2006.
performed with great attention. Besides the general Bucur-Horváth, I. & Bulbuk, M. Original solutions for struc-
demands concerning functionality, as well as structural tural and functional rehabilitation of masonry buildings.
reliability, the rehabilitation process has to respond to In Görün Arun (ed.), Studies on Historical Heritage; Proc.
some specific requirements regarding authenticity. In intern. symposium; Antalya, 17–21 September 2007.
this sense, the rehabilitation decision and technical Istanbul: Yildiz Technical University, TA-MIR, 2007.
174
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The present paper focuses on the experiences gained, methods used and attitudes assumed while
formulating a management system for the Champaner-Pavagadh Archaeological Park. It brings out the conflicts
which occur when there are several stake holders managing the site; who have good intention but narrow rather
limited purview of action and vision; and ponders whether solutions to these conflicts can be prioritised with
mutual consent as a strategy for effective conservation management of the site.
175
Figure 4. Quickbird satellite image of the site.
Forest Department has under its purview a large area the Gaben Shah Vav. (Refer Figure 6). It protects only
of the site, mainly the Pavagadh Hill and the buried those buildings that are more than a 100 years of age
Sultanate city. The forest act is very powerful and and only the ones in stone! Its law does not allow for
has successfully controlled development in the for- protection of significant buildings like Iteri Masjid,
est land thereby protecting the authenticity of the site. Rani no Mahal, etc.; simply because they are brick
But, at the same time, its obligation of planting fresh structures, in spite of their being important knowledge
saplings every year has proved to be a big threat to resource!
the archaeological heritage. The roots of the planta- The District Collector and the DDO once spent
tions are slowly harming the vulnerable heritage buried enormous money and energy constructing the water
below the earth. (Refer Figure 5). The overgrown for- facility at a heritage site. But it happened to fall within
est also makes accessibility to most structures very the 300 meter boundary of ASI and had to suffer
difficult leading to their subsequent deterioration. demolition even before it started functioning. These
ASI protects, out of the identified 114 structures, uncalled-for conflicts and a constant struggle for asser-
only a mere 55 and that too in isolation. It has cre- tion of one authority over the other are is not helping
ated islands of protected territories within the entire the site, rather they are wasting resources. A system
Archaeological Park. Paradoxically enough, it was the is needed which enables co-ordination among various
ASI which nominated this place as a site to WH com- government departments.
mittee; but when it comes to protection it can not see Gram Panchayat promotes programmes for peo-
beyond its 100/300 m absurd irrational law. ple’s development like repairing of pilgrims’ path,
To add to that the State Department of Archaeol- creation of tourism oriented business opportunity, etc.
ogy has just one monument under its jurisdiction i.e., Recently, Gram Panchayat has constructed a school in
176
Figure 7. Rampant growth of Kalikamata temple Precinct.
177
Figure 11. Champaner village.
Figure 9. Malik Sandal Vav before repair work.
178
efforts go into pleasing them often overlooking the Bayley, Sir Edward C. 1970. The local Mohammadan Dynas-
vulnerability of the site. ties, Gujarat, New Delhi: S. Chand and Co. Ltd.
Brown, P. 1996. Indian Architecture – Islamic Period, Bom-
bay: D B Taraporewala Sons & Co. Pvt. Ltd.
Burgess, J. & Cousen, H. 1975. Architectural Antiquities of
4 REASONS FOR OCCURRENCE OF Northern Gujarat, Delhi: Archaeological Survey of India.
CONFLICTS Burgess, J. 1994. Muhammadan Architecture of Bharuch,
Cambay, Dholka, Champanir, and Mahmudabad in
Main conflicts arise: Gujarat, New Delhi: Archaeological Survey of India.
Department of Landscape. 2001. Champaner-Pavagadh
1. Due to non availability of complete information Archaeological Park Plan, Urbana Champaign: University
regarding various issues to each stake holder. of Illinois.
2. Due to vast amount of cultural resources lying un- Chavan, A. R. 1966. The Flora of Pavagadh, Baroda: Depart-
possessed and hence unattended by any agenda of ment of Botany, Faculty of Science, M. S. University.
any stake holder. People for Heritage Concern. 2001. Cultural Resource
3. Due to pilgrim/tourist oriented opportunities which Information System – Inventory of Built Heritage of
make the site economically self-sustained but also Champaner-Pavagadh, Baroda: Heritage Trust.
Faridi, Fazlullah L. 1990. Mirat-i-Sikandari – A Study of
highly vulnerable. Medieval History of Gujarat, Gurgaon: Vintage Books.
4. Due to the presence of a living village whose aspi- Modi, Sonal M. 2002. Water Intelligent City, Land-
ration of a better life-style can never be overlooked scapes of Water, History, Innovation and Sustainable
on any grounds whatsoever. Design, Politechnico di Bari – Facolta di Architet-
tura, Dipartimento di Scienze dell’Ingehneria Civile e
dell’Architettura (ICAR), Bari, Regione Puglia, Italy.
5 PROPOSED MANAGEMENT SYSTEM Goetz, H. 1949. Journal of the Gujarat Research Society,
Volume XI, Issue No. 2.
Joshi, G. 1999. Pavagadh Darshan, Ahmedabad: Chiragh
Hence a four point agenda has been formulated Printers.
which enables management of Information, Cul- Mehta, R. N. 1979. Champaner – An Experiment in Medieval
tural Resource, Visitor/Pilgrim, Site or the Physical Archaeology, Baroda: Ajanta Publications.
Resource. Mehta, R. N. 1979. Champaner – Ek Adhyayan, Baroda:
The Management plan proposed is an attempt at Maharaja Sayajirao University.
prioritising actions and at reducing conflicts if not Mendonca, Francis A. 1981. Chronica do Reyno de Gusarate,
doing away with them altogether. Govt of Gujarat, with Ed., Misra S. C. & Mathew K. S., Department of History,
the constant persuasive efforts of Heritage Trust, has Faculty of Arts, Maharaja Sayajirao University, Baroda.
Modi, S. 2004. Impressions of a Forgotten City – Archi-
recently instituted a legal authority with such inten- tectural Documentation of Champaner-Pavagadh, New
tions. But it is still a long way before various stake Delhi: Archaeological Survey of India & Heritage Trust
holders open up their constricted vision and co-operate Baroda.
towards a peaceful co-existence. Convenient misinter- Misra, S. C.1985. Tarikh-i-Mahmus Shahi, Baroda: Depart-
pretations of the authority by lazy officials are leading ment of History, Faculty of Arts, The Maharaja Sayaji Rao
to several bottlenecks in the development procedures, University of Baroda.
yet all efforts are being made to ensure a continued Patel, G. D. 1972. Gazetteer of India – Panchmahals District,
collaborative effort between Panchayat, government Gujarat Ahmedabad: Government Publications Depot.
bodies and professionals. Sandesara, B. J. 1973. Translation of Gangadhar Prani-
tam, Gangdas Pratapvilasa Natkam, A Historical Sanskrit
Play, Composed 1449, Baroda: Oriental Institute.
Thakur, N. 1987. Champaner- draft Action Plan for Integrated
6 CONCLUDING REMARK Conservation, Baroda: Heritage Trust.
Thakur, N. 2000. Note on the legal Status of Champaner,
It is indispensable that Conservation is inte- Unpublished.
gral to an Urban Planning Process enabling a Thakur, N. 2000. Participatory Conservation Collaborative:
Progress-Oriented-Approach because… Heritage is A Regeneration Programme for Champaner-Pavagadh,
Unpublished.
living… ageing… and… it wants to let fade away… to Thakur, N. 2000. Archaeological Park as a tool for Inte-
give way to a people’s right to human living grated Protecting Heritage Management with Planning
conditions. Process: The case of the deserted 15th century capital site,
Champaner-Pavagadh, Gujarat, Goa.
Tripathi, K. 1971. Pravasdham Pavagadh; Baroda: Diamond
BIBLIOGRAPHY Publication.
Watson, J. W.1877. The Indian Antiquary, A Journal of
Alfieri, Bianca M. 2000. Islamic Architecture of the Indian Oriental Research, Vol. VI.
Subcontinent, Ahmedabad: Mapin Publishing Pvt. Ltd.
179
History and documentation
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
A. Kutnyi
Department of architectural history, building archeology, heritage conservation, Munich Technical University,
Munich, Germany
ABSTRACT: From 1927–1929, the renown Russian engineer Vladimir Grigorevic Suchov (1853–1939)
designed and built several towers for the Nigres power transmission line. As one of the last and most refined
hyperboloid steel lattice towers, they represent the culmination of his lifelong quest in the development and
optimization of this structural system.
1 INTRODUCTION
183
Figure 2. Plan view of the 5 hyperboloid sections.
184
Figure 5. Assembly of the structure using the telescope
method. (Graefe, 1990).
185
assumptions, steel member design as well as the design This procedure is repeated with the respective over-
of the riveted connections and the foundations. turning moments to design the L-shaped verticals of
Referring to load combinations given by the the higher hyperboloid sections.
Russian building code, Suchov states that the high- Finally, the riveted connections of the first verti-
est forces in the tower are caused if “the transmission cal of each section are designed. The combined shear
lines aren’t ruptured, no ice build-up exists, and the and bearing pressure appears to be at the maximum
wind pressure is 250 kg/m2 ”. 800 kg/cm2 for 7/8 inch rivets. 7/8, 3/4 and 5/8 inch
The design wind pressure of 250 kg/m2 – constant rivets are used for the connections according to the
over the height of the structure – is used to calculate calculations.
the resulting forces on the different hyperboloid sec- In addition, the calculations entail the design of
tions and at the base. This design pressure is applied anchor bolts, the steel members at the top, the stabil-
on all the vertical members, using their unprojected ity against overturning of the structure as well as the
length times the width of the steel angles as the design of the concrete foundation. There is no struc-
reference area. tural design of ring elements or lattice girders included
To account for wind shielding at the sides of the ring in the original calculations. (Suchov 1927)
elements, Suchov uses the following formula with the
diameter D and the width of the steel angle b to cal-
culate the resulting horizontal wind force on one ring. 4 NEW ANALYSIS
186
Table 2. Resulting wind pressures c∗f q(ze ) on the steel
members according to DIN 1055-4.
4.3 Stability
To assess the ultimate bearing load of the tower, it
was critical to study its stability behaviour. Therefore,
a geometrical non-linear stability analysis had to be
performed. For that reason, it was essential to find
suitable imperfection shapes.
Following recommendations by Graf, the imperfec-
tion shape of grid shell structures should be deter-
mined by scaling the first eigen-value mode shape
under ultimate load. The first oscillation period was
determined in the modal analysis as 1.02 seconds.
In the case of a tower-like structure, the eigen-value
mode shapes are obviously extremely from the buck-
ling shape of the perfect geometry. The structure of Figure 6. Buckling shape, based on the perfect geometry.
the first hyperboloid section proved to be the most The plan view shows the first hyperboloid section only.
critical. The buckling shape shows six evenly around
the perimeter line distributed bulges, which are con-
fined to the lower half of the section. This buckling
shape of the perfect geometry was scaled and imposed dividing the E-modulus of elasticity by the coefficient
on the structure as the imperfection shape. The max- γM = 1.1. The load increase factor of the imperfect
imum deformation was selected rather conservatively geometry was determined as 4.39, compared to 4.87 of
as 150 mm, a deformation considered to be visible to the perfect geometry. This is a rather modest decrease
the naked eye. compared to other grid shell geometries.
According to DIN 18800-2, the stiffness of struc- Hyperboloids and some other anticlastic formed
tures susceptible to buckling has to be reduced by shells are typically less sensitive to imperfections, due
187
Table 3. Comparison of forces in the first hyperboloid
section under different loadings.
LC 1 −30 KN 20 KN –
LC 2 −42 KN 29 KN –
LC 3 −393 KN −245 KN 11 KNm
+309 KN +303 KN
LC4 −132 KN −39 KN 3 KNm
+48 KN +99 KN
LC 5 −281 KN −169 KN 7 KNm
+219 KN +212 KN
LC 6 −246 KN −131 KN 8 KNm
+167 KN +184 KN
188
depending on their twisting angle. This imposes a ver-
tical distortion of the lattice girder at the end of each
section.
The discontinuity of the verticals between the dif-
ferent hyperboloid sections results in a decrease of
the shear stiffness, causing large deformations in these
areas. The lattice girders are subject to linearly increas-
ing compression forces on the front face and tension
forces on the opposite side, thus balancing the normal
forces of the inclined verticals below.
The bending action of the tower causes the lattice
girder to ovalize, thereby inducing bending moments
around the strong axis of the member. (fig. 8) In addi-
tion, the girders are even more affected by the push-
pull action of the sidewise verticals, acting around their
weak axis. The ring elements adjacent to the lattice
girders get affected by the distortion in this area as
well, causing overstress due to bending.
Based on the current analysis and load assump-
tions, the structure would not be adequate to sustain
the 50-year wind, due to buckling of the verticals in
the first section. The structure including transmission
lines would only satisfy code requirements if the fac-
tored wind loads are reduced by 76%. Even the current
condition without the transmission lines would not be
sufficient to sustain the 5-year wind.
Despite of some minor local overstressing, almost
all other members of the structure are suitable (load
case 3).
5 CONCLUSION
189
Innsbruck, in cooperation with Rainer Barthel, Murat Gioncu, N.B. 1992. Instability behaviour of single layer retic-
Gappoev, Uta Hassler, Alexander Kolesov, Igor Molev, ulated shells. In: International journal of space structures.
Manfred Schuller, and Tatjana Vinogradova. Vol.7. No 4.
Graefe, R. 1990. Die Kunst der sparsamen Konstruktion.
Stuttgart: Deutsche Verlags Anstalt.
Graf, J. 2002. Entwurf und Konstruktion von Translationsnet-
REFERENCES zschalen. Stuttgart: Verlag Grauer.
Schueller, G. 1981. Einführung in die Sicherheit und Zuver-
De Vries, P. 1999. Morphology and structural behaviour of
lässigkeit von Tragwerken. Berlin: Verlag Ernst und Sohn.
the hyperbolic lattice. In: 4. International colloquium on
Suchov, G.1927. Structural calculations of a 128 m tall tower.
structural morphology. Delft: University of Technology.
city archive of Nizhniy Novgorod.
DIN 18800. 1990. Stahlbauten.
DIN 1055-4. 2005. Einwirkungen auf Tragwerke –
Windlasten.
190
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The recent investigations on the Brunelleschi’s Cupola of Florentine Cathedral were devoted
above all to identify masonry structure and to analyse the construction of the two domes (inner and external shells)
in connection with the real building problems found in the works. Exact information on the characteristics of brick
masonry has been obtained. Other important data have been supplied by georadar investigations, demonstrating
that the inner dome is made of two (intrados and extrados) curtains with interposed a different masonry. These
new studies have been put in relation with the surveys of the dome, reaching original conclusions, far from the
common mythicized interpretations of Brunelleschi’s Dome. In the construction a fundamental role was played
by the executive slowness.
1 INTRODUCTION
191
Figure 2. Cast of the brick masonry taken after the removal
from dome intrados of some plaster with Zuccari’s mural
paintings. A black vertical line marks the corner of the dome.
Figure 3. The survey of part of dome intrados masonry
points out discontinuities in herring-bone bonds pattern at
as the year of completion of the second dome segment. the corner. Bricks different in size were used, frequently cut
So, including the year 1430, the dome portion between to fit them to the space left between the herring-bone bonds.
the second and third corridors was built in fifty-eight
months.
The construction of the upper section, between the reduction of the employed masons due to the less space
third corridor and the lantern, started in July 1433, was available on the scaffoldings, it is evident the increased
concluded by March 1435, as the following year the speed in the construction of the upper part of the dome.
serraglio (the lantern ring base) was begun. So this It could seem contradictory that works proceeded
final segment was built in twenty-one months. quicker just where the brick beds inclination was
It has been possible to evaluate the average raise of greater. However at this level a centring was probably
the wall per month in each of the three dome sections, laid resting on a system of beams inserted in open-
putting in relation the building period with the height ings correspondent to the small passages opened to the
of the built masonry. For the first section of the dome, inside of the cupola. The openings were then partially
high 5.50 m and built in 39 months, this average value walled up to build the present oculi.
corresponds to a 14 cm masonry per month; for the
second section (12 meters in 58 months) the value is
nearly 21 cm per month, in the third section (15 meters 3 THE HERRING-BONE APPARATUS
in 21 months) 71 cm per month. On this basis, the brick
courses laid each month would result about two, three Most of the recorded herring-bone bonds is located
and ten respectively in the three sections. between the second and third corridor, where a very
Limiting the working period to ten months per year, irregular laying of the bricks and relevant variation of
due to weather conditions, the values of dome ele- their inclination are visible.
vation are nearly the same: respectively 17, 27 and After the dome intrados survey, the zenith angle
94 cm masonry per month in the three sections. The of the herring-bone traces has been measured. It has
results would not be different from the previous ones. been verified a very wide range of the inclination (from
In the years 1422–1430 the dome works proceeded circa 31◦ to 55◦ ) among the eight faces of the dome and
very slowly: up to the third corridor the wall was raised among the herring-bone bonds of the same face, with
less than a brick course per week; on the contrary, a the prevalent inclination from 36◦ to 40◦ . It depends on
relevant increase is recorded in the upper part built the different way of laying the bricks, using more or
between 1433 and 1435. less quantity of mortar in the joints, as an evidence
It is necessary to consider that above the third cor- of the unequal way of working of the different teams
ridor the relevant narrowing if the dome reduces the of masons.
built masonry volume to nearly 1/3 in comparison with In particular, in the South-East face of the dome
the masonry of the second section. However, even con- much more inclined herring-bone bonds than else-
sidering such a difference, which caused a consequent where have been recorded. It seems to prove the
192
Figure 4. The eight faces intrados (by stereophotogrammetric survey) with the traces of herring-bone lines to point out their
different inclination, ranging from 31◦ to 55◦ .
193
bricks, sudden changes in their inclination, splitting of
herring-bone bonds, . . .).
In the case of herring-bone bond crossing the dome
masonry through its total thickness following a unique
curve centre, very different dimensional data should be
expected: two adjacent herring-bone bonds distant 60
cm on the intrados of the inner dome, on its extrados
the distance should be 65 cm and on the outer dome
respectively 68 and 70 cm, distances only limitedly
greater than the ones recorded on the inner intrados.
The dimensional heterogeneity that has been recorded
in the building is totally different, even if the rele-
vant slowness of the works should have made easy to
control the geometry of the herring-bone lines.
Figure 6. The georadar diagram of the inner dome dis- For these reasons it seems improbable that the
plays two symmetrical discontinuities (at the intrados and
the extrados) in the brick masonry, and a different inner core.
herring-bone bonds cross radially the whole of the
dome following a unique alignment to the centre. It is
more reasonable that the intrados and extrados herring-
bone bonds of the inner dome are not continuous, but
belonging to two different masonry layers. The exis-
tence of two distinct layers, the inner and the outer
parts of the wall, has been proved by georadar inves-
tigations, and can also be supported by considerations
relating to the building technique used in the works.
Moreover, it seems proved that the herring-bone bonds
of outer dome are totally independent from those of the
inner dome.
194
Figure 8. The accurate brick pattern visible in the serraglio
at the top of the dome (left) and in the passages through
the ribs (right). Here the flat brick line carved to suggest a
different pattern is pointed out.
Figure 9. Side wall of a passage to an oculus open to the It is meaningful the superimposition of the detailed
inner space of the dome. Basing on georadar investigations survey of the just described extrados masonry of
(diagrams on the right) the very accurate visible wall appara- the outer shell on the measured drawing of the sub-
tus is just a facing, uneven and different from the remaining
dome masonries.
horizontal cantilever-arches located on the intrados
between the corner and intermediate ribs.
Above all it has to be underlined the peculiarity
of their masonry structure: in the lower part head-
brick masonry was represented by that one of the pas- ers are laid to form a segmental arch thick one brick,
sages. On the contrary, it seems even possible that the while above this arch some courses of stretchers fol-
passage facing was placed during further restoration low the brick laying of the ribs and of the outer dome.
works. The bricks of these cantilever-arches, however laid,
Next to the serraglio the situation is the same: the are connected to the ribs and the outer shell; those
intrados masonry of the inner dome is very confus- ones above the segmental arch are laid together with
ing, with irregular and segmented herring-bone bonds, the bricks of the shell (at least at the lower levels,
whilst in the passage at the base of the serraglio the as proved by the presence of herring-bone bonds in
visible brick masonry is as accurate and refined as the some of them). The segmental arches continue beyond
one just examined in the lower passages. So this kind the extrados up to the external surface of the shell,
of masonry appears once more to be only a facing, as documented by the superimposition of the upper
and this is also confirmed by the outer face of the cantilever-arches and the extrados survey. The plaster
same serraglio wall, visible at the top of the space absence in the extrados of the outer dome gives the
between the two domes: under the plaster a usual thick opportunity to identify nearly the total straight-arch of
joint masonry, the same found everywhere but in the a cantilever, and partially a second straight-arch placed
passages, emerges. at the same level on the opposite side.
195
The reducing cross section of those arches, typi- herring-bone row and in the sub-horizontal masonry
cal of cantilevers, and the consequent wider section between them as well as.
in correspondence of the corner ribs, transfers on the A further aspect has to be kept in mind. At the intra-
latter ones most of the stresses. Furthermore, it can be dos the dome corner, measured from the gallery to the
imagined that the thickening of the cantilever-arches base of the lantern oculus, is only a little more than
at the corner rib could allow to cross them diagonally 39 meters, while at the middle of a face the corre-
and to make possible the laying of a continuous iron sponding length is 3 meters shorter. In the works, this
tie, shaped as a multi-sided polygon and so much more different length (nearly 10%) required a greater num-
effective. ber of brick courses in the corners, and this necessarily
caused further irregularities.
During the construction these work adjustments
6 THE CORDE BLANDE were made easier by the masonry discontinuity given
by the herring-bone bonds, that caused the included
The corde blande, the curved laying of brick courses, parts of masonry to be independent the one from the
progressively increasing from the base to the top of other. For this reason such adaptations are nowadays
the dome, are well documented by the survey of the nearly invisible when examining the wall.
dome extrados. This special laying system was caused The marble mouldings at the external base of the
by works needs, as it was necessary to give continuity lantern are apparently curved, and so they seem to
to the differently inclined brick beds in the two cor- reply the curved line of the corde blande. In fact,
ner ribs and in the central part of a single face. The the masonry which supports them is straight and the
consequent curved laying of the bricks was got and curved line of the mouldings is caused by the brick
made progressive not only with varied thickness of masonry settling, which was not followed by a cor-
the mortar joints, but introducing discontinuities in the respondent subsidence of the marble covering of the
corner ribs. Here the joints between the elements
are extremely reduced and so subject to only lim-
ited dimensional reductions for plastic deformation or
shrinkage. On the contrary, these phenomena are rele-
vant in all the walls of the dome, built with thick joint
masonry (Petrini 1989).
Figure 12. Detail of a cantilever-arch. The vertical bricks at the intrados (left) form a segmental arch supporting the above
masonry, laid on sub-horizontal curved beds (a corda blanda) homogeneous with the outer dome masonry.
196
now their presence has been put in relation with the extrados, indicates that probably this element is the
necessity to fasten intrados suspended scaffolding, or anchor of an iron tie crossing radially the main shell.
to fix corner wooden centrings (Dalla Negra 1995). Broader investigations to find out iron presence in
The same works praxis has been proved to be previ- the dome were carried on by metal-detectors by Italian
ously utilized in the construction of aisles and nave of Army (Compagnia Genio Guastatori della Brigata
the church, where iron elements used to fix centerings Motorizzata “Friuli”). Exterior wall stripes and floors
have been found in transverse arches and diagonal ribs were examined at the level of the three corridors. Geo-
of the nave vaults. radar investigations made later on the inner dome
Some iron bars inserted in the dome masonry are revealed signs, probably related to the presence of
visible. They could be part of a more extended tying iron bars.
system, which probably has played a relevant role in Basing on these investigations, it comes out that
building static condition. the builders kept in mind the different behaviour of
At the dome spring, just above the inner gallery, the two shells, placing reinforcing iron bars mainly
where should be located the first stone tie, many eyed in the corridors. In any case it has been ascer-
iron bars stick out of masonry among stone cantilevers. tained the conspicuous use of radial iron bars in the
The bars probably cross nearly the whole of the dome stone masonries at the dome spring. Here the pres-
masonry. ence of large stone elements could have required a
Detailed observation has been made at the level of widespread use of iron cramps, unnecessary in brick
the third corridor in the space between the shells at masonry.
the North-Eastern side of the dome. Here many iron The late introduction of the cantilever-arches, start-
elements are visible: a radial bar laid aside a central ing from the second corridor, probably suggested also
rib; two transversal bars inserted at the intrados of the usage of almost continuous perimetral iron bars.
stone ties; a third bar, protruding from the masonry This kind of hooping, present in a more recogniz-
of the inner dome in coincidence of its set back. In able form at this level, seems absent above the third
this last case, the bar position, almost parallel to dome corridor.
Figure 13. Survey of the extrados summit of the outer North-Eastern shell. The projection on the outer surface of inner ribs
and canti-lever-arches points out the exact coincidence between the intrados arch of the latter and the vertical bricks of the
survey. The herring-bone bonds, though irregular, also go on intermediate ribs.
197
The use of metallic connections along the rib cor-
ners is probably a consequence of the difficult toothing
of ribs and herring-bone masonry of the dome. Any-
how all that proves, even if in outstanding structures
as the dome and nave of Santa Maria del Fiore, the
widespread usage of iron reinforcements in Medieval
and Early-Renaissance buildings.
REFERENCES
Dalla Negra R. (ed.) 2004. La Cupola di Santa Maria del
Fiore. Città di Castello: Sillabe.
Dalla Negra R. 1995. La cupola del Brunelleschi: il cantiere le
indagini, i rilievi. In Acidini Luchinat C. & Dalla Negra R.
(eds.). La cupola di Santa Maria del Fiore. Il cantiere di
Figure 14. Masonry at the connection of the outer shell, restauro, 1980–1995: 21–22. Roma: Istituto Poligrafico e
corner rib and cantilever-arch. Everywhere very thick mortar Zecca dello Stato.
joints are visible. Giorgi L. & Matracchi P. 2006. Santa Maria del Fiore, fac-
ciata, corpo basilicale, cupola. In Rocchi 2006: 277–324.
Petrini G., 1989. Su alcuni recenti rilievi architettonici per il
restauro dei monumenti. In XY. Dimensioni del disegno,
V, n. 11–12, pp. 147–149.
Rocchi Coopmans de Yoldi, G. (ed.) 2006. S.Maria del Fiore.
Teorie e storie dell’archeologia e del restauro nella città
delle fabbriche arnolfiane. Firenze: Alinea.
Rocchi, G. et al. 1988. S.Maria del Fiore: Rilievi, documenti,
indagini strumentali. Interpretazione. Il corpo basilicale.
Milano: Hoepli.
Saalman, H. 1980. Filippo Brunelleschi. The Cupola of Santa
Figure 15. Survey detail of the outer shell at the base of the Maria del Fiore: 263. London: A. Zwemmer.
lantern. At the centre, the marble moulding bends fitting to
the slight lowering of the dome. At the corners, where the
marble covering of the rib act as a strut, a hollow space (in
black) is formed.
Figure 16. Near the edge between the rib (left) and the extra-
dos of the inner dome (right) a metal bar comes out the set
back. The bar is probably the anchor of an iron tie crossing
the main dome.
198
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Wells Cathedral: The crossing and the central tower. A new hypothesis on the
constructional sequence and dating of the central tower
G. De Blasi
Department of Art, Science and Technology (DASTEC), University of Reggio Calabria, Italy
ABSTRACT: The focus of the present study is the crossing and the central tower of Wells Cathedral. An
analysis of the published research show that this portion of the Cathedral has been inadequately treated, and
even the two most recent monographs by Sampson and Rodwell add relatively little about the inception, the
heightening of the central tower, and the insertion of the strengthening of the crossing. The present study seeks
to advance understanding of the monument in two ways: a detailed architectonic survey of the central tower and
in-depth analysis of the biographical sources and documentary evidence.
1 INTRODUCTION
199
Figure 3. The structural section of the first row of arches
(left) and the structural section of the second row of arches.
200
Figure 5. The second row of arches elevation (left), and
plan (right). It is possible to see the difference in the masonry
Figure 7. Two portion of the plan at the third level: left the
blocks between the lower and upper part.
section is about the grids level, right the level above the floor.
201
Figure 8. The third level in section and elevation. The sec-
tion drawing enhances the peculiarity of the structure: two
shells divided by an empty space and connected every 7
blocks by a transversal one. The elevation shows signs of
a precedent roof system.
202
the tower is more complex outside than inside. Taken
together with the fact that during the external survey
some marks of whitewash were found, this could indi-
cate that the masonry was intended not to be exposed
to weather. The same kind of washes has been found
on the central tower of Salisbury Cathedral, indicating
a relatively common practice of the Middle Ages. It is
uncertain if they belong to the original building or if
they constituted a sort of ‘restoration’work, conducted
after the construction of the tower. However, an expla-
nation of the presence of the whitewash could be that in
this way the central tower became a sort of landmark,
visible from very far. This hypothesis could explain
the greater complexity and richness of the external
decorative apparatus in respect to the internal one. To
sum up it is possible to affirm that the piers belong to
the Decorated Period and the walling up is definitely
Perpendicular.
In conclusion the analysis of the architectonic sur-
vey on the central tower leads to the identification of
Figure 10. External elevation at grids level. The drawing
enhances the richness of the decorative apparatus.
the following constructional phases: (1) first row of
arches; (2) second row of arches till the third course
above the annulet; (3) the upper part of the second row
of arches; (4) the heightening of the central tower up to
looses its regularity and the blocks of stone are smaller
the present spring line of the lancets; (5) modification
and less regular than those of the piers. This part is
of the crown of the tower, walling-up of the bay and
composed in total by 6 courses of stone blocks. The
insertion of the grids.
first three are higher than the others. Between the third
and fifth course there are the putlog holes. At present
they support no beam. The last course presents a
cavetto moulding. It constitutes a frame at the connec- 3 CRITICAL READING OF THE
tion between the masonry and the roof. Regarding this BIBLIOGRAPHICAL SOURCES
element Harvey (1996) gives an explanation, which
perfectly matches with the fact that this part was built Both Sampson (1998) and Rodwell (2001) place the
after the fire of 1439. Indeed, he shows that this course construction of the central tower at the end of the third
was built with stone blocks of the old parapet, the phase (1200–05), before the ‘Interdict Break’. Accord-
design of which is a peculiar characteristic introduced ing to the data from the survey this is right as afar as
by Thomas of Witney, (Morris 1996). the first section is concerned – i.e. from the extrados of
The previous considerations relating to the walling the crossing vault to the top of the first level of arches.
up and the alterations of the roof indicate that the In the preceding section it has already explained that
‘grids’ were added during or after the walling-up. In the first section and the lower part of the second section
fact if they were introduced in 1356 when the window present both the wall blocks tooled in diagonally, while
bays between the piers were totally open, they would all the shafts moulding have a vertical tooling. As the
have been visible from outside and would not have any diagonally tooling is generally dated before the ‘break’
connection with the double shell wall structure, and and the vertical one after, it is very probable that the
so creating incongruence. Moreover, one of Harvey’s decorative apparatus was finished after the ‘break’.
arguments in favour of their insertion in 1356 is the However, these two sections differ also stylistically.
fact that they present similar mouldings to those of the It has already been established that the upper part of
scissor-arches. In fact, the architectonic survey shows the second level of arches belongs to a different build-
that their mouldings are similar partly to the piers that ing phase. Nevertheless, even the first section differs
bear them, partly to the external piers. Consequently, from the lower part of the second section. The main dif-
taking Harvey’s hypothesis to the extreme conclusion, ference is constituted by the evident fact that the first
they should belong to the building phase of 1315–22, section has four bays per side. Moreover, it has already
and that is clearly impossible. Furthermore, the grids pointed out that the nave roof started initially at a lower
present the ‘interrupted ogee moulding’, which was level. It is possible that after the ‘break’the elevation of
already noted in the description of the external piers. the central tower was altered and probably not totally
Finally it is necessary to take into consideration finished. The upper part of the second row of arches
the fact that the decorative apparatus of this part of level could have been built at the beginning of the 14th
203
century, at the start of the heightening of the central masonry of inadequate strength: the four piers were
tower. The further discrepancy between the lower part well built and solid. No deflection – a common fault
and upper part of the second row of arches level can at crossing – can be seen opposite the thrusts of the
be explained with the damage due to the earthquake in lower arcades. So the problem had to do nothing to with
1248. The construction of the upper part of the second crushing and bending. It would seem that, in contrast
level of arches increased the total height of the cen- to the south-east pier founded on solid rock, the other
tral tower. These two facts, i.e. the earthquake and the three had inadequate foundations and sank to different
increasing of the tower height, can explain the change depths through the subsoil. To this unequal settlement
of the level of the nave roof, because it belongs to the the inevitable moving stresses set up by wind-pressure
same building phase of the upper part of second level must have contributed.”
of arches. Moreover Sampson (1998), speaking of the east
To sum up, the first level of arches is dated 1200– nave, has pointed out that “Colchester . . . .observed the
05; the lower part of the second level of arches and way in which the six eastern nave piers on the south
the decorative apparatus of the first level of arches is side all lean away from the crossing, whereas those
dated after the mid-nave break; the upper part of the to the west of the mid-nave break are vertical. In the
second level of arches has to be dated at the beginning triforium, from the third bay west to the break, extra
of the 14th century just before the heightening of the relieving arches have been built in the rere-arches of
central tower. the opening.”
After the death of Bishop Jocelin in 1243, work Consequently the insertion of the scissor-arches
on the cathedral came to a stop until 1286, when the was due to the subsidence, the wind-pressure and the
construction of the Chapter House begins. Evidently fact that the crossing probably with a spire increased
its construction, finished in 1306, took precedence in the instability of this part of the structure. The added
respect of the completion of the tower for practical load due to the heightening of the central tower
reason, and probably funds did not permit both the increased a problem already present, but it was not
constructions to be undertaken at the same time. More- the main cause of the threatening collapse. Some kind
over, stylistic analysis on the upper part of the second of reinforcement was inserted, i.e. “moulded orders
level of arches dates this part at the beginning of the inserted within the arches next to the crossing in the
14th century, in sympathy with these considerations, easternmost arches of the nave and the innermost
even if all the scholars agree in attributing these first arches of each transept on the west side”, (Harvey
two sections to its inception in the first decade of the 1996).
13th century. Harvey has noted that these mouldings are similar
The heightening of the central tower was carried to those of the Lady Chapel, and dates the insertion of
between 1315–22. Harvey (1996) suggests the name these elements just after the completion of the height-
ofThomas of Witney as a possible master mason for the ening of the central tower, i.e. by c. 1325. Moreover,
heightening of the central tower. The study of Morris in another paper on the state of conservation of Wells
(1991) on the involvement of Thomas of Witney at cathedral, he and Colchester (1981) note that “between
Exeter, Winchester and Wells, based on stylistic analy- the vaults and the roof, above each pier, is a project-
sis, gives further evidence in this direction. He isolates ing block of toothed masonry for abutment.” They
some specific mouldings used by Witney at Exeter explain these elements as the start of temporary stone
Cathedral, such as the wave, the undulating mould- flying buttresses, which were never built. Similar fly-
ing and the use of the ogee moulding. All these kind of ing buttresses were introduced “at the north-west and
mouldings are present on the external fronts of the cen- south-west corners of the crossing from the clerestory
tral tower. However, Thomas of Witney is more famous to main arcade of nave and transept”, (Harvey and
as a structural and architectural specialist, (Morris Colchester 1981). In fact, both the mouldings and the
1991), and these decorative details could have been flying buttresses have to be read as a first attempt to
the production of the masons of his workshop. solve the problem of instability at the crossing, conse-
All the scholars agree that the central tower was quently dating them after the statement of 1338. This
transformed between the 1315 and 1322 and this means that the scissor-arches were inserted in a sec-
caused the sinking of the western piers. Only recently, ond time when these first remedies failed to prevent
Sampson (1998) has pointed out that “settlement was the collapse of the tower.
occurring at the time of the first building, not just in Historians agree that the insertion of the scissor-
the 1330 . . .”. In a documentary statement of 19 May arches started ca. 1338. Only Draper (1981) raises
1338 the church is described as inordinately fractured some doubts on this date, and analysis of the com-
and greatly deformed, by the Chapter, [R.i., 201 (Cal. plex structure of the scissor-arches, indeed makes
I, 239)]. it very difficult to agree with the 1338 dating. As
However, Harvey and Colchester (1981) affirm that Draper has pointed out hyperbole is frequent in the
“the defects were not due to the added load of upon medieval documents. Consequently the statement “at
204
the Chapter meeting of 19 May 1338 as enormiter con- minimum 5–6 cm outside and 1–2 cm inside. The dif-
fracta and confracta et enormiter deformata” cannot ference between inside and outside could be caused
be taken as a certain date. Moreover, it cannot be for- by the diffuse erosion of the external stone blocks.
gotten that before the insertion of the scissor-arches This fact can constitute a verification of the theory
the flying buttresses named above were introduced. already exposed that the distress of the crossing was
All these considerations with the addition of the prob- present at its inception and not because of the height-
lem of the Black Death spread all over Europe in ening of the central tower. The added load of the
1347, suggest the possible conclusion that the scissor- central tower increased the instability of the two piers
arches were built later. Harvey (1996) agrees with of the crossing. The fact that the east nave was leaning
these observations, but the analysis of the mould- was a consequence of the instability of the crossing.
ings convinced him that the scissor-arches could not The first attempt after the heightening of the central
have been constructed after the 1340s. However, the tower was the introduction of the flying buttresses at
study of Morris (1996) on this topic regarding the clerestory level. However, the problem was more com-
scissor-arches of Salisbury Cathedral shows that this plex because the piers needed to be braced in all three
consideration is not valid. A possible starting date can directions, to counteract the thrust toward south and
be 1356 when the Abbot in Glastonbury conceded north-west. The solution was partly already experi-
forty loads of stone from the quarries at Doulting “ad mented at Salisbury. Morris (1996), in his study on
reparacionem magni campanili ecclesieWell”, (Draper the scissor-arches of Salisbury Cathedral has proved
1981). Harvey (1996) has always related this with that they are earlier than to those of Wells, dating them
the construction of the ‘grids’. The architectonic sur- c. 1320–30. Consequently they become the prototype
vey has shown that these elements have been built for those of Wells cathedral. Moreover he explains the
later, together with the walling up of the window bays apparent discrepancy of the more complex moulding
of tower. Consequently, Draper (1981) is right when system of Salisbury strainer arches by differences in
he relates this document to the construction of the the economic resources of the two cathedrals. How-
scissor-arches. ever this cannot be the only explanation. In fact, the
Regarding the scissor-arches it is necessary to Wells scissor-arches have a more complex architec-
understand their architectonic and structural role, even tonic structure. The addition of complex moulding
if this part was not included in the survey. From the would be a disturbing element in the whole balance
architectonic point of view, they can be regarded as and harmony of the structure. Indeed the peculiarity
flying buttresses. In fact, each scissor-arches consists of Wells scissor-arches is constituted by their solution
of two great pointed arches, which are connected by to a complexity of formal architectonic and structural
the respective vertices. However, it is sufficient to problems in a compromised context. The solution was
divide them along a vertical axis instead of a hori- necessarily different from other similar cases, and it is
zontal one. In this way we obtain a sinusoidal curve. its peculiarity that encouraged the necessary exper-
Moreover, if we isolate the crossing piers from their imentation to answer the problem with a unity of
more complex context, it can be easily understood that purpose.
the formal solution could be a sort of flying buttresses
applied from inside instead of the outside. As far as
the formal definition is concerned, there are a lot of 4 CONCLUSION
examples in the supports of the windows and openings
of most Gothic buildings of the period, i.e. Glouces- The findings emerged from the critical reading of the
ter Cathedral, Salisbury Cathedral, Exeter Cathedral, architectonic survey and their comparison with anal-
etc. Moreover, from the 1230s, the use of the scissor ysis of the documentary and bibliographical sources
bracing in roof carpentry became common, (Fletcher invite a new proposal for dating the constructional
1976). sequence relating to the central tower of Wells Cathe-
From the structural point of view, the suggestion dral. To summarize there are 8 building phases of the
that the added weight of the tower was not the main tower: Phase 1 – 1200–05, corresponding to the present
cause of the distress of the crossing and the conse- first row of arches level, probably without the deco-
quent insertion of the scissor-arches may none the less rative apparatus; Phase 2 – after the mid-nave break
remain valid, as the main factor of instability of the c. 1219, second level, corresponding to the present
crossing was due to subsidence and it occurred just lower part of the second row of arches. In this period
after the building of the crossing and the first east bay the decorative apparatus was carried out to the third
of the nave, (Sampson 1998). block above the ring of the annulet shaft; Phase 3 –
However, it is possible to give another explana- beginning of the 14th century, upper part of the sec-
tion. The architectonic survey executed on the external ond row of arches level, to present stringcourse, which
and internal fronts of the central tower shows a very divides it from the grids level; Phase 4 – 1315–22,
interesting fact. The deviation from the plumb line is heightening of the central tower, with probably a spire;
205
Phase 5 – after 1338, insertion of the flying buttresses the west scissor-arch, counterbalances the horizontal
at the clerestory level; Phase 6 – from 1356 onwards, emphasis of the decorative apparatus of the nave. Thus
insertion of the scissor-arches; Phase 7 – after the fire they are the perfect answer to a compromised situation,
of 1439, walling up, insertion of the grids and transfor- in which the pre-existent part had to be altered the least,
mation of the upper part of the tower and roof; Phase and the new insertion had to harmonise with the rest.
8 – c. 1475 onwards, construction of the present fan Despite the contribution of no less than twelve gener-
vault above the crossing by William Smyth. ations of master masons, Wells Cathedral maintains a
The solution at Wells has to be read in the context of unique and impressive coherence.
the complex transformation of the eastern end, i.e. the
crossing, the central tower, the quire, the presbytery
and the Lady Chapel. The aim behind this transforma- REFERENCES
tion was the attempt to canonize William de Marchia
and transform Wells Cathedral into a place of pilgrim- Draper P., (1981) ‘The sequence and the dating of the decorate
work at Wells’, Medieval Art and Architecture at Wells
age. The construction of the West Front in the 13th
and Glastonbury, The British Archaeological Association
century totally hid from the sight the first stage of the Conference Transactions, (1978), Leeds, 30–35.
central tower. Even the building of the second row of Fletcher J., (1979), ‘Medieval Timberwork at Ely’, Medieval
arches did not modify the predominant presence of the Art and Architecture at Ely Cathedral, British Archaeo-
façade. The heightening of the central tower aspired to logical Association Conference Transaction for the Year
make it a landmark, visible from faraway. From this 1976, Leeds, 58–70.
the possibility that the external fronts were covered Harvey J.H., (1996), ‘The Building of Wells Cathedral, II:
with whitewash assumes an outstanding significance. 1307–1508’ Wells Cathedral: A History, edited by L.S.
However, the canons did not succeeded in obtaining Colchester, Wells, Open Book, 76–101.
Harvey J.H. & Colchester L.S., (1981), ‘Wells Cathe-
the canonization of William de Marchia, and the peti-
dral: Architecture and Conservation’, Transactions of the
tion was abandoned after the 1329. Nevertheless, the Ancient Monuments Society, New Series, Volume 25,
building project was carried on with the completion 1981, 104–112.
of the eastern arm. The insertion of the scissor-arches Morris R. K., (1996), ‘The Style and Butressing of Salisbury
can be read as part of the project. Draper has pointed Cathedral’, Medieval Art and Architecture at Salisbury
out that “ at Wells in the 13th century, the transept were Cathedral, British Archaeological Association Confer-
shat off from the laity by the pulpitum on the west side ence Transactions for the year 1991, Leeds, 46–58.
of the crossing because the stalls extended beneath Morris R. K., (1991), Thomas of Witney at Exeter, Winch-
the crossing and the north transept was used as the ester and Wells’, Medieval Art and Architecture at Exeter
Cathedral, British Archaeological Association Confer-
chapter house.” Taking this into account, with the addi-
ence Transactions, Leeds, 1991, 57–84.
tional fact that the eastern end was intended to be the Reid R.D., (1973), Wells Cathedral, Wells.
shrine of the de Marchia, the scissor-arches may also Rodwell W., (2001), Wells Cathedral. Excavations and struc-
have assumed the role of a screen, which separated the tural studies, 1978–93, London, English Heritage.
canons from the laity. They in effect acted as a sort of Sampson J., (1998), Wells Cathedral. West Front Con-
visual and physical frame between the two parts of the struction, Sculpture and Conservation, Stroud, Sutton
cathedral. Moreover, their presence, especially that of Publishing.
206
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M. O’Sullivan
University of Manchester, Manchester, UK
T. Swailes
University of Manchester, Manchester, UK
Senior Engineer, Arup, UK
ABSTRACT: Wrought-iron was the dominant structural framing material from 1850 to 1890. With similar
properties to early mild steel, it is more variable, creating uncertainty in the assessment of existing structures
where sampling and testing opportunities are limited. This paper considers the factors that determine how
historical test data may be used to better inform the assessment of wrought-iron in structures. Strength values for
wrought-iron lie between wide limits. Sources of variability include test methods used, the ‘grade’ or ‘quality’
of wrought-iron, and the type of structural element tested or from which samples have been taken. Yield point
became of great importance when the use of elastic stress analysis for beam design became commonplace and
by 1879 Watertown Arsenal in the USA was publishing stress-strain data. Ten years later testing wrought-iron
was a routine part of the education of engineers at universities.
207
properties deteriorate with further working. Exces-
sive working of iron may have occurred when rolling
complex shapes.
208
became the standard value for all structural work.
(Colburn 1863). French engineers worked to a value of
3.8 ton/in2 (59 N/mm2 ) as the maximum design stress
(The Engineer 1863).
These figures were based on tests that did not
directly measure elastic limit. For example, when
David Kirkaldy conducted his investigation into the
properties or wrought-iron and steel in 1858–61 at
the Napier shipbuilding firm, his instruments could
not measure elastic limit. When Napier and Sons dis-
continued Kirkaldy’s testing program he resigned his
position with the firm so that he could devote his entire
time to materials testing. He designed a new form of
testing machine and set up Britain’s first commercial
testing works in 1865 (Smith 1980). Records of his
tests conducted in 1866 show that by this time Kirkaldy
was able to measure elastic limit.
209
Figure 3. United States Testing Machine at Watertown
Arsenal (The Engineer 1890).
210
Table 1. Summary of tensile test data on British and
American bar iron. Elastic limit values are represented in
Figure 6.
211
Table 2. Summary of test data represented in Figures 8
and 9.
212
Table 4. Summary of tensile test data on angle and tee iron.
(O’Sullivan 2007).
Range 124–253
Mean 197 Figure 10. Modulus of elasticity of bar wrought-iron
Standard deviation 13 (O’Sullivan 2007).
the slag inclusions are excessively elongated test sam- prone to brittle fracture at normal temperatures. Impact
ples can exhibit lower compressive strengths due to the test data indicates that the toughness of wrought is
ferrite matrix being divided into columns which can quite variable. Charpy values for wrought-iron from
buckle prior to yielding of the ferrite (Gordon 1988). an American truss bridge were in the range 34–144
Joules (Sparks 1998) while Charpy values for mate-
rial from another American bridge were in the range
13 BRITTLE FAILURE OF WROUGHT IRON 10–80 Joules. (Green 1999). For a rolled wrought-iron
beam tested at UMIST the Charpy values were quite
Past failures of wrought-iron structural elements low, 10 Joules for the flanges and 23 Joules for the web
have included suspension bridge chain links. William (Steude 2000).
Kirkaldy investigated the failure of a tie-bar from
Charing Cross Station roof in 1905 (Swailes 2005).
More recently a beam failure in a building in Leeds 14 MODULUS OF ELASTICITY
was reported (Bland 1984).
The consequences of failure are more severe for Tensile tests on various American, British and Nor-
some structures than others and some structures are wegian wrought-irons were compiled to produce the
more important. To take account of this in assessment histogram of values for modulus of elasticity shown
different factors of safety could be applied to different in Figure 10. The mean value is 197 kN/mm2 which is
types of structure. However, for reasons of simplicity close to the BD21 value of 200 kN/mm2 . Outlying val-
and lack of sufficient data, use of a single factor of ues are a consequence of experimental measurement
safety is prevalent. error.
Fatigue failure of wrought-iron is an area that
deserves further investigation. Cullimore conducted
useful work on fatigue and determined that the fatigue 15 CONCLUSIONS
limit of wrought iron can be taken to be about one-third
of the ultimate tensile strength (Cullimore 1967). The quality of wrought-iron is highly dependent on
Lack of toughness rather than strength has been manufacturing practices. Firstly the skill of the pud-
attributed to various failures of structural elements. dler dictated composition while the care exercised
Wrought iron from the S.S. Great Britain (Morgan in the forge in piling and rolling iron determined
1996) and Walnut Street Bridge in the U.S. (Green the microstructure of the metal. Wrought-iron was a
1999) showed a high ductile-to-brittle transition tem- batch produced metal, which resulted in considerable
perature indicating that wrought-iron is potentially variation in mechanical properties.
213
Of the various impurity elements carbon and phos- Green, P.S., Connor, R.J., Higgins, C. 1999. Rehabilitation
phorus exhibit the greatest influences by causing of a nineteenth century cast and wrought iron bridge.
embrittlement if present in excess quantity. Tests have Proceedings of the 1999 Structures Congress. Pages
shown excess phosphorus to result in significant loss 259–262.
Higgins, J.H. 1930. Investigation of the effect of phospho-
of impact resistance. rus in wrought iron. Contained in Appendix II of the
High tensile strength should not be used as a mea- report of committee A-2 on wrought iron, Proceedings
sure of quality. Ductility must also be taken in account of the annual meeting of the American Society for Testing
so that together with strength it is toughness that should Materials, Part 1, June 23-27, 1930, Pages 159–186.
be used as the measure of quality. Hutchinson, E. 1879. Girder-making and the practice of
The UK Highway Standard BD21 quotes 220 N/mm2 bridge building in wrought-iron. London: E. & F.N. Spon.
as the characteristic yield strength of wrought-iron. Johnson, J.B. 1939. Johnson’s materials of construction’
The tests examined in the present research so far indi- Rewritten and revised by M.O. Withey & J. Aston, 8th
cate the characteristic yield strength of plate iron to be Edition, New York: John Wiley & Sons Inc.
Jeffrey, R., Seager, J. & Woodhead, J.H. 1959. The mechani-
185 N/mm2 , and characteristic elastic limit of bar iron cal properties of wrought iron. Safety in Mines Research
to be 154 N/mm2 . Establishment, Research report No. 154, Ministry of
Power.
Kirkaldy, D. 1876. Results of an experimental inquiry into the
ACKNOWLEDGEMENTS
relative properties of wrought-iron plates manufactured at
Essen, Rhenish Prussia, and Yorkshire, England. London:
The authors are grateful to Cass Hayward & Partners Kirkaldy Testing and Experimenting Works.
and to Network Rail for sharing the findings of a study Kirkaldy, W. G. 1891. Illustrations of David Kirkaldy’s system
of wrought-iron stresses. of mechanical testing. London: Sampson Low. (located in
The Watertown Arsenal and M.I.T. test reports were ICE archive)
located by Institution of Civil Engineers Archivist Kontos, N. 1996. Investigation of wrought iron as a structural
Carol Morgan. material. MSc Dissertation. UMIST.
Lanza, G. 1887. Notes on the Engineering Laboratories of the
Massachusetts Institute of Technology. Minutes of the Pro-
REFERENCES ceedings of the Institution of Civil Engineers, Part 1. Vol.
91. Session 1887-1888. Pages 347–365. Paper no. 2245.
Barlow, P. 1867. A treatise on the strength of materials. Lanza, G. 1912. The progress in testing full-size pieces under
London: Lockwood and Co. practical conditions, together with locomotive testing in
Bland, R. 1984. Brittle failure of wrought-iron beams. The the United States. Journal of the Franklin Institute. Vol
Structural Engineer. Vol.62A. No.10. October 1984. Let- CLXXIV. No.6. December 1912. Pages 607–620.
ter to Verulam. Page 327. Marshall, C.A. 1887. Compressive strength of steel and
Colburn, Z. 1863. On the relation between the safe load and iron. Transactions of the American Society of Civil Engi-
the ultimate strength of iron. The Engineer. March 6th neers, Vol. XVII, paper no.364, August 1887, New York:
1863. Pages.136–137. American Society of Civil Engineers. Pages 53–110.
Cullimore, M.S.G. 1988. Fatigue and fracture investigation Morgan, J.E. 1996. The wrought-iron of the S.S. Great
carried out on the Clifton Suspension Bridge’, Proc. ICE. Britain. Royal Institute of Naval Architects, confer-
Part 1. 84. (April 1988) pages 309–329. ence proceedings, Historic Ships Design, Restoration &
Cullimore, M.S.G. 1967. The fatigue strength of wrought iron Maintenance.
after weathering in service. The Structural Engineer. Vol. M.I.T. 1897. Tests of engineering materials made at the
45. No.5. pages 193–199 Massachusetts Institute of Technology, 1887–1897, Com-
Department of Transport. 2001. The assessment of highway piled from Technology Quarterly, Society of Arts, Mas-
bridges and structures, Departmental standard BD 21/01. sachusetts Institute of Technology. (Produced by M.I.T. in
Fairbairn, W. 1864. Iron its history, properties, & processes book form. A copy is located in the ICE archives).
of manufacture. Edinburgh: Adam and Charles Black. O’Sullivan, M. 2007. First year PhD report. University of
Gale, W.K.V. 1977. Historic Industrial Scenes Iron and Steel. Manchester.
Buxton: Moorland Publishing Company. Pugsley, A.G. 1944. The history of structural testing. The
Gale, W.K.V. 1969 Iron and Steel. London: Longmans, Green Structural Engineer, Dec 1944. Pages 492–505.
and Co. Ltd. Rawdon, H.S. & Epstein, S. 1924. The nick-bend test for
Gale, W.K.V. 1963. Wrought iron: A Valediction. Transac- wrought iron. Technological papers of the Bureau of Stan-
tions of the Newcomen Society. Vol.36. pages 1–11. dards No. 252, part of Vol.18. Washington: Dept. of
Gale, W.K.V. 1964. The rolling of iron. Transactions of the Commerce, Government printing office.
Newcomen Society. Vol.37. Pages 35–46. Skelton, H.J. 1924. Economics of iron and steel. 2nd Edition.
Gibbons, C.H. 1934. History of testing machines for materi- London.
als. Transactions of the Newcomen Society, Vol.15, Pages Smith, D. 1980. David Kirkaldy (1820–1997) and Engi-
169–184. neering Materials testing. Transactions of the Newcomen
Gordon, R.B. 1988. Strength and Structure of Wrought Iron’, Society, Vol.52. Pages 49–65.
Archeomaterials Vol. 2, No. 2, Pages 109–137. Sparks, S.P. & Badoux, M.E. 1998. Non-destructive eval-
Gordon, R.B. 1996. American Iron 1607–1900. Baltimore: uation of a historic wrought-iron truss bridge in New
The Johns Hopkins University press. Braunfels, Texas. APT Bulletin. Vol. 24, No.1. Pages 5–10.
214
Steude, T. (2000), ‘The strength of wrought iron and early The Tech. 1928. Obituary of Gaetano Lanza. The Tech. Friday.
mild steel beams’, MSc Dissertation, UMIST. March 23rd. 1925. Vol. XLVIII. No.19. Cambridge: Mas-
Sutherland, R.J.M. (ed). 1997. Structural Iron, 1750-1850. sachusetts Institute of Technology.
Aldershot: Ashgate Publishing ltd. Thorne, R. (ed). 2000. Structural iron and steel 1850–1900.
Swailes, T. & Marsh, J.O. 1998. Structural appraisal of iron- Aldershot: Ashgate Publishing Ltd.
framed textile mills. Institution of Civil Engineers Design Thorneycroft, G.B. 1850. On the manufacture of malleable
and Practice Guide. London: Thomas Telford Ltd. iron; with the results of experiments on the strength
Swailes, T. & Marsh, J. 2005. Development of long-span iron of railway axles. Proceedings of the Institution of Civil
roof structures in Britain. Proceedings of the Institution Engineers. Vol.9. Pages 294–300, Paper No. 828.
of Civil Engineers, Structures and Buildings 158. October Turner, T. 1908. The Metallurgy of Iron. 3rd Edition, Charles
2005. Issue SB5. Paper No.13908. Pages 321–339. Griffith & Company Ltd. Strand.
Swailes, T. 2006.Scottish Iron Structures. Guide for practi- Unwin, W.C. 1910. The testing of materials of construction.
tioners. Edinburgh: Historic Scotland. 3rd Edition. London: Longmans, Green & Co.
The Engineer. 1863. The elastic limit of iron. The Engineer, Walker, R. 2002. The production, microstructure and proper-
March 6th 1863, Pages 139–140. ties of wrought iron. Journal of Chemical Education, Vol.
The Engineer. 1888. The Watertown Arsenal Testing 79, No. 4, April 2002.
Machine. The Engineer, January 20th 1888, Page 54–55. Watertown Arsenal. 1883-1893. Report of tests on the
The Tech. 1925. Professor-Emeritus Gaetano Lanza. The strength of structural material made at the Watertown
Tech. Monday. September 28th. 1925. Cambridge: Mas- Arsenal, Massachusetts. (Reports appeared as annual
sachusetts Institute of Technology. volumes).
215
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
L. Sorrentino
Structural and Geotechnical Engineering Department, Sapienza University, Rome, Italy
D. Bruccoleri
Architect, Guidonia, Italy
M. Antonini
Architect, Terni, Italy
ABSTRACT: A sequence of ground shakings occurred from October, 1831 to March, 1832 in Central Italy. Due
to the high intensity and the repetition of the events, the Basilica of Santa Maria degli Angeli, Assisi was damaged
severely, suffering the collapse of the vaults of central and left naves. After the first emergency interventions,
engineer- architect Luigi Poletti (1792–1869) was appointed to design the retrofitting and restoration works.
From an architectural point of view the works carried out by Poletti were rather respectful, but from a structural
point of view they introduced interesting innovations. Contrary to the collapsed construction, Poletti rebuilt the
roof over the central nave with an original raised tie beam and iron braced truss, not resting on the below vault,
while the vault had now a tapered transversal section. Poletti’s solution are considered from a structural point of
view, performing numerical simulations by means of limit analysis and finite element models, with reference to
his original drawings, to previous and following earthquake damages, and to contemporary technical literature.
217
In 1645–48 the right transept was built, but the left
one was raised in 1650–52. Even though the construc-
tion of the dome started in 1662, just a year later was
stopped by a harsh controversy (both monetary and
technical) between the master builder and the Friars.
The works restarted between 1668 and 1674, when they
were stopped due to an out-of-plumb of the façade.
The damage was ascribed to the too limited thickness
of its upper section, and the void created by the stairs
in the lower one. Between 1674 and 1675 the façade
was strengthened and partially rebuilt. Only after that
the works on the dome started again and were com-
pleted on September 25, 1677. Then, the bell tower
was built in 1679–85, thus completing the church with
the exception of the façade (Bartelli 1989).
218
The Basilica suffered cracks in the choir apse, in the
transept, in the cupola, in the lateral naves, and in
some chapels. The dome and the main nave were only
slightly damaged (Boschi et al 1998, 164). Architect
Giovan Battista Tiberi suggested extensive use of iron
ties (Boschi et al 1998, 165), especially in the lateral
naves and in the chapels. His proposals have been par-
tially carried out, avoiding the visible ties, and locating
them in the attic (refer to pictures in Lunghi & Lunghi
1989, 194–198; and description in Boschi et al 1998,
166).
Shear and spalling cracks occurred in the lantern
above the dome by the April 19, 1984 earthquake
(Boschi et al 1998, 214), whose felt intensity was
VII MCS (Guidoboni et al 2007). Such damages
induced careful provisional and permanent interven-
Figure 2. The left and central naves of the Basilica after tions (Menegotto 1993 and 2003), which have lived
the March 15, 1832 shock (di Boveglio & d’Isola Maggiore through the following seismic event.
1834, pl. 7). The 1997 Umbria-Marche seismic sequence has
also damaged the Basilica: the façade (rebuilt for
architectural reasons during the 20th century) rotated
although this would have reduced the drum of the
slightly, cracks were detected on the lateral naves
dome.
vaults and at the junction between naves and dome
On August 22 engineer architect Luigi Poletti
(refer to: Capalbini et al 1999, where also a complete
(1792–1869) is first mentioned. Poletti was a pre-
geometric survey of the basilica is available).
eminent practitioner at that time (Dezzi Bardeschi
Although from an architectural point of view, Poletti
et al 1992) and during his professional lifetime was
restoration was substantially faithful to the 16th cen-
frequently involved in interventions after catastrophic
tury design (Perilli 1842, 7; Vignoli 1989, 152), from a
events (Reale et al 2004).
structural point of view he introduced important inno-
Poletti recommended to shore the choir’s vault, the
vations which will be discussed in the next section. It
right nave’s arches, and the three arches between the
is important to stress that the portions reconstructed
naves and the transept.
under his supervision survived all the subsequent
Between 1832 and 1835 the master builder
events with only minor damages (Boschi et al 1998,
appointed to execute his instructions added iron rings
151). As will be shown, the strategy pursued by Poletti
to the dome, repaired two pillars in the right nave and
was not to recur to special earthquake-resistant solu-
tear down the upper section of the façade.The interven-
tions. Instead he tried to reduce the vulnerability of the
tions on the two pillars were not faultless. According
edifice first of all by means of rule-of-art construction
to Perilli (1842, 5) the original masonry of the pil-
techniques, such as solid brick masonry instead of rub-
lars was a rubble core one, with a thin external brick
ble core masonry. Secondly he saved weight on vault.
masonry skin, and the master builder did simply repair
Finally, he reduced the thrust exerted by horizontal
them. The description of the construction technique is
structures, both vault and roof.
very interesting, since it may help to explain the severe
damages observed in the building.
After many discussions in 1836 Poletti was con-
firmed in his role of designer and Mollari was
3 STRUCTURAL EVALUATION OF THE
appointed as supervisor of the execution. Between
INTERVENTIONS
March and September of the same year the right nave
was torn down and rebuilt (with solid brick masonry,
3.1 The tapered vault
Perilli 1842, 8) up to the arches, while the vault was
completed between April and June 1837. Between July Unfortunately there is no graphical evidence about the
and October 1836 the demolition of the top of the construction shape of the pre-1832 vault. However, in
façade was completed. di Boveglio & d’Isola Maggiore (1834, pl. 7; Figure 2)
The central nave vault was rebuilt in 1838. The the mark of vault with the triumphal arch has a constant
construction of the roof took place in 1839. Finally, thickness. Moreover, Perilli (1842, 6) explicitly states
the façade was partially rebuilt between March and that the previous vault had a constant thickness.
August, 1840. Poletti’s drawing dated July 19, 1836 shows the
On February 12, 1854 another earthquake hit the cross section of his new vault (Figure 3). Although
area (VII felt intensity in Assisi, Guidoboni et al 2007). this is of some interest, it has dragged little attention
219
so far. Poletti designed a tapered vault (i.e. a vault Architect Carlo Fontana (1638–1714), in is famous
with non constant thickness). The actual vault geome- treatise on the Vatican Temple (1694, 361–367) sug-
try is slightly different from the one originally drawn gested a partially different rule for pointed domes,
(Lunghi & Lunghi 189, 194–195). The vault has a using four different centres, at the same height, in order
crown thickness of 32 cm, and (through three 16 cm to get a thickness decreasing from the lantern to the
steps) a haunch thickness of about 80 cm. Therefore, drum.This rule was based on the observation of several
the span (at springers) to crown thickness ratio is about existing domes and on the exam of previous geometric
57.7, the span (at haunches) to crown thickness ratio rules (Huerta 2004, 272).
is about 45.7, while the crown to haunch thicknesses The first reference to tapered arches and barrel
ratio equals to 2.5. vaults is probably due to the French military engineer
It will be shown that such a profile has several Amédée-François Frézier (1682–1773). In his treatise
advantages and it is therefore legitimate that Poletti (1737–1739, 2: 87) he describes a method to draw
purposely pursued it in order to reduce the vault’s arches of non constant thickness based on the first
thrust, considered among the reasons of the 1832 scientific theories on arches due to La Hire, Parent
collapses. and Couplet (for such theories refer e.g. to: Benvenuto
There is no much attention in the historical literature 1991, 321–326, 331–336, 338–344). Frézier’s method
to tapered arches or vaults. Probably the first refer- is rather similar to that by Serlio and Palladio, since
ence is due to architect Andrea Palladio (1508–1580) the intrados and extrados curves are obtained consid-
in a 1567 appraisal on the new Cathedral of Brescia in ering two centres of different height. Once the crown
Northern Italy (Puppi 1988, 123–125). Writing about thickness has been established, their positions are fixed
the dome, Palladio states that the thickness should either by fixing the springer thickness or by lowering
grow from the lantern to the drum in order to reduce the the centre of the intrados curve by a fraction of its
load at mid-span. Such thickness reduction is obtained radius. In Frézier’s examples the springer/crown thick-
by using two spheres, of diverse radii, with their cen- ness ratio equals 3 and extrados/intrados radius ratio
tres at different heights. According to Huerta (2004, equals 7/6. The French engineer gave also rules for
200–201), Palladio used this method to design all the vaults bearing heavy loads (e.g. bridges), vaults bear-
domes in his treatise (Palladio 1570), and the same was ing small loads (e.g. vaults loaded by some wooden
done previously by architect Sebastiano Serlio (1475– truss), and vaults subjected to self weight. In the
1554). Palladio’s appraisal was reported by Zamboni latter case he recommends a thickness at the crown
(1778) and later by Rondelet (1832–1835, Book 9, equal to 1/24 of the span, a thickness which should
Sec. 6, Ch. 4, Notes and Pl. 195), so it is possible that be doubled in the 30◦ section close to the haunches
Poletti was aware of it. (2: 96–97).
Figure 3. Cross section highlighting the tapered vault, dated July 19, 1836 (BCALPM 1836, detail). Courtesy of: Biblioteca
Civica d’Arte “Luigi Poletti”, Modena, Italy.
220
In 1748 in an unpublished manuscript (Huerta 2004, In order to better understand such favour for
358–360) the French engineer Jean Rodolphe Per- tapered vaults as well as Poletti’s design, his vault
ronet (1708–1794) suggested to design bridges with is analysed here by means of the safe theorem of
a thickness doubling from the crown to the haunches. limit analysis (Heyman 1966). Two lines of thrust
In the case of semi-circular arches the crown thick- are drawn, minimising either maximum top and bot-
ness should be equal to 5/144 of the span plus 1 foot. tom stresses or the thrust exerted on the haunches
A rather similar rule was suggested by the French (Méry 1840). In the second case maximum com-
Engineer Bernard Forest de Bélidor (1697–1761) in pression stress was assumed 2.2 MPa (OPCM 2005).
1750–1753 (2: 445). Three different geometries are considered: (1) Poletti’s,
Frézier’s and Perronet’s approach was followed by (2) constant thickness t = 0.54 m giving a vault of
the French architect Jean Baptiste Rondelet (1734– same weight, (3) constant thickness 1/3 bigger than
1829) in his famous treatise (French edition 1812– Poletti’s crown thickness (t = 0.43 m), thus following
1817, Italian translation 1832–1835). Therefore, he Rondelet’s abovementioned tables. Two load cases are
recommends to design arches of non constant thick- taken into account: (1) self-weight only, (2) self-weight
ness by lowering the centre of the extrados curve (Pl. plus a 18.5 kN load applied at mid span over a 0.5 m
27 and Book 9, Sec. 6, Ch. 1, Art. 1), and to differenti- length (representing the load of the roof resting on
ate the crown thickness as a function of the span and of the vault, whose load-per-unit-area has been estimated
the type of live load (heavy, medium and zero; Book 3, equal to 2.0 kN/m2 ).
Sec. 3, Ch. 1). However, also based on some exper- The analyses are performed with a computer code,
iments he performed, the thicknesses suggested are drawing the aforementioned thrust lines and the asso-
sometimes much smaller. For the case at hand, and ciated stress fields (Dr. Cesare Tocci, personal com-
considering an average stone, he suggests a thickness munication). Sample outputs are presented in Figure 4
of 25 cm at the crown which should be doubled at the and Figure 5.
haunches, by means of a linear increase. Moreover, In Table 1 is presented the comparison between
Rondelet clearly states that a tapered vault is much Poletti’s vault and the two vaults of constant thickness.
more convenient than a constant-thickness one. As a In terms of optimal stress Poletti’s solution grants a
matter of fact, he writes that a semi-circular arch of 3 geometrical safety coefficient (Heyman 1982, 24)
constant thickness subjected to its self-weight needs a with a 62.0 kN thrust. The 0.54 m constant thickness
thickness equal to 1/17 of the span, while a tapered one vault gives the same geometrical coefficient with a
needs a crown thickness which is one fifth of the previ- much higher, 74.6 kN, thrust.
ous (ibidem). Furthermore, he declares that a tapered The 0.43 m constant thickness exerts a little smaller
vault will exert a much lower thrust of the abutments horizontal force, 62.0 kN, but with a much lower, 2.5
than a constant thickness one of equal span (Book 9, geometrical coefficient.
Sec. 6, Ch. 1, Art. 2; and ibidem Ch. 3). In this last In terms of minimum thrust, Poletti’s vault gives
chapter Rondelet gives tables for the crown thickness again the best value, 53.6 kN, with the 0.43 m vault
of semicircular vaults with horizontal extrados, with granting a similar 54.6 kN. However, such figure is
solid infill up to half height and thickness constant also due to the lower self-weight. Therefore, the ratio
or tapered. The crown thickness of the constant one between horizontal / vertical components is 0.64 in the
is one third bigger than the crown thickness of the
variable vault.
Rondelet’s approach had a great fortune, as evi-
denced by other widespread manuals of 19th century
(Breymann 1885, Vol. 1, Ch. 8, Sec. 9 and Pl. 100).
The vaults drawn by Rondelet or Breymann are
meant to have a smooth thickness increase. Such a
result can be obtained if the vault is meant to be Figure 4. Poletti’s vault: optimal thrust line under self
made of cut stones. However, if bricks are used (as weight.
in Santa Maria degli Angeli) it is easier to has discrete
steps in the cross section. As far as it is known to the
authors, such case is rarely considered in the litera-
ture. Körner (1895, 286) recommends to avoid such a
design, since the thrust line will be very close to the
extrados of central (thinnest) segment, while cracks
and bulging will appear on the two sides of the vault
close to the first thickness variation. As Rondelet and
Breymann he endorses an even growth of the cross Figure 5. Poletti’s vault: stress field associated to the
section. optimal thrust line.
221
Table 1. Analysis of Poletti’s tapered vault compared to two Table 2. Analysis of Poletti’s tapered vault compared to two
constant thickness vaults. Self weight only considered. constant thickness vaults. Self weight + roof.
Minimum Stress Thrust Stress Thrust Stress Thrust Minimum Stress Thrust Stress Thrust Stress Thrust
222
to tension if the restraint gives way laterally 2 cm only.
Such a lateral spread is not unrealistic both due to
lack-of-fit problems, masonry local settling under the
truss thrust, deformation of the unrestrained principal
rafters ends. As a matter of fact, modelling such unre-
strained end as a clamped inclined beam subjected to
the whole vertical reaction of the masonry, the lateral
deflection will be approximately 6.5 cm. It is possi-
ble that Poletti expected such a lateral spreading and,
therefore, he added the braces. The presence of the
braces is able to markedly reduce the horizontal thrust
exerted by the truss (Table 3), a problem which Poletti
must have taken account of, since the roof has been
considered a reason of the collapse, and since this
was the same reason of the different French, English
and American solutions previously mentioned. Fur-
thermore, the additional struts remove roughly 80%
of the vertical reaction from the principal rafters, and
thus reduce the risk of their excessive bending as well
as they transfer the load much further down the wall.
It is interesting to note that as the sagging increases
maximum bending moments in principal rafter, raised
tie beam, and strut tend to become more and more sim-
ilar (Table 3). Although he certainly did not perform
Figure 6. Design drawing of the new roof, dated September explicit calculations, Poletti must have been aware that
10, 1838 (BCALPM 1838, detail). Note the iron braces and the braces will have induced deflections in the raised
tie. Courtesy of: Biblioteca Civica d’Arte “Luigi Poletti”, tie and in the struts. That is probably the reason why he
Modena, Italy. gave to the three fir elements the same cross section.
Finally, it is interesting to note that, according to the
Italian standards (DT206 2006; UNI 11035-2: 2003),
are prolonged to the principal rafters, thus obtaining a
the design ultimate moment (capacity) in the case of
scissors-braced truss.
permanent loads is equal to 128 kN m. Nonetheless,
Poletti’s roof solution seem to have dragged more
it is true that some of the principal rafters and of the
attention than the vault one, although such interest has
extended ashlar struts had to be reinforced by means
not always brought to a correct interpretation of its
of L-shaped steel profiles (Lunghi & Lunghi 1989,
features.
196–197).
According to Lunghi & Lunghi (1989) Poletti’s
iron braces are useless, since “assuming on both sides
hinges as restraints for the structure, this is fully com-
pressed and therefore the braces are not reacting” 4 CONCLUSIONS
(p. 193). Although not clearly stated by the authors,
they seem to have assumed fixed restraints. Under such Confronted with the need to retrofit a pre-eminent
hypothesis it is true that the inclined iron elements building while restoring (not reconstructing) Santa
are compressed. However, if a slight lateral sagging is Maria degli Angeli Basilica (Assisi, Central Italy) after
assumed then this is not true anymore. the severe seismic sequence of 1831–1832, architect
A similar phenomenon has been observed in the Luigi Poletti did not to recur to special earthquake-
analysis of cross vaults. If they are examined assuming resistant solutions. Instead he tried to reduce the
fixed restraints almost no tension is observed (Alexan- vulnerability of the edifice first of all by means of
der et al 1977). However, if the supports can give way, rule-of-art construction techniques, such as solid brick
elements previously compressed will be subjected to masonry instead of rubble core masonry. Secondly
tension (Barthel 1989). he saved weight on the vault of the central nave and
The roof is analysed here with a linear elastic reduced its thrust assuming a tapered cross section. As
finite element code (CSI 2004), under self-weight only has been shown in the paper, by means of numerical
(a 1.35 partial safety factor has been considered, for simulations of thrust lines and associated stress fields,
later code verifications), and assuming an increasing such solution is more effective compared to uniform
lateral sagging on both sides. Although the considered thickness ones, having the same weight or complying
values are still rather small compared to the free span with recommendations of the contemporary technical
of the roof (18.45 m), Poletti’s iron braces are subjected literature.
223
Table 3. Analysis of Poletti’s roof truss under self-weight. (WOB = Without Braces;
WB = With Braces)
Poletti tackled also the problem of the design of riforme di Archi, e Volti da farsi nella ricostruzione della
a roof not resting on the vault and not concealing medesima.
the dome. Therefore, he designed a king post truss BCALPM 1838, Inv. 549. Disegno della nuova incavallatura
with raised beam, additional extended ashlar struts and del tetto della navata di mezzo della Madonna degli Angeli
presso Assisi, 10 settembre 1838.
iron braces. As shown in the paper, the problem of a Bélidor, B.F. 1737–1753. Architecture hydraulique, ou L’arte
roof truss built on a vault which projects itself above de conduire, d’elever, et de menager les eaux pour les
the wall top has been dealt with both in actual build- differens beisoin de la vie. 2 vols. Paris: chez Charles-
ings and in the technical literature. Nonetheless, Poletti Antoine Jombert.
solution is original and effective, as proven in the paper Benvenuto, E. 1991. An introduction to the history of struc-
by means of a finite element model. The additional tural mechanics. New York: Springer.
struts reduce the vertical reaction at the foot of the Boschi, E., Guidoboni, E., Ferrari, G. & Valensise, G. 1998.
principal, thus limiting the bending deflection, while I terremoti dell’Appennino umbro-marchigiano, area sud
the braces are effective in case of a lateral spreading orientale dal 99 a.C. al 1984. Bologna: Compositori.
Breymann, G.A. 1885. Trattato generale di costruzioni civili
of the structure and in reducing the thrust exerted on con cenni speciali intorno alle costruzioni grandiose.Vols.
the walls. 1–2. Milano: Francesco Vallardi.
More convincing than any numerical simulations, Capalbini, P., Ripa di Meana, D. & Salvatici, B. 1999. Basil-
Poletti’s retrofitted elements survived with minor dam- ica patriarcale della “Porziuncola”: progetto di ripristino
ages all the following earthquakes. statico e funzionale a seguito degli eventi sismici del
26-09-1997 e successivi. Ingegneria Sismica 16(2): 5–16.
Courtenay, L.T. 1985. Where roof meets wall: structural inter-
ventions and hammer-beam antecedents, 1150–1250. In P.
REFERENCES Long (ed.), Science and Technology in Medieval Society,
98–124. New York: New York Academy of Sciences.
Alexander, K.D., Mark, R. & Abel, J.F. 1977. The Struc- CSI (Computers and Structures Inc.) 2004. SAP2000,
tural Behavior of Medieval Ribbed Vaulting, Journal of rel. 9.03. Berkeley, CA, USA: CSI.
the Society of Architectural Historians 36(4): 241–251. Dezzi Bardeschi, M. (ed.) 1992. Luigi Poletti architetto
ASPSF (Archivio di Stato di Perugia Sezione di Foligno) (1792–1869). Bologna: Nuova Alfa Editoriale.
1833. Archivio Moderno, Fabbriche, I, Titolo XII, 1001. di Boveglio, B. & d’Isola Maggiore, M. 1834. Compendio
Lettera di Antonio Rutili-Gentili, Ingegnere della prefet- storico del Perdono di Asisi e della chiesa detta Porziun-
tura delle acque di Foligno, al Gonfaloniere di Foligno: cola: cui aggiungesi un maggiore ragguaglio della Basil-
descrizione e commento del progetto di restauro del ica che la circonda, della caduta di questa, e dell’attuale
Palazzo Comunale proposto da Luigi Poletti, Foligno 20 sua riedificazione. 13. ed. Asisi: dalla tipografia Spariglia.
Aprile 1833. DT (DocumentoTecnico) 206 2006. Istruzioni per il Progetto,
Bartelli, C. 1989. La nuova facciata. Il progetto di Dante l’Esecuzione e il Controllo delle Strutture di Legno. Roma:
Viviani (1913) e gli interventi di Cesare Bazzani (1925– Consiglio Nazionale delle Ricerche.
30). In F.F. Mancini & A. Scotti (eds), La Basilica di Santa Frézier, A. F. 1737–1739. La théorie et la pratique de la coupe
Maria degli Angeli. 1. Storia e architettura: 207–231. de pierres et des bois pou la construction des voûtes et
Perugia: Electa Editori Umbri Associati. autres parties des bâtiments civils et militaires, ou traité de
Barthel, R. 1989. Analysis and Structural Behaviour of stéréotomie à l’usage de l’architecture. Strasbourg/Paris:
Masonry Cross Vaults. In C.A. Brebbia (ed.), Struc- Charles-Antoine Jombert.
tural Repair and Maintenance of Historical Buildings, Guidoboni, E., Ferrari, G., Mariotti, D., Comastri, A.,
119–127. Basel: Birkhäuser. Tarabusi, G. & Valensise, G. 2007. CFTI4Med, Cata-
BCALPM (Biblioteca Civica d’Arte “Luigi Poletti” Modena) logue of Strong Earthquakes in Italy (461 B.C.-1997) and
1836, Inv. 551. Spaccato per traverso della Basilica di Mediterranean Area (760 B.C.-1500). Roma-Bologna:
S. Maria degli Angeli presso Asisi coll’indicazione delle INGV-SGA. http://storing.ingv.it/cfti4med.
224
Heyman, J. 1966. The stone skeleton. International Journal emendata ed accresciuta dall’Autore, a cui fa seguito
of solids and structures 2(2): 248–279. un’appendice. Roma: P. Aurelj.
Heyman, J. 1982. The masonry arch. Chichester: Ellis Hor- Puppi, L. (ed.) 1988.Andrea Palladio: scritti sull’architettura,
wood Series in Engineering Science. 1554–1579. Vicenza: N. Pozza.
Huerta, S. 2004. Arcos, bóvedas y cupolas: Geometria y Reale, C., Scheibel, B.,Vignoli, F., Decanini, L. & Sorrentino,
equlibrio en el cálculo traditional e estructuras de fábrica. L. 2004. Il Regolamento edilizio di Norcia del 1860:
Madrid: Instituto Juan de Herrera. fra storia sismica e storia urbanistica. In S. Lagomarsino
Körner, C. 1895. Gewölbte Decken. In J. Durm, Hand- (ed.), L’ingegneria sismica in Italia; Proc. 11th Nat. Conf.,
buch der Architektur, Part 3, Tome 2, Ch. 3. Darmstadt: Genova, 25–29 gennaio 2004. Padova: SGE.
Bergsträsser. Rondelet, G. 1832–1835. Trattato teorico e pratico dell’arte
Lunghi, A. & Lunghi, E. 1989. Il terremoto del 1832 e la di edificare. 5 vols. Mantova: Negretti]. (Original French
ricostruzione di Luigi Poletti. In F.F. Mancini & A. Scotti edition: 1812–1817).
(eds). La Basilica di Santa Maria degli Angeli. 1. Storia UNI (Ente Nazionale di Unificazione) 11035-2: 2003. Legno
e Architettura, 189–206. Perugia: Electa Editori Umbri strutturale – Regole per la classificazione a vista secondo
Associati. la resistenza e i valori caratteristici per tipi di legname
Menegotto, M. 1993. Seismic Repair and Upgrading of a strutturale italiani.
Dome Lantern in Assisi. Structural Engineering Interna- Vignoli, F. 1989. La fabbrica alessiana e il cantiere di Santa
tional 3(1): 34–36. Maria degli Angeli dal XVI al XIX secolo. In F.F. Mancini
Menegotto, M. 2003. Lanterna di Santa Maria degli Angeli, & A. Scotti (eds), La Basilica di Santa Maria degli Angeli.
Assisi, Perugia. In P. Rocchi (ed.), Trattato sul consolida- 1. Storia e architettura, 97–159. Perugia: Electa Editori
mento, C 486–489. Roma: Mancosu. Umbri Associati.
Méry, E. 1840. Sur l’équilibre des voûtes en berceau. Annales Yeomans, D.T. 1991. British and American solutions to a
des Ponts et Chaussées 19(I sem.): 50–70. roofing problem. Journal of the Society of Architectural
OPCM (Ordinanza del Presidente del Consiglio dei Ministri) Historians 50(3): 266–272.
n.3431 del 3 maggio 2005. Ulteriori modifiche e inte- Zamboni, B. 1778. Memorie intorno alle pubbliche fabbriche
grazioni all’OPCM n.3274 del 20 marzo 2003. Gazzetta più insigni della città di Brescia. Brescia: Vescovi.
Ufficiale, 10 maggio 2005, n. 107.
Perilli, S. 1842. Relazione storica sul risorgimento della
Basilica degli Angeli presso Asisi. Edizione: 2. ed.
225
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Robert S. Cotta
Senior Structural Engineer, Linea 5, inc., Boston, MA, USA
MSc Conservation of Historic Buildings, University of Bath, Bath, UK
ABSTRACT: The 14th century spire of Salisbury Cathedral, when built, achieved unprecedented heights
in stone masonry construction. This accomplishment was made possible through several ingenious structural
systems that used iron and timber systems to supplement the traditional methods of stone masonry construction of
the time. Its survival over 700 years, though, has not been without challenges and today the engineering systems
of the tower and spire provide a superb example of evolved technologies that assure the structural integrity of
the 123 meter high structure. The actual theoretical basis for the contributions of the structural systems in the
Salisbury tower and spire will be explored in this paper with a goal of identifying the specific critical roles
of individual systems based on historic structural behavior and analysis based on twentieth century structural
concepts.
227
2 THE HISTORICAL CONTEXT FOR THE century, there is little or no written documentation
SALISBURY CATHEDRAL SPIRE and the history must be conjectured primarily by an
assessment of the surviving physical built structure.
2.1 The historical problems Salisbury Cathedral, though, is fortunate to have a
comparatively extensive documentation of its history
The spire and tower of Salisbury Cathedral have pre-
since the 17th century that provides insight into the
sented problems for builders, clerks of the works,
conditions noted and the solutions implemented to
architects and engineers through the centuries. These
maintain the structure through the past 350 years. Two
problems appeared from the very start when the deci-
documents in particular are landmarks in the history of
sion was made to construct a stone tower and spire
English building archaeology and conservation. The
of unprecedented height on a supporting structure at
first of these is Sir Christopher Wren’s assessment
the great crossing that from all available evidence
on the state of the Cathedral prepared for Dr. Seth
was never intended to support a tower of any signif-
Ward, Bishop of Salisbury, in 1668. This document
icant height. The consequences of this decision have
is considered to be the first reasonably comprehen-
impacted almost every structural component from the
sive evaluation of a historic medieval ecclesiastical
supporting foundations to the top of the spire. The
building in England. The second document is titled
impacts include the following:
A Series of particular and useful Observations, Made
• In the original cathedral structure, minimal foun- with great Diligence and Care, upon that Admirable
dations under the great crossing piers have visibly Structure, the Cathedral Church of Salisbury, writ-
settled, especially at the south-west pier, resulting ten by Francis Price, the Clerk of the Works for the
in an overall ‘lean’ to the tower and spire to the Cathedral from 1738 to 1753. Not only did Price
southwest. describe the building in words, but he also produced
• The crossing piers have a visible bow, a situation detailed drawings based on careful measurements, pro-
that is believed to have prompted the installation of viding at the time a unique snapshot of the Cathedral
the strainer arches at the great crossing. at that specific point in time.
• Buttressing, much of which appears to have been
added as part of the tower construction, was prob-
ably installed in response to outward thrust from 2.2 The structural systems supporting the base
the added tower weight that caused shifting of tri- of the tower
forium piers, cracking of walls and vaulting, and
The concerns with the foundations and supporting
deflections of transept piers that can still be seen in
pillars at the great crossing of Salisbury Cathedral
the Cathedral today.
have been the topic of extensive discussion, conjecture
• The base of the tower up to just above the main
and concern over the years. Some of the conjecture
roof ridges is a comparatively light arcaded structure
has been based on incomplete evidence or concern
with comparatively thin two-foot thick outer walls
extrapolated from experience at other medieval cathe-
with multiple penetrations. Reinforcing of this level
drals. Until the 1980’s, when Peter Taylor was able
has been implemented on several occasions over the
to make specific measurements of the pier founda-
centuries and includes iron framework ties, infilled
tions at the main crossing, the actual foundation sizes
stair towers and wall openings, and stiffening walls
were unknown quantities that many had presumed
between the arcade columns and the outer walls.
were insufficient. Measurements taken in the 1980’s
• Extended periods of neglect have resulted in failure
by Peter Taylor have been used to evaluate the footing
of iron reinforcing at the base of the spire, result-
situation and using soil mechanics theory developed
ing in stone damage and the need for a reworking
in the twentieth century have for the first time been
or replacement of the iron reinforcing and major
able to provide a theoretical basis for explaining the
restoration of damaged masonry on at least two
marginal adequacy of the footings under these piers.
occasions.
Taylor notes that the historic concern with the delicacy
• The spire itself has suffered frequent damage from
of the situation has perhaps saved the spire and tower
storms and lightning, resulting in cracks and dis-
from disaster, as this current theoretical explanation
placed units in the masonry and, in at least one
highlights the importance of maintaining the confin-
instance, fire damage to the internal timber fram-
ing soils around the foundations, a contribution that
ing. There is also a visible bend twelve meters from
would have been disrupted by any attempt to excavate
the tip of the spire in the southwest direction that is
and enlarge the foundations. (Taylor, 1988, p. 3)
of unknown origin.
Similarly, the bow of the pillars at the great cross-
The structural behavior and conditions behind these ing has been the topic of much discussion over the
problems have been considered by caretakers of the centuries. The bow and associated cracking and dis-
cathedral fabric over the centuries and solutions have placement of arcade piers at the triforium level are
been implemented. For early work prior to the 17th generally believed to have originated when the tower
228
and spire were added to the original tower base. Much Iron, though, does corrode and when exposed to
of the engineering that can be found within the main weathering conditions will require significant restora-
body of the Cathedral to counteract forces imposed tion or replacement on more frequent intervals than
by the weight of the tower and spire, including scis- stone and such replacements and restorations pro-
sors arches at the east crossing and 112 buttresses at vide opportunities for utilizing improved technologies.
the arcade and triforium levels (Price, 1753, p. 35), Iron-based technologies have also evolved over the
were believed to have been added during or shortly centuries, and thus have also been favored for added
after construction of the tower and spire in the early retrofits to address deficiencies that develop over the
14th century. The exception to this are the east-west years.
strainer arches at the great crossing, which appear
to have been installed approximately 100 years after
the spire addition. The contribution of these strainer 3.2 The metal tension framework and gallery
arches, which are supposedly restraining through com- infill in the original lantern tower
pression bearing, has been called into question over the
The best evidence that a conscientious engineering
years based on visible gaps between the arches and the
effort was involved in the raising of the tower and spire
crossing piers. (Price, 1753, p.53) Strengthening mea-
at Salisbury is the elaborate medieval wrought iron
sures have been recommended as recently as the late
framework that is located in the original lantern tower
twentieth century to stiffen the crossing pillars and
of the Cathedral. The medieval ironwork in the lantern
increase the capability to resist buckling forces, but
was later supplemented by diagonal ties installed under
no specific deficiency has been identified (except the
George Gilbert Scott and his engineer, F. W. Shields in
visible bow). The current approach is to monitor the
the 19th century.
situation to ascertain if the bow is a manifestation of
The original lantern at the base of the tower – the
an active condition, or an adaptation of the building
portion from the eaves of the main roofs up to the
structure to stresses from loads imposed in the past
clear horizontal parapet band – has been identified by
that have found a steady state of equilibrium.
Tatton-Brown and other sources as clearly not designed
for the superimposed load of a soaring tower and spire.
(Tatton-Brown, 1991, p. 328) Originally, the lantern
was believed to have been designed to support only a
3 ENGINEERED SYSTEMS IN THE TOWER
short timber spire or pyramidal roof and was intended
AND SPIRE
to be open to the space below with large wall openings
to bring exterior light into the center of the building.
3.1 The historic importance of the tower and spire
The load of the tower and spire carried by the
structural systems at salisbury
lantern, estimated to be 6000 tons at this level, exerts an
It is the systems employed in the tower and spire of imposed stress of 1.724 N/mm2 on the approximated
Salisbury that demonstrate concerted efforts to address effective bearing area (Reeves, Simpson & Spencer,
identifiable structural behavior and that have a his- 1992, p. 381), which is well within the minimum
tory of upgrade and improvement over the centuries. 16.5 N/mm2 compressive strength of the Chilmark
This is in part due to the fact that from the beginning limestone of which it is built. What the iron tie system
the structure of the tower and spire was more than at this level does accomplish, though, is to effec-
just a structural system of stone masonry acting in tively reduce the unbraced length of the wall and
compression bearing to transfer loads to the ground. corner column elements, limiting the unbraced length
Salisbury Cathedral, partially due to the inadequacy for outward buckling at these corner stair towers to
of the base of the tower for the support of such a tall the height of the arcade, and keeping the theoreti-
tower and spire, was unique among medieval cathe- cal unbraced length to thickness ratios within limits
drals in making extensive use of an elaborate iron considered acceptable by twentieth century code stan-
framework in conjunction with stone masonry as a dards. Inward buckling of the stair towers is resisted
permanent structural system. Most ironwork used in through bearing of the masonry wall construction
medieval buildings elsewhere comprised simple short on the four sides, while outward buckling forces are
bar cramps between stone blocks or tie rods across resisted by tension in the iron framework (Figure 1,
arches, for which the primary purpose was believed to Force Fm) that transfers the load into compression
be to stabilize the masonry during construction until bearing against the depth of the walls at the adjacent
the lime mortar had set. The one other known cathedral corners. The lantern is thus reinforced, for purposes
of comparable vintage with extensive permanent iron- of buckling, so that the unbraced length is limited
work structures was the Freiburg Münster in Germany, to the height of the arcade. The intent of this frame-
where embedded iron rings were used to achieve the work appears to be valid and for over five hundred
open tracery spire that was a model for numerous later years performed admirably in providing the confining
continental spires. hoop stress to allow the thin (by medieval standards)
229
the framework into the interior of the tower as an
exposed framework, and any thermal growth can be
taken by elastic buckling and/or the play in the hook
and eye connections between segments without impos-
ing excessive stress on the stone. The amount of iron
in contact with the stone (at the corners) is dimension-
ally small, thus limiting thermal loads that may impose
tensile stresses on the stone.
230
It is the ‘iron bandage’ at the base of the Salis-
bury spire that has been most subject to the pitfalls
of using iron in exposed conditions. Price reported
that the band at the base of the spire, even though
protected in a recess covered with lead sheet, had
been swollen due to corrosion, lost material and splin-
tered the stone. (Price, 1753, p. 38) Price added an
iron band at the base of the spire in approximately
1740 to supplement the original medieval band, both
of which were replaced in 1968 when the stainless
steel band and concrete ring beam system. (Tatton-
Brown, 1991, p. 342) The medieval iron band, a system
that is essential to the structural integrity of the spire,
did however function adequately for approximately
400 years and Price’s additional ‘bandage’ provided
another 200 years of service – an impressive dura-
tion for the pre-industrial wrought iron encased in lead
system.
231
bolts that pass through the saddles and the spire shell
and are anchored by bearing plates at the exterior of the
spire corners. (Taylor, 1988, p. 8) The system has the
advantage of minimizing the exterior exposed compo-
nents (which can be of high grade corrosion resistant
materials) and placing the primary structural compo-
nent in the protected interior, where the material can
be selected for its superior structural properties (i.e. a
very high Young’s modulus value).
232
the timber scaffolding increases the factor of safety at of steel tube frames that fit against the interior face of
12 meters from the top from 1.57 to 2.29, values that the spire at several elevations. These frames serve the
represent a modification of the factor of safety from purpose of stiffening the spire and providing a second
a marginally acceptable value (∼1.5) to a value that structural system to temporarily carry loads of the spire
would be considered good design practice (>2.0) by above to allow stone replacement during restorations.
current standards. In a sense, they are a permanent installation of ‘tempo-
In 1738, Francis Price reported that the central piece rary shoring’. Taylor specifically designed the frames
of the timber scaffolding had become disconnected with a geometry provided by the slope of each face
from the capstone approximately at the level of the that prevents the structure from becoming a mecha-
weather door. This defect was immediately repaired, nism. The frame is redundant structurally and contains
but Price surmised that this defect may have con- an ‘in-built limit to the amount of movement of which
tributed to the upper bend in the spire that was of it is capable, thus resisting any irresponsible attempt
such concern to Wren in his 1668 report. (Price, 1753, to ‘lift’ the spire.’ (Taylor, 1988, pp. 8–9) In keeping
p. 59) Calculations show that the spire could accom- with appropriate principles of historic building conser-
modate design wind conditions without the benefit of vation, the frame installations are reversible and are of
the additional ‘prestress’ provided by the dead weight a technology that reflects the date of installation.
of the scaffold. However, there is no way of know-
ing precisely the nature of the wind loads experienced
by the spire at 123 meters above ground, an elevation
where wind speed, gusts and other factors can differ 4 CONCLUSIONS REGARDING THE
radically from the conditions at ground level. Even EFFECTIVENESS OF SALISBURY’S
over a limited time period exposed to less than maxi- STRUCTURAL SYSTEMS
mum wind load conditions, fatigue behavior can cause
deterioration of all structural systems, including stone. The structural systems for the tower and spire at Salis-
A marginally acceptable structure will be more sus- bury Cathedral fall into two distinct groups. The first
ceptible to crack propagation at minor defects under group is the supporting base from the foundations up
repeated cyclical loadings. The ‘bend’ is noted to be to to the original lantern level on which the 85 meters
the southwest by Price and Wren, and this would be the of tower and spire were constructed. The problems
side typically subject to tensile stress by the prevailing experienced through this zone have primarily been
wind direction. A temporary disconnection may have attributable to the fact that the original construction
begun this process of fatigue deterioration by allowing at and over the crossing was not intended to carry
the incremental opening and closing of fine cracks on such a great mass of stone. The structural systems in
the southwest side, causing deterioration of the stone, this area are essentially retrofit solutions of buttress-
which was fortunately arrested by the reinstatement of ing at the base to resist outward thrust at the crossing
the weighted capstone system. pillars and iron ties in the lantern to resist outward
Less apparent is the damping contribution of this buckling from the high levels of compression stress
framework that would work to temper resonant vibra- on the comparatively thin supporting sections. These
tions of the spire in gusty high wind conditions – systems appear to be effective and necessary solutions
a known phenomenon in tall thin structures. Wren to resist the thrusts and structural actions resulting
designed a comparable pendulum structure suspended from this large load. Less conclusively addressed,
from the capstone in the spire in Chichester in 1721. though, have been concerns about the bowing and set-
It is possible that Wren was inspired by what he tlement of the pillars. Numerous attempts to assess
saw inside the Salisbury spire during his review of the conditions at the foundations and at the pillars
the fabric in 1668. (Tatton-Brown, 1991, p. 345.) If have developed a tenuous explanation for the structural
intended, Wren’s engineering basis for the damping behavior based on advanced soil mechanics theory and
contribution of this suspended framework was at best that the supporting base for the spire has found a stable
instinctual. Given what is known of structural behavior equilibrium.
under gusty wind load conditions from 20th century The structural systems at Salisbury Cathedral that
research, the damping contribution of the pendulum can be considered most applicable to other structures
could be as considerable as the dead load prestress in are the various iron bands and framework systems that
reducing the magnitude of the wind-induced bending tie together the tower and spire structures and resist
stresses and associated long term fatigue-like tensile the outward thrust forces generated on these struc-
stresses experienced by the spire during high-wind tures. The original medieval iron frameworks in the
storm events. lantern and at the base of the spire are quite elegant
There is one final system that was installed as part in their minimal mass, simple connection joints and
of the major restoration of the Salisbury spire in the concealed design. Successful modern replacements,
late 1980’s/early 1990’s. Peter Taylor designed a series most notably Taylor’s tie ring above the openings at
233
the base of the spire, follow similar concepts while REFERENCES
incorporating appropriate updated materials improve-
ments. What these tie systems accomplish is to provide Cocke, T. & Kidson, P. 1993. Salisbury Cathedral: Perspec-
the mechanism to internally counterbalance outward tives on the Architectural History. London: HMSO.
Fritz, H. 1926. Der Turmhelm des Münsters zu Freiburg im
thrust loads and avoid the need for external buttress- Breisgau. Dissertation (Doktor-Ingenieurs). Technischen
ing or other visually intrusive modifications. Properly Hochschule zu Karlsruhe, Germany.
designed to not impose additional stresses due to Heyman, J. 1995. The Stone Skeleton – Structural Engi-
differential thermal movement and with adequate pro- neering of Masonry Architecture. Cambridge: Cambridge
tection from weather effects that cause corrosion, these University Press.
metal tie reinforcements are structural solutions that Price, F. 1997 Facsimile Edition of original 1753 Edition. A
minimally intrude on the historic fabric. Series of particular and useful Observation, Made with
Metal bands along the height of the spire, such as great Diligence and Care, upon that Admirable Structure,
the ten copper bands of the Salisbury spire and the The Cathedral-Church of Salisbury. And over, Hampshire:
Salisbury & Stonehenge Edition Facsimiles.
iron rings embedded in the horizontal stone bands Reeves, J., Simpson, G. & Spencer, P., 1992. Iron Rein-
of the Freiburg Münster address less straight-forward forcement of the Tower and Spire of Salisbury Cathedral.
but important issues. These bands are not theoretically Archaeological Journal, 149, pp. 380–406.
required due to any dead or wind load stresses on the Tatton-Brown, T. 1991. Building the tower and spre of Salis-
shell of the respective spires, but they do effectively bury Cathedral. In: L.T. Courtenay, ed. The Engineering
consolidate the spire cones to distribute stresses so of Medieval Cathedrals. Aldershot, Hampshire: Ashgate
that local irregularities do not propagate and become Publishing Limited, pp. 327–349.
a major structural problem. Tatton-Brown, T. 1996. The Archaeology of the Spire of Sal-
The other critical structural system in the Salisbury isbury Cathedral. In: L. Keen and T. Cocke, eds. Medieval
Art and Architecture at Salisbury Cathedral, Proceedings
spire is the weighting of the top portion of the spire of The British Archaeological Association Conference,
through connection of the internal scaffold system to 15–18 July 1991 Salisbury. Leeds: W. S. Maney and Sons
the capstone to provide adequate dead load resistance Limited, pp. 59–67.
to wind loads through the upper portion of the spire. Taylor, P. 1988. The Tower and Spire of Salisbury Cathedral.
This system supplements the metal cramp connected Association for Studies in the Conservation of Historic
stones to provide a sufficiently solid structure at the Buildings Trans., 12, pp. 3–10.
top to counteract high wind load stresses and thin Wren, C. 1668. Sir Christopher Wren’s Report on Salisbury
geometrical cross sections. Although specific histori- Cathedral for Dr. Seth Ward, Bishop, 1668. In: The Wren
cal evidence is insufficient to prove direct cause and Society, ed.The EleventhVolume of theWren Society 1934.
Oxford: Printed for the Wren Society at the University
effect, it does appear possible that the temporary loss Press, pp. 21–26.
of this connection could be a cause for the visible bend
at 12 meters from the top of the spire.
234
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The authors report results of an extensive survey of baroque timber roof constructions in Bavaria.
It turns out that the majority of these roofs lack a tie-beam at their base, because the roofs harbour masonry
vaults which rise high into the roof space. In the church roofs analyzed, different historical techniques have
been identified for carrying the thrust of the roof and the vault in the case of a missing tie-beam. We analyze
these techniques both from a historical point of view (development and spread of technological knowledge by
carpentry treatises) and from a structural point of view, including the structural assessment in an exemplary case.
235
The main girders of the roof are constituted by two in 17th century Bavarian architecture) was already a
superimposed trussed frames. The lower one is char- reaction to immediate damage or not. However, it is
acterized by two rows of timber columns rising directly sure that considerable deformation set in immediately
above the main arches of the church. This part is well after the completion of the construction. This is tes-
stiffened by various braces rising parallel to the rafters. tified by the deformed shapes of the aisle vaults, the
The upper storey of the roof, by contrast, makes use of outward tilt of the circumferential walls, as well as a
a “liegender Stuhl” structure, the most characteristic mess of historical rehabilitation structures above the
element of 17th and 18th century roofs in Germany. vault (cf. Figs 2 and 3). These include scissor braces
The roof accommodates the rising barrel vault of the intended to carry the thrust of the roof, as well as
central nave by simply leaving out the tie-beam in the another pair of scissors anchored to the circumferential
central section. walls, obviously meant to carry the vault thrust. From
However, in the intervals between the lunettes of the a modern point of view, it is obvious that the slowly
main vault, projecting beam ends suggest that, origi- hardening historical mortar would have prevented the
nally, the principal transverse trusses of the roof in aisle vaults from performing their intended function
fact did have continuous tie-beams (cf. Fig. 3). There as abutments, and that movements would slowly come
is no immediately obvious explanation for this feature. to a standstill only as the mortar reached its final stiff-
However, it turns out that this tiny peculiarity – easily ness. Most of the thrust is now carried by the conical
overlooked – is intimately connected to the construc- lunettes. On the other hand, the scissor braces presum-
tion process of the vault and roof, and combines well ably did not do a great service towards stopping the
with the following observations on the vaults, as we process of deformation.
shall see: Anyhow, a large percentage of the present-day
deformation must have occured immediately after
– The longitudinal barrel vault of the central nave is completion, since the stucco work of the vault already
slightly pointed in cross-section
– The lunettes of the central barrel vault rise consid-
erably (1 m circa) above the capstone of the main
arches
– There is no transverse arch, neither in the nave nor
in the aisles, neither on the extrados or intrados of
the vault
– The vaults of the aisles are rounded transverse barrel
vaults, connected to each other by conical lunettes
in longitudinal direction
– In cross-section, the profile of the vaults over the
aisles is visibly distorted where the adjacent conical
lunettes meet each other.
All these observations can be accommodated in the
following hypothetical account of the vault erection:
First, the outer walls of the church and the two arcaded
walls separating the naves were built. At this point,
there was no intermediate transverse connection or
stiffening between the free-standing, slender arcades
and the outer walls. Next, the roof was erected. The
roof made use of continuous tie-beams as anchoring
beams in the main trusses. This ensured temporal sta-
bilization of the free-standing, very slender masonry
structures (even the outer walls are very slender struc-
tures, exhibiting huge windows, but no external or
internal buttresses). After roof erection, the vaulting
was started with the transverse barrel vaults of the
aisles, the anchoring beams of the roof still being in
place. After completion of the transverse barrel vaults,
the builders obviously trusted them to carry the thrust
of both the roof truss and the central barrel vault,
and dared to cut the anchoring beams. Then, finally, Figure 3. Roof structure, Tuntenhausen. Arrow indicates
the central barrel vault was built. One might speculate location of originally continuous tie-beam cut away prior to
whether the unusual shape of the pointed vault (unique vault erection in the nave.
236
takes account of the deformed vault shapes, and is also Technically even more demanding is the church type
fastened to the vault in such a way as to effectively developed in Bavaria during the 18th century, namely,
hide already existing cracks. Therefore, the authors felt churches which embed an octogon for the congrega-
entitled to conclude that the present-day cracks present tion in a rectangular outer shape, the vault being a
no hazard for the structural stability of the church, even cupola which is hidden beneath an ordinary ridge roof.
though some of the cracks are up to 10 mm wide. These cupolas easily reach 20 m spans (e.g., parish
As a remark concluding the case study, we would church of Murnau, see below), pushing the required
like to note that the timber columns of the roof are all roof span to the limit of traditional carpentry.
tilted. However, this effect seems to be mostly inde- The authors of the present paper have conducted,
pendent from the problems with the vaults. Rather, the during the last two years, an extensive survey of
upper storey of the roof, the “liegender Stuhl”, leans baroque roofs in southeastern Germany (Bavaria),
towards the east due to a lack of stiffening. The stiff with the goal of identifying typical solution strate-
construction of the lower storey prevented the upper gies followed by carpenters in that region to cope
part from toppling, even though the construction of with the problem of large span and the challenges
the hip roof over the ambulatory is very defective. Fur- of fairly low pitch and missing tie-beam. Around
thermore, it is worth noting that the horizontal beam at 40 church roofs have been inspected, covering the
mid-elevation of the lower storey of the roof (cf. fig- entire period from 1600 to 1800. Our study comple-
ure 2) cannot be considered as a raised tie-beam, being ments the survey conducted by Sachse in the region
connected to the columns only by tenon-and-mortise directly adjacent to our area in the west (Sachse 1975).
connections. The full results of the survey will be published in a
dedicated monograph (Holzer & Köck 2008). In the
remainder of the present paper, we present typical con-
structive strategies encountered during the survey, and
2 CHALLENGES OF THE BAROQUE ROOF
we address the question how much information the
contemporary carpenters could possibly have gained
The roof of Tuntenhausen is just one of the start-
from the comparatively rich production of carpenter’s
ing points of the development of the baroque roof
treatises published in German language. Also, we dis-
in Bavaria. Obviously, it did not work quite as well
cuss issues of the mechanical behaviour of the typical
as it should have. The reasons for the failure of
structures encountered.
the structural concept at Tuntenhausen and the chal-
As late as 1814 (!) Franz Sax, of Vienna, Austria,
lenge of similar situations elsewhere are quite obvious:
published a section of a barrel-vaulted church with a
Baroque roofs in southern Germany typically exhibit
roof truss in his treatise on Bau-Technologie und Bau-
a pitch of around 45◦ . Contemporary carpenter’s trea-
Oekonomie (Sax 1814, cf. Figure 4). The lower storey
tises call this pitch the “German” pitch. It is a pitch
of the roof structure is constituted by a high “liegender
that is considerably lower than the 60◦ pitch of typical
Stuhl”. The only precaution taken to carry the thrust of
gothic roofs. The low pitch aggravates problems with
the roof is a raised tie-beam, connected to the braces
thrust, particularly in situations where continuous tie-
of the “liegender Stuhl” by means of iron clamps.
beams are not feasible, and the traditional carpentry
techniques that had been developed over the centuries
in the area north of the Alps for the high-pitch gothic
roofs were ill adapted to the new challenge of low-
pitch roofs, as evidenced by many 17th century roofs,
including the one of the Hofkirche at Neuburg on the
Danube (1615), the lowest of the Bavarian roofs of
the time.
Furthermore, brick barrel vaults with small lunettes
became the standard vaulting solution in Bavarian
baroque architecture. Unlike the adjacent regions in
southwestern Germany, Bavarian baroque builders
used timber “lattice” vaults only with reluctance, with
only very few exceptions, mostly cupolas. In addition,
virtually all the brick vaults of south-eastern Bavarian
baroque churches rise high enough into the attic space
to prevent continuous tie-beams.
At the same time, even minor church buildings of Figure 4. Design of a church roof accommodating a
the 17th and 18th century easily reach the 16.50 m masonry barrel vault, from Franz Sax, Bau-Technologie, 1814
span which is a typical upper threshold on late gothic (detail from plate V). A raised tie-beam is secured by iron
church nave widths (Augsburg, St. Ulrich, 1499). clamps.
237
Figure 5. View below the raised tie-beam in the roof of the Figure 6. Design of a roof without a tie-beam, from Franz
church atAltenerding (1721). The long tie-beam in the middle Sax, Bau-Technologie, 1814 (detail from plate X). Diagonal
of the picture is not a continuous one, but also stops at the braces attached to king post.
extrados of the masonry vault (covered by thermal insulation).
238
Figure 8. Joint of diagonal braces with king post.
Munich-Perlach (1728). Viewed from the back side. Front
side shows a dove-tailed lap joint and additional iron straps.
Figure 10. Solution for the trefoil timber barrel-vault as
suggested by Johann Wilhelm (1649).
239
Figure 12. Section of the roof at Baumburg (1756).The scis-
sor-braced roof accommodates a longitudinal barrel vault, not
Figure 11. Scissor braces in the roof of the parish church shown here.
at Murnau (1721), employed to accommodate the cupola of
20 m diameter inside the roof.
church with a longitudinal barrel vault covering the
nave; as such it reflects the typical situation of Bavarian
found in the church of Murnau (1717–21). However, baroque. Secondly, the joinery of the roof exhibits a
this church was a completely singular structure at the marked “engineering” character: The scissor-braces
time of its building. The “nave” of the church is con- are not connected to the other timber members by tra-
stituted by a Bavarian variant of the Roman Pantheon, ditional joints, but rather, exclusively by iron bolts.
namely, an octagonal space with a diameter of 20 m. On the other hand, all the less important joints follow
A perfect sphere is inscribed into the nave, the princi- tradition, even to such an extent that lap joints with
pal corniche coinciding with the equator of the sphere. a curved end – a characteristic element from the ver-
Strangely, the huge hemispherical cupola (one of the nacular architecture of the region – abound in the roof
largest baroque cupolas in Germany, surpassing the truss.
one at Weingarten, Swabia, cf. Kutnyi & Wiesneth
2007, with a diameter of 15 m, and approaching the
cupolas at Ettal and Dresden, both with a diameter of
approximately 25 m) is not visible from the outside, but 3 STRUCTURAL ASSESSMENT OF SCISSOR
hidden under an ordinary ridge roof. Accommodating TRUSSES
such a large cupola in the attic posed an unprecedented
challenge to the carpenter. Stress analysis of historical roof trusses poses spe-
In Murnau, the carpenter resorted to scissor braces cific problems which are otherwise not common in
connecting all pairs of rafters to each other in the area structural engineering. As opposed to the analysis of
where the cupola protrudes into the roof space (see modern constructions, the idealization of the structure
Fig. 11). Also, he added a bridge-like truss in the plane is not straightforward, in particular because the stiff-
of the rafters to carry the load from the “open” part of nesses of the joints and the support conditions are not
the roof to the parts with continuous tie-beams. Last well defined. On the other hand, these modeling deci-
not least, both the brick cupola and the “hole” cut into sions have a decisive influence on the results, as we
the base of the roof are reinforced by closed rings of tie- shall demonstrate by a little example.
beams at various levels. The Murnau roof is one of the In the following, we present some preliminary
most remarkable feats of Bavarian baroque carpentry results from the analysis of a principal girder of the
and would deserve a monographical study in which we Baumburg roof under symmetric loading. Very similar
cannot give here. analyses have also been conducted by Lewandoski &
Only by the middle of the 18th century, scissor Levin (2003) for historic American scissor trusses.
braces had developed into a commonplace solution for Although these 19th century American trusses are
the problem of interrupted tie-beams. Regularly, the only superficially akin to our baroque trusses, but dif-
south-eastern Bavarian architect Johann Alois Mayr fer conceptually in many details, our results compare
(1723–1771) employed scissor-braces whenever there well to those already reported by Lewandoski & Levin
was a need to use discontinuous tie-beams. His church (2003).
roofs at Baumburg (1756), Marienberg (1764) and In the finite element analyses presented, we have
Kirchweidach (completed only after Mayr’s death, in considered the “liegender Stuhl” girders as frames
1774) all employ scissor-braces. The case of Baum- with rigid corners. The rafters are in contact with the
burg (Fig. 12) is particularly interesting for two rea- tilted columns of the “liegender Stuhl”. All traditional
sons. Firstly, Baumburg is a standard “wall-pillar” timber joints have been considered as perfectly hinged,
240
Figure 13. Baumburg roof. Normal forces obtained for a Figure 14. Baumburg roof. Normal forces obtained for a
model based on supports that are allowed to slide horizontally. model based on fixed supports. Thick lines: compressive
Thick lines: compressive forces. Thin lines: tensile forces. forces. Thin lines: tensile forces.
241
Holzer, S. & Köck, B. 2008. Barocke Bautechnik. Gewölbe, Rosenheim 2007. Staatliches Hochbauamt Rosenheim,
Kirchendachwerke und Kuppeln in Südbayern. Regens- report on the stuccos surfaces of Tuntenhausen, unpub-
burg: Schnell & Steiner, to appear in 2008. lished personal communication.
Kutnyi, A. & Wiesneth, A. 2007. Die Kuppelkonstruktion der Sachse, H.-J. 1975. Barocke Dachwerke, Decken und
Basilika in Weingarten. Preprint, Lehrstuhl Bauforschung Gewölbe. Berlin: Mann.
und Denkmalpflege, Tech. Univ. Munich. Sax, F. 1814. Bau–Technologie und Bau–Oekonomie. Vol. III:
Lewandoski, J. & Levin, E. 2003. Historic American roof Die Zimmermanns–Kunst. Wien: Doll.
trusses. Part I: Scissor trusses. With an appendix: Historic Walter, C. 1704. Architectura Civilis. Augspurg: Wolff.
scissor truss analysis. Timber framing (69): 4–17. Wilhelm, J. 1649: Architectura Civilis. Frankfurt: Author.
242
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: This document presents partial results of a much broader research project that has recognized
historical and technical aspects pertaining to a set of 34 historic brick arch bridges, 20 of which still exist; all
with diverse geometric characteristics and singular dimensions. The methodological process is exposed as was
carried out with one of them and conclusions are presented, which can be extended to almost the totality of the
cases studied for the purpose of offering analysis alternatives to professionals interested in the rehabilitation and
conservation of the South American architectural and engineering patrimony.
1 INTRODUCTION
243
Figure 2. Lateral rendering of the brick arch bridge over the
Cauca River in the city of Popayán. Drawing by: E. Sotelo.
244
Figure 5. View of the main arch of the bridge over the Cauca Figure 6. View of lower portion of the ring on the second
River in the city of Popayán (2007). arch of the bridge over the Cauca River, in Popayán, show-
ing the differential degree of deterioration of the constitutive
In another document kept in the Nation’s General bricks (2007).
Archives (Bogotá) there is testimony relating how, dur-
ing a first stage of the work begun in 1769, the bridge tropical regions, which contributes to the concentra-
collapsed at the precise moment of closing one of the tion of distinct living organisms upon bridge surfaces
brick arches – due to the lack of technical direction. (Fig. 6).
The final construction was left to Francisco Basilio de The macro structure is still in service to vehicular
Angulo and Josep Hidalgo de Aracena. traffic up to 5 Tons, as indicated to users. Maintenance
Once in service and for decades, the bridge over work is necessary, as is the replacement of ceramic
the Cauca River was always considered one of the pieces that have disappeared by using new materi-
best in the region and the nation; for example, English als with similar physical and mechanical properties
colonel J. P. Hamilton briefly mentions it in his book to those used in its construction.
(Hamilton, 1827) as part of his journey between
Popayán and Cali, just prior to the earthquake of 1827
that affected the region and destroyed several temples
and caused severe damage to the bridge’s main arch, 3 CHARACTERIZATION OF THE CERAMIC
rendering it useless. PIECES ON THE BRIDGE
There is evidence (Zawadsky, 1929) that on at
least two occasions local construction workers tried to 3.1 Mechanical properties
rebuild it, but with no satisfactory results given that in According to the results of the historical research, the
both cases, upon removing the wooden trusses placed bricks on the bridge over the Cauca River were made
under the vault, the materials fell and were dragged in artisan manner with local clays as raw material, the
away by the river. It was not till 1840 that Polish same craftsmanship still used in the fabrication of all
engineer, Estanislao Zawadsky, directed the definite types of pieces, with 40 × 20 × 10 cm standard dimen-
reconstruction work of the bridge. sions. The furnaces are still simple constructions that
use a combination of mineral coal and vegetable lum-
ber as fuel, reaching firing temperatures between 750
2.3 Current state of conservation
and 1100◦ C. Production periods for the bricks vary
A visual inspection carried out in 2006 permitted ver- between 8 and 12 days.
ifying that the bridge has not been altered by recent Due to analysis of specimens obtained from direct
structural interventions, except for the restitution of extraction of the masonry of the bridge, it was possible
the roadway surface in asphalt. A masonry plasterwork to perform physical and mechanical characterization
made with Portland-type cement mortar during the of the clay pieces by following recommendations con-
first half of the 20th century is in poor state of conser- tained in the ASTM C133-97 (American Society for
vation (Fig. 5) and only remains on the upper portion of Testing and Materials, 2003) norms. Some of the
the bridge’s tympanums. On the lower part of the rings results obtained in seven of the samples are shown
there is evidence of an accelerated degradation process in Table 2.
of the bricks by weathering of the clay-like material In spite of its low resistance capacity, there is no
due to prolonged exposure to environmental humidity physical evidence of fissures or crushing caused by
and to the presence of the type of vegetation found in compression efforts.
245
Table 2. Results of the cold compression test applied sam-
ples taken from bricks on the bridge in Popayán over the
Cauca River.
Sample 1 – Figure 8
Element Wt %∗ At %∗∗
CK 11.09 18.25
OK 40.31 49.81
NaK 01.87 01.61
MgK 01.03 00.84
Figure 7. SEM/EDX micrograph (162x) of the organic por- AlK 13.74 10.07
tion of the surface of the construction work. Source: Plasma SiK 23.94 16.85
Physics Laboratory, Universidad Nacional de Colombia at MoL 02.98 00.61
Manizales, Colombia. KK 00.45 00.23
CaK 00.57 00.28
Other physical analyses carried out permitted con- TiK 00.55 00.23
cluding that the bricks on the bridge have high porosity FeK 03.47 01.23
(39.4–40.4%), which could not always be a charac-
∗
teristic of the original sample, but could be related Percentage related to the total weight of the sample.
∗∗
to natural physical wear and to the environmental Percentage related to the atomic weight of the sample.
conditions.
of masonry located at the base of the main arch of
3.2 SEM analysis the bridge. This analysis permitted a first approach
Scanning Electron Microscopy (SEM) micrographs to understanding the structural characteristics of the
with 162x magnifications (Fig. 7) applied over surface material. The results obtained from one of the sam-
planes of the bricks reveal hyphae and the forma- ples are expressed in Figure 8, and also related on
tion of mycelium of the fungus that has grown due Table 3. Silica and oxygen contents are notable in the
to long-term environmental exposure. The porosity of total weight of the sample, slightly close to elemental
the bricks facilitates the presence of these organisms. content of the material currently used in the production
Visual inspection permits deducing that the most of ceramics.
affected areas are found on the borders of the struc-
ture and are probably caused by the incompatibility of
3.3 XRD analysis
the mortar joints with the bricks, along with the water
percolation effect for many years. A study of the mineralogical composition of three
Simultaneously, SEM analysis techniques were masonry samples was conducted via X-ray Diffraction
applied on samples extracted from an inner portion (XRD) tests to identify the origin of the raw materials,
246
Table 4. Quantification of crystalline phases present in the
samples via XRD.
% of % of % of
compounds compounds compounds
Compound in sample 1 in sample 2 in sample 3
247
Figure 10. First vibration mode of the bridge at 1839 Hz. Figure 12. Tenth vibration mode of the bridge at 3138 Hz.
248
is abundant bibliography on the issue, as well as case
studies where the MFE has been applied. One of
the great advantages is that, additionally, it permits
answering a fourth question: how will it behave? This
is an important nucleus of the problem: foresight of
the interrelationships between the existing object and
the different alternatives to intervene upon it and how
to obtain a guarantee in each of them.
Thus, precisely knowing the behavior modes, under-
standing their matter, and predicting the types of
responses to future actions are today the indispensable
means of intervention upon the historical patrimony.
The research project exposed here points, in the long
term, toward the concrete design of actions leading
to the recovery and adaptation of the greatest pos-
sible number of cases, involving the work of local
Figure 13. Tension forces on the clef of the main arch of the
bridge over the Cauca River produced by seismic accelera-
communities.
tion, under the framework of the seismic spectrum estimated
for Colombia.
ACKNOWLEDGEMENTS
tension phenomena with values around 75% above the The authors thank the support of the Dirección de
maximum crushing values presented by its static con- Investigaciones (DIMA) at Universidad Nacional de
dition, aside from such forces being concentrated in Colombia in Manizales, for the financial support to
the zone of the main arch. carry out this project. The authors also recognize the
collaboration received from professors A. Devia and
O. Correa from the same institution.
5 CONCLUSIONS
249
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
O. Niglio
University e-Campus, Civil Engineering, Italy
ABSTRACT: The studies on the mechanic behaviour of domes, based on the static analysis of arches and vaults,
started in the first half of the XVII century. At the end of 1741, when the Pope Benedetto XIV commissioned an
assessment to three mathematicians Roger Joseph Boscovich, François Jacquier e Thomas Le Seur, belonging
to the Repubblica Romana dei dotti. There where serious worries on the static conditions of the San Pietro
vault where important cracks could be appreciated. Some interesting studies were developed on this problem,
even followed by learned and heated arguments, that lead to compile treatises that systemized the knowledge on
this argument. The historical documents even report the opinion of other experts, for example that one of the
mathematician Giovanni Poleni from Padua. On spite of that, the study of the three previous mathematicians
was different from the others carried out before because it had an important innovation: the assessment was
accomplished, perhaps for the first time, using a scientific criterion in order to interpret the mechanic behaviour
of an architectural structure. Its historical importance resides on the scientific theoretical conceptions used to
analyze a structural problem that were completely different from the empirical rules, at most of geometric nature,
used before. The principle of virtual works was further applied in the assessment. It was used to define the size
of the metal rings for the drum of the vault. Some researchers intend this assessment as the moment when the
change from the engineering based on artisan traditions to the engineering based on the application of new
scientific theories has been accomplished.
251
Figure 1. Saint Peter’s dome (Curcio, 2003).
Figure 2. Section of Saint Peter’s dome (Curcio, 2003).
252
3 THE ASSIGNMENT OF THE ASSESSMENT
4 DIAGNOSIS
5 AETIOLOGY
The first part of the study is dedicated to a detailed
description of the dome and the creation of a detailed The second part of the assessment is dedicated to iden-
geometrical survey. This instrument of knowledge is tifying the causes responsible for the instability. The
subsequently perfected by superimposing the existing authors formulate a graphic diagram to show how the
pattern of cracks. movements may have occurred. The cracks are inter-
The precise representation aims to lay the basis for preted as hinges around which the parts of stonework,
analysing the loads and interpreting the relative move- which is not cracked, considered non-deformed, have
ments between the various structural parts, which, rotated.The kinematical interpretation of the pattern of
according to current terminology, could be consid- cracks identified enables the three Mathematicians to
ered macro-elements. Even the variations in width of exclude a number of hypotheses formulated by others.
the cracks along the development of each of them In the absence of cracks, which can justify them,
are evidenced, with the clear intention of representing they exclude that the causes of the instability can be
253
causes are responsible for pushing the drum out-
wards: the weight of the small dome and the weight
of the ribs including the gores of the dome. Similarly
two horizontal forces create resistance: the force of the
rings and that of the support made up of the base, drum
and buttresses.
To assess the weights of the structures the three
Mathematicians weigh a mass of travertine and one
of stonework. Thanks also to the geometric data of
the single parts surveyed, they succeed in proving
that the total weight of the dome is equal to about
56 000 tons.
Whereas, to asses the force corresponding to the
iron rings the three Mathematicians appeal to the
treatise Coesione de’ corpi solidi by Pietro Ban Muss-
chenbroek deducing that the traction resistance of the
first ring corresponds to 114 t and that of the second
to 95.
Figure 4. Description of the deformation mechanism (1742, After observing that the variation in the length of
Parere di Tre Matematici, Bibliotheca Hertziana, Rome). a circular chain increases in the proportion of 2π in
relation to the variation of the radius, the Three apply
the PVW equalling all of the positive and negative
works made by the elements in play. The positive work
attributed to the subsiding of the foundations. In sub- is produced by the weights of the macro-elements,
stance, they think that the weight of the small dome, which represent the damaged parts of the dome and
the ribs and the double cap have weighed down caus- small dome. The negative work is determined both by
ing the drum to move outwards. As for the iron rings the resistance opposed by the drum in contrasting the
and possible damage, the three Mathematicians, still deformations outwards, and by the resistance available
with an elasticist mentality, claim that it would not be in the rings on the various levels. This latter work, how-
possible to know if, and to what extent the metal rings ever, is assessed incorrectly: as the concept of potential
are truly effective. This is because they cannot be seen elastic energy was not yet clear, it is calculated directly
directly as they are inserted into the stonework, nor is it considering the resistance on breaking.
possible to know the tightening tensions. Moreover, the After obtaining the values of the forces (thrust
iron could have suffered thermal deformations chang- and resistance), the problem of balance, however, is
ing its ring strengths, while some of the rings could handled strictly in terms of energy.
even have broken. The process adopted by the three Mathematicians
to solve the problem, despite containing a number
of imperfections, is daring and modern. The Mathe-
maticians renounce the use of processes based on the
6 PROGNOSIS
polygons of the forces, and refer to a method, the PVW,
mentioned previously by René Descartes in his prin-
The assessment continues defining the process that
cipal work published in 1637 (Descartes, 1637) and
the three Mathematicians intend to use to assess the
subsequently perfected by J. Bernoulli.
quantity of actions associated with the kinematical
In this way the three Mathematicians succeed in
mechanism and the tensional state of the rings whose
proving that the weight of the small dome and the dome
scarce efficiency appeared to be responsible for the
exert a force H on the impost ring of the dome dis-
instability.
tributed as internal pressure p (H = 2πR × p), whose
In terms of scientific innovation this is the topical
total value results from the following relation, obtained
moment of the assessment because the Three decide
equalling the virtual works considered (Fig. 5):
to apply a process of calculation based on the PVW. In
fact, the instability refers to the excess weight that bur-
dens, above all, the upper parts of the dome, pushing
the drum outwards, and to the scarce ringing resistance where G indicates the weights of the small dome and
of the lower buttresses. the portions of the dome, where v indicates the lower-
At this point the three Mathematicians consider the ings of their centres of gravity and where h indicates
results obtained by Philippe de la Hire and Couplet the horizontal opening of the impost.
(Benvenuto, 1981) concerning the behaviour of arches The resistance W, which contrasts the thrust H is
and vaults. They come to the conclusion that two made up partly of the resistance exerted by the rings
254
Consequently, the cause of the instability is
attributed to this imbalance. The rings, which had
been laid during the construction are therefore unable
to contrast the pushing action of the structures. In
substance, the three Mathematicians conclude observ-
ing that the upper part tends to move inwards under
the action of the loads of the small dome, while the
lower part tends to move outwards developing traction
tensions in the rings.
In the light of the knowledge relating to the mechan-
ics of structures, developed later on, the adopted
process does not appear without imperfections. First
of all, the work carried out to dilate the rings, calcu-
lated as work of extending an equivalent straight rod,
is assessed considering a constant force applied from
the beginning of the elastic deformations, unduly asso-
ciating a size growing elastically with a static one.
Moreover, no reference is made to the work of elas-
tic deformation of the “macro-elements” according to
which the kinematism of the instability was examined,
nor of the anelastic deformations of the unbalanced
areas. On the other hand, as we have already seen,
Figure 5. Analysis of the balance dealt with by applying a mechanical theory for structures had not yet been
the PVW. developed to the point of being able to assess these
aspects correctly. Therefore, the daring choice of a pro-
cess of theoretical calculation applied for interpreting
mechanical phenomena remains significant.
7 THERAPY
255
8 CONCLUSIONS
REFERENCES
action of the drum is clearly less than that of the small
Rare and archival manuscripts.
dome; therefore they can be preserved in the state they Boscovich, R.G. & Al. 1742. Parere di tre Matematici sopra i
still are today. However, they suggest filling in a num- danni che si sono trovati nella cupola di S. Pietro sul finire
ber of spaces, which support the vaults and indicate dell’Anno MDCCXLII. Dato per ordine di nostro signore
other rings to be inserted level with the drum. Papa Benedetto XIV. Venice (printed in 1943).
Lastly, the three Mathematicians claim that all of Cosatti, L. 1743, Contignationes ac pontes Nicolai Zabaglia.
the other solutions they had heard, and which had Rome: Nicolò Paglierini
been suggested to them, were superfluous for resolv- Descartes, R. 1637. Discours de la Méthode de bien conduire
ing the static problem of the dome. The six rings alone sa raison et chercher la vérité dans les sciences; plus la
Dioptrique, les Météores et la Géométrie, qui sont des
and the various careful interventions planned would
essais de cette méthode.
have undoubtedly contributed to improving the situ- Galilei, G. 1638. Discorsi e dimostrazioni matematiche
ation. In fact, the other solutions would have given a intorno a due nuove scienze attenenti alla mecanica et
load of approximately 950 tons in relation to a total i movimenti locali.
weight of the dome of over 56 000 tons, so with rather Poleni, G. 1748. Memorie istoriche della Gran Cupola del
insignificant added value. Tempio Vaticano e de’ danni di essa, e de’ ristoramenti
The indicated project solution applying a scientific loro divise in libri cinque alla santità di nostro signore
process made it possible to propose work that was Papa Benedetto XIV. Padua.
not invasive, respectful of the Michelangelesque build- Ackerman, J.S. 1968. L’architettura di Michelangelo. Turin:
Einaudi.
ing and its decorations and aesthetics. The assessment
Baggio C.& Da Gai E. 2000. Tra differenza ed inno-
generated great controversy from well-known scholars vazione: la meccanica in architettura, in G. Curcio
including Poleni himself, who had also been com- and E. Kieven (ed.), Storia dell’architettura italiana, Il
missioned to make an analysis of the same problem Settecento. Milan: Electa Mondatori.
(Poleni, 1748; Baggio & Da Gai, 2000). Boscovich’s Buccaro A. 1988, Aspetti della cultura tecnico-scientifica
work was later praised by C.L. Navier (Navier, 1829) in epoca vanvitelliana: dall’architetto allo scienziato-
who recognised its originality. artista, in Tecnologia Scienza e Storia per la con-
The reinforcement work was carried out under servazione del costruito. Florence: Callisto Pontello
the supervision of Luigi Vanvitelli (Buccaro, 1988) Foundation.
Barrow, J. D. 1992. Perché il mondo è matematico, Rome-
who applied the three Mathematicians’ proposal, and
Bari: Laterza Editore.
Master Nicola Zagaglia was responsible for the organ- Beltrami, L. 1930. Relazione delle indagini e dei lavori di
isation of the site (Cosatti, 1743; Zander 1991). restauro alla Cupola Vaticana dal maggio 1928 a marzo
The assessment described in this memorial is illus- 1930. Vatican City: Vatican Printing Works.
trated in detail in the book Dall’ingegneria empirica Benvenuto, E. 1981. La scienza delle costruzioni ed il suo
verso l’ingegneria della scienza (Niglio, 2007). sviluppo storico, Florence: Sansoni Editore.
256
Capecchi, D. 1999. Il Principio dei Lavori Virtuali da Niglio, O. 2007, Dall’ingegneria empirica verso l’ingegneria
Aristotele a Bernoulli, Naples: Luda Editore. della scienza. La perizia di tre Matematici per la Cupola
Capecchi, D. 2002. La storia del Principio dei LavoriVirtuali, di San Pietro (1742), Padua: Il Prato Editore.
Benevento: Hevelius Edizioni. Rocchi Coopmans de Yoldi, G. 1996. S.Pietro. Arte e Storia
Conforti, C. 1997. Lo specchio del cielo : forme signifi- nella Basilica Vaticana. Bergamo: Bolis Editore.
cati tecniche e funzioni della cupola dal Pantheon al Timoshenko, S. P. 1953. History of Strength of Materials.
Novecento, Milan: Electa Mondadori. London: McGraw-Hill.
Curcio, G. 2003. Il Tempio Vaticano 1694. Carlo Fontana. Timoshenko, S.P. 1956. Scienza delle Costruzioni. Turin:
Milan: Electa Mondadori. Andrea Viglongo & C. Editori.
Di Pasquale, S. 1996. L’Arte del Costruire tra conoscenza e von Halász, R. 1969. La prefabbricazione nell’edilizia indus-
scienza. Venice: Marsilio Editore. trializzata. Milan: I.T.E.C.
Navier, C. L. 1829. L’application de la Mécanique a Zander, G. 1991. Storia della Scienza e dellaTecnica Edilizia.
l’etablissement des constructions et des machines. Paris. Rome: Multigrafica Editrice.
257
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The Gesù Nuovo church – the seat of the Society of Jesus in the capital of the Spanish viceroyalty –
presents the opportunity for a significant case-study related to the connections existing between static history
and present problems in the structural field. Archival documentation allows the reconstruction of the various
hypotheses being made, starting from the 17th century, in relation to the stability of the building. The research
highlights a historical structural “weakness”, referring, in particular, to one of the pillars in the crossing and
to the arches weighing onto it. Making a close study of the interpretations, given from the 18th century, by
royal engineers until recent years, the study aims at correlating the hypotheses expressed in the past with the
present structural conditions of the church. The “historical” issue of the cross damages is studied carefully in the
conclusions in order to emphasize the necessity of a well-considered integration between the analysis of present
conditions and a complex system of multidisciplinary skills, aiming at a full comprehension of the causes of
architecture instability.
1 INTRODUCTION
The local Soprintendenza [regional board of the
Ministry of cultural and environmental heritage]
1.1 Ancient and new structural interventions
recently restored the four pillars, the tambour and
in the Gesù Nuovo church
the arches of the church’s crossing. It is interesting
One of the restorer’s main tasks is to make a close to notice that, over the centuries, crushes and dam-
investigation of the various phases of a building’s con- ages have been interpreted using the best technicians
struction and transformation. The knowledge of all of each age. The first dome’s double calotte col-
the previous consolidation or “embellishment” inter- lapsed after the 1688 earthquake and afterwards some
ventions, permits the planning of a restoration project hypotheses were made about the characteristics of the
as compatible as possible with the conservation of an crush. The dome was then rebuilt with the plan of
artefact. the architect Arcangelo Guglielmelli with its weight
This research has been carried out on a monument discharging onto the four 16th century pillars under
of considerable importance, the Gesù Nuovo church the tambour. In particular, the first pillar on the left
in Naples. It aims not only at illustrating its history showed, over the centuries, serious symptoms of insta-
but also at assessing the damage, caused by opera- bility. Also on this occasion, the specialists, who were
tions conducted during the past centuries, to which, the most qualified in each period, were engaged in a
in different ways and using the technological skills of debate about the causes of the damage.
the various ages, there has been an attempt to find a Luigi Vanvitelli, leading the debate, hypothesized
remedy. that the cause of the damage to the left pillar and of
Even in this specific case the research has con- the complex cracking pattern of the church’s crossing
firmed the conviction that each building’s consolida- was due to the hurried repair of building damage after
tion is absolutely not separate from the more complex the 1688 earthquake.
operation of restoration. This operation is based on In conclusion, during the 18th century, two tech-
historical knowledge and uses all possible technical nical fronts were formed: one believed that damage
investigations to reach a correct diagnosis of the build- was due to the weight of Guglielmelli’s dome, the oth-
ing damage and to preserve, as much as possible, the ers that they were due to the covering pressure on the
material with which it is constructed. Moreover, con- vertical walls.
sidering the static model as a testimony of a civilization The remedy consisted in the replacement of the
and a historical document, any project must be based masonry dome with a light lathwork vault supported
on a precise understanding of it. by wooden trusses. Nevertheless, considering the issue
259
with a view that ignores the historical perspective, if of its urban significance and role. The vertical
such a drastic solution, implying the destruction of perimetric structures of the palace were preserved
a significant constructive element of the monument, and underfoundated (Pirri 1970, p. 116, p. 121, note
is to be censured, the subsequent substitution can be 58) while the excavation of the four crossing pillars
judged much more seriously. In 1973, the Genio Civile started in 1584, probably with the center of the cross-
(Ministry of Public Works) replaced the light lathwork ing planned in correspondence to the palace courtyard
vault with a reinforced concrete one, weighing on the (Schinosi 1706–1711, lib. V, cap. VI, p. 464).
masonry structures below. The reconstruction of the building phases is facil-
In the period after World War II the use of reinforced itated by information given by the Jesuit chronicler
concrete became more and more frequent, especially Giovan Francesco Araldo at the end of the 16th cen-
in structures that were disguised; this happened in the tury (Divenuto 1998), together with historical data
Gesù Nuovo church where the inside of the concrete provided by documentary sources preserved in the
cap had been coated with stucco. Nevertheless, in this Jesuit archive in Rome (Pirri 1970; Bösel 1985, vol. I,
case, with great probability, the remedy has been the pp. 416–419). Through Araldo’s narration it is possible
cause of further disruptions for which, in recent years, to understand that the foundations of the four pil-
there has been an attempt to find a solution. Fortu- lars were realised between November 1584 and March
nately, today the instruments allow technicians to make 1585, reaching the “monte, et ferma pietra”, that is the
diagnoses, which are also supported by numerical cal- tuff in the subsoil (Divenuto 1998, p. 238). Reading the
culations, that are much more precise than in the past; Jesuit chronicler, we learn that the four pillars on which
still, as in this case, the attention and sensitivity of the the dome had to be placed were built in only three
restorers must ensure that the proposed solutions do months, from March to June 1585 (Divenuto 1998,
not involve any more drastic substitution of parts. The p. 240 and p. 243), covering their tuff basements with
solutions should instead imply operations for struc- piperno blocks. An interruption of approximately ten
tural improvement with the addition of elements which years in the building of the church followed the estab-
are compatible with the ancient structures. lishment of the pillars so that the construction was
started again only in 1594 with work in the tribune
and in its side chapels (Divenuto 1998, p. 372); con-
2 THE BUILDING OVER TIME. struction times slowed down again in 1596 because
CONSTRUCTION AND RESTORATIONS a serious blaze burned in the church and because of
BETWEEN THE 16TH AND 18TH the death of Valeriano in July. The construction of the
CENTURIES crossing was started again only when the tribune was
completed. The erection of the four arches discharg-
2.1 The building site and the pre-existing ing on the pillars – built about ten years before – and,
structures finally, of the bottom ledge of the tambour concluded
this building phase.
The study of events of the construction of the Gesù
The church of Gesù Nuovo was inaugurated in
Nuovo church provides significant grounds for reflex-
October 1601 (Errichetti 1963, p. 183, note 2) without
ion referred to multiple topics which are related to
its central dome: about thirty years passed from the
the construction and to the transformation of architec-
inauguration to the beginning of the construction of
ture since the 16th century. The analysis of historical
the Jesuit dome in 1629 (Errichetti 1963, p. 177).
sources suggests an articulated combination of inter-
pretations of damage and, consequently, of methods of
2.2 The domes of the Gesù Nuovo church in the
consolidation proposed during past centuries. At the
seventeenth century
same time, the examination of the planning choices
which have been adopted in the substitution of parts – Although it is scarce, useful information about the
the dome in the crossing, particularly – highlights how construction of the dome can be deduced from the
the integration of lacking parts has been developed seventeenth century cartography and, moreover, from
during each historical phase with a sincere expres- the description of the structure and of its dimensions
sion of the contemporary “state of technique” and the given by the Neapolitain guide Carlo Celano in 1692.
culture of planning. The first dome of the Gesù Nuovo church had a tam-
Since its establishment, the church of the seat of the bour that was marked externally with buttresses with
Society of Jesus in the capital of the Spanish vicereign volutes (Figs 1 and 2). Two calottoes weighed on the
was planned in correspondence with the pre-existing tambour and they were separated by an interspace
architecture of the 15th century palace of the prince of about seven metres high in correspondence with their
Sanseverino (De Frede 2000). keystone; in addition, a staircase leading to the lantern
External facades were re-used in Giuseppe Vale- on top was constructed between the two vaults (Celano
riano’s project along three sides of the building for 1692, p. 44). The lantern, consisting of a cylinder
economic reasons and, moreover, for the recognition of masonry surrounded by eight grey tuff columns,
260
cathedral of Naples in the same year (Russo 2007),
refused any hypothesis of a fall “on perpendicolo”
of the Gesù Nuovo dome as had been advanced by
contemporary “mathematicians”; on the contrary, he
emphasised the rotation suffered by the vaulted parts
and by the lantern towards the west transept side. As
the report of Lucchese makes clear, the four arches,
the pendentives and the tambour remained “firm and
intact” (ATSG, A/22, foll. 66v–67); a circumstance
that would not have arisen, according to Lucchese, in
case of a collapse only in a vertical direction.
The Jesuits attended to the repair of the tambour,
to the construction of the transept left vault again and
Figure 1. Naples, the Gesù Nuovo church in the map of B.
to closing the crackings with lime and bricks in only
Stopendael (1653).
six months (Errichetti 1963, p. 177). Thereafter, the
collapsed parts were rebuilt following the plan of the
Neapolitan architect Arcangelo Guglielmelli between
1692 and 1693 (Amirante 1990, p. 247 and p. 334).
As a drawing of 1769 clarifies (Sasso 1856, tav. 13;
Errichetti 1963, p. 178) (Fig. 3), a single vault con-
cluded by a smaller lantern was built in order to replace
the complex double dome of the beginning of the cen-
tury (Fig. 4); its weight loaded, of course, on the pillars
and arches dating, as said before, back to the end of
the sixteenth century.
Figure 2. Naples, plan of Gesù Nuovo church and Professed 2.3 From the interpretation of structural damage to
House (17th century). The drawing shows the double calot- planning choices. Debate and construction in
toes structure of the first dome of the church (Bösel 1985, the eighteenth century
tav. 271).
Just over sixty years after the rebuilding of the vault
balustrades and vessels, produced a significant weight and of its lantern, the dome by Guglielmelli and, in
on the lower calottoes; as we learn from Celano, the particular, the first pillar on the left required the atten-
lantern was the first element to crack shortly after its tion of Jesuits because of a worrying cracking pattern.
completion so that it required the replacement of a A monitoring of this pillar was undertaken in 1767
stone column with a brick one. using “swallow tails of marble” (Carrafiello 1995,
According to the above mentioned seventeenth cen- p. 358) in correspondence with the above-mentioned
tury author, this intervention caused a “weak point” pillar; at the same time, a first consultation with the
which was at the origin of the collapse of the Jesuit royal architect Luigi Vanvitelli was carried out in order
double dome, which occurred during the earthquake of to have suggestions about appropriate remedies (Di
1688. Therefore, such a hypothesis, empirically tend- Stefano 1973, p. 232).
ing to highlight the role played by the lantern in the The expulsion of the Jesuits from the Kingdom of
collapse of the dome, had already been advanced in an Naples, which occurred in the same year, and the sub-
“official report of damages” produced approximately sequent settlement of Reformed Franciscans in the
one month after the earthquake; in fact, this document church caused an interruption in the control of pro-
stressed that “the dome fell beginning to crumble from gression of the crackings and in Vanvitelli’s task. Only
the lantern”, striking in its collapse the lateral chapels from 1769 onward did the attention of engineers and
of the transept (Cantabene 2004, p. 58).Through ref- religious focus again on the conditions of the build-
erence to the documentary sources, it is possible to ing as the survey of 1769 by the architect Giuseppe
suppose that the seismic tremors caused the loss of the Mauro demonstrates (Fig. 3). Beginning from 1769
two vaults and that of the lantern, while they spared the the most accredited Parthenopean technicians of the
seventeenth century tambour which remained largely second half of the eighteenth century were involved in
in place.This fact is confirmed by the description of the the issues of the church. Among these, in particular, a
dynamics of the collapse provided by the Neapolitan decisive role was played in the choices by Ferdinando
architect Giuseppe Lucchese in 1708; this technician, Fuga, who replaced Vanvitelli as the trusted techni-
engaged in the analysis of crackings of the double cian of the Franciscans. Considering “azzardosissima”
dome of the Treasury of St.Gennaro Chapel in the the solution of the rebuilding only the cracked pillar,
261
Figure 4. Naples, the Gesù Nuovo church. Hypothesis about
the longitudinal section before the demolition of the second
dome (from Guerra 1967, Fig. 5).
262
exponent of an emerging “theory of vaults” in the
Neapolitan context, showed himself to be significantly
abreast of studies conducted in other parts of Europe
in the report he presented on that occasion (Guerra
1967, pp. 392–393); this, in particular, was in relation
to the equilibrium of vaults and arches, as the quota-
tions from de la Hire and Belidor demonstrate. This
theoretical and cultural background was not enough,
however, to avoid the answer, articulate and pragmatic,
that Luigi Vanvitelli presented to the Secretariat of the
Royal House in 1772 (Di Stefano 1973, pp. 233–234):
sharing the idea of intervention of Ferdinando Fuga
and removing the issue from “mathematical” interpre-
tations, Vanvitelli brought the problem of the Gesù
Nuovo church back to the direct examination of the Figure 5. Naples, the Gesù Nuovo church. The longitudinal
phenomena.The architect attributed the main causes of section after 1786 (from Guerra 1967, fig. 6).
the pillar’s cracking and the causes of the damage still
underway to the seismic tremors and to the excessive
speed of past “restorations”.
The position taken on the issue by Galiani seems
equally interesting: a meditated approach to the con-
structive and dimensional characteristics of the build-
ing emerges from the report submitted by him in 1774,
especially when the technician highlighted the exis-
tence of planning “repentances” in correspondence to
the cross. He reported, in particular, the existence of
a system of brick arches that had been constructed
higher than the four tuff arches and connected with
them through masonry plugs of poor quality. Galiani,
like Lamberti and Gioffredo, excluded the possibility
of an excessive load transmitted from the top dome
on the four pillars and he rather attributed the dam-
age to the thrusts of coverage structures on vertical
walls (Carrafiello 1995, pp. 355–373). Despite the
above described opposing positions, the majority of
technicians involved in the question ended up attribut-
ing the damage underway to the excessive weight of
the dome of Guglielmelli and to the poor quality of
the construction; so, the demolition of the tambour
and the higher vault was proposed in 1774. Started
only in 1786, the intervention was carried out under
the direction of the royal engineer Ignazio Di Nardo
and it involved the replacing of the seventeenth cen-
tury structures with a lighter “incannucciata” bowl
(Fig. 5), sustained by a system of upper wooden beams
(Errichetti 1963, pp. 181–183). Figure 6. Naples, the Gesù Nuovo church. The iron scaffold
Di Nardo’s intervention concluded, so, a significant used to build the fourth dome (1972) (ANSI).
phase in the history of the church’s domes, character-
ized by the progressive reduction of covering masses
made in 1973 by the Ministry of Public Works,
and, consequently, of weights.
through its local office of Genio Civile (Fig. 6). The
intervention was carried out to remedy the progressive
deterioration, aggravated by the effects of the 1961
3 THE FOURTH DOME OF THE CHURCH
earthquake, of the 18th century fake lathing vault (Iap-
pelli F.S.I. 1993). At a closer look the present dome,
3.1 Damages and interventions of consolidation
covered with decorative stuccoes in late-baroque taste,
The Gesù Nuovo “dome” that today dominates the is a plate of reinforced concrete leaning on a circular
crossing of the church is the result of a substitution masonry tambour whose base is inscribed in a square.
263
As an application of the widespread theories of that
moment on the use of reinforced concrete shells, the
“camouflage” solution adopted by the Genio Civile
complied with both structural and formal require-
ments. There was a foretaste of this choice in the
postwar institutional culture that looked favourably
on solutions that recompose architectural elements
of buildings highly compromised by war damage, –
with particular regard to the replacement of wooden
roofing – making use of the innovative technology of
suitably disguised reinforced concrete.
In a 1949 letter the Soprintendente of Campania
Monuments, Antonino Rusconi, informing the Jesuit
rector of the church about the conditions of the wooden Figure 7. Naples, the Gesù Nuovo church. The concrete
truss above the dome, calls for a review of the struc- truss in the project of 30th october 1987 (ASBANa 8/158,
ture of the cover and also, for a replacement of the Ministry of Public Works. Lavori di ripristino urgenti alla
chiesa del Gesù Nuovo danneggiata dagli eventi sismici del
lathing vault with a masonry or reinforced concrete
23/11/1980 e 14/02/1981).
one (ASBANa, Gesù Nuovo 8/158, draft n. 5846 of
the 9/12/1949). Therefore, concern about the condi-
tion of the dome is closely connected, as early as the
first post-war period, to the conditions of the wooden
trusses of the coverage.
At the same time, symptoms of masonry crushing
begin to cause worry when Rusconi brought to the
attention of the authorities to the fact that, in the accu-
mulated delay in the repair of the war damages, the
pillars of the crossing were showing a phenomenon of
expulsion of marble coatings (ASBANa, 8/158, letter
of 21/06/1952). The Genio Civile officer in his turn
(ing. L. Persico) highlighted the necessity to uncover
the pillars and, over all, to proceed to the restoration of
the dome to its former condition. This “ripristino” was
truly urgent, given the detachments of large plaster and
stucco pieces from the present dome’s structure, which Figure 8. Naples, the Gesù Nuovo church. Hand sketch of
was detrimental to the public safety, and given also the the covering above the choir (ASBANa, ing. Ettore Minervini
progressive degradation of the wooden truss. Report s.d.).
Although the cover replacement in 1973 has clearly
increased the heaviness on the walls, in the light of (Fig. 7), in addition to the spherical cover, would soon
recent investigations regarding the structure, which are show signs of its presence and, although in the past
discussed below, it has been observed that, during the the walls had already shown symptoms of a lack of
earthquake of 1980, the concrete cover played a joint cohesion, in the few years after the construction of the
function of the masonry structures in response to the new coverage, worrying signals alarmed the experts
pressures of the undulatory waves (ASBANa, Report involved.
of ing. Ettore Minervini, 20/09/1999). In 1993, a study group, appointed by the Soprinten-
In 1987, the condition of the rotting beams and of denza Generale agli Interventi Post Sismici in Cam-
the secondary frame of the roof above the crossing pania e Basilicata to draw up an executive draft
suggested the replacement of the wooden roof with a aiming at the recovery and restoration of Monumental
reinforced concrete one. Concrete slide walls, 25 cm Public Heritage of Naples Historical Center, indi-
thick, are projected above the cover to provide the cated a pattern of damage, due to crushing problems,
connection between all the longitudinal stringcourse bringing to light the need, impossible to postpone,
attaching at the roof ’s impost. The walls are configured to systematically investigate the static conditions of
at the intrados according to the shape of the vault that the elevated masonry walls (ASBANa, Gesù Nuovo
covers the crossing and at the extrados according to the 8/158, 01/09/1993, Min. BB.AA.CC., Progetto esec-
pitched roof. These walls lean on a re inforced concrete utivo per il restauro della chiesa del Gesù Nuovo).
stringcourse whose section was to be, according to the Five years later both the central organs of the Ministry
plan, 130 cm × 80 cm (ASBANa, Gesù Nuovo 8/158 (ASBANa, 14/10/1998, letter of Central Inspector
project). The weight reached by the new structure arch. R. Di Paola) and the local Soprintendenza
264
(ASBNa Report No. 103 of 20/5/1998) admitted the The question, also raised at the ministerial level
urgency of a plan of investigations aimed at under- (arch. R. di Paola) of whether these injections could
standing the overall structural situation of the church have the effect of burdening the nucleus of the pilaster
with particular reference to the elements that showed rather than consolidating it and risking an increase in
signs of instability: the four pillars of the crossing, the stress compression, has conducted to the use of very
dome’s pendentives and the tambour. fluid mortars compatible with the masonry materials,
These surveys, consisting in horizontal and inclined also able to exclude chemical reactions with the salts
logging, endoscopic investigations, non-destructive present in masonry themselves and avoid harmful reac-
investigations on the walls using flatjacks, topographi- tions to the marble cladding (ASBANa, 20/05/1998,
cal surveys with electromagnetic waves for the spatial Report n. 103). This first intensive work on the pil-
measurement of cracks and empty spaces, with the lars was followed by operations aimed at consolidating
return of data collected using surface georadar (GRP), the trabeation above the St. Ignatius chapel, above the
showed that the four pillars, despite a constructive nave on the side of the Moscati chapel and the tambour
symmetry between them, were characterized by a under the “dome” (ASBANa report n. 93, 26/05/1999).
remarkable diversity of the composition of walls and During the execution of these operations, signs of
by a lack of continuity, with special anomalies found in crushing have been noticed at the keys of the arches of
the south west pillar (ASBANa, Georadar report Tec- the transept, of the aisle, of the apse together with any
noin srl). Along with these investigations the engineer other secondary arches. In addition, the lower edge of
Ettore Minervini, entrusted by the local Soprinten- the tambour, once perfectly horizontal, was now like a
denza, performing the necessary calculations on the wavy line with the lowest points on the keys of these
structures of the crossing, determined the tensions of arches (ASBANa Report n. 38 3/4/2000). Therefore,
medium compression to which pillars are subjected: in the further static intervention of consolidation, on
the basis of what has already been done on the pil-
• Total weight of the structure weighing on the four
lars, executive priority has been given to saturation of
pillars: 13,398 t;
cracks as a preparatory operation.
• Media-resistant surface of the pillars: A. tot
4 × 4 × 4 mt = 64 square meters;
• Pressure-average = 20.9 kg/cm2 . (ASBANa, ing. 3.2 The conservation of existing constructive
E. Minervini report s.d.) systems between innovation and tradition
Combining the diagnostic results with structural Currently, the work of the Soprintendenza is focus-
calculations, engineer Minervini pointed out that the ing on the saturation of cracks that span the entire
pillars, though endowed with a certain wall thick- vault on the main aisle. In this case the operation has
ness, had a resistant surface reduced only to the outer been preceded by a consolidation of the wooden roof
layer of masonry, an outer peel of “tufo-pipernoide” (ASBANa Report n. 109, 14/06/1995) which shows
(tuff-piperno), 50 cm thick; while the inner core of how, in recent years, the attitude is to prioritize the
masonry bag showed a low degree of cohesion. The permanence of traditional structural systems through
intervention, aimed at stemming an obvious crushing consolidation. This aims at the preservation of the role
phenomenon that was underway, had become, at the of structural components of the architectural building.
end of the twentieth century, urgent. It is worthwhile to point out that the intervention
The occasion of a project for a pedestrian cross- carried out on the coverage correspondent to the choir
ing below the church provided the opportunity to also goes in the direction of the conservation of pre-
investigate the church underground. The results of sur- existing constructive systems. The constructive system
veys conducted by the Soprintendenza were reassur- produced up to now, while showing clear signs of a
ing about the compactness of foundational structures worrying history of damage, is material proof of a
and allowed the exclusion, as a concurring cause, of local building tradition that is also rich in historical
any possible sinking of the foundations (ASBANa, significance. The planned coverage is made up of a
29/09/2003 Geognostic investigations, Tecnoin srl). sequence of parallel walls shaped according to the
The solution reached by the Soprintendenza’s tech- trend of the pithed roof. Each septum is composed of
nicians aimed to pursue a widespread consolidation a masonry arch that is slender compared to the larger
of the building, which is now historicized as a mixed mass of the triangular wall itself, which, departing
structure of masonry and concrete, at the same time from the underlying barrel vault, discharges its weight
improving the masonry response to the weight that on the raising walls and partially on the same vault’s
the enormous coverage of the dome exercises on ele- buttresses.
vated walls. A system – of which ample use was made Due to a arch shaped central opening on the top of
in other parts of the structure affected by crushing – the central triangle, laid upon the arch below, worry-
of armed injections with mortars injected at low ing cracks and disconnections have showed at the top
pressure was used in the four pillars of the crossing. and on the sides walls. In order to ease the burden of
265
the cover above these walls, which are about 20 metres Montini, R.U. 1956. La chiesa del Gesù. Napoli: Azienda
apart, and to help the relative stability avoiding trau- Autonoma di Soggiorno Cura e Turismo.
matic interventions, supporting structures have been Errichetti, M. 1963. La cupola del Gesù Nuovo. Napoli
projected. In particular, two Polonceau metal profiles nobilissima II (5): 177–184.
Pane, R. 1966. Antico e nuovo (Incredibilia sed vera:
in contact with the opposite sides of each of the tri- la ricostruzione della cupola del Gesù Nuovo). Napoli
angular walls, properly connected, will be burdened nobilissima 4: 161.
with the coverage weight and they will support the Guerra, G. 1967. La cupola del Gesu Nuovo: problemi statici
cemented points of the same walls as well. These metal e curiosità storiche. Atti dell’Accademia Pontaniana 16:
structures will unload the entire weight on two string- 384–398.
courses in reinforced concrete, shaped so as to assist Pirri, P. 1970. Giuseppe Valeriano S.I. architetto e pittore,
the masonry structures and to contain a channel for the 1542–1596. Roma: Institutum Historicum S.I.
disposal of rainwater. The mechanical properties of the Di Stefano, R. 1973. Vanvitelli ingegnere e restauratore. In
steel profiles made this solution acceptable, because AA.VV. Luigi Vanvitelli. Napoli: Edizioni Scientifiche
Italiane.
they are relatively thin with respect to the high mass Errichetti, M. 1974. La chiesa del Gesù Nuovo. Campania
of wall that they support, with the function they serve Sacra.
in full view (ASBANa Report n. 134 del 21/10/1996). Bösel, R. 1985. Jesuitenarchitektur in Italien (1540–
The interventions that this review briefly describes 1773). Wien: Verlag der Österreichischen Akademie der
constitute the testimony of how, over the last 15 Wissenschaften.
years, thanks to experiences that have shown their Amirante, G. 1990. Architettura napoletana tra Seicento e
ineffectiveness over time, the path to improving the Settecento. L’opera di Arcangelo Guglielmelli. Napoli:
preexistent structural condition has been undertaken, Edizioni Scientifiche Italiane.
trying to avoid the wishful thinking of radical substi- Iappelli, F.S.I. 1994. Committenza, danni, restauri. Societas
3: 68–74.
tutions. At a time when the architectural heritage is Carrafiello,T. 1995. Berardo Galiani intendente d’architettura
subject to a close evaluation in terms of seismic vul- (1724–1774). Archivio Storico per le Province Napole-
nerability we expect to submit these choices to a safety tane. CXIII: 245–380.
assessment, as required by the recent “Guidelines for Schiattarella, A. & Iappelli, F. S.I. 1997. Gesù Nuovo.
the evaluation and reduction of seismic risk to cultural Castellammare di Stabia: Edizioni Eidos.
heritage” approved in Rome on June 15, 2007. Divenuto, F. 1998. Napoli l’Europa e la Compagnia di Gesù
nella «Cronica» di Giovan Francesco Araldo. Napoli:
Edizioni Scientifiche Italiane.
4 CONCLUSIONS De Frede, C. 2000 (ried.). Il principe di Salerno Roberto San-
severino e il suo palazzo in Napoli a punte di diamante.
Through the examination of past and recent interven- Napoli.
tions on a complex architecture, the authors aim at Cantabene, G. 2004. Il terremoto del 1688 a Napoli. Nuove
focusing on present conservation issues of the Gesù acquisizioni documentarie. Ricerche sul ‘600 napole-
Nuovo church in Naples. Whereas in past centuries tano: 58.
Casiello, S. (ed.). 2005. Le cupole in Campania. Indagini
the solution to structural problems tended toward the
conoscitive e problemi di conservazione. Napoli: Arte
massive substitution or elimination of parts, questions Tipografica Editrice.
of consolidation must take into consideration today a La Regina, F. 2005. Il tema delle cupole nella conservazione
“constructive requirement”; this means the necessity del patrimonio architettonico. Questioni di metodo per
of preserving a diachronic architectural system must la conoscenza e l’intervento. In Stella Casiello (ed.), Le
be taken into consideration. As described in the paper, cupole in Campania. . ., : 107–114.
traditional and “modern” solutions contribute to the
mentioned requirement in a unique way. ABBREVIATIONS:
266
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The Zeytinburnu Military Factory, a large complex, was built with the purpose of modernizing
the army as a result of the reform movements of 1826. The necessary materials and machines for the factory
whose construction is known to have continued between 1845–1846, were brought from London by Barutçubaşi
Hoca Ohannes Dadyan Efendi. British engineer William Frin was responsible of the construction and assembly
of the factory. The factory was established on a wide area between the Sea of Marmara and the railroad. A map
dated 1918 shows the factory buildings and contains information about their functions. Eight buildings/building
groups of the factory are still extant. They show variation in the roof structures and their supports. Three of
the 4 cranes in one of the buildings were removed. The factory buildings that house original examples of the
period’s construction techniques must first be documented in detail and then measures must be taken for their
conservation.
267
pumps, equipment for small enterprises and parts 2.1 Building no. 1
for guns were manufactured in addition to military
The building shown as the “muhafaza bölüǧü
equipment (Anonim, 1995).
koǧuşları” (guards’ division barracks) on the map
In the second half of the 19th century the factory
dated 1918 is located next to the entrance gate of the
was transformed into a solely arms and ammuni-
factory area. The single storey rectangular building has
tion manufacturing plant and at this time the factory
round headed windows and doors. The building with
known as the “Zeytinburnu Silah Fabrikası” (Zeytin-
plastered and painted façades is covered with a hipped
burnu Military Factory) manufactured goods for the
roof. The wall thickness of the solid block masonry
“Tophane-i Amire” (Imperial Canon Factory). Goods
building varies between 65 cm and 87 cm.
manufactured at the factory were presented at the
Sultanahmet exhibition in 1861. A technical school
in connection with the factory was established at 2.2 Building no. 2
this time (Anonim, 1995). Grooved canons and 140
firearms of different size and make manufactured at The structure shown as the “büyük havuz” (large tank)
this factory were exhibited at the “Istanbul Interna- on the map dated 1918 is located at the west of the
tional Exhibition” of 1863 (Önsoy, 1988). Special iron factory area. It consists of a superstructure with a cir-
posts for street lighting, iron pipes, ploughs, gardening cular plan scheme over an almost square base. The
tools, drawgear were also manufactured at the factory base section contains a wide round headed opening,
(İlgürel, 1989). which is used to enter the lower level section of the
Production at the factory gradually decreased due to structure. The base of the solid block masonry struc-
its being far from a rich mine, its inefficient operation, ture is in alternating courses of ashlar and brick. The
competition with Europe and the political problems of top section with a circular plan scheme is in brick. On
the state and it was closed down during World War I the façade, there is moulded dressing between the base
and the armistice period (İlgürel, 1989). The factory and upper section. There are 154 cm long metal ties on
went back to operation in 1936 and was turned over to the façades of the base. The structure is now used as a
the “Ordonance” class of the army in 1948. Today it water tank (Fig. 5).
provides service as the “1st Army Maintenance Centre
Commandership” and the maintenance and repairs of
army equipment as well as the production of some 2.3 Building group no. 3
equipment is executed here. According to the map dated 1918, this group of build-
ings located in the north-south direction in the centre
of the factory area consists of 5 adjacent one-storey
units. From north to south, the first unit housed the
2 ARCHITECTURAL CHARACTERISTICS “bakır hâdde-hânesi” (copper rolling mill) and the
“şâh-merdân ocakları ve buhâr kazganları” (pile driver
The Zeytinburnu Military Factory is located over a furnaces/drop forges and steam boilers), the 2nd unit
large area of 128.121 m2 in the west of the city housed the “demir hâdde-hânesi” (iron rolling mill).
between the Sea of Marmara and the railroad line. The Part of the east facing section of this unit is demol-
Prime Ministry Ottoman Archives has various docu- ished. The 3rd unit contains the “buhâr kazganları”
ments related to the construction, repairs and use of (steam boilers), “pirinç kılıç-hânesi” (brass sword ate-
the factory buildings for the period between 1261– lier) and the “maden döküm-hânesi” (metal foundry).
1334/1849–1916. The German Union Map (“Alman The 4th unit has the “kılıç-hane” (sword atelier),
Sendika Haritası”) dated 5 October 1918 shows that “buhâr kazganları” (steam boilers) and “kundak fab-
the factory consisted of a number of detached and rikası tesviye-hânesi (gunstock works fitting shop) and
adjacent buildings (Fig. 2). The captions on this map the 5th unit houses various functions defined as the
contain information on the different units of the time “büyük çark-hane” (major machine shop) and “serî’
and their functions. However, no information could atışlı kundak fabrikası” (repeat fire gun stock shop).
be obtained about the original plans of the mentioned Today, the dimensions of units 1–3 are
buildings. 27.70 × 71.20 m and they are divided into two spaces
Eight buildings/building groups of the factory have with a wall extending in the north south direction.
survived to the present. Some of the buildings in the The main walls of the solid block masonry units are
factory area as well as the quay by the sea and the boat 114–118 cm thick. The roofing with wooden struc-
landing no longer exist (Fig. 3). The quay was filled ture rests on the main walls. The windows and doors
in during the 1950–1960s, and the factory’s connec- are round headed. The space in the east section con-
tion with the sea was cut off with a seaside road. The tained 4 cranes of which only one exists today (Fig. 1).
buildings that have survived have to a large extent pre- The 4th and 5th units in solid block masonry
served some of their original façade characteristics and are 49.60 × 52.75 m and have a wall thickness of
construction techniques. 115–126 cm. In this group of buildings only the space
268
“kılıçhane” (sword atelier) in the west direction is cov-
ered with a masonry barrel vault. Tie beams have later
been placed in the vault that has windows on its east
façade. The other units are covered with wooden roof-
ing supported by wooden posts with a cross section of
36 × 35 cm (Fig. 6). The windows and doors of these
units are round headed and the façades are plastered
and painted.
Figure 2. The German Union Map (“Alman Sendika Haritasi”) dated 5 October 1918 (Factory Archive).
269
Figure 4. The building group no. 4 (IRCICA Archive).
map dated 1918 these units are designated as “şarjör shop, dynamo shop, bullet and cartridge case work-
ve mukavvâ kutu dâiresi, kovan birinci ameliyyât shop, explosives shop, zinc can workshops, cartridge
dâiresi, kovan ikinci ameliyyât dâiresi, dinamo dâiresi, factory, bullet and jacket workshop, bullet presses,
kurşun ve kovan-hâne dâiresi, imlâ dâiresi, çinko cap/primer workshop).
sandık dâireleri, fişek fabrikası, kurşun ve gömlek The first section of this building group that can be
i’mâlât-hânesi, kurşun presleri, kapsül i’mâlât-hânesi” divided into two is located on the southeast side and
(charger and cardboar box shop, cartridge case initial is 16.50 × 133 m and the second section which is on
operations shop, cartridge case secondary operations the north-west side is 38.25 × 95.25 m. Some units
270
Figure 7. Wooden truss roof structure in the building group
no. 4 (2001). Figure 8. Detail of an “I” shaped steel beam and wooden
beam in the building group no. 4 (2001).
consist of large single volume spaces adjacent and
parallel to one another. The façades of the buildings
with depressed arched doors and windows are plas-
tered and painted. Photographs in the archives show
that the tile covered units in the southeast have roof
windows placed so as to catch the northern light.
The main wall thickness of the group of buildings
in solid block masonry varies between 50 and 96 cm.
The measurable bricks of the wall are 10 × 22 × 6 cm.
In various units of the structure there are load bear-
ers with cross sections of 20 × 35 cm, 24 × 24 cm and
26 × 27 cm and cast-iron columns with a diameter of
21 cm. Wooden and metal truss roof structures are
used in the units with gable roofs (Figs 7–12). The
cracks formed on the walls of the buildings during the
Istanbul earthquake of 1999 were repaired and covered Figure 9. Metal truss roof structure in the building group
in 2003. no. 4 (2001).
271
Figure 10. Wooden truss roof structure and wooden posts
in the building group no. 4 (2004).
272
The factory reflects the modern and traditional
construction technology of the 19th century with its
original wood and iron structures, masonry technique
and materials and its production activities. Therefore,
it is a group of buildings that should be conserved for
their value for the archaeology of industry. Changes
and additions were made in the functions of the build-
ings with time, therefore, the buildings now contain
different materials and construction details. Metal roof
structures as well as timber structures dating of the last
quarter of the 19th century are of major importance. As
mentioned earlier, an important number of the factory
buildings where the materials and structural systems
of the period were used to their maximum potential
are still standing and functioning.
The factory built to manufacture a large portion of
the arms and military equipment for the Ottoman army
is also of a documentary nature with the equipment in
its various buildings. Therefore, the buildings as well
as the equipment they contain must be documented
in detail. The expansion of the factory, the traditional
and modern manufacturing methods, assembly and
operation schemes, as well as the machines, work-
benches, tools and equipment imported from Europe,
the foreign experts, foremen and administrators who
have worked at the factory, the relations established
with the Krupp and Karlsruhe factories in Germany
are all matters that must be explored in the archive
documents.
The existing factory buildings have preservation
Figure 13. Metal window frames in the building group no. 4 problems arising from changes of function, earthquake
(2001). and environmental factors. The factory buildings that
have retained their original construction techniques to
The factory’s boat landing on the Marmara Sea shore a large extent and the original equipment in the build-
and the railroad that was built by the northwest of ings must be repaired before they lose their originality,
the factory in the end of the 19th century were used and must be reviewed within the context of sustainable
to transport the raw goods and products. The factory preservation.
buildings that have survived to the present are still used
as the maintenance workshops for military equipment.
Unfortunately, no conservation decision has been REFERENCES
given for the factory buildings that have survived and
whose connection with the sea was terminated with Anonim. 1995. Zeytinburnu Silah Fabrikası. Dünden Bugüne
İstanbul Ansiklopedisi, 8: 488. İstanbul: Kültür Bakanlıǧı
the building of the shore road. The only measured
ve Tarih Vakfı Ortak Yayını.
drawings for documentation purposes are those of Çiftçi, A. 2004. 19. Yüzyılda Osmanlı Devleti’nde Askeri
a building with an administrative function (Building Mimari ve İstanbul’da İnşa Edilen AskeriYapılar. İstanbul:
No. 7). Yıldız Teknik Üniversitesi Fen Bilimleri Enstitüsü
The observations and evaluations made on the exist- Doktora Tezi (yayınlanmamış).
ing factory buildings within the scope of this work İlgürel, M. 1989. Zeytinburnu’nda Bir Demir Fabrikası. Tarih
have remained limited as the said buildings are being Boyunca İstanbul Semineri 29 Mayıs-1 Haziran 1988
used for military purposes. Though information on the Bildiriler, İstanbul: Edebiyat Fakültesi Basımevi.
functional distribution of the factory buildings can be Önsoy, R. 1988. Tanzimat Dönemi Osmanlı Sanayii ve Sanay-
ileşme Politikası, Ankara: Türkiye İş Bankası Yayınları.
attained from the map dated 1918, an absolute eval-
Tuğlacı, P. 1993. Osmanlı Mimarlıǧında Balyan Ailesinin
uation has not be possible as the original plans and Rolü, İstanbul: Yeni Çıǧır Kitabevi.
documents could not be reached.
273
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
A.A. Abdessemed-Foufa
Department of Architecture, Faculty of Engineering Sciences University of Blida, Algeria
D. Benouar
Department of civil engineering, Faculty of Civil Engineering, University of Algiers, Algeria
ABSTRACT: This research work presents the consequences of the 1716 Algiers earthquake. The damages
records are collected from the ottoman files, consular mails, scientists and travellers of the 18th century. The
1716 Algiers earthquake is one of the most significant historical event having affected the city (I = IX MSK).
Indeed several sources describe it and give significant information about the scope of the damage recorded. These
entire documentaries sources and in particular those of Ottoman regency were carefully studied and analyzed
allowing us to have supplement information on this earthquake. This research work makes it possible to have
a clear vision on the damage as well as on the vulnerability of the predominant constructions in Algiers. The
information obtained, constitute an excellent damage database which will make enable us to work out the future
seismic scenarios to protect cultural heritage of the old nuclei of Algiers and eventually other historical sites
with the same characteristics.
275
vulnerability. As a broad categorization we can group
buildings in five grades.
Grade 1: Negligible to slight damage (no structural
damage, slight non-structural damage).
Grade 2: Moderate damage (slight structural damage,
moderate non-structural damage)
Grade 3: Substantial to heavy damage (moderate
structural damage, heavy non-structural damage)
Grade 4: Very heavy damage (heavy structural dam-
age, very heavy non-structural damage)
Grade 5: Destruction (very heavy structural damage)
(See annexes).
1.4 Quantities
Figure 1. Map of Algiers (After anonymous 1830).
The use of quantitative terms (“few”, “many”, “most”)
provides an important statistical element in the scale. 2 THE MEDINA OF ALGIERS AT 1716
It is necessary to confine this statistical element to
broad terms, since any attempt to present the scale as In 1716, Algiers showed the typical characteristics
a series of graphs showing exact percentages would be of a medieval Moslem Maghreb cities such as Fez
impossible to apply in practice and would destroy the (Morocco) and Tunis (Tunisia) (Le Tourneau 1949).
robustness of the scale. The definition of quantity has Indeed, the blocks are of various sizes and contain a
been presented, very deliberately, in graphical format several number of houses. The parcels are fully built;
to emphasise the way these numerical categories are the houses are overlapping and leaning against each
blurred rather than sharply defined. In such a case as other forming a compact unit. In addition, a great
a precisely determined quantity falls into an overlap- number of narrow and short streets were covered by
ping area, the user should consider the implications galleries on top of which the houses extended and thus,
of classing it as one category or the other, in terms of created roofed passageways called “Sabat”. In Algiers,
which would be more consistent with any other data just a main street crossed the city from the south to the
available for the same place. north in the lower part of the city. Algiers was sur-
rounded by walls built before 1517, was pierced by
five gates with twelve batteries (figure 1).
1.5 Assessing intensity from historical records
The term “historical data” is frequently used to
mean descriptions of earthquake effects from his- 3 TYPOLOGY OF CONSTRUCTIONS AT THE
torical records, that is, written sources prior to the 18TH CENTURY
instrumental period (before 1900). Historical accounts
often report in detail damage to special monumental According to various historical sources (Haedo 1578–
buildings (castles, churches, palaces, towers, pillars, 1580, Shaw 1808 & De Grammaye 1670) in Algiers,
mosques, and so on). Less frequently do they report all houses look alike, most of them have a central
the effects on ordinary buildings, which are the only patio or square courtyard surrounded by rooms. This
ones which can be used within the framework of the type of house is called “Dar”. A gallery supported
scale. With regard to ordinary buildings, the vulnera- by columns with horseshoe pointed arches surrounds
bility classes of traditional houses range in most cases this courtyard called “wast al dar”. A great number of
from A to B, even to C and D (wooden structures. With the palaces and houses have three levels. Houses with
regard to ordinary buildings, the vulnerability classes courtyard develop another type of dwelling, which is
of traditional houses range in most cases from A to called “Dar shebak”, house with metal grid, where
B, even to C and D (wooden structures). Very little dimension of the courtyard is reduced. This grid rep-
is known from the general literature about building resents only a tiny vertical opening for daylight. Fur-
types in Europe and Maghreb up to the 17th century, thermore, we can find houses called “Alwi”, because
except for the obvious facts that people used the mate- of constraint of the site giving narrows plots. So, they
rials nearest to hand, and that the richer the owner, the do not neither “patio” nor grid. As a consequence, this
better-built and better-maintained his house was likely houses open towards external sance. These are gener-
to be. But in the Middle Ages, certainly, most houses ally situated on the border of the streets (figure 2). The
in many parts of Europe were made of wood, and in number of houses was estimate at 5000 (De Paradis
the Maghreb they were made of stone and adobe brick. 18th century) and are about 800 today (Cellule Casbah
276
Niveau Stah
Niveau Shin
4
6
5
10
17
17
4
17
6
17
16
6
7
6
7
17
4
5
17
17
17
17
Sedda
Niveau S'hin
17
7
17
17
4
8
10
8
17
6
4
9
2
8
6
15
1
6
7
7
4
18
7
17
17
Sedda
17
Sedda
277
3 cm
3 cm
3
3
3
3
Brick
Lime Mortar
Figure 3b.
55 40
Section
0 10 20 50 cm
40
55
278
Anonymous manuscript no date, French Archives 6.2 Destruction of the walls
1716, Gazette de France 1716, Comelin 1720 &
According to the ottoman files (1716), many houses
Ibn Redjeb 1740). Not only the poorest houses but
were damaged and repairs were conducted.
also solid construction, such religious one as the
The grade of damage 3 was assessed.
mosques (Great Mosque of Algiers, Sha’ban Khudja
mosque) suffered extensive cracking of walls and par-
tial collapse of cupola, therefore they needed to be
repaired in 1735 (Devoulx 1870 & Barges 1877) and 6.3 Rupture of floors
palaces as Dar Aziza where the first floor collapsed,
and the great palace where walls suffered extensive According to the ottoman files (1716), related to the
cracking. cadastral acts, it was revealed that many first floors
On February 26, a violent commotion, damaged of the houses collapsed following the earthquake.
what resisted to the first shock. Comelin (1720) reported on the event: “. . .The house
For this earthquake, Roussel (1973) have assigned of the ambassador of France was one of the most beau-
an intensity I = X MM, while Ambraseys and Vogt tiful of Algiers. It did have three floors before the last
(1988) have assigned that this intensity could not be earthquake, now only two floors remain”.
reached because no solid construction was damaged The grade of damage 4 was assessed.
and they give I = 7 MSK. In this study, we assess inten-
sity I = 9 EMS, on the basis of the assumption that the
houses stock and the mosques are respectively equally
7 DAMAGE TO MONUMENTAL BUILDINGS
distributed between vulnerability classes A and B and
(MOSQUES AND PALACES)
vulnerability classes D.
An anonymous map (Algiers 1830), in absence of
There is no detailed report on damaged monumental
any map dating from the 18th century, can be used as
buildings. The information is consigned in the reg-
a reliable reference for the earthquake situation.
isters of the religious buildings (claim of real estate).
These registers defer all the operations of maintenance,
which took place regularly on the damaged buildings
6 DAMAGE TO MINOR BUILDINGS by the catastrophes (earthquake, floods, war. . .). It
(HOUSES) result from it that only two damaged mosques by the
1716 earthquake had been described:
Damage description for the earthquake of February 3, The great Mosque of Algiers (Djama’al kabir) pre-
1716 is available for 240 minor buildings (houses) sented several cracks on the north-west walls or the
and 4 monumental buildings (mosques and palaces). “qibla” walls (Devoulx 1946), and the reparations took
For all these constructions, they are detailed and con- place only in 1732 (Barges 1877).The grade of damage
cerned damage caused by this earthquake. All of 2/3 is assessing.
them have been classified in term of EMS-98 damage The cupola of Sha’ban Khudja mosque collapsed.
grades. The grade of damage 3/4 is assess.
According to the various historical sources (cited It was said that many palaces were destroyed or
above) and in particular, those of the ottoman files, severely damaged. The first one related is “Dar ‘Aziza”
various damages due to earthquake were emphasized palace, which was the French Embassy. The Consul
and they were of three types: Clairembaut (French Archive 1716) and later Comelin
(1720) described the damage. All the higher floor of
this large house felt down. The damage was thus signif-
icant in this residence. The degree of damage assessed
6.1 The total collapse of the houses
was 4. The second one is “Dar al Sultan” called “Al
According to the analysis of the ottoman files (doma- Djenina” palace, which was at that time the residence
nial acts 1716) [8], Comelin (1720), Delphin (1922), of the regency and the government palace. Accord-
Shaw (1808) and Carette (1850), 200 houses collapsed, ing to Klein (1937), this palace suffered many cracks
most of the dwellings were ruined and part of the city that have constrained the “Dey Ali Shaush” to leave
was thrown to the ground. According to Tassy (1830) precipitately the palace with its court to find a pro-
and Burzet (1866–1869), the country houses or houses visional refuge in “Burdj Mulay Muhammad” called
of the fahs around Algiers collapsed completely at a later “fort de l’étoile”. The grade of damage 2/3 is
distance of about 3 km around the city. Carette and assessing.
Burzet reported that the aftershock of 26 February Synthetic information: collapsed, destroyed, several
added damaged to most of the houses, which did not cracks, referring to cumulative effects of the 1716
collapse totally after the main shock. earthquake, are available for the remaining 240 houses,
The grade of damage 5 was assessed. 2 religious buildings and 2 palaces.
279
Table 1. Assessment and damage grade of minor building After the 1716
Algiers earthquake.
ID. ID.
Type of Slightly Destruction Several Total Total
damages damages of floors cracks collapse damages
Grades 2 4 3/4 5
Houses 68 24 15 8 130
Houses 86∗ 13∗ 9∗ 7∗ 115
Total 162 37 24 14 245
% 67.5% 15.40% 8.33% 5.40%
∗
indicate the houses, which has been not located in the map
Table 2. Assessment and damage grade of important building after the 1716 Algiers earthquake.
Great x – – –
Mosque of
Algiers
Shaban – x – –
Khudja
Mosque
Dar Aziza – – x –
Palace
Dar al Sultan x – – –
Palace
Zone 1
- 10
- -
?
?
?
Zone 2
39
-
?
- ?
-
Zone 6
37
?
Zone 5
?
8
?
?
-
Zone 3 Zone 4
19 2
Area 1
10 houses damaged
Area 2
39
Area 3
19
Figure 5.
280
D am age Grade EM S-98
5
4/5
4
3/2
D am ages not located or without inform ation
inform ation
Damage grade 5 is assigning when the description This data, though not useful for detecting possible
clearly reported “total collapse”. amplification or desamplification areas, can provide
Grade 4 is assigning when the sources simply useful informations on the vulnerability of the serv-
reported “collapse”. ing buildings and serve as an input for evaluating the
Grade 2/3 is assigning when the sources reported ground shaking.
“slightly damage”.
The summary of damage is reported in Tables 1
and 2. Annexes:
The identification of the buildings is performing on
the historical map of Algiers (figure 5).
8 CONCLUSION
281
Consular mail 1716. French Consuls Correspondence at
Algiers 1642–1792, Foreign Affaires B1 120, 1716–1720
F◦ 3-4-5 février 1716, F◦ 6-7- 21 février 1716, F◦ 8-9-10
mars 1716, Unpublished sources.
De Gramaye, J.B. (1670). Alger XV–XVII siècles. Journal de
Quantities after the EMS 98. Jean Batiste Gramaye Evèque d’Afrique. Translated by
Ben Mansour A.H. 1998. Paris. Ed Cerf.
De Haedo, D. (1578–1581). Topographie et histoire d’Alger.
Translated by Monnereau & Berbrugger. Transcripted by
Rebahi. A. Alger. Ed G.A.L.
Delphin, G. 1922. Histoire des Pachas d’Alger de 1515 à
1745, Extrait d’une chronique indigène, Extrait du Journal
Asiatique (Avril-Juin 1922 et Janvier-Mars 1925), 216–
217 ; Paris. Imprimerie Nationale.
Devoulx, A. Alger. 1875. Etude archéologique et
topographique sur cette ville aux époques romaines (Ico-
sium), arabe (Djazâir Beni Mezghanna) et turque (El
Djazâir). Revue Africaine; 19 ; 295–332 and 385–428.
Devoulx, A . 1846. Histoire d’Alger. Manuscrit n ◦ 3213. f ◦
120. National Algerian Library.
Devoulx, A . 1870. Les édifices religieux de l’ancien Alger.
Alger. Ed Typographie Bastide.
Gazette de France 1716. Press.
Grüenthal,G. 1998. L’Echelle Macrosismique Européenne.
Cahier du Centre Européen de Géodynamique et de
séismologie. Ed Française A. Levret.
Guidoboni, E et Boschi E. 2001. Catania terremoti e lave del
mondo antico alla fine del novecento. Instituto Nazionale
di Geofisica e Vulcanologia, SGA Storia Geofisica Ambi-
ente. Bologne. Ed Compositori.
Ibn Redjeb, H. 1740. Chronicle in arabic, translated by
Delphin, G, 1922.
Klein, H. 1937. Feuillets d’El Djazair. Alger. Chaix.
Moroni A, Azzaro R, Stucci M, Barbano M.S Albini P.
1999. Damage in Catania on the occasion of the 1693
earthquakes.The Catania Project, Earthquake damage sce-
narios for a high risk area in the Mediterranean; GNDT.
Damages classification after the EMS 98. Pessina. Ed E Facioli and V. 119–125.
Ottoman Archive in the National Archives of Algiers. Unpub-
REFERENCES lished sources.
Ottoman Archives in the CAOM of France. Unpublished
Ambraseys, N et Vogt, J. 1988. Materials for the investiga- sources
tion of the seismicity of the region of Algiers. European Roussel, J. 1973. Les zones actives et la fréquence des séismes
Earthquake Engineering; 3; 16–29. en Algérie 1716–1970. Bull. Soc, Hist. Natur ; Afrique du
Anonymous manuscript (without date and title), (in Ara- Nord ; 64 ; 3 ; 2.
bic), manuscript n◦ 1637, National Library of Algiers; Shaw, T. 1808. Travels or observations relating to several
Unpublished sources. parts of Barbary and the Levant. Edinburgh. Printed by
Bargès, J.J.L (Abbé). 1877. Complément de l’histoire des J.Ritchie.
beni Zeiyan, rois de Tlemcen. Paris. Ed Ernest Leroux, Tassy, L. De. 1830. Histoire d’Alger et du bombardement de
Burzet, (Abbé) Histoire des désastres de l’Algérie, 1866, cette ville en 1816. Paris.Librairie Piltan.
1867, 1868 ; 1869.Alger, Imprimerie CentraleAlgérienne. Tourneau (Le) R. 1949. Fès avant le protectorat. Etude
Carette, A.E.H. 1850. Algérie. L’univers ou histoire et économique et sociale d’une ville de l’occident . Publié
description de tous les peuples de leurs religions, mœurs, par l’Institut des Hautes Etudes Marocaines.
coutumes. Paris, Ed Firmin Didot Frères. Venture De Paradis, J-M. 1983. Tunis et Alger au XVIIIXème
Comelin, P. 1720. Voyages pour la rédemption des captifs aux siècle. Mémoires présentées par Joseph Croq. Dijon. Ed
royaumes d’Alger et de Tunis. Sindbad.
282
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
B.T. Oliveira
Instituto do Patrimônio Histórico e Artístico Nacional – IPHAN, Ouro Preto, MG, Brazil
ABSTRACT: This paper deals with the creation of the Morro da Queimada Archaeological Park in Ouro Preto,
MG – Brazil. The Morro da Queimada constitutes an invaluable archaeological site, being an excellent sample of
the first architectonical typologies in town, for it keeps registers of the gold exploitation period at the beginning
of the 18th century and the vestiges of one of the most dramatic moments in the Brazilian colonial history. This
project is coordinated by IPHAN and is supported by national and international institutions, at the municipal,
state and federal levels, as well as religious bodies, NGOs and community associations. It was formally proposed
by the Museu de Arte Sacra do Carmo, based at the Paróquia de Nossa Senhora do Pilar and was approved
by the Conselho Nacional de Incentivo à Cultura – CNIC (National Committee for the Promotion of Culture),
under the auspices of MinC – the Federal Ministry for Culture of Brazil.
1 INTRODUCTION
283
Figure 2. Aerial view of Morro da Queimada in the
mid-decades of the 20th century (IPHAN archives, Ouro Figure 3. Irregular occupation of Morro da Queimada
Preto, MG, Brazil). (IPHAN archives, Ouro Preto, MG, Brazil).
According to Diogo deVasconcelos, Felipe dos San- growth of the City of Ouro Preto. In the absence of
tos was arrested on either the 19th or 20th of July, urban planning, the site was occupied, the ruins being
received a summary judgement and was executed. The used by the population for building material, and the
other heads of the revolt were transferred to Rio de surviving structures employed as foundations for new
Janeiro and there condemned to exile in Lisbon, Por- constructions (Fig. 3). The lack of protection for the
tugal. At the time, the Governor, the Count of Assumar, archaeological remains on Morro da Queimada in
stated that even the very stones of the Arraial of Ouro Ouro Preto constitutes one of the more serious cases
Podre plotted against the Portuguese Crown. of neglect of cultural heritage by the various levels of
As a result of the Sedição de Vila Rica in 1720, the government.
establishment of the proposed Royal minting houses In April, 2003 a technical mission from UNESCO
was postponed for four years, and Minas Gerais on a visit to the city identified the preservation of the
became an administrative region independent of the archaeological site at Morro da Queimada as one of
Capitania of São Paulo. the measures necessary to halt the alarming deteriora-
Morro da Queimada today constitutes a priceless tion of the cultural and environmental heritage of Ouro
archaeological site, a concrete record of the first urban Preto.
architectonical typologies of Minas Gerais, evidence
of the gold rush at the beginning of the 18th cen-
tury, and a remnant of one of the most dramatic 3 THE RECOVERY PROJECT
moments in the history of Colonial Brazil. Besides
the ruins of old buildings, in the area may be found the Following this warning, the Instituto do Patrimônio
remains of the many gold mines that dot the hillside: Histórico Artístico Nacional-IPHAN (National Insti-
hollowed shelters carved out of rock, long galleries, tute for Historical and Artistic Heritage) established
ventilation shafts, and entrance tunnels. Still found on work groups (historical research, property issues,
this archaeological site are mundéus, reservoirs con- Housing, physical planning, and community rela-
structed for the washing of ore, small dams, water tions), necessary for the preparation of a pilot project
diversion channels, and hydraulic systems used for for an archeological park in the area.
transporting slurries of water and mud mixed with gold This project sought to create a substantial and posi-
(Fig. 2). tive impact on the City of Ouro Preto by means of the
following:
– expansion of historical research and the creation of
2 THE DETERIORATION OF MORRO programs for archaeological excavations, facilitat-
QUEIMADA ing a wider knowledge of mining history and the
material culture of the period;
In recent decades, Morro da Queimada as an archae- – protection and classification of the ruins dating
ological site has become corrupted due to the chaotic from the earliest settlements in Ouro Preto (Fig. 4);
284
Figure 4. View of part of the ruins at Morro da Queimada
(IPHAN archives, Ouro Preto, MG, Brazil). Figure 6. Creation of the Ecomuseum (IPHAN archives,
Ouro Preto, MG, Brazil).
Figure 5. Aerial view of the park: area of 124.88 hectares Figure 7. View of the ruins on Morro do Queimada, in
(IPHAN archives, Ouro Preto, MG, Brazil). the, in the mid 20th century (Photo from Luiz Fontana IFAC
archives, Ouro Preto, MG, Brazil).
285
Figure 9. View of a mine entrance, 1990s (Photo from
Figure 8. Demolition of irregular constructions (IPHAN Germano Neto, Ouro Preto, MG, Brazil).
archives, Ouro Preto, MG, Brazil).
286
Figure 10. View overlooking Ouro Preto from Morro da
Figure 11. View of the Peak of Itacolomi from Morro da
Queimada (IPHAN archives, Ouro Preto, MG, Brazil).
Queimada (IPHAN archives, Ouro Preto, MG, Brazil).
University of Ouro Preto – UFOP, State Institute for The creation of the archaeological park of Morro da
Historical and Artistic Heritage – IEPHA, the State Queimada is one of the most important recent under-
Forestry Foundation – IEF, the State Public Fund- takings in Ouro Preto, and the realization of this project
ing Agency, Municipality of Ouro Preto – PMOP, will be as important to the city as the Acropolis is for
Municipal Chamber of Ouro Preto – CMOP, the Gor- Athens or the Palatine Hill and the Ancient Forum are
ceix Foundation, parish of Nossa Senhora do Pilar, for Rome.
parish of Santa Efigênia and the Sociedade Soto Zen The creation of the archaeological park of Morro
do Brasil. In addition, there is support from various da Queimada will establish a setting favorable to the
NGO’s, such as the Associação de Proteção Ambien- recovery of the cultural and environmental heritage of
tal Ouro Preto – APAOP (Ouro Preto Environmental the city, a new start in the place where the city began:
Protection Association), the Amigos do Patimônio Cul- the Arraial of Ouro Podre where once Paschoal was
tural e Natural de Ouro Preto – AMO Ouro Preto master (Meirelles, C, p. 55).
(Friends of the Cultural and Natural Heritage of
Ouro Preto) and the Federação das Associações de
Moradores de Ouro Preto- FAMOP (Federation of REFERENCES
Residents Associations of Ouro Preto).
This project is coordinated by IPHAN, developed Barker, P. 1981. Tecniche dello scavo archeológico. Milano,
longanesi & C.
in partnership with The Federal University of Ouro Bediaga, B. (org.). 1999. Diário do Imperador D. Pedro II,
Preto – UFOP, Municipality of Ouro Preto – PMOP, viagem a Minas Gerais Vol. 24, – primeira parte, 26/03 a
Municipal Chamber of Ouro Preto – CMOP, man- 19/04 de 1881. Petrópolis, Museu Imperial.
aged by the Museu de Arte Sacra do Carmo, based De Varine, H. O Tempo Social. 1987. Rio de Janeiro, Eça
at the Paróquia de Nossa Senhora do Pilar, and spon- Editora.
sored by the Programa Petrobras Cultural, the Caixa Meirelles, C. 1989. Romanceiro da Inconfidência. Rio de
Econômica Federal (Federal Development Bank) and Janeiro, Editora Nova Fronteira.
Novelis Brasil Ltda. Mello, S. de. 1985. Barroco Mineiro. São Paulo: Editora
Morro da Queimada is a site endowed with rare nat- Brasiliense. Belo Horizonte: Editora Rona, 1979.
Vasconcelos, D. de. 1974. História Antiga de Minas Gerais.
ural beauty, from where it is possible to obtain a unique Vol. II. Belo Horizonte: Editora Itatiaia.
view overlooking the historical centre of Ouro Preto Vasconcellos, S. de. 1977. Vila Rica. São Paulo: Editora
(Fig. 10), and the Peak of Itacolomi, which guided the Perspectiva.
first European settlers, the bandeirantes, to the area, Vasconcellos, S. de. 1979. Arquitetura no Brasil: sistemas
and, still to this day, a symbol of the city (Fig. 11). construtivos. Belo Horizonte: Editora Rona.
287
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Structure as a document
Jinisha Jain
Conservation Architect, Delhi, India
ABSTRACT: Known as the ‘Old Fort’, Purana Qila, the first Mughal capital city, is situated on the eastern
edge of Delhi, along the river Jumna. Humayun, the second Mughal emperor (1530–38 A.D.) began constructing
a walled city and fortress on this site in 1533 A.D., and named it Din-Panah, or ‘Refuge of Religion’. He was
temporarily deposed by Sher Shah Sur (1538–45 A.D.), who extended the Old Fort but the fort was finally
completed by Humayun (1555–56 A.D.), the founder king. The comeback of Humayun and physical additions
to the site which can visibly be discerned but not assigned to any one ruler with confirmation compound
the problems of authorship and the historic sources in this respect fail to give any conclusive or corroborative
evidence. The fort walls are over one mile long, and contain three gates, a triple storey octagonal structure
called Sher Mandal and a mosque called Qila-I-Kunha Masjid meaning the ‘Mosque of the Old Fort’. The paper
focuses on Qila-I-Kunha and uses documentation of this structure as a tool to unveil the mystery.
289
color scheme and according to another interpretation
none of the structures in Purana Qila belong to Sher
Shah, except for minor additions and repairs. There
is no epigraphical evidence to support either claim.
In any case, Sher Mandal and Qila-i-Kuhna Masjid,
are the two surviving structures of certain complex-
ity, within the fort ramparts from the mid sixteenth
century, which command interest and consideration
as they mark an important and decisive stage in the
development of the subsequent styles and features of
Islamic architecture in India.
290
completion of work, but the fact that it was itself domes, a mosque typology that had developed in the
included within the ramparts, suggest that not major precedent Sayyid and Lodhi periods. It occupies an
recycling of materials from this site was possible). area of approximately 51 m by 13.5 m, with a total
• According to the Tarikh-I-Khan Jahan, another height close to 20 meters, inclusive of the dome.
important work on Afghan history, Sher Shah’s son The mosque is built in core masonry of stone rubble
Salim Shah Sur (1545–54 A.D.) is stated to have bonded with lime mortar and with fascia of finished
built a wall encircling the city of Humayun, which blocks of Delhi quartzite and red sandstone. The inter-
falls perfectly in place as both Tarikh-I-Daudi and nal space, a large rectilinear hall is divided laterally
Tarikh-I-Shershahi say that the outer fortification through massive load-bearing stone masonry arches
had begun by Sher Shah Sur but could not be into a distinct number of five. The external eastern
completed by him (thus completed by his son).s façade correspondingly consists of an arrangement of
• The features of Purana Qila, which have been iden- five archways with the central one being sunk in an
tified as, ‘Mughal’, could also date from Akbar’s arched recess contained within a pronounced rectangu-
period because we learn from Akbaranama that in lar frontage, the Liwan. The liwan is ornamented with
spring 1560 A.D., Shihabuddin Ahmad Khan, the intricate stone relief and intarsia in multi-coloured
governor of Delhi, in anticipation of an attack by schemes-red and yellow sandstone, marble and black
Bairam Khan, proceeded to strengthen the fort of slate. At the rear of liwan rises a characterristic Lodhi
Delhi and to repair its towers and walls. (maramat- dome with small minarets called guldastas, literally
i-burj-o-bara). the flower bouquets, flanking the drum and with a
• Zafar Hasan, a 19th C archeologist admitted that finial, kalash on the top. The scheme of five also
the difficulty in judging how much of the citadel translates internally as an arcade of mihrabs, one in
(Purana Qila) is the work of Humayun and how each bay, along the qibla or the west facing wall. The
much of Sher Shah. However, he states that accord- mihrabs are designed on the same general principle
ing to general opinion the walls and gates are as the recessed arches in the eastern façade; however
considered to be the work of Humayun and mosque they exhibit very fine workmanship. The system of
(Qila-I-Kunha) and Sher Mandal, of Sher Shah. roof support consists of domes, cross-rib vaults and
• Recently, J. Burton-Page and R. Nath have sug- semi-domes with sqinches, stalactites and corbelled
gested that none of the Delhi monuments tradition- pendentives in the phase of transition. Certain innova-
ally ascribed to Sher Shah were actually built during tive roofing devices have been employed and there are
his reign, much less patronized by him. three roofing techniques which have been used. The
• Architectural Historian Catherine Asher, assigns the central bay has a single masonry dome with multiple
Qila-I-Kunha masjid as indeed the mosque of Sher openings in the drum for ventilation and is decorated
Shah- a visual symbol of royal aspirations of the with colored tile work. Squinch is employed in the
ruler. phase of transition. The penultimate bays on either
• M.C Joshi, a modern historian, acknowledges side have what appear to be flat domes or vaults. The
minor contributions of Sher Shah and through end bays, which are oblong unlike the central and the
re-examination of literary and existing epigraphi- penultimate bays which are somewhat squarish have an
cal proofs credit most of the surviving structures to unusual roofing mechanism. The last bay shows two
Humayun. half arches which spring tangentially from the shoul-
ders of the lateral arches which define this bay. This
results in creation of three spaces to be spanned with
4 ON STRUCTURES WITHIN: QILA-I-KUHNA a cross-rib arrangement. Thus the central space of the
MASJID end bay has a shallow dome, while the spaces towards
the western and eastern arcade are covered through
Purana Qila is a subject that has been dealt with by semi-domes, nim-gumbads resting on the suspended
many architectural historians, researchers and other arches which in turn rest on the main lateral arches.The
scholars time and again. It is humble to acknowledge phases of transition are covered through beautifully
that the results or conclusions can never be path- carved corbelled pendentives. Attached to the ends of
breaking in the sense that there are eventually three the rear wall are two octagonal turrets, which are func-
possibilities one encounters: the structures were built tional as well as structural. Oriel windows, decorative
by Humayun, the structures were built by Sher Shah brackets, sunshades or chajjas and narrow turrets with
and the third-the structures were built in parts by both. taper, reminiscent of Tughlaq architecture- an early
Nevertheless, it is the approach, a fresh perspective phase, on each side of the central rectangular bay both
and the rigor of the quest that has kept the intrigue on in the front and rear are other architectural features
the subject going. which are used in the edifice. Hexagrams or six pointed
The Qila-I-Kuhna mosque conforms to the archi- star motifs along with colored stone inlay of vari-
tectural tradition of a five-bay mosque with triple ous geometric patterns form the chief ornamentation.
291
The mosque had private entrances on the north and • M.C Joshi opines that it is not unlikely, that this
south side for the use of the royal family, which are mosque was designed and founded by Humayun
now kept closed. The mosque is generally accepted to but was completed by Sher Shah, who seems to
represent the culmination of its type and as also the have been responsible for building the upper part
prototype for the more refined subsequent typologies of the super structure including the central dome,
of Mughal mosques. crenellated parapet of the roof- also encountered in
The mosque traditionally is considered to be Sher his Narnaul and Sasaram monuments- and intro-
Shah Sur’s royal chapel. In the absence of histori- ducing structural ornamentation including blue tile
cal inscriptions on the Qila-I-Kuhna, the traditional work, colourful gilded paintings, lotusbud fringe in
view of the mosque’s patron is based on the near arches, and ornate mihrabs. Perhaps the Sur ruler
contemporary histories of the Sultan Sher Shah Sur. also built its roof flanking the central dome and
Following presents abstracts from these histories and the two sidedomes, the bases of which are still par-
their interpretation by some of the recent historians. tially extant. The unfinished marble facing of the
• Both the Tarikh-I-Shershahi by Abbas Khan and front, with rich inlay designs can in no case be older
Abdulla’s Tarikh-I- Da’udi state that among the than the time of Akbar (1556–1605 A.D.), because
very first of Sher Shah’s accomplishments upon the ornate geometric patterns as well as their tech-
assuming the Delhi throne in 1540 A.D. was the niques of execution are purely Mughal in character
construction of his own capital, which included two and absent in Sur or Sultanate monuments.
• Architecturally, this edifice expresses the logical
fortified enclosures, the city walls and the royal res-
idence, the area known today as the Purana Qila. development of the five bay structural model repre-
(The destruction of Humayun’s capital in Delhi is sented by Masjid Moth (Lodhi) and Jamali-Kamal
not implied directly in any of the texts). The erection (Baburi) mosques; yet the employment of pointed
of a magnificent Jami mosque within the Purana arches, nim-gumbads (half domes) in it supporting
Qila is cited by both Tarikhs. Moreover, Abdullah the roof of the side- bays and pietra dura certainly
in the Tarikh-I-Da’udi states that this mosque was suggests a strong ‘Mughal’ association.
completed quickly, implying its full construction
during Sher Shah’s reign.
• Tarikh-I-Shershahi says, ‘The former capital city of
5 CRITICAL REVIEW AND ANALYSIS OF THE
Delhi was at a distance from the Jumna, and Sher ARCHITECTURE
Shah destroyed and rebuilt it by the bank of the
Jumna, and ordered two forts to be built in that city, We have already come across the inadequacy of the
with the might of a mountain, and loftier in height; archeological, historical, epigraphical and toponymic
the smaller fort for the governor’s residence; the (study of names and places; Sher Mandal as an exam-
other, the city wall to protect it; (literally, Jahan- ple would bring to mind association with Sher Shah
panah) and in the governor’s fort he built a Jama but most of the scholars agree with Humayun being
masjid of stone, in the ornamenting of which much the actual founder) proofs in the case of Purana Qila
gold, lapis lazuli, and other precious articles were as demonstrated above.
used. But the fortifications round the city were not This paper focuses on the documentation, architc-
completed when Sher Shah died.’ tural and structural analysis and subsequent interpre-
• It seems unlikely that Abbas Khan, who was com-
tation of one structure in Purana Qila, Qila-I-Kunha
missioned by Akbar to write the Tarikh-I-Shershahi, Masjid to trounce the other research inadequacies.
would attribute the mosque within the Purana Qila Herein are presented the author’s views on the
to Sher Shah if the actual patron was either Akbar authorship of this structure:
himself or his father, Humayun. • The planning, masonry and the same proportions in
• According to Catherine Asher, the literary evidence all the pylons inside the mosque clearly illustrate
referring to Sher Shah’s patronage of the Qila-I- it was conceptualized as a five bay mosque, thus
Kuhna mosque is hard to dispute. The mosque, then, thwarting any possibilities of the extreme end bays
must have been built soon after 1540 A.D. when being added later, although the change in propor-
Sher Shah ascended the Delhi throne. A compar- tions of the elevational arches of these bays is odd
ison of central archways with bracketed entrances and not seen in any of the previous prototypes or
of Ibrahim Sur’s tomb (Narnaul, Haryana), which other contemporary mosques of the period, though
were built by Sher Shah according to an inscriptions the central liwan is almost always elevated.
thereon, and those of Purana Qila further supports • The internal staircases housed within the end bays
the above point of view. In fact, like the Sur monu- were planned to reach the terrace of the turret and
ments of Sasaram, the mausoleum of Ibrahim Sur is then further into the drum gallery to finally the ter-
an improved and a refined version of Bara Gumbad race. This is more pronounced in the light of the
or other square Lodhi tombs. fact that the mosques of Moth-ki-Masjid and Jamali
292
Figure 3. East elevation.
293
people (and not just scholars) instead of adopting a
linear method of dispersing information. Qila-I-Kunha
Masjid shows innovations, which became prototypes
for emulation for the later buildings. It set precedents.
In the absence of sound epigraphical, historical or any
other proof, the building itself has served to be a stand-
ing document and has revealed significant information
on the basis of which we can infer that Purana Qila is
one of the few sites where the two different architec-
tural traditions of two rival camps-the Mughal and the
Sur have managed to co-exist.
REFERENCES
Abdullah. Tarikh-I-Daudi. S.A. Rashid (ed). Aligarh.
Figure 4. Structural analysis of Purana Qila. Abu-l-Fazl. Akbarnama Vol. I. 1897–1939. (Translated from
the Persian by H. Beveridge) Calcutta: Asiatic Society of
Bengal.
where Sher Shah Sur had spent much his time in Ahmad Khan Matta, Basheer. 2005. Sher Shah Suri- A Fresh
youth. (Although neither of them show the clas- Perspective. Oxford University Press.
sicism achieved in the mihrabs of Qila-I-Kunha B.Asher, Catherine. 1977. The Mausoleum of Sher Shah Suri.
with exquisite inlay work, which probably could be In Artibus Asiae Vol. XXXIX: 273–98.
ascribed to Mughals- Humayun). B. Asher, Catherine. 1982. The Qil’a-i-Kuhna Mosque: A
• Humayun after recapturing Delhi Sultanate must Visual Symbol of RoyalAspirations. In Chhavi II: 212–17.
have contributed in introducing pietra dura, such Varanasi: Banaras Hindu University.
B. Asher, Catherine. 1988. Legacy and Legitimacy: Sher
as in the liwan, in the mihrabs, in the brackets
Shah’s Patronage of Imperial Mausolea. In Katherine P.
supporting the pendentives, in the frieze with crenel- Ewing (ed), Shari’at Ambiguity in South Asian Islam:
lation motif at the rear, etc- much of which was left 79–97. Berkeley: University of California Press.
unfinished because of his untimely death. The con- Brown, Percy. 1975. Indian Architecture (Islamic Period): 26.
tributions made by Akbar if any are not clear. Minus Bombay: D.B. Taraporevala Sons and Co. Pvt. Ltd.
the sophisticated inlay work done in parts the whole Fergusson, James. 1910. History of Indian and Eastern
building indeed begins to look more like a struc- Architecture. London.
ture built in Suri time, with more dominant play of Hasan, Maulvi Zafar. 1997. Monuments of Delhi, Vol. I-IV.
quartzite and red sandstone carved cladding just like New Delhi: Aryan Books International.
Khan Sarwani, Abbas. 1952. Tarikh -I- Sher Shahi or Tufhat-
the tombs in Sasaram and Narnaul.
I-Akbar Shahi d. 16th C. (Translation in Posthumous
It is interesting to note that many Timurid ideas papers of H.M Elliot, John Dowson (ed). First Ed., 1871.
and a new aesthetic sensibility had been introduced by Second Ed. 1952).
Babur, whom Sher Shah Sur had personally served. Khwand, Amir. 1940. Qanun-I-Humayuni d. 1535. (Transla-
tion by Baini Prashad). Calcutta: The Asiatic Society.
Thus the influence of Mughal ideas or assimilation of
Koch, Ebba. 1991. Mughal Architecture: An Outline of its
their aesthetic sensibility by a later dynasty shall not History and Development (1526–1858). Munich: Verlag.
be a surprise. Ni‘mat-Allāh ibn Habı̄b-Allāh Haravı̄. Tārı̄h i Khān-Jahānı̄,
Thus it can be disputed that the mosque in its d. 1024/1615. (Translation). Collection of Oriental Works.
entirety was conceptualized and built by the Sur ruler, Qanungo, K.R . 1965. Sher Shah and His Times. Orient
and only non-structural modifications such as replac- Longmans Limited.
ing the cladding, carving and introducing fine persian R. Nath. 1978. History of Sultanate Architecture. New Delhi:
inlay work were made by the Mughal rulers. Abhinav Publications.
R. Nath. 1995. Studies in Medieval Indian Architecture. New
Delhi.
Sharma,Y.D. 2001. Delhi and its Neighbourhood. New Delhi:
6 CONCLUSION Archeological Survey of India.
Srivastava, Ashirvadi. 1950. Sher Shah and his Successors.
There is a Suri tradition of architecture different from Agra.
both its predecessor and successor dynasties. The idea Stephen, Carr. The Archaeology and Monumental Remains of
has been to integrate such research and documentation Delhi.
with conservation works to be able to determine the Tūzuk-i Jahāngı̄rı̄ or Memoirs of Jahāngı̄r. 1909–1014.
right kind of significance for the structure and also to (Translated by Alexander Rogers and Henry Beveridge)
aid in preservation of historic proof while opening up London: Royal Asiatic Society.
Yadgar, Ahmad. Tarikh-i- Salatin-i- Afghani (Translation).
possibilities for multiple interpretations for different
294
Monitoring, non destructive evaluation
and testing
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
R. Capozucca
Università Politecnica delle Marche, Ancona, Italy
ABSTRACT: An investigation on historic masonry wall models – in scale 1/3th – characterised by two leaves of
brickwork masonry and intermediate mortar has been carried out in the laboratory. Experimental results obtained
by tests on wall models under compression and cyclic shear loads are discussed and compared to theoretical results
obtained by a non linear procedure of calculus based on a tie & truss modelling for multiple-leaf masonry wall.
297
Figure 1. Masonry wall model.
2.2 Set-up and shear cyclic test 3.1 Theoretical modelling of the HURM model
Two specimens S1, S2 were tested under combined Masonry is a material which exhibits distinct direc-
vertical load and horizontal shear. A system of jacks tional properties due to mortar joints. In general, the
298
Transducer 1
40
F (kN)
30
20
10
0
-1 -0,5 -10 0 0,5 1 1,5
-20
-30
-40
(a) x (mm)
Transducer 1
40
F (kN)
30
20
10
0
-1,5 -1 -0,5 -10 0 0,5 1 1,5
-20
-30
-40
(b) x (mm)
299
Figure 7. Theoretical modelling of HURM model: (a) tie
and truss modelling of web surface; (b) geometric section of
HURM model web; (c) tie and truss modelling of web.
Figure 9. Typical experimental horizontal force vs
deflection.
300
evaluated at the end of linear elastic phase; Fi = lateral
load at ith step – have been calculated by experimental
data of test on S2 specimen.
A linear relation for the damage coefficient d has
been evaluated with m = 1.37:
301
Bernardini A. et a1. 1980. A comparison of laboratory Pina-Henriques J. et al. 2004. Testing and modelling of
test methods used to determine the shear resistance of multiple-leaf masonry walls under shear and compression,
masonry. Proc. 7th World Conf. on Earthquake Eng., Proc. SAHC, Padua, vol. II 299–312.
Istanbul, vol. 7: 181–184. Sinha B.P. & Hendry A.W. 1969. Racking tests storey height
Calvi B.M. et al. 1985. Random cyclic behaviour of rein- shear wall structures with openings subjected to precom-
forced masonry under shear action. Proc. 7th Int. Brick pression, Proc. Int. Conf. on Masonry Struct. Systems
Masonry Conf., Melbourne. Texas 1969, USA, 192–199.
Capozucca R. & Sinha B.P. 2004. Strength and Behaviour Hendry A.W. & Sinha B.P. 1971. Shear tests on full scale
of Historic Masonry under Lateral Loading. 13th IBMaC, single storey brickwork structures subjected to pre-
Amsterdam, vol. 1: 277–284. compression, Civil Engineering and Public Works Review,
Capozucca R. & Sinha B.P. 2005. Evaluation of shear strength 1339–1334.
of historic masonry. 5th Int. Conf.AMCM, Gliwice, Poland Yokel F.Y. & Fattal S.G. 1976. Failure Hypothesis for
(in CD-ROM). Masonry Shear walls, Jour. of the Struct. Division ASCE,
Capozucca R. & Sinha B.P. 2007. Shear Strength of Historic 102, n. ST3: 515–532.
Masonry. Int. Symposium Studies on Historical Heritage, Tomazevic M. 1977. Seismic design of masonry structures.
Antalya, Turkey, 435–440. Construction Research Communications Limited ISSN
EC6. 1995. Design of Masonry Structures, Part 1-1: general 1365-556, Earthquake Eng. and Struct. Dynamics, 88–95.
rules for buildings – rules for reinforced and un-reinforced Turnesec V. & Cacovic F. 1971. Some experimental results on
masonry ENV 1996 1-1: Bruxels: CEN. the strength of brick masonry walls, Proc. 2nd Int. Brick
Hamid A.A. & Drysdale R.G. 1980. Behaviour of Brick Masonry Conf., Stoke on Trent 1971, UK, 149–156
Masonry under combined shear and compression loading.
Proc. 2nd Canadian Masonry Symp., Ottawa.
302
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
O. Corbi
Department Structural Engineering, University of Naples Federico II, Naples, Italy
ABSTRACT: The paper focuses on the analysis of masonry vaults: on one side it addresses some of the
geometrical features involved in the treatment of shells, which usually implies not negligible difficulties in
handling the vaults’ problem due to the differential nature of their geometry, and, on the other side, it outlines
possible approaches to the static analysis of masonry vaults. To this regard, the overall static approach should
be conceived in such a way to allow the selection of membrane stress surfaces able to equilibrate the applied
loads and to satisfy the admissibility conditions of the masonry material. The paper, then, deals with these two
features, outlining a general approach for analytically handling the problem of static analysis of vaults.
Figure 1. Correspondence between the two metrics for representing the mid-surface of a shell of general form relevant to the
reference systems (Ox1 ,x2 ,x3 ) and (Ou1 ,u2 ).
303
single-valued and continuously differentiable vector Substitution of ui = ui (t), in Eq. (2) gives
function r(u1 ,u2 ) depending on two variables u1 and r = r[u1 (t), u2 (t)], which is the equation of a curve
u2 belonging to a bi-dimensional space U. on the surface in Eq.(2), as shown in Figure 2. The
The components of r(u1 ,u2 ) are the coordinates vector
of the point P with respect to any reference frame
Ox1 x2 x3 .
By denoting the first derivative of r(u1 ,u2 ) with
respect to ui by ri (u1 ,u2 ), and, marking by “×” the
vector product, it is assumed that coincides with the tangent to this surface curve .
The tangents to all surface curves (Mitrinovic
& Ulcar 1969), which pass through a fixed point
(u1 ,u2 ) of the surface, lie in the tangent plane
(Fig. 3) to the surface at that point, which con-
The equation of the surface, as regards to cartesian co-
tains r1 (u1 ,u2 ) and r2 (u1 ,u2 ). After denoting by R the
ordinates xi in the reference Ox1 x2 x3 , can be expressed
position vector of any point other than P on the tan-
in the parametric form
gent plane, the equation of the tangent plane is the
following
Figure 2. Surface normal vector N and tangent vector t of the surface curve in P.
304
The tangent surface of a curve , r = r(s), is the surface On a surface with the first fundamental form
generated by lines tangent to . It has the equation gik dui duk , let consider the two curves u = u(t) and
v = v(t). If they intersect at a point, then the tangents
to these two curves at the point of intersection define
an angle α between the unit tangent vectors tu and tv ,
where λ is a parameter and t is the unit vector tangent which is defined by (Fig. 7)
to . The moving of a straight line (generator) in any
direction along a curve (directrix) generates a ruled
surface. In the case when every tangent plane to a
surface is tangent to a generator, the surface is called
a developable surface. where
The differential of arch length ds of a curve ui = ui
(t) (i = 1, 2) on the surface r = r(u1 ,u2 ) may be cal-
culated by means of the first fundamental form of the
surface, as follows
is the determinant of the metric tensor g containing the
gik components, which is a symmetric second-order
tensor of the type
where gik = ri · rk .
Using the summation convention, the expression of
the first fundamental form of the surface (Mitrinovic
& Ulcar 1969) can be written in the form
thus representing the first fundamental tensor of the
surface.
Let consider the net formed on the surface by the
For the surface in the form x3 = z(x1 ,x2 ), assuming coordinate curves u1 = const and u2 = const (Fig. 4).
u1 = x1 , u2 = x2 , one has that One can evaluate the area dA of the surface elements
d as
305
Figure 5. Normal section curve n and surface curve : Figure 6. Principal curvatures of the surface at P.
relevant normal vectors and curvatures.
Since the normal vector N to the surface normal sections. If this is not the case, then the normal
r = r(u1 ,u2 ) has the direction of the vector r1 × r2 , curvature kn at the point P has two extreme values
one has which determine two tangents at P.
The stationary values of kn are the principal curva-
tures of the surface at P and the corresponding tangents
are the principal directions at P.
The principal curvatures are denoted by k1 and k2
Let consider now a normal section n of the sur- and are the roots of the equation
face at P (Fig. 5), i.e. the intersection of a surface
with a plane which passes through the surface nor-
mal at a point P of the surface (thus containing the
unit vector normal N to the surface at that point). The
normal curvature of the normal section at P is denoted
by kn = 1/ρn with ρn the radius of curvature. Denoting by φ the angle between any direction at P
According to the Meusnier Theorem, one can relate and the principal direction at P corresponding to “1”
the radius of curvature ρ of the generic surface curve (Fig. 6), the normal curvature kn in the given direction
[ui = ui (t) (i = 1,2)] with curvature k = 1/rρ at the at P is defined by Euler’s formula, as
point P and the angle θ between the normal vectors n
and N at P as follows
306
Figure 8. Normal section of a shell of revolution and
principal curvatures.
307
whence, after some algebraic operations, one gets
308
of Figure 10 yield, as usual, three equations of Baratta & Corbi 2005a, 2006, 2007, Heyman 1977),
equilibrium. showing to produce in its numerical implementation
The external load which may be acting on the ele- results in excellent agreement with experimental data
ment has been resolved into the three perpendicular (Baratta & Corbi 2005b, 2006b).
components pϕ , pθ , pn ., while stresses are those shown In order to undertake this approach the expression
in the figure, yielding, after some algebraic develop- of the Complementary Energy embedded in a masonry
ments, the three equilibrium conditions respectively in vault element should be evaluated. In case of NRT
the tangential direction (direction of θ), in the tangen- assumption, one should consider that the generic ele-
tial direction (direction of ϕ), and in the radial direction ment appears to be partially resistant; this means that,
(i.e. normal direction), as follows denoting by u the distance of the membrane surface
from the upper profile of the vault element, the posi-
tion of the neutral surface (i.e. characterized by null
stresses) is located at a distance of 3u from the extrados
of the element.
For a bi-linear distribution of the normal stress on
the volume element, when neglecting the shear stress
component, the elastic energy can be calculated as
follows
309
So the stress function should be modeled in such a In the second part one outlines some discussion on
way to recognize, after imposing the remaining equi- possible approaches to the analysis of masonry vaults,
librium conditions and admissibility, the membrane also as regards to the possibility of obtaining analyti-
stress functions that can be assumed as the membrane cal expressions of the solutions in terms of membrane
surfaces relevant to different vault geometries. surfaces for different vault typologies.
Some practical applications relevant to the cases
of the barrel vaults with indefinite length, the barrel
vaults confined at their extremity cross-sections, and REFERENCES
the spherical domes are developed in details in papers
by the author (Baratta & Corbi 2007a, 2007b; Baratta Baratta, A. 1984. Il materiale non reagente a trazione come
modello per il calcolo delle tensioni nelle pareti murarie.
et al. 2008).
J. Restauro. 75/76: 53–77.
To this regard, one should emphasize that this is Baratta, A. 1991. Statics and reliability of masonry structures.
not at all a trivial objective to be pursued. Even sim- In F.Casciati & J.B.Roberts (eds), Reliability Problems:
ple applications, which means the specialization of the General Principles and Applications in Mechanics of
described approach to vaults of simple shapes, require Solids and Structures. Udine, CISM.
a pretty hard work for setting up and checking the Baratta, A. & Corbi, O. 2005a. On variational approaches
expressions of the relevant stress functions. in NRT continua. J. of Solids and Structures. 42:
As a counterpart, this approach allows to obtain 5307–5321.
solutions in analytical form, which represents its major Baratta, A. & Corbi, O. 2005b. Fibre Reinforced Com-
posites in Civil Engineering. Experimental Validation of
result.
C-Fibre Masonry Retrofit. Intern. Journal of Masonry
Thereafter the solution for each case can be com- International, British Masonry Society. 18(3): 115–124.
pleted by setting up the energetic approach as men- Baratta, A. & Corbi, O. 2006a. Analysis of masonry vaulted
tioned in the above. systems: the barrel vaults. In Structural Analysis of His-
Definitively this approach is pretty complex since torical Constructions, New Delhi.
it requires to hypothesize and test a number of analyt- Baratta, A. & Corbi, O. 2006b. Frp Composite Reinforce-
ical functions for any vault typology, which represents ments on Masonry Vaults: Effectiveness and Reliability.
a consistent effort even for simple vaults geometries; In A. Mirmiran & A. Nanni (eds) Composites in Civil
anyway it has the big advantage and original research Engineering., Miami: 367–370.
Baratta, A. & Corbi, O. 2007a. Towards a new theory for
result, never available in literature before, to produce
masonry vaults assessment. 7ICM, the 7th International
the explicit analytical expressions of the stress and Masonry Conference. London, U.K.
membrane functions for the single cases, i.e. to give Baratta, A. & Corbi, O. 2007b. Basic Theory For No-Tension
the solutions in analytical form. Vaults: Statics And Strength. AIMETA’07, 18◦ Congresso
Italiano di Meccanica Teorica ed Applicata, Brescia.
Baratta, A., Corbi, I. & Corbi, O. 2008. Stress Analysis of
4 CONCLUSIONS Masonry Structures: arches, walls, and vaults. SAHC08,
VI International Conference on Structural Analysis of
The paper presents some of the fundamental features Historical Constructions, Bath, UK.
Heyman, J. 1977. Equilibrium of shell structures, Oxford
involved in the study of masonry vaults.
University Press 1977.
The first part of the paper is devoted to show the Mitrinovic, D.S. & Ulcar J. 1969.Differential geometry tuto-
main difficulties encountered in the description of the rial texts. Wolters-Nordhoff Publishing.
vaults geometries and to introduce the differential tools Ugural, A.C. 1999. Stresses in plates and shells. McGraw-
for handling this problem. Hill.
310
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
G. De Roeck
Catholic University of Leuven, Belgium
A. Campos-Costa
LNEC, Lisbon, Portugal
ABSTRACT: The paper aims at exploring damage assessment in masonry structures at an early stage by
vibration measurements. For this purpose, one approach is proposed combining global and local ND methods.
To further evaluate the approach, one masonry tower in Portugal was studied together with one wall model in the
laboratory. The model was built as reference, undamaged, state. Afterwards, progressive damage was induced
and sequential modal identification analysis was performed at each damage stage, aiming to find adequate
correspondence between dynamic behavior and internal crack growth. The paper presents all the analyses carried
out with the aim to detect and locate the damage by means of vibrations measurements.
1 INTRODUCTION see Montalvão et al. (2006), but there are only a few
papers on the application to masonry like structures.
Damage on masonry structures mainly relates to An important task before damage can be identified
cracks, foundation settlements, material degradation from vibration characteristics is the study and subse-
and displacements. When cracks occur, generally they quent elimination of the environmental effects Peeters
are localized, splitting the structures in macro-blocks. (2000).
Dynamic based methods to assess the damage are
attractive to this type of structures due to the present
requirements of unobtrusiveness, minimum physical 2 PROPOSED APPROACH FOR DAMAGE
intervention and respect of the original construc- IDENTIFICATION
tion. The assumption that damage can be linked to
a decrease of stiffness also seems to be reasonable to The current practice of structural health monitoring is
this type of structures. based mainly on periodic visual inspections or con-
The present paper deals with the problem of dam- dition surveys but, during the last decade, software
age identification by using Global and Local damage and hardware developments made continuous moni-
identification techniques. It is advantageous to have toring possible, Chang et al. (2003). Typically, one can
two categories for damage identification: (a) the vibra- install hundreds of sensors in a structure and read the
tion based damage identification methods, currently data in real time. Therefore, the present focus of inter-
defined as Global methods, because they do not give est is what type of information is important from the
sufficiently accurate information about the extent of structural point of view and how should the data be
the damage, but they can identify its presence and processed and stored for damage analysis, Londoño
define its precise location (e.g. Chang et al., 2003); and (2006)?
(b) the methods based on visual inspections or experi- Given the latest advantages in the field of damage
mental tests, such as sonic tests, radar tests, radiograph identification analysis and concerning monitoring, it
and thermal field methods (e.g. Doherty, 1987), also is proposed to monitor historical masonry structures
called as Local methods. in four phases (Ramos, 2007):
Many methods are presented in literature for 1. The first phase is the data collection of the
damage identification based on vibration signatures, structure, including the historical information,
311
geometrical and topographic survey, damage sur-
vey, the mechanical characterization of materials
with Non Destructive (ND) tests, a global dynamic
modal test and a numerical model analysis for static
and dynamic calibration. This is the first approach
to the structural behaviour in the assumed healthy
condition at time “zero”;
2. In the second phase the health monitoring plan can
resort to a limited number of sensors (e.g. a pair of
reference accelerometers, strain gauges at critical
sections, temperature and humidity sensors, etc).
Data should be stored periodically and the mon-
itoring system should be able to send an alarm.
Environmental and loading effects should be stud- Figure 1. Mogadouro Clock Tower: (a) and (b) general view
ied and the presence of damage should be detected of the tower with severe damage; and (c) the tower after the
by the global modal parameters; strengthening works.
3. In the third phase, after alarm triggering, a full-scale
dynamic survey with more sensors and measur- Due to lack of maintenance, severe damage was
ing points should be performed. In this phase the observed in the tower in 2004. The damage was char-
“health condition” of a structure is studied with acterized by cracks, material degradation, out-of-plane
more detail. Damage identification methods should displacements, loss of material in some parts, and bio-
be applied to the structure after filtering the envi- logical growth, see Figure 1a. The most severe damage
ronmental effects. The aim of the dynamic methods was a pair of cracks in the East and West façade, see
is to confirm and locate the (possible) damage in a Figure 1b, that divided the box cross section of the
global way; tower in two separated U bodies, leading to a lower
4. In the last phase, a local approach with visual and safety condition.
complementary ND tests should be performed to A geometrical survey of the structure with pho-
locally assess the damage and classify it.This can be togrammetry software allowed registering all anoma-
carried out with sonic test or radar tests, depending lies and dimensions. The dimensions of the structural
on the access conditions of the structure. This local survey have an error about ±5 cm.
approach can give a better definition of damage. To re-establish the tower safety, consolidation
The global and local approach should be considered works were carried out in 2005. The remedial treat-
as complementary tasks. For the case of historical con- ment included: injections for the walls consolidation,
structions these two approaches seem to be suitable, replacement of deteriorated materials and the installa-
since they are ND procedures to evaluate the health tion of tie-rods at two levels with very light prestress-
conditions. ing stress. Figure 1c presents the final appearance of
The following sessions present one case study the tower after strengthening.
and one laboratory simulation of the University of
Minho, where the first three phases of the preceding
3.1 Modal identification before and after the
methodology were already applied.
rehabilitation works
In the case of the Mogadouro Tower, damage was
evident and a dynamic based damage identification
3 STRUCTURAL ASSESSMENT analysis to detect its presence was not necessary. Nev-
ertheless, the dynamic based methods could be used
For the first phase of the proposed approach it is pre- to locate and to assess the damage. Therefore, two
sented the case study of Mogadouro Clock Tower. The dynamic modal identification tests were performed
Mogadouro Clock Tower is located inside the cas- before and after the structural rehabilitation. The aim
tle perimeter of Mogadouro (see Figure 1), a small was to compare the response of the tower with different
town in the Northeast of Portugal. It has a rectangu- structural conditions in order to localize the damage,
lar cross section of 4.7 × 4.5 m2 with and interior hole to archive qualitatively the extent of damage present
of 2.5 × 2.3 m2 , and a height equal to 20.4 m. It was before the rehabilitation works, and to evaluate the
built in stone masonry with granite and shale, with dry efficiency of the rehabilitation works.
joints at the corners and with mortar joints in the cen- Table 1 and Figure 2 present frequencies and mode
tral part of the walls. The walls have, on average, 1 m shapes, as well the Modal Assurance Criteria (MAC)
of thickness and are composed by an irregular bond of values, for the two structural conditions. On average,
stones. the frequencies increase 50%.
312
Table 1. Estimated natural frequencies values.
Before After
313
Table 2. Results of the updating analysis.
314
Figure 5. Environmental effects of the dynamic response. Figure 6. Numerical simulation: (a) ARX model; and
(b) the residuals distribution.
After 22nd of October the frequency values decrease
linearly with humidity almost constant. This indicates
that the structure absorbs water and the mass changes, variation, but it doesn’t account the water absorption
reducing the frequency values because the two quan- phenomenon, because the relative humidity varia-
tities are inverse related. It can also occur that water tion does not totally represent that change. Further-
leads to a stiffness reduction, at least in the lime mor- more, the damage is detected by frequency shifts that
tars. The inverse drying phenomenon also happens significantly go outside the confident intervals ci.
during the hot period. Concerning the numerical frequency simulation, it
Since the environmental and the loading variables should be stressed that the calibration period of one
are changing the modal parameters of the tower, an year might be not enough for having a tuned model. A
attempt to model the dynamic response according to longer period of, at least, three years should be used
the three variables was carried out. It was decided to for calibration in order to have a reliable model for
use a procedure similar to the one used by Peeters frequency prediction.
(2000) with AutoRegressive output with an eXoge-
neous input parts models (ARX models). Here, ARX
models with multiple inputs and a single output (MISO 5 DAMAGE LOCALIZATION
models) were computed in MatLab (2006), function
arx, to model each frequency value. The multivari- For damage localization (the third phase of the pro-
able ARX model with n inputs u and one output y posed approach) a laboratory simulation of a wall
is presented by: model is presented here. The replicate of histori-
cal constructions was built as reference, undamaged,
state. Afterwards, progressive damage was induced
and sequential modal identification analysis was per-
where Aq is a scalar with the delay operator q−1 , Bq is a formed at each damage stage, aiming at finding ade-
matrix 1 × n, and e is the unknown residuals. For dam- quate correspondence between dynamic behavior and
age detection, confidence intervals were establish and internal crack growth.
by analyzing the outliers is possible to detect damage. The wall model was built with clay bricks with
Figure 6 shows for the first natural frequency the 210 × 105 × 55 mm3 , handmade in the Northern area
fitting model through the normalized frequency and of Portugal. The adopted clay bricks have low com-
simulated errors with the 95% confidence intervals pression strength and the adopted Mapei®mortar for
ci. In general, the model represents the frequency the joints has low mechanical properties for the joints,
315
Figure 7. Wall model: (a) general view; and (b) dimensions.
316
be neglected in the subsequent damage identification
analysis.
317
Table 3. Frequency results for ambient excitation [Hz].
†
Mode RSC DSI-C DSII-C DSIII-C DSIV-C
318
masonry-like structures. The methodology was then Doherty, J.E. 1987. Nondestructive Evaluation, Handbook on
applied to the Clock Tower of Mogadouro, Portugal, Experimental Mechanics, A.S. Kobavashi Edt., Society
and to a laboratory wall specimen. for Experimental Mechanics
From the presented analyses, it seems that the Farrar, C.R. and Doebling, S.W. 1998. Damage Detection
and Evaluation II – Field Applications to Large Struc-
proposed methodology is useful and applicable to tures, Modal Analysis and Testing, Júlio Silva and Nuno
historical masonry structures. The natural frequency Maia (Editors), NATO Science Series E, Vol. 363, Kluwer
observation seems to be a reliable approach for Academic Publishers, London, pp 345–378
damage detection, while the mode shape curvatures Londoño, N.A. 2006. Use of Vibration Data for Struc-
changes are reliable quantities in the case of damage tural Health Monitoring of Bridges, PhD Thesis, Carleton
localization. University, Ottawa, Canada
MATLAB 2006. MATLAB User Manual, Release 7.2, The
MathWorks, USA
Montalvão, D.; Maia, N.M.M.; and Ribeiro, A.M.R. 2006.
REFERENCES A Review of Vibration-based Structural Health Moni-
toring with Special Emphasis on Composite Materials,
Chang, P.C.; Flatau, A.; and Liu, S.C. 2003. Review Paper: The Shock and Vibration Digest, Vol. 38, No. 4, pp.
Health Monitoring of Civil Infrastructure, Structural 295–324
Health Monitoring, Vol. 2 (3), pp. 257–267 Peeters, B. 2000. System Identification and Damage Detec-
Choi, S.; Park, S.; and Stubbs, N. 2005. Non-destructive Dam- tion in Civil Engineering, PhD Thesis, Catholic University
age detection in Structures using Changes in Compliance, of Leuven, Belgium
International Journal of Soils and Structures, n.42, pp. Peeters, B.; and De Roeck, G. 1999. Reference-Based
4494–4513 Stochastic Subspace Identification for Output-Only
DIANA 2006. DIANA-9 Finite Element Analysis, User’s Modal Analysis. Mechanical Systems and Signal Process-
Manual – Release 9, TNO, Netherlands ing, 13(6), pp. 855–878
Doebling, S.W.; Farrar C.R.; Prime, M.B.; and Shevitz D. Ramos, L.F. 2007. Damage Identification on Masonry Struc-
1996. Damage identification and health monitoring of tures Based on Vibrations Signatures, PhD Thesis, Univer-
structural and mechanical systems from changes in their sity of Minho, Portugal (www.civil.uminho.pt/masonry)
vibration characteristics: a literature review, Los Alamos SVS 2006. ARTeMIS Extractor Pro User Manual, Release
National Laboratory, NM, 132 p. 3.5, Structural Vibration Solutions, Aalborg, Denmark
319
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: In the paper, the authors present a synthetic overview of some results obtained by means of a
number of theoretical and experimental studies developed on some classical masonry typologies such as arches,
panels and vaults. The proper implementation of the analyzed structural problem, specialized to the specific case,
which derives from the extension of classical structural approaches to structures made of masonry material, is
shown to provide a reliable approach to the problem itself, also in the case of some reinforcement, as demonstrated
by experimental data which are in perfect agreement with numerical results.
321
2,3 m
0,124 m 1
0,200 m
3 2
2,23 m
4
architrave
1,322 m FRP
0,382 m
0,775 m 0,750 m
Transducer Strain-gauge
(a) (b)
Figure 1. (a) Masonry panel geometry, (b) with the applications of a light reinforcement by means of FRP strips.
In this regard, a special formulation of Limit Anal- to be adopted if compared with the displace-
ysis for No-tension structures has been performed, ment/strain approach, whose number of governing
allowing the set up of theorems analogous to the basic variables is higher and, moreover, increasing with
kinetic and static theorems of classical Limit Analysis; the order of discretization (Baratta & I. Corbi 2003,
thus, one can establish efficient procedures to assess 2004, 2006).
structural safety versus the collapse limit state (see e.g. The following section reports some results show-
Como & Grimaldi 1983) by specializing and applying ing how the proper implementation of the described
fundamental theorems of Limit Analysis to NRT con- theoretical approach for classical masonry structural
tinua (Baratta 1991, Baratta & O. Corbi 2003a, 2005a, typologies such as panels, arches and vaults, produces
Bazant 1996, Como & Grimaldi 1983, Khludnev & results that are in a very good agreement with exper-
Kovtinenko 2000). In details, the individuation of the imental data, demonstrating the overall reliability of
collapse (live) load multiplier for NRT continua can the mentioned approach, for whose details one should
be referred to the approaches relying on the two main refer to cited references.
limit analysis theorems: As shown in the following, results can also be suc-
cessfully extended to the case of reinforcements with
– the static theorem; fiber-reinforced polymers (FRP).
– the kinetic theorem.
322
jointed to each other by a pozzolana mortar in order to be recorded in the consolidated case with comparison
confer a light additional resistance to the masonry; to the unconsolidated case.
the masonry itself is characterized by unit weight To this regard, the pretty light type of reinforcement
γ = 10300 Nm and Young modulus E = 5.5 GPa. As allows to read the influence of even a small provision
regards to the loading condition, a varying force is on the panel response, which, on the counterpart,
applied in the middle of the left side of the panel, cannot be expected to be macroscopic.
in such a way to mitigate the proneness of the panel One should emphasize that the first objective of this
to sliding of bricks with respect to each other, and application is, then, to show the sensitivity of the NRT
some loading/unloading cycles are developed up to model even to small changes in the structural response,
the collapse condition. very differently from the elastic model, which, on the
Once reached the crisis, the panel is reinforced contrary, for the specific case, is unable to detect any
by directly laminating on the masonry some FRP difference in the behaviour of the wall. A number of
strips according to the provision scheme shown in Fig- more effective reinforcements have also been tested
ure 1.b, at the same time with the impregnation of by the authors obviously resulting in more appreciable
the fibers by means of a special bi-component epoxy results and a much higher performance (Baratta & I.
resin, and a further experimental investigation is devel- Corbi 2006).
oped on the reinforced structure by re-executing some In the specific case, one can notice that, with refer-
loading/unloading cycles. ence to the same load intensity [e.g. in correspondence
The adopted reinforcement, produced by FTS, of the load value 3000 N in Figures 2.a–c], lower dis-
is a BETONTEX system GV330 U-HT, made of placements can be recorded in case of FRP insertions.
12 K carbon fiber, jointed by an ultra light net Moreover, the increase of the overall stiffness of the
of thermo-welded glass. The mechanical character- panel results in a higher loading capacity with respect
istics of the employed carbon fibers are: tensile to the not-reinforced wall. In particular the trend of
limit stress σfrp = 4.89 GPa, elastic modulus in trac- each curve, shows that it is closer to the x-axis (repre-
tion Efrp = 244 GPa, limit elongation σfrp = 2%. The senting the load variable), thus indicating an increase
FRP strip is characterized by thickness of 0.177 mm in the stiffness which is also related to an higher
and depth of 200 mm. The induced displacements collapse value of the load.
at some selected points [the transducers 1, 2, 3
and 4 in Figure 1.a] of the panel both for the not
reinforced and for the lightly reinforced panel are
2.2 Experimental/theoretical comparison
recorded by a monitoring equipment consisting of: 4
transducers, placed at different locations of the panel Actually the application of the general theory of NRT
in order to record the absolute displacements, and 15 structures to the considered case of the masonry panel,
strain-gauges, arranged in 3 blocks of 5 strain-gauges, also in the presence of FRP reinforcements, can pro-
in such a way that each block is devoted to record duce numerical results which are in good agreement
the related strain situation. In details two transduc- with the results obtained by the above reported exper-
ers are located horizontally at two different heights imental campaign (Baratta & I. Corbi 2006). The
on the panel right side (transducers 1 and 2), and two specialization of the general problem to the case of
are placed in correspondence of the opening, one in masonry walls requires the definition of a discrete
horizontal position at the top of the left side of the model coupled to the real structural model, the set up
hole (transducer 3) and the other one under the archi- of the energetic problem (in the case of masonry pan-
trave, which is devoted exclusively to control the panel els the potential energy approach is to be preferred)
deflection (transducer 4). The displacements s(mm) for the discrete problem, which, for masonry material,
versus the varying force F(N) monitored by the trans- results in a Non Linear programming problem to be
ducers during the experiment in the not-reinforced solved by means of Operational Research tools, and,
and in the reinforced case with some horizontally finally, the search of the numerical solution of the set
applied C-FRP strips are shown in Figures 2.a–c up OR problem by means of a suitably implemented
and 2.d–f respectively, as regards to the first load- calculus code (Baratta & I. Corbi 2004, 2006).
ing cycle. By the diagrams in Figure 2, which report Once followed the above described steps, the
the displacements s(mm) vs the varying force F(N) numerical results can be compared to the ones coming
read by the transducers 1–3, some considerations can out from the experimental investigation, for the final
been made. validation of the theoretical set up.
With reference to the panel’s reinforcement by For the specific case one may compare the results
means of the application of some C-FRP strips, the relevant to the first loading cycle with those related
major effect of the C-FRP intervention is the reduction to experiments. As shown in Figures 2, the theoretical
of the stress in the masonry. In general lower displace- data (continuous lines) are in good agreement wit the
ments at the locations monitored by the transducers can experimental ones (dotted lines) both as regards to the
323
Transducer 1 Transducer 2 Transducer 3
(a) (b) (c)
Experimental data Experimental data Experimental data
Theoretical results 1 Theoretical results 1 Theoretical results
1
0,8 0,8
0,8
0,6 0,6
u3 (mm)
0,6
u2 (mm)
u1 (mm)
0,4 0,4
0,4
0,2 0,2
0,2
0 0
0 0 100 200 300 400 500 600
0 100 200 300 400 500 0 100 200 300 400 500 600
F (Kg) F (Kg) F (Kg)
0,8
0,8 0,8
0,6
0,6 0,6
u1 (mm)
u2 (mm)
u3 (mm)
0,4
0,4 0,4
0 0 0
0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500
F (Kg) F (Kg) F (Kg)
Figure 2. Comparison between the numerical (continuous line) and the experimental (dotted line) at the monitored positions
1, 2, 3 for the not reinforced panel (a, b, c) and for the reinforced panel (d, e, f).
not reinforced case (Figs 2.a–c) and to the consolidated In this case, the sliding between bricks are reduced
case (Figs 2.d–f). and do not influence the overall characteristics of
In the first unconsolidated case the masonry deformability and stiffness of the masonry panel.
exhibits a behaviour which appears lightly stiffer than
in the theoretical model: this effect is maybe due to
3 TESTS ON PROTOTYPES OF MASONRY
the micro-fractures present at the first stage of the
ARCHES
computational procedure, which are probably absent
in the real behaviour of the masonry. The transducers
3.1 Experimental investigation
2 and 3 show an overall pretty good agreement between
numerical and experimental data even if also other This section reports some of the results of the wide
phenomena as sliding between bricks, micro-fractures, experimental campaign developed at the Laboratory
etc., should be taken into account, which cause the not of Materials and Structural Testing of the University of
perfect agreement of the diagrams relevant to the first Naples “Federico II” on masonry arches, consolidated
transducer 1 (Figs 2.a–c). or not by means of FRP strips (Baratta & O. Corbi
It is indeed because of these reasons that the numer- 2003b, 2005b).
ical/experimental agreement is higher, almost perfect, The geometry of the portal arch (Figs 3 and 4)
in the reinforced case (Figs 2.d–f). is symmetrical and is characterized by span
324
d 10 d9 d8
d 11 d7
d 12
d 13 d6
d 14 d5
d 21 d 22 d 23 d d4
24 d 25
d 15 d 20 d 26
I1 d 19 d3
d 27
d 16 d 18 d 28 d2 F
d 17 d 29
G1
E1
d 30 d 1
T1 T2
Dial Gauge G1
Inclinometer I1
Transducers T1,T 2
Extesemeter E1
Deformometric cells dk, k=1…30
325
10 ∆ u (mm) – 1 under the load cell on the intrados of the right pile
(where shear occurs),
Experimental Data – 1 at the bottom of the right pile on the extrados.
8 Trend Line The collapse is reached at F∼800 N with an increase
in the loading capacity of the portal arch of approx-
imately 10 times with respect to the unconsolidated
6 case. The experimental force-displacement diagram is
Numerical reported in Figure 7.
Data
4
3.2 Experimental/theoretical comparison
Actually the application of the general theory of NRT
2
structures to the considered case of the masonry portal
arch, also in the presence of FRP reinforcements, can
produce numerical results which are in good agree-
0 F (N) ment with the results obtained by the above reported
0 20 40 60 80 100 experimental campaign (Baratta & O. Corbi 2005a,
2005b, 2007).
Figure 5. Unreinforced portal arch: pile displacement u The specialization of the general problem to the case
versus load F-numerical/experimental comparison.
of masonry arches requires the definition of a discrete
model coupled to the real structural model, the set up
of the energetic problem (in the case of masonry arches
the complementary energy approach is to be preferred)
The collapse condition is reached at F∼80 N; the
for the discrete problem, which, for masonry mate-
low failure value of the force shows that, due to the
rial, results in a Non Linear programming problem
chosen elliptical shape of the arch, the funicular line
(which in the specific case can be reduced to a Linear
compatible with the applied loads and admissible (i.e.
Programming problem) to be solved by means of Oper-
interior to the arch profile) is already very close to the
ational Research tools, and, finally, the search of the
upper and lower bounds of the arch profile at the rest
numerical solution of the set up OR problem by means
condition.
of a suitably implemented calculus code (Baratta &
The experimental force-displacement diagram is
O. Corbi 2003a, 2003b, 2005a).
reported in Figure 5.
Once followed the above described steps, the
After reaching the collapse condition, the portal
numerical results can be compared to the ones coming
arch is then unloaded in order to be prepared for the
out from the experimental investigation, for the final
subsequent experimental tests on FRP reinforcements.
validation of the theoretical set up.
After completing the unloading process, the portal arch
Numerical investigation on the portal arch model
is prepared for laboratory tests on FRP reinforcements,
experimentally tested results in the possibility of
which are finalized to the evaluation of the benefits
appreciating the skill of the NRT model to capture
induced on the model response by the application of
the major features of the structure behaviour. More-
carbon fibre strips.
over also the correct modelling of the reinforcement
The reinforcement consists of a continuous FRP
and of its coupling with the main structure can be
strip bonded on the extrados of the arch. Since the
evaluated. Figure 5 reports the numerical/experimental
collapse mechanism of the not reinforced simple por-
comparison relevant to the right pile top displacement
tal arch is characterized, as described in the above, by
u (mm) versus the varying load F (N) for the considered
the formation of two intrados hinges at the reins of the
un-reinforced arch.
arch, corresponding to the fractures d4 –d5 and d12 –d13
A very good agreement between the numerical and
at the extrados, the major effect of this intervention
experimental data can be observed. The calculus code
is supposed to be the prevention of these fractures,
is demonstrated to be able to capture the behaviour of
and, therefore, a wide increase in the model loading
the portal arch following the whole loading path up to
capacity.
collapse; Figure 6 depicts the collapse mechanism of
The funicular line is now free to exceed the lower
the structure as it appears directly from the calculus
contour of the portal arch cross section.
code, clearly due to the formation of four hinges: one
In this case the critical condition is related to the
at the keystone on the extrados, two at the reins on
activation of a collapse mechanism composed by four
the intrados, one at the bottom of the right pile on the
hinges, distributed as follows:
extrados.
– 1 at the top of the left pile on the intrados, Moreover one reports in Figure 7 the numeri-
– 1 at the keystone on the extrados, cal/experimental comparison relevant to the right pile
326
Figure 6. Unreinforced portal arch: picture of the collapse Figure 8. Portal arch with extrados reinforcement: picture
mechanism captured from the calculus code. of the collapse mechanism captured from the calculus code.
10 u (mm)
8 Experimental Data
Trend Line
6
Numerical
4 Data
327
dy the length of the corresponding sides on the ele- where z0 and z1 are arbitrary ordinates, conditioned by
ment A B C D projected in the xy-plane, and ϕ and θ the fact that z(t) should be contained in the interior of
denote the angles formed by the meridian sides AB and the profile of the vault.
DC of the element with the x-axis and by the parallel After this result, it is possible to calculate the
sides AD and BC with the y-axis, respectively. As con- internal forces Nx ≤ 0, Ny = Nxy = 0 and Nx ≤ 0,
cerns equilibrium, hypothesizing that the vault is in a Ny = Nxy = 0
membrane state of stress, a correspondence can be set
between forces acting on the element ABCD (stresses
Nx , Ny , Nxy = Nyx and applied load for unit area, px ,
py , pz ) and projected forces acting on the associated
element A B C D (Nx , Ny , Nxy = Nyx and px , py , pz ) It is also possible to realize that the equilibrium
in the xy-plane (Baratta & O. Corbi 2007). solution allows the structure to behave as a sequence
In absence of horizontal loads and if the vertical of identical independent arches. From this result, one
load is not dependent on “y”, as it happens when the may refer to the results reported in the previous section
vault is subject to only vertical loads due to the self- for the portal arch model, reinforced or not with some
weight (i.e. pz = pz (x) ≥ 0), and assuming that the vault FRP strips, whose analytical problem implementation
has an indefinite length in the direction y, equilibrium has been shown to give theoretical results in perfect
may be expressed in the form agreement with the produced experimental data, also
exhibiting very effective results in the reinforced case.
5 CONCLUSIONS
which reduces the problem to the determination of
The paper reports some results proving the successful
stress function ψ(y).
application of a correct theoretical treatment, based
Assuming that the directrix curve of the vault is
on the NRT material assumption, of structural prob-
a circular arch (Fig. 10) of radius R, with constant
lems relevant to classical masonry typologies such as
thickness “s” and unit weight γ, and imposing suit-
arches, walls and vaults.
able constraint conditions, one yields the final solution
The set up of the general energetic approach for
(Baratta & O. Corbi 2007)
analyzing masonry structures under live loads, its
specialization to the relevant discrete models, the
implementation of ad hoc built up calculus codes are
demonstrated to produce numerical results in very
good agreement with data produced by experimental
with investigation.
One should emphasize that, differently from many
models which require a number of parameters allow-
ing a certain adaptation of the shape of the numerical
curve to the experimental one, the NRT model has the
big advantage that the only mechanical parameter to
be evaluated is the masonry elastic modulus. Since the
tuning of the theoretical model is pretty simple, there
would be no possibility to force it to produce theo-
retical results fitting with such a good agreement the
experimental data, because the tuning operation itself
cannot influence the shape of the numerical diagram
but only the displacements scale.
As a point of fact, the sensitivity of the mod-
elling to material assumptions reduces to the inverse
proportionality between the material elastic modulus
and displacements, without any influence on the load
capacity and on the evolution of displacements with
the loads.
Actually the extension to the case of some rein-
forcement directly applied on the masonry can also be
Figure 10. Cross section of a barrel vault with circular arch studied by properly modeling the reinforcement itself
generatrix. and its connection with the masonry.
328
The theoretical/numerical agreement, which is an Civil and Structural Engineering Computing. Egmond
original result also for the case of masonry con- Aan Zee, Netherlands.
structions with FRP reinforcements, demonstrates that Baratta, A. & Corbi, O. 2003b. The No Tension model for the
the overall approach is reliable for the treatment analysis of masonry-like structures strengthened by Fiber
Reinforced Polymers. J. Masonry Intnl. 16 (3): 89–98.
of masonry constructions also in the presence of Baratta, A. & Corbi, O. 2004. Applicability of Duality Theory
consolidation interventions. to L.A. Problems with Undefined Flow Law. Proc. 1st
Int. Conf. RRTEA Restoration, Recycling and Rejuvation
Technology for Engineering, Cesena.
REFERENCES Baratta, A. & Corbi, O. 2005a. On variational approaches in
NRT continua. J. Solids and Structures. 42: 5307–5321.
Baratta, A. 1984. Il materiale Non Reagente a Trazione come Baratta, A. & Corbi, O. 2005b. Fibre reinforced composites
modello per il calcolo delle tensioni nelle pareti murarie. in civil engineering. Experimental validation of C-Fibre
J. Restauro, 75: 53–77. masonry retrofit. J. Masonry Intnl. 18(3): 115–124.
Baratta, A. 1991. Statics and reliability of masonry structures. Baratta, A. & Corbi, O. 2007. Towards a new theory for
In General Principles Applications in Mechanics of Solids masonry vaults assessment. Proc. 7Th Int. Masonry Conf.
and Structures. CISM, Udine: 205–235. London.
Baratta, A. & Corbi, I. 2003. Investigation of FRP consol- Baratta, A., Vigo M. & Voiello G. 1981. Calcolo di archi in
idated masonry panels. Proc. 9th Int. Conf. Civil and materiale non resistente a trazione mediante il principio
Structural Engineering Computing. Egmond Aan Zee, del minimo lavoro complementare. Proc. 1st Nat. Conf.
Netherlands. ASS.I.R.C.CO. Verona.
Baratta, A. & Corbi, I. 2004. Iterative procedure in No- Bazant, Z.P. 1996. Analysis of work-of fracture method for
Tension 2D problems: theoretical solution and exper- measuring fracture energy of concrete. J. Engineering
imental applications”, In G.C.Sih & L.Nobile (eds), Mechanics ASCE 122 (2): 138–144.
Restoration, Recycling and Rejuvenation Technology for Como M. & Grimaldi A. 1983. A unilateral model for Limit
Engineering and Architecture Application, Aracne Ed: Analysis of masonry walls. In Unilateral Problems in
67–75, Bologna. Structural Analysis, Ravello: 25–46.
Baratta,A. & Corbi, I. 2006. On the reinforcement of masonry Heyman, J. 1966. The stone skeleton. J. Solids and Structures,
walls by means of FRP provisions. In A. Mirmiran & 2: 269–279.
A. Nanni (eds) Composites in Civil Engineering., Miami, Khludnev, A.M. & Kovtunenko, V.A.. 2000. Analysis of
2006. cracks in solids. Applied Mechanics Review, 53 (10).
Baratta, A. & Corbi, O. 2003a. Limit Analysis of No Tension Rao, S.S., Optimization: Theory and applications, Wiley &
bodies and non-linear programming. Proc. 9th Int. Conf. Sons, New Delhi, 1978.
329
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The typical failure modes of the Taiwanese historic timber frame temples (Dieh-Dou buildings)
are observed and classified from existing reports and surveys undertaken after the 1999 Chi-Chi earthquake.
The results of a series of laboratory full-scale joint tests are used to set two types of failure criteria, joint failure
and material failure. Static elastic and nonlinear step by step analyses are then performed after simulating the
Dieh-Dou structure with a FE software to assess their behaviour under lateral load. The results show that although
the static analysis can reasonably simulate the overall behaviour of the structure, is only with the step-by-step
analysis, where the correct failure sequence is identified by progressive removal of the elements damaged or
failed according to the failure criteria, that the response of the structure is determined in the correct way. The
critical elements of the frame are then identified, providing important indications for repairing or strengthening
of these structures for the future.
Figure 1. Schematic illustration of a Dieh-Dou timber frame modified from FHCR (2005).
331
Table 1. Level of damage of Dieh-Dou buildings after Chi-
Chi earthquake.
Damage
level Failure mode %
332
As the bearing capacity of the structure as a whole
is not impaired, damage level 3 includes eaves rupture
(Figure 4), masonry wall cracks and columns sliding
from the original position: these are not considered an
immediate threat to collapse but reduce the structures
bearing capacity. Finally, Dieh-Dou frames with in
plane or out of plane leaning or collapse are classified
as damage level 4 or 5, respectively.
333
of the in plane beams connecting to the out of plane
brackets by means of dove tail joints.
Apart from the complicated Dou-Gon joint set,
there are other more common beam to column joints
in the Dieh-Dou buildings. The stiffness of these joints
of these mortise tenon type was assumed in relation to
the connection geometry as proposed by Chang (2006)
in which equation (1) is for beams continuous through
column and equation (2) is for flat cut beam face to
face inside the column.
Figure 6. FE model.
Cw : column width
Bw : beam width
Bd : beam depth dove tail joint (then relying on the timber to timber
E⊥ : modulus of elasticity of timber perpendicular to friction, as expected).
grain. Assuming proper dove tail connections which are
normally guaranteed by accurate craftsmanship, a
Two type of analysis were performed with this FE value of 8000 N was set to define the failure due to
model, elastic and nonlinear step by step. In the elas- element pull out in the FE model. Once the forces are
tic analysis a total lateral load equal to 0.33 mg, FL , is over this value in the model, the element capacity of
applied in correspondence of each purlin. The advan- transmitting axial force is considered lost and the cor-
tages of linear elastic analysis is the possibility of responding degree of freedom is released on the beam
getting a rapid result and make a quick judgement of element in the FE model.
the capacity of the structure under a load magnitude of Differently from pull out tests, the rotational tests
the order of the maximum which can be applied to the showed that the results, in terms of stiffness, depend on
structure by the strongest earthquakes. However, ele- the magnitude of the vertical load. However, although
ments are likely to fail during the earthquake, induced the rotational stiffness were different, the maximum
loads causing loss of structural stiffness and leading moment resisted by the joint was found to be always
the frame to a diverse distribution of load. A nonlinear around 1000 Nm (Tsai & D’Ayala & Lewis 2006),
step by step analysis therefore was performed, where after which the capacity dropped. For the beam-
the model was modified at each load step according to column joints an elastic limit value of 1500 Nm was
the criteria explained below. Figure 6 shows the finite taken from Chang (2006). If the bending moment in
element model of the frame. the model was found to exceed that of the criteria
mentioned above at some point in the analysis, then
the joint will be considered failed in bending, which
3.2 Failure criteria
implies its capacity to transmit moment to the neigh-
The two categories of failure leading to the failure bour elements is lost. In this case, the beam element is
modes previously described, namely joint failure and released in rotation. Finally, the foot of the column has
material failure, are detailed below. a shallow pin into a stone base and relies on friction to
Three types of joint failure modes can be further resist lateral action.
grouped as pull out and bending failure (applying to The coefficient of static friction were calculated
horizontal members) and column failure. from laboratory tests of the timber pieces, obtaining
A series of pull out tests were first performed in values of 0.53 and 0.58, with an average of 0.55, in line
laboratory with full scale samples of one layer of with codes indications (which range from 0.5 to 0.6 for
Dou-Gon joint. The results indicated that the maxi- friction of timber to timber in dry specimens). Divid-
mum pull out forces at failure were similar and did ing the resultant base shear applied at the bottom of
not depend on the magnitude of vertical load applied the column by the vertical load applied, if the value is
(with a value varying between 8000 N and 9000 N) over 0.55, then the columns are assumed to be sliding
if the dove tail connection was good, but reducing to from their original position and so no more horizontal
3000 N and 1500 N when applied loads were 6500 N force could be transmitted to the ground anymore. Due
and 3250 N respectively, in the case of poor fit of the to the particular connection with the ground described
334
above, it is further assumed that the columns are able and the failure criterion (Miyamoto al, 2004), and is
to rotate and do not transmit moment. matched by the observation of the real situation after
For material failure, ombined bending and tension, the earthquake (Tsai & D’Ayala & Lewis 2006). For
combined bending and compression, and shear stress, the step by step analysis, a total of 30 steps were per-
are examined. The material capacities are taken from formed until the collapse criterion was reached. The
the material tests (Tsai & D’Ayala & Lewis 2006) and lateral load at the thirtieth step was 80% of FL. In both
from the Wood Handbook (1999). The failure criterion cases (elastic and step by step analysis), the columns
for bending stress was set at 4.6 E + 7 (N/m2 ), for shear did not reach the failure criterion. At the seventeenth
at 5.5 E + 6 (N/m2 ) and for tension and compression step (50% of FL ), the roof apex has already met the
at 4.5 E + 7 (N/m2 ) and 2 E + 7 (N/m2 ) respectively. damage criterion. The structure drifts results are shown
For the elements subject to axial compression and in Figure 8.
bending, the interaction formula of British Standard When compared with the results of the elastic anal-
(BS 5268-2; 2002), shown as equation (3), is used as ysis where the full lateral load was applied, the step by
failure criterion: step analysis shows a considerable decrease of over-
all stiffness. This is because during these 30 steps
several joints lost their rotational stiffness and sev-
eral beams pulled out and lost bending transmitting
capacity, rendering the structure more deformable.
The result of elastic analysis compared with the
where σm,a, is the applied bending stress; σm,adm, is evidence after the earthquake (Tsai & D’Ayala &
the permissible bending stress; σc,a, is the applied Lewis 2006), confirmed that finite element model
compression stress; σc,adm, is the permissible com- can catch the structural behaviour under horizontal
pression stress (including buckling); σe is the Euler forces.
critical stress. However, the physical tests showed that the max-
Equation (3) is for permissible stress design. How- imum pull out load in Dieh-Dou joint is 8000 to
ever, because the present defined analysis of Dieh-Dou 9000 N, and in the linear analysis results of FE model,
building focuses on failure, the values above are there are several elements already over 8000 N, imply-
inserted instead of permissible values. The value K12 ing that several joints could have suffered element
can be found in Table 22 of the British Standard (BS pull out leading the frame to a diverse distribution
5268-2; 2002). of load which may also cause additional decrease
If the resulting stresses at one step are over the of capacity. The step by step analysis proved to be
values set above, then the corresponding element is more conservative than the elastic analysis. Further-
regarded as failure and completely deactivated in the more, in the FE model analysis, timber material were
corresponding part of the FE model. assumed in good condition, which may not always be
the case.
3.3 Results
In the elastic analysis the maximum load FL is applied 3.3.2 Vulnerability
all at once, while the step by step analysis consists In order to find out the critical elements that may
of a series of models in which 10% of the FL is ini- govern the structural behaviour, the maximum hori-
tially applied and further increases of 10% of FL are zontal displacements at each load level are reported in
added if the results of the previous step shows that no Figure 9 for the step by step analysis.
damage occurred. Once elements are found to reach It can be noted that until the horizontal load is below
the failure criteria they are treated accordingly and the 20% of FL , the structure is stiffer; after this level the
same load level reapplied until no further element fail- displacement rate increases and is maintained constant
ure is attained before increasing the load again. The until failure. This implies that the elements and joints
step by step analysis stops when the roof apex lateral whose capacity is lost before 20% of FL are critical to
displacement are over the collapse criteria proposed the stiffness of the Dieh-Dou frame.
by Miyamoto et al 2004 for Asian timber structures. In figure 10 the elements that failed up to 20% of FL
The procedure of analysis implemented in the pro- are marked. It is seen that they represent 8 main beams
gram is shown in the flow chart of Figure 7. The and corridor beams ends, together with 2 binding ele-
results of both analyses are discussed in the following ment ends, and they failed mostly for loss of rotational
section. stiffness. Three main beams span the main ceremonial
space of the Dieh-Dou temples (see Figure 1). The
3.3.1 Structure drift results show that the second and third main beams,
In the static elastic analysis, the displacement of the failed at both ends, can be the first critical elements
roof apex is 55.2 mm, which lies between the damage under earthquake.
335
Figure 7. Flow chart of step by step analysis.
11
10
9
x displacement (cm)
8
7
6
5
4
3 step by step result
result of static analysis
2
damage criterion
1 failure criterion
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Step
336
Figure 11. Element failed up to 80% of FL .
337
Hsu, U.-C., 2002, Investigation project of historic build- seismic diagnosis, seismic retrofit of Japanese temple,
ing – Chang family temple in Nan-Tou, Nan-Tou County, Proceeding of 13th World Conference on Earthquake
Taiwan, in Traditional Chinese Engineering, August 1–6
Huang, P. &. Shu, M. S., 2001, A study on seismic failure Tsai, P.-H. & D’Ayala, D. & Lewis, G., 2006, Finite Element
and maintenance of historic buildings. I: Bamboo and Modelling of Dieh-Dou Buildings in Taiwan, Structural
timber construction, Council for Cultural Affair, Taipei, Analysis of Historical Construction V. 2. pp. 1297–1304.
in Traditional Chinese Delhi: MacMillan India
King W.S., Yen J.Y., Yen Y.N., 1996, Joint characteris- Wood Handbook, 1999, US department of Agriculture Forest,
tics of traditional Chinese wooden frames, Engineering USA
Structures, Vol. 18, No. 18, pp. 635–644.
Miyamoto S., Miyazawa K., Irie Y. Wu J., Nomata Y. &
Goto, O. 2004, A study on seismic performance and
338
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The restoration of historical buildings is a delicate, complex process, in that newly developed
strengthening materials, whose long-term effects have not yet been fully tested, may interact adversely with the
materials making up the masonry. In the analysis of the behaviour of masonry structures and their constituent
materials, increasing importance has been assumed by the study of the long-term evolution of deformation and
mechanical characteristics, that may be affected by both loading and environmental conditions, especially in
the case of structural restoration by means of innovative materials, which is compounded by the problem of
the durability of the support-mortar system. The purpose of this investigation is to develop a methodology to
be used in laboratory tests as a preliminary design stage for structural interventions, in order to pre-qualify the
strengthening mortars and be able to formulate a judgement as to their compatibility and long-term behaviour
when applied to historical masonry walls.
1 INTRODUCTION
339
is often difficult to analyse and has to be assessed case Table 1. Elastic modulus and compressive strength of
by case. materials.
Through a rather fast laboratory procedure
Eaverage σaverage %σ
(Bocca & Grazzini 2004), this method supplies useful Material (N/mm2 ) νaverage (N/mm2 ) (6 months)
indication for selecting, from a range of alternatives,
the product that is best in keeping with the mechan- Mortar A 6208 0.12 8.27 −7.50
ical characteristics of the historical material, thereby Mortar B 7534 0.19 10.91 +111.55
avoiding the errors associated with materials that are Mortar C 12678 0.23 10.34 +146.39
not mechanically compatible (Valluzzi et al. 2002). By Mortar D 12274 0.32 24.95 +57.47
focusing on the evolution of the deformation param- Historical 4099 0.08 8.09 –
eters in accordance with recent damage models, it is brick
possible to compare the characteristics of the materi-
als, assess their interaction and fatigue behaviour. An
important part was devoted to the execution of cyclic
tests, using new monitoring criteria, leading to a more
comprehensive understanding of the fatigue phenom-
ena that jeopardise the brick-mortar system. Special
attention was devoted to thermo-hygrometric aspects,
which are often overlooked, but whose effects on the
masonry system are often significant enough to com-
promise the validity of strengthening interventions.
2 EXPERIMENTAL CAMPAIGN:
INSTRUMENTATION AND MATERIALS
340
Table 2. Mechanical characteristics during maturation: Table 3. Mechanical characteristics during maturation:
mortar B. mortar D.
σ σ % E E% σ σ % E E%
Test Piece Days (N/mm2 ) (28-d) (N/mm2 ) (28-d) Test Piece Days (N/mm2 ) (28-d) (N/mm2 ) (28-d)
341
Figure 7. Apf and Apstf compute.
342
Table 5. Brittleness index during maturation: mortar B. Table 7. Brittleness index after freez.-thawing cycles:
mortar D.
Apf Apstf Ib %
Piece d (KNmm) Ipf Ipstf | Ccs | Ib (28-d) Ib %
Piece Cycles Ipf Ipstf | Ccs | Ib (0-cycles)
MB 4.3 28 24.64 7.73 0.761 0.239 14.27 0.011
MB 6.2 28 21.61 16.03 0.574 0.426 11.86 0.007 Average 28 days 0.073
MB 7.3 28 20.64 14.31 0.591 0.409 14.96 0.009
Average 28 0.009 – MB1.2 25 0.811 0.189 135.52 0.110
MB5.2 25 0.689 0.311 106.02 0.073
MB 3.3 90 15.00 3.39 0.816 0.180 151.89 0.124
MB 1.3 90 13.46 4.10 0.767 0.233 151.39 0.116 Average 25 0.092 +25.08
MB10.2 90 17.12 7.42 0.698 0.302 72.41 0.051
MB3.2 50 0.894 0.106 279.54 0.250
Average 90 0.097 +995
MB4.2 50 0.721 0.279 83.58 0.060
MB 2.2 150 12.19 2.60 0.824 0.176 177.03 0.146 MB7.1 50 0.718 0.282 83.71 0.060
MB 6.1 150 15.20 3.31 0.821 0.179 199.05 0.163 Average 50 0.123 +68.73
MB 6.2 150 18.82 3.22 0.854 0.146 206.54 0.176
Average 150 0.162 +1732
MB 2.3 210 10.71 1.81 0.855 0.145 316.93 0.271
MB 8.1 210 12.85 3.50 0.786 0.214 213.83 0.168 Table 8. Results of preliminary static tests on mixed test
MB 3.1 210 14.12 1.11 0.927 0.073 910.29 0.844 pieces.
Average 210 0.428 +4741
Pmax σmax σaverage E
Series Test piece (KN) (N/mm2 ) (N/mm2 ) (N/mm2 )
The pieces of mortar B tested within 28 days show a 4 RESULTS OF THE EXPERIMENTAL
low brittleness index which initially justifies the duc- CAMPAIGN: MIXED TEST PIECES
tile behaviour. But during the maturation there is a
remarkable increase of Ib , about 4700%. Such ten- 4.1 Static tests
dency finds on facts in the diagrams which show an
always steeper slope unloading branch as it evolves Compression tests were performed in order to deter-
the maturation of material (Fig. 4). On the contrary, mine the failure load (table 8). Others mixed test
mortar D shows at 28 days a less ductile behaviour, pieces were subjected to 28 freezing-thawing cycles.
but during the maturation the increase of brittleness Of special interest is the medium and great increase
is modest (50–200%) so it highlights a much more in strength observed in CL and DL series after the
constant behaviour in the time. freezing-thawing cycles (table 9).
343
Figure 9. AL series cyclic tests: max vertical deformations.
344
Figure 13. AL series cyclic tests: volumetric deformation.
Figure 12. DL series cyclic tests: max vertical deforma-
tions. Table 10. Analysis of the data.
stage II) were worked out. Through a linear regression Test Log
on the logarithmic scale (Grazzini 2004), it is possible piece n ∂εv /∂n LogN (∂εv /∂n) Nthe
to plot the data in a diagram to obtain an analytical
relationship (10) between secondary creep variations, AL01 22380 0.0270 4.350 −1.569 25583
AL05 53465 0.0198 4.728 −1.703 32029
∂εv / ∂n, and the number of cycles (N) to fatigue failure:
BL03 100000 0.0047 5.000 −2.323 90605
BL05 100000 0.0040 5.000 −2.398 102716
BL06 100000 0.0024 5.000 −2.612 147056
CL05 461 5.1818 2.664 0.714 555
CL09 1223 2.5110 3.087 0.400 941
CL10 15835 0.0501 4.200 −1.300 16294
In a majority of cases, especially in the case of the DL03 1149 0.4704 3.060 −0.328 3187
test pieces which failed to complete the cyclic test, the DL05 100000 0.0015 5.000 −2.813 206028
agreement is very good. DL06 100000 0.0070 5.000 −2.155 68328
The goal is to focus on the deformation response BL04 40993 0.0340 4.613 −1.469 21612
of the material subjected to fatigue tests consisting BL09 360 9.4729 2.556 0.976 358
of different cyclic load combinations. This requires CL04 100000 0.0035 5.000 −2.454 112832
considering additional factors affecting deformation CL07 46622 0.0192 4.669 −1.717 32795
behaviour, as a study of these parameters is needed DL09 100000 0.0025 5.000 −2.594 142671
DL10 100000 0.0113 5.000 −1.947 48171
to complete the analysis of all the mechanical prop-
erties and be able to predict the fatigue response
of the material. A valid correlation was established
between secondary creep rate (∂εv /∂n) during stage II than the actual value, corresponded to the time when
and fatigue life (number of cycles to failure, N). By a significant variation in trend was recorded, i. e. for
performing a certain number of cycles on the material piece AL05 in figure 13 (table 10): horizontal defor-
until it reaches the stage where deformations increase mations growing to a significant extent or volumetric
at a constant rate, it is possible to predict fatigue life deformations shifting to a negative sign (test piece
with a good degree of approximation. propensity to bulge due to poor vertical collaboration
Failure occurs when a deformation limit (corre- or detachment at the interface between the two mate-
lated to the loading level) is reached, after which the rials). The methodology and the numerical analysis
volume begins to increase; if the deformation rate is proved very sensitive to the initial signs of weakening
too slow, the material does not reach the limit value of the brick-mortar system, indicating clearly the onset
during cyclic loading and the values of volumetric of a crisis due to fatigue (Grazzini 2004, 2006).
deformation remain positive.
In some samples, the theoretical value Nthe was
4.3 Brittleness index: comparison between static
found to be lower than the value obtained from lab-
compressive tests, before and after
oratory tests. From a case-by-case analysis it can be
freezing-thawing and loading cyclic
seen that relationship (10), which is a function of the
secondary creep rate, is able to indicate the onset of the The figures 15–18 illustrate the static tests performed
crisis of the brick-mortar systems immediately preced- on mixed test pieces. As for static strength after
ing the final value of the testing cycles. The analysis freezing- thawing cycles, the values obtained for a
of the horizontal and volumetric deformations of these majority of the series remained within the average
test pieces, in fact, demonstrated that the theoretical of the two preliminary tests, save for the series DL
value obtained from expression (10), that was smaller (Fig. 18), which displayed a considerable increase in
345
Figure 14. AL series static tests.
Figure 17. DL series static tests.
5 CONCLUSIONS
346
Table 11. Brittleness index. REFERENCES
Apf Apstf Anzani, A Garavaglia, E., Binda, L. 2006. Time dependent
Piece (KNmm) Ipf Ipstf | Ccs | Ib Ib % Behaviour of Historic Msonry: a probabilistic model,
Proc. of the 7th IMC, London, UK, 50, CD-ROM.
AL02 48.61 50.34 0.491 0.509 103.62 0.051 Binda, L., Saisi, A. 2002. Conservazione degli edifici: il
AL04 128.06 217.99 0.370 0.630 50.86 0.019 miglioramento strutturale e la compatibilità tra i materiali
Average AL (virgin) 0.035 e la struttura originaria (in italian), L’Edilizia, 2, 34–41.
AL03 46.64 18.27 0.719 0.281 252.31 0.181 Bocca, P., Grazzini, A. 2004. Long term mechanical inter-
AL06 46.16 11.58 0.800 0.200 278.34 0.223 action between strengthening material and pre-existing
AL08 33.13 21.20 0.610 0.390 64.68 0.039 structures, Proc. 1th Int. Conf. Innovative Materials and
AL07 38.10 41.96 0.476 0.524 74.53 0.035 Technologies for Construction and Restoration, Lecce,
Average AL after 25 freezing-thawing cycles 0.120 +342 Italy, 2, 176–190.
BL01 64.12 31.16 0.673 0.327 230.85 0.155 Bocca, P., Carpinteri, A., Valente, S. 1988. On the appli-
BL02 56.01 24.10 0.699 0.301 54.75 0.038 cability of fracture mechanics to masonry, Proc. of the
Average BL (virgin) 0.096 8th Int. Brick and Block Masonry Conference, Dublin,
1027–1039.
BL07 52.16 50.53 0.508 0.492 58.22 0.030 Carpinteri,A., Bocca, P., Lacidogna, G., Grazzini,A., Masera,
BL10 57.73 30.80 0.652 0.348 62.00 0.040
D. 2006. Damage evaluation by acoustic emission in brick-
Average BL after 25 freezing-thawing cycles 0.035 −63.5
work structures under variable amplitude loading, Proc. of
BL03 25.91 26.46 0.495 0.505 103.94 0.051 the 7th IMC, London, UK, 32, CD-ROM.
BL05 48.07 26.14 0.648 0.352 175.10 0.113 Grazzini, A. 2004. Experimental analysis of the durability
BL06 25.55 28.61 0.472 0.528 68.22 0.032 of masonry strengthening works, PhD thesis, Dep. of
Average BL after 100000 loading cycles 0.065 −32.3 Structural Engineering , Turin Polytechnic, Italy, pp. 156.
CL01 21.18 25.92 0.450 0.550 23.67 0.011 Grazzini, A. 2006. Experimental techniques for the evalua-
CL02 55.15 17.18 0.762 0.238 215.20 0.164 tion of the durability of strengthening works on historical
Average CL (virgin) 0.087 masonry, Masonry International, 19 (3).
Krajcinovic, D. 1996. Damage mechanics, Ed. Elsevier,
CL06 56.30 16.46 0.774 0.226 331.05 0.256
Amsterdam, The Netherlands.
CL08 13.32 26.89 0.331 0.669 35.43 0.012
Average CL after 25 freezing-thawing cycles 0.134 +154 Minh-tan, D., Chaallal, O., Aitcin, P.C. 1993. Fatigue
behaviour of high-performance concrete, Journal of Mate-
CL04 12.32 22.62 0.353 0.647 59.14 0.021 rials Civil Engineering ASCE, 5, 96–111.
Average CL after28 freez-t. + 100000 loading cycles −75.8 Mu, B., Shah, S.P. 2005. Fatigue behavior of concrete sub-
DL01 25.81 17.71 0.593 0.407 124.19 0.074 jected to biaxial loading in the compression region,
DL02 30.30 28.11 0.519 0.481 40.49 0.021 Materials and Structures, 38, 289–298.
Average DL (virgin) 0.047 Mutluturk, M., Altindag, R., Turk, G., 2004. A decay func-
tion model for the integrity loss of rock when subjected
DL07 76.17 23.96 0.761 0.239 309.12 0.235 to recurrent cycles of freezing-thawing and heating-
DL08 56.94 25.05 0.695 0.305 194.16 0.135
cooling. Int. J. of rock mechanics and mining sciences, 41,
DL04 52.37 12.13 0.812 0.188 203.18 0.165
Average DL after 25 freezing-thawing cycles 0.178 +379 237–244.
Taliercio, A., Gobbi, E. 1996. Experimental investigation on
DL05 59.80 91.16 0.396 0.604 204.22 0.081 the triaxial fatigue behaviour of plain concrete, Magazine
DL06 42.80 28.10 0.604 0.396 81.05 0.049 of Concrete Research, 48, 157–172.
Average DL after 100000 loading cycles 0.065 +138 Valluzzi, M.R., Binda, L., Modena, C. 2002. Experimental
DL09 44.92 45.33 0.498 0.502 88.00 0.044 and analytical studies for the choice of repair techniques
DL10 33.05 42.60 0.437 0.563 89.28 0.039 applied to historic buildings, Materials and Structures, 35,
Average DL after 28 freez-t. + 100000 load cycles 0.042 −10.6 285–292.
Web site: www.lavenaria.it
347
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Salvador Ivorra
Departamento de Ingeniería de la Construcción, OO.PP. e I.U., Universidad de Alicante, Alicante, Spain
Francisco Pallarés
Departamento de Física Aplicada, Polytechnic University of Valencia, Valencia, Spain
ABSTRACT: The bell-tower of Santa Justa y Rufina is located in the historical city of Orihuela (Alicante-
Spain). Its construction dates from the XV century, having suffered several changes later. The monumental group
was declared National Monument by the Spanish government in 1971. For the study of the structural behaviour
of the tower it becomes necessary to know its dynamic response influenced by two important aspects: a) the
construction is located in a high seismic risk area, and b) the tower is endowed with six bells that rotate on their
axis according to the Spanish characteristic system at a speed between 20 cycles/min and 40 cycles/min. With
the purpose of studying its dynamic response a monitoring program has been carried out using accelerometers
with the aim of determining the bending and torsional natural frequencies of the tower as well as approaching
the value of the structural damping ratio of the construction. Once the experimental results were analyzed, a
numerical model of the tower was carried out. This model has been calibrated using the experimental results
to assess the seismic vulnerability of the structure to earthquakes that currently the Spanish Standard states for
this area, as well as to evaluate the effect of the swinging of the bells after the future restoration planned. The
overall elastic modulus and joint rigidity of the tower and the nave of the church have been considered as control
parameters. A foundation of 1.5 m height was considered; the connection between the tower and the foundation
has not been considered rigid.
1 INTRODUCTION a stairway that gives access to the body of bells. All the
structure is made in masonry, arranging brick masonry
1.1 General with ashlars, maintaining a constant thickness of 1.5 m
along the whole first body of the tower. The total height
This paper presents an initial work made on the bell-
is 25.5 m. The bell room reaches the height of 35.5 m
tower of Santa Justa y Rufina church in Orihuela
with 1.0 m wall thickness. Seven windows are inside
(Alicante-Spain).This structure is a gothic belfry made
this body (in each wall) to accommodate the bells with
in XV century that possesses five bells. Some restora-
approximately 5.5 m2 each window.
tion works are planned on these bells and it is necessary
The tower has a rigid joint with the lateral walls of
to know de dynamic characteristics of the tower to dis-
the main body of the church. This joint can be observed
card future problems originated by a possible dynamic
in the west, south and east sides of the tower. The joint
interaction between the natural frequencies of the
in the west wall of the nave reaches the height of 18 m,
tower and the horizontal forces introduced by the bells
although the joint of the east wall reaches 9 m high.
on the tower.
(Figure 3)
349
Figure 3. (a) Simplified model. (b) Model to simulate with
Figure 1. (a) West façade. (b) East façade. FEM.
350
Figure 5. Accelerometers arrangement in the belltower.
(a) Acquisition system. (b) Accelerometer no. 3.
(c) Accelerometer no. 7.
Frequency Mode
Accelerometer Direction (Hz) classification
the arrangement commented, belltower vibrations in of construction. The average damping ratio used in the
the E-W and N-S directions could be determined. The present work for the masonry belltower is 0.0159.
dynamic data obtained from the ambient vibration
have been registered by a Kyowa PCD-320 equip-
ment with a sample rate of 200 Hz. Fig. 5 shows the 3 BELL FORCES
acquisition system and some accelerometers. Tem-
porary acceleration movements have been analysed On the windows of the belfry five bells are located (see
with DAS-100a Kyowa software to obtain frequency Figure 8). These bells swing according to the Spanish
results. system: A counterweight provides a high level of bal-
Power spectra responses in E-W and N-S direc- ance (see Figure 7) and the bells, directly anchored
tions have been obtained from ambient vibrations, so on the tower windows, rotate continuously in the same
the modal parameters can be concluded as shown in direction.
Table 1 and Figure 6. The characteristics of these bells have been deter-
The structural damping ratio is obtained from the mined according to the works of Heyman & Therefall
results in Ivorra (2006), since the structures involved (1976), Ivorra & Llop (2002) and Ivorra et al. (2005);
have very similar characteristics and analogous period they are presented in Table 2.
351
Figure 7. Bell Santas Justa y Rufina. Big rotating bell.
352
Table 2. Characteristics of the bells of the tower.
1
Distance between centre of rotation and centre of gravity.
2
Maximum horizontal force divided by total weight.
3
Maximum vertical force divided by total weight.
353
Table 3. 3D model. Modal participating mass ratios.
Axis
Direction Rotation-axis
Natural
Mode Freq. N–S E–W R(N−S) R(E−W) RZ
classification (Hz) % % Axial % % %
REFERENCES
Bachmann, H., Ammann, W., Deischl, F., 1997. Vibration
Problems in Structures: Practical Guidelines. Springer
Verlag, Berlin, 50–55.
Casolo, S., 1998.A three-dimensional model for vulnerability
analysis of slender medieval masonry tower. Journal of
Figure 12. Power spectrum of horizontal force of bell Earthquake Engineering, Vol. 2, No. 4, 487–512.
number 5. Gentile, C., Saisi, A. (2007). Ambient vibration testing of
historic masonry towers for structural identification and
damage assessment, Construction and Building Materials
5 CONCLUSIONS Vol. 21, 1311–1321.
Ivorra, S., Pallarés, F. (2006). Dynamic investigations on a
A theoretical and experimental dynamic investigation masonry bell tower. Engineering Structures, Volume 28,
of a historic masonry bell-tower is described in the Issue 5, April 2006, Pages 660–667.
paper. The following conclusions can be drawn from NCSE-02. Norma de Construcción Sismorresistente Española.
the study: Parte General y Edificación (Spanish Standard). Ministe-
rio de Fomento. 2002.
1. A simplified and low-cost method is described to Heyman, J.,Therefall, B.D. Inertia Forces due to Bell Ringing,
evaluate the dynamic effect generated by the swing International Journal of Mechanical Sciences, 18, 161–
of bells on the bell tower. 164, 1976.
354
Ivorra, S., Llop F. Determinación de algunas características Schutz, K. G., 1994, Dynamische Beanspruchung von
físicas de una campana. In Proc. of the XIV Congreso Glockentürmen. Bauingenieur 69, Springer-Verlag,
de Conservación y Restauración de Bienes Culturales. 211–217.
Valladolid, p. 891–900 2002[In Spainsh]. Selby, A. R., Wilson J. M., 1997. Dynamic behaviour of
Ivorra, S. et alter (2005) Dynamic forces produced by swing- masonry church bell towers, Worldwide Advances in
ing bells. Meccanica. International Journal of the the Structural Concrete and Masonry. In: Proceedings of the
Italian Association of theoretical and applied mechanics. CCMS Symposium, Chicago, ASCE, NewYork, 188–199.
Ed. Kluwer Academic Press pub. Vol. 41- 1, 47–62, 2006 Wilson, J.M., Selby A., 1993. Engineering a Cathedral,
DIN 4178: “Glockentürme: Berechnung und Ausfhrung”; London, Thomas Telford Ltd., London, 77–100.
1978.
355
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
The analysis of the strains and stresses of the reinforced brick vaults
ABSTRACT: The article specifies the aspects of the analysis of strains and stresses of the reinforcement of
the brick vaults. It introduces the problems of restoring and reconstruction of the vaults and presents various
ways of reinforcing these types of brick constructions applied to the dorsal side, minding at the same time that
frescos and other decorations are normally applied to the intrados side. What is more, the analytical methods and
numerical methods (FEM) used in the analysis of the construction of arches, vaults and domes, are introduced
together with the findings of the research conducted within the range of technical scale in the laboratory of the
Engineering Institute of the Technical University of Wroclaw. In addition, the measurement of the effectiveness
of various types of the adopted reinforcements was also made (inserted bar steels, grids and FRP tapes).
Keeping and maintaining the monumental construc- Making the simulation of the masonry construction,
tions, particularly the arches and vaults, which were there are three major groups of analyses: linear anal-
either partly or totally devastated, is the problem which ysis, nonlinear analysis and limit analysis. It is also
is encountered by more and more people dealing with worth mentioning that despite the differences resulting
construction reinforcement. from the fact that various types of methods gener-
The effective accomplishment of this task requires ate different safety coefficients and different final
an extensive knowledge about the origins of the effects results, there are no ample reasons for choosing the
which contribute to deteriorating the construction’s one particular method over the others – during the
properties, and both the knowledge about the old con- researches, it was decided that the simplest type of
struction methods and modern methods which enable analysis will be applied, namely the linear analysis.
one to prevent those effects and to restore the objects Such a decision was made because of the applicabil-
within the scope of their former carrying capacity and ity of this type of analysis in the simple constructing
stiffness. solutions.
Implementing the modern methods of restoring and
reinforcing the brick constructions or their specific
elements (e.g. arches, vaults or domes), leads to the 2.1 The analytical models of calculation
considerable decline in costs, allowing at the same time A mathematical analysis of the curved brick construc-
saving the culture-precious objects. tions placed an important position in the deliberations
Reinforcing and restoring the brick constructions of mathematicians and construction theoreticians for
of the arches and vaults are the difficult tasks which a long time. What is more, many instructions sub-
usually require the individual approach to each case. stantially based on the engineering practice were
The analysis of the masonry constructions (espe- established.
cially the monumental arches, vaults and domes) is Associating the practical knowledge with the theo-
a very complicated topic because of many problems retical deliberations, enabled one to the better under-
resulting from the description and from the simulation standing of the problem of the static work of arches,
of this type of constructions. Moreover, the numerical vaults and domes. Making the analysis of the current
approach is also a complex problem (simulation of the state of knowledge about estimating the capacity of
material which in fact behaves in a nonlinear way), as arches, vaults and domes, one can easily notice that
well as the limited information about the mechanical one of the major problems appearing on the level of
properties of the materials and the complexity of the the estimation of carrying capacity is accepting the
geometrical construction. static scheme of the curved brick construction with
357
application of the proper method of calculating the
internal forces in constructing elements.
On the way to constructive deliberations about
the statistics of the curved masonry constructions,
the systematization of those deliberations seems to be
intentional. However, this systematization should be
different from that which dominates over the histori-
cal aspects of the subject. Here, the static scheme of the
work of the groups, justifies the division into groups
which can be attributed to a certain group. The knowl-
edge of the static scheme enables the calculation of
internal forces which are present in elements of the
arches, vaults and domes as well it facilitates carrying
out the analysis of their work in respect to construction Figure 1. Research model of an arch.
security and in case of a failure in accepting the proper
method of strengthening.
However, the static calculations are based on vari-
ous simplified assumptions which are not truly real.
Therefore, such calculations cannot be made with
full precision. Uncertainty of the static calculations
becomes problematic when weight is taken into con-
sideration. The construction weight (especially in rela-
tion to monumental buildings) and snow and wind
burdens, are rather unknown and depend on other
factors which cannot be precisely calculated. Even
model-scale experiment cannot ensure certainty.
The problem which ought to be precisely analysed is
the issue of static calculations of the damaged arches,
vaults and domes. It must be specified how the various
types of damage (i.e. loosening of squinch from arch Figure 2. Real view of one of the examined arches –
rib, change of construction geometry) influence the arch A2.
way the internal forces flow through the curved brick
constructions and safety of the entire construction.
material. After a profound study of the subject’s
Although not all the mathematical considerations can
literary sources, it was decided that the masonry
be easily applied, there are a few considerations which
material will be treated as homogenous and isotropic
highly influence the analysis of the curved masonry
material as it has got the same physical properties in
constructions’ behaviour. Moreover, further solutions
all directions.
relate to various types of the curved brick construc-
tions, not only to the one described type of construction
(i.e. arches).
The group of major analytical methods com- 3 DESCRIPTION OF TEST MODELS
prises of:
Laboratory tests were conducted on 7 brick arches
– graphical method of establishing the line of pressure (0.12 m thick, 0.77 m wide, 4 m span and 2 m radius
in the arches, each – as in Fig. 1 and Fig. 2). Barrel arches built of
– three-jointed arches calculation method, materials of characteristics similar to those of mate-
– Heyman’s non-moment method applied to vaults rials in historical objects (full ceramic bricks of class
and domes. 100 and pure lime mortar). Static pattern of support
All of the numerical analyses of the monumental and loading realized in the same way in all the arches
buildings related to the static analysis, are encumbered tested.
with a basic error, namely it is very difficult or almost All of the arches models underwent monotonically
impossible to describe the resistance and deformabil- increasing static loading which was realized on the
ity of materials in present masonry constructions. If entire width of the arch, 1/3 by an actuator until the
this was possible, it would enable taking the correct destruction of the arch.
calculation model. Each arch was reinforced in a different way:
The main problem of such approach is the accep- – A1 model – the arch without any reinforcement –
tance of the most precise and efficient model of witness arch,
358
Figure 3. Model and arch A1.
– A2 model – the arch reinforced with FRCM – Figures 3–5 present a diagrammatic way of enforc-
Ruredil X Mesh C10 carbon fibre screen set in the ing compared with a real view of another arch.
Ruredil X Mesh M 25 mortar on the side of the
ridge,
– A3 model – the arch reinforced with 2 CFRP S&P 4 LABORATORY FINDINGS
Lamelle CFK 150/2000 carbon tapes (100/1.4 inter-
section) placed in a 40 cm axial spacing and with In the course of the researches, the tracks of the static
FRCM material – Ruredil X Mesh C10 carbon fibre balance for all the arches were marked out by mea-
screen set in the Ruredil X Mesh M 25 mortar on suring the intensity of the force loading the laboratory
the side of the ridge, models. Figure 7 presents those tracks.
– A4 model – the arch reinforced with FRCM – The reinforcing materials were supposed to be
Ruredil X Mesh C10 carbon fibre screen set in the set by the dorsal side of the arches or inside the
Ruredil X Mesh M 25 mortar on the side of the ridge arches’ section. This type of reinforcement is domi-
and on the intrados side, nant in case of the historical arches and vaults which
– A5 model – the arch reinforced with ø 8 steel bars, are vividly decorated on the intrados side. In the
set in furrows made in the ridge surface of brick subject’s literary source, one cannot find any simi-
arch, put ins with glue made on the basis of Epidian lar cross-sectional researches with so many methods
5 epoxide resin, of reinforcing the curved brick constructions which
– A6 model – the arch reinforced with 2 CFRP S&P are acceptable with respect to the doctrine in the
Lamelle CFK 150/2000 carbon tapes (100/1.4 inter- constructional conservation of the historical brick
section) glued in a 40 cm axial spacing on the side arches, especially with respect to FRCM reinforce-
of the ridge, ments (with the use of screens e.g. carbon screens
– A7 model – the arch reinforced with a girder made placed in a mineral die-block). The laboratory find-
of GL24 glued wood, joined with the brick arch by ings showed that the reinforcement favorably affects
the use of ø 8 steel anchors glued on the side of the the carrying capacity and stiffness of the examined
ridge. arches.
359
Figure 5. Model and arch A5.
Figure 7. Track of the static balance for arches A1, A2, A5 and A6.
Figure 8 presents the laboratory findings of all the model arch. Table 1 shows the proportional growth of
7 models of arches with specification of the ana- force.
lyzed arches: A1, A2, A5, and A6. One can also The displacements of the arches at the application
notice the considerable growth of the destructive force points with the different levels of loading were
force in all the reinforced arches comparing to the also marked (see Figure 9).
360
Figure 8. Frontier amts of destructive force Fmax in the specific laboratory models.
Table 1. Growth of destructive force Fmax. reinforced by the CFRP tapes of steel bars alone. The
change of the schemes of destruction and growth of
Failure force Fmax Growth of failure force the carrying capacity of the particular models of arches
Arch [kN] [%]
prove the usefulness of the applied reinforcements and
A1 2.882 – the good quality of their workship.
A2 10.866 377.03 In the course of the researches, also the defor-
A5 19.314 670.16 mations were measured (arches A5 and A6) and the
A6 16.588 575.57 normal stresses as well as tangential stresses in the
adhesive-bonded joints, carbon tapes (A6) and steel
bars (A5) were assigned. The results of those measures
are shown on the Figures 14 and 15 (normal stresses)
Figures 10–13 show the real schemes and models and on Figures 16 and 17 (tangential stresses).
of destruction of the examined arches. One should
also notice the transformation of those models of
destruction. Starting from the typical destruction tak- 5 NUMERICAL FINDINGS
ing place in the arch A1 and described in literature
(the model with 4 articulated joints), the next arches It is generally acknowledged that for the detailed
the forming articulated joints change their place (they description of the functioning and destruction of
are formed within bearings and under the place of masonry constructions, it is necessary to adopt the
application force. One can also observe the shearing calculative models with respect to nonlinearity and
of the arches on the bearings’ areas. In the practical brittleness of walls. However, in the case the analy-
use of constructions, there are also very massive walls sis of the static work of the examined arches, it was
at the places of bearings, which are much higher than decided to adopt much less complicated linear-elastic
arches and which have the additional loadings (e.g. models. This was mainly the result of the fact that
loadings of pinnacles). Those walls prevent the arches one of the assumptions was to enable the analysis of
from a thrust so as the case of shearing in bearing the effectiveness of the reinforcements in the common
cannot occur. engineering papers.
The reinforced models gain much bigger deflections On the bases of the created numerical models of the
than the witness arch, even with an additional loading. arches, the series of numerical calculations in the area
It relates to especially those models of reinforcement of flexibility were made with respect to resistance and
which involve using the carbon fibre screens where strength. Basing on those calculations, the contour line
the deformation and destruction proceeds slowly and of deformations and level of stresses for the particular
smoothly. calculative models was received. Figure 18 shows the
There is no any rapid movement related to separa- examples of the deformations in the arch A5 which
tion of the reinforced material as it is with the models arose as a result of unloading on 1 kN level. Figure 19
361
Figure 9. Relation: force “F” – displacement “u” at the loading application point.
Figure 10. Model and real scheme of the destruction of the arch A1(force: 2,882 kN).
Figure 11. Model and Real scheme of the destruction of the arch A2 (force: 10,866 kN).
362
Figure 12. Model and Real scheme of the destruction of the arch A5 (force: 19,314 kN).
Figure 13. Model and Real scheme of the destruction of the arch A6 (force: 16,588 kN).
Figure 14. Normal stresses “σ” in the steel bar at points 1–10 in the sequent amts of the loading with “F” force in the arch A5.
363
Figure 15. Normal stresses “σ” in the carbon tape at points 10–13 in the sequent amts of the loading with “F” force in the
arch A6.
Figure 16. Tangential stresses “τ” in the adhesive-bonded joint between points 1–10 in the sequent amts of the loading with
“F” force in the arch A5.
Figure 17. Tangential stresses “σ” in the carbon tape between points 10–13 in the sequent amts of the loading with “F” force
in the arch A6.
364
Figure 18. Deformations in the numerical model of the arch A5 – the loading level – 1 kN.
Figure 19. Comparison of the deformations at 5 rodman points in the arch A2 (for the real model – a full line and for the
numerical model – dashed line.
shows the comparison of the curves of deformations (Figure 10–13). It shows that the static work of the rein-
at rodman points for the real model (a full line) and forced arches changed and it also proves the need for
the numerical model (dashed line). the applied reinforcements. One should also consider
the schemes of destruction of the joints between an
enforcing material and an arch:
6 CONCLUSIONS
– within the A2 arch it was noticed that the carbon
It is easy to notice that the schemes of destruc- fibre screen loosened from masonry construction
tion in all of the reinforced arch models changed in with brick elements – it shows that destruction
comparison with the A1 arch which was not reinforced (discerption) took place inside the brick,
365
– within the A5 arch it was noticed that steel bars – the serving facility of FRCM systems in arches and
loosened from masonry construction with brick ele- vaults constructions was confirmed,
ments – it shows that destruction (discerption) took – the research proved possibility and usability of joint
place inside the brick and glue composition fulfilled application of CFRP tapes and FRCM system glu-
its role, ing techniques and, in result use of the positive
– within the A6 arch it was noticed that the FRP tape properties of both aforementioned systems,
loosened from the masonry construction it shows – the research proved the possibility and usability
that destruction (discerption) took place inside the of the application of steel bars glued inside the
brick. construction (however, it seems to be very time-
consuming and labour-consuming),
All reinforced models of arches considerably
– the research proved the possibility and usability
increased their carrying capacity. As it is presented
of the application of glued wood elements (joined
in the picture 8, the increase of carrying capacity is
with the construction by the glued steel anchors)
very huge, ranging from 3,77 (A2 arch) through 5,76
in order to stiffen the arch layers (it should be
(A6 arch) up to 6,70 (A5 arch).
noticed that there are some difficulties with match-
The conclusion of the performed research on the
ing of the curvature of enforcing elements made of
models in technical scale shows, it may be stated that:
glued wood to the curvatures of arches and vaults
– all research models were performed with refer- constructions),
ence to the practical use of analysed techniques in – the intersections of CFRP tapes and steel bars are
conservation of the masonry constructions, very rarely used; the best solution of this problem
– the research proved the possibility and usability seems to be their preliminary conjunction which
of FRCM enforcement application, that is the car- enables the substantial use of their carrying capacity
bon fibre screens submerged in mineral mortar what further results in increasing the effectiveness
in order to enforce the historical curved brick of reinforcement. However, such solution is barely
constructions, feasible in this type of constructions.
366
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: In the present paper, we describe the results obtained from double flat-jack tests performed
varying the size of the masonry prism involved in the test. In addition, not only the deformations have been
acquired, but also the acoustic emissions (AE), in order to get information about local cracking during the test.
We present a meso-scale numerical model of the test, where every brick of the masonry is modeled in the details.
Discrete cracks can arise both in the mortar joints and in the brick units. A good correlation is found between
the amount of cracking simulated numerically and the experimental acoustic emissions for different prism sizes.
The model is also able to catch the decrease in the compressive strength with increasing size. It is not possible
to obtain an easy direct relation between the acoustic emission and the amount of cracking; nevertheless, it is
possible to state that the two quantities are proportional to each other when increasing sizes are considered.
367
in a structure and the corresponding activity recorded
on masonry elements of different sizes, tested to failure
by means of double flat-jacks.
368
Figure 2. Acoustic emission measurement system.
369
Figure 6. Combined flat-jack test and AE monitoring.
370
Figure 9. Double flat-jack test on Volume 2: cumulative
number of AE events (2) versus cyclic loading (1).
371
Figure 11. Finite element mesh adopted for Volume 1
exploiting symmetry (crf. Shaded area in Figure 7a). Mesh
and materials (a); loads and boundary conditions (b).
372
5.0
y = 0.6216x - 0.081
2.0 2
R = 0.893
1.0 Nmax
Cracked elements
0.0
3.9 4.0 4.1 4.2 4.3 4.4 4.5 4.6
log V/cm3
600
500
Cracked elements
400
y = 0.0248x + 75.802
2
300 R = 0.93
Figure 13. Crack patterns due to flat-jack pressure in the
three specimens. 200
100
373
In fact, the calculated coefficient of linear regression Brindley, B.J., Holt, J., and Palmer, I. G. 1973. Acoustic emis-
is equal to 0.93. sion. III: The use of ring-down counting. Non-Destr. Test,
Finally, let us observe that the slope of this linear 6(5):299–306.
relation depends on the discretization of the finite ele- Carpinteri, A., and Bocca, P. 1991. Damage and Diagnosis
of Materials and Structures, Pitagora Editrice, Bologna,
ment model. On the other hand, refining the mesh (e.g. Italy.
dividing by two the linear size of each element) does Carpinteri, A., and Lacidogna, G. 2002. Structural monitor-
not change sensibly the exponent in Figure 14. ing and diagnostics by the acoustic emission technique:
Scaling of dissipated energy in compression. Proc., 9th
Int. Congress on Sound and Vibration (ICSV9), Orlando,
6 CONCLUSIONS Fla., Paper No. 166.
Carpinteri, A., and Lacidogna, G. 2003. Damage diagnosis
A numerical simulation of an innovative double flat- in concrete and masonry structures by acoustic emission
jack test combined with AE has been proposed. The technique. J. Facta Univ., 3(13):755–764.
Carpinteri, A., and Lacidogna, G. 2006. Damage monitoring
numerical results agree rather well with the experi- of a masonry building by the acoustic emission technique.
mental evidences, both in terms of the estimated com- Materials & Structures, 39:161–167.
pressive strength and of the crack pattern. The model Carpinteri, A., and Pugno, N. 2002a. Fractal fragmenta-
is also able to catch the decrease in the compressive tion theory for shape effects of quasi-brittle materials in
strength with increasing size. compression. Mag. Concrete Res., 54(6):473–480.
In addition, the number of Acoustic Emissions is Carpinteri, A., and Pugno, N. 2002b. A fractal comminution
put into relation with the number of Gauss points in approach to evaluate the drilling energy dissipation. Int.
the finite element model where cracking takes place. J. Numer. Analyt. Meth. Geomech., 26(5):499–513.
A good correlation is found between the amount of Carpinteri, A., and Pugno, N. 2003. A multifractal comminu-
tion approach for drilling scaling laws. Powder Technol.,
cracking simulated numerically and the experimental 131(1):93–98.
acoustic emissions for different prism sizes. Carpinteri, A., Invernizzi, S., and Lacidogna, G. 2005. In
It is not possible to obtain an easy direct rela- situ damage assessment and nonlinear modeling of an
tion between the acoustic emission and the amount historical masonry tower. Eng. Struct., 27:387–395.
of cracking; nevertheless, it is possible to state that de Witte, F.C., and Schreppers, G.J. DIANA Finite Element
the two quantities are proportional to each other when Analysis User’s Manual, TNO DIANA BV, Delft, The
increasing sizes are considered. Netherlands.
Gregorczyk, P., and Lourenço, P.B. 2000. A review on flat-
jack testing. Engenharia Civil UM, 9:39–50.
REFERENCES Holroyd, T. 2000. The acoustic emission and ultrasonic
monitoring handbook, Coxmoor, Oxford, U.K.
Anzani, A., Binda, L., and Mirabella Roberti, G. 2000. The Kaiser, J. An investigation into the occurrence of noises in
effect of heavy persistent actions into the behavior of tensile tests, or a study of acoustic phenomena in ten-
ancient masonry. Materials & Structures, 33:251–261. sile tests. Ph. D. dissertation. Munich (FRG): Technische
ASTM. 1991a. Standard test method for in situ compressive Hochschule München; 1950.
stress within solid unit masonry estimated using flat-jack Ohtsu, M. 1996. The history and development of acoustic
measurements. ASTM C1196-91, Philadelphia. emission in concrete engineering. Mag. Concrete Res.,
ASTM. 1991b. Standard test method for in situ measurement 48(177):321–330.
of masonry deformability properties the using flat-jack Pollock, A.A. 1973. Acoustic emission. II: Acoustic emission
method. ASTM C1197-91, Philadelphia. amplitudes. Non-Destr. Test., 6(5):264–269.
Binda, L., and Tiraboschi, C. 1999. Flat-jack test as a slightly Rossi, P.P. 1982. Analysis of mechanical characteristics of
destructive technique for the diagnosis of brick and stone brick masonry tested by means of in situ tests. In: 6th Int.
masonry structures. International Journal for Restoration Brick/Block Masonry Conf., Rome, Italy.
of Buildings and Monuments, 449–72.
374
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
L. Binda
Department of Structural Engineering, Politecnico di Milano, Italy
ABSTRACT: In the present paper the results from a recent monitoring campaign and numerical analysis
performed on the ancient Cathedral of Syracuse in Sicily are presented. The Acoustic Emission (AE) technique is
adopted to assess the damage pattern evolution. The localization of the propagating cracks is performed using six
synchronized AE sensors. A clear correlation between the regional seismic activity and the AE acquisition data is
shown. In fact the AE count rate presents peaks corresponding to the main seismic events. In addition, a numerical
analysis of the Cathedral is presented. The nonlinear Finite Element model is particularly refined to account for
the cracking in the most damaged pillar. Some recent seismic events in the area acted as crack propagators. The
crack occurrence obtained from the numerical analysis agrees quite well with the crack localization provided by
the AE monitoring.
375
Figure 1. The Cathedral and its evolution.
376
Figure 5. View of the four sides of the monitored pillar. In
the figure the various materials making the pillar are reported.
377
Figure 8. Differential (a) and cumulated (b) number of AE
oscillations during the monitoring time on pillar no. 19. The
Figure 6. AE signal identified by the transducer.
most relevant seismic events, with the local magnitude value,
occurred during the same period are indicated in the graph.
378
was observed by the author during an earlier moni-
toring process carried out with the AE technique at in
site which was also subject to seismic activity (Carpin-
teri & Lacidogna 2006b, 2007). Thus, it may be stated
that whilst the AE technique finds it primary appli-
cation in non destructive tests, it is also important to
explore its potential as a monitoring tool in structures
exposed to particular environmental conditions, such
as seismic activities. As already pointed out by the
authors, in fact, the two phenomena – acoustic emis-
sions on a structural scale and seismic activities on a
territorial scale – appear really similar and mutually
correlable.
The first stage of the localisation method consists in The time dependence of the structural damage
recognising the data needed to identify the AE sources, observed during the monitoring period, identified by
followed by the triangulation procedure. During the parameter η, can also be correlated to the rate of prop-
first stage, the groups of signals, recorded by the var- agation of the microcracks. If we express the ratio
ious sensors, that fall into time intervals compatible between the cumulative number of AE counts recorded
with the formation of microcracks in the volume anal- during the monitoring process, N , and the number
ysed, are identified. These time intervals, of the order obtained at the end of the observation period, Nd , as a
of micro-seconds, are defined on the basis of the pre- function of time, t, we get the damage time dependence
sumed speed of transmission of the waves (P) and the on AE:
mutual distance of the sensors applied to the surface
of the material. It is usual to assume that the ampli-
tude threshold of 100 µV of the non-amplified signal is
appropriate to distinguish between P-wave and S-wave
arrival times. In fact, P-waves are usually character- In equation (1), the values of Ed and Nd do not nec-
ized by higher value signals. In the second stage, when essarily correspond to critical conditions (Ed ≤ Emax ;
the formation of microcraks in a three-dimensional Nd ≤ Nmax ) and the td parameter must be construed as
space is analysed, the triangulation technique can be the time during which the structure has been mon-
applied if signals recorded by at least five sensors fall itored. By working out the βt exponent from the
into the time intervals. Thus, with this procedure it data obtained during the observation period, we can
is possible to define both the position of the micro- make a prediction as to the structure’s stability con-
cracks in the volume and the speed of transmission ditions. If βt < 1, the damaging process slows down
of P-waves. The localisation procedure can also be and the structure evolves towards stability conditions,
performed through numerical techniques using opti- in as much as energy dissipation tends to decrease;
misation methods such as the Least Squares Method if βt > 1 the process diverges and becomes unsta-
(LSM) (Carpinteri et al. 2006a, 2007a). ble; if βt ∼
= 1 the process is metastable, i.e., though
In the present work, applying the localisation pro- it evolves linearly over time, it can reach indifferently
cedure more than 50 AE sources have been localised either stability or instability conditions (Carpinteri &
with an high confidence level. Considering previous Lacidogna 2006b, 2007; Carpinteri et al. 2007b).
applications of the AE technique carried out by the During the observation period, which lasted 121
authors, the approximation for elements with large days for the monitored pillar, the number of AE counts
size is about ±10 mm (Carpinteri et al. 2006b). The was ∼ = 4300 (Fig. 8). In order to obtain indications on
localised sources and the cracking pattern for pillar the rate of the damage process, as given in Equation 1,
no.19 are represented in Figure 9. It can be noted that the data obtained with the AE technique were sub-
the localised sources are concentrated near the more jected to best-fitting in the bilogarithmic plane. This
visible crack paths. The localisation of these source yielded a slant βt ∼= 0.98 as shown in Figure 10. The
concentration (Fig. 9) and the oscillation counting result confirm that the damage process in the pillar
(Fig. 8) denounce that the pillar is subject to a damag- is in metastable conditions according to a quasi-linear
ing phenomenon in slow but progressive evolution. progression over time.
379
Figure 10. Evaluation of damage, βt exponent for pillar 19.
6 NUMERICAL SIMULATION
380
Figure 12. Finite element mesh of the pillar (a). Details of Figure 14. Principal compressive stress (a). Deformed mesh
the masonry inserts (b). under the seismic acceleration (b).
381
correspond both to diffuse cracking in the contin- Binda, L., Cantini, L., Condoleo, P., Saisi, A. & Zanzi, L.
uum elements in the sandstone blocks and opening 2006. Investigation on the Pillars of the Syracuse
or sliding of the discrete interface between blocks. Cathedral in Sicily, 3-Day Int. Conf. Structural Faults &
Repair, M.C. Forde (Ed.), Engineering Technics Press,
Edinburgh, CD-ROM: 1–12.
Binda, L., Casolo, S., Petrini, V., Saisi, A., Sanjust, C.A. &
7 CONCLUSIONS Zanzi, L. 2007. Evaluation of the Seismic Vulnerability
of the Syracuse Cathedral: Investigation and Modelling,
Proc. of Int. Symposium Studies on Historical Heritage,
The evolution of damage in a pillar made of calcareous Yildiz Technical University, Istambul, Turkey, 683–690.
stone blocks that is part of the vertical bearing structure Binda, L., Saisi, A. & Tiraboschi, C. 2000. Investigation
of the Syracuse Cathedral was evaluated using the AE Procedures for the Diagnosis of Historic Masonries, Con-
technique. The data collected were analysed in order struction and Building Materials, ed. Elsevier-Norwich
to interpret the evolution of damage and to determine 14 (4): 199–233.
the positions of AE sources within the pillar. From Binda, L., Lualdi, M., Saisi, A. & Zanzi, L. 2003. The Com-
the charts plotted for the differential and cumulative plementary Use of on Site Non Destructive Tests for
functions of the AE signal counts it can be seen that the Investigation of Historic Masonry Structures, Proc.
the pillar is actually undergoing a damage process. 9th North American Masonry Conference, (9NAMC),
978–989.
Moreover, by applying the AE source localisation pro- Brindley, B.J., Holt, J. & Palmer, I.G. 1973. Acoustic
cedure it was possible to identify ca. 50 emission points Emission-3: The Use of Ring-Down Counting, Non-
within the pillar. Within the stone blocks to which the Destructive Testing 6: 299–306.
sensors had been applied, the points were seen to con- Carpinteri, A., Invernizzi, S. & Lacidogna, G. 2008 Crack-
centrate along the cracks that could be discerned more ing simulation of brick-masonry subjected to the double
clearly on the surface. The identification of these emis- flat-jack test. Proc. of Structural Analysis of Historical
sion sources together with the oscillation counts shows Constructions, Bath 2–4 July 2008.
that the pillar is indubitably undergoing a slow but Carpinteri,A. & Lacidogna, G. 2006a. Damage Monitoring of
incessant damage process. an Historical Masonry Building by the Acoustic Emission
Technique, Materials and Structures 39: 161–167.
Beside the experimental research, using a numer- Carpinteri, A. & Lacidogna, G. 2006b. Structural Monitoring
ical simulation based on the Finite Element Method, and Integrity Assessment of Medieval Towers, Journal of
the most damaged pillar has been modelled in details, Structural Engineering (ASCE) 132: 1681–1690.
providing an accurate description of the blocks inter- Carpinteri, A. & Lacidogna, G. 2007. Damage Evalua-
action by means of discrete interface elements. The tion of Three Masonry Towers by Acoustic Emission,
numerical analysis has been performed considering Engineering Structures 29: 1569–1579.
the structure subjected to seismic horizontal loads. By Carpinteri, A., Lacidogna, G. & Manuello, A. 2007a. Dam-
this model the crack nucleation is found in the same age Mechanisms Interpreted by Acoustic Emission Signal
positions identified through the AE monitoring. In this Analysis, Key Engineering Materials 347: 577–582.
Carpinteri, A., Lacidogna, G. & Niccolini, G. 2006a. Critical
way the numerical model confirms the results obtained Behavior in Concrete Structures and Damage Localization
experimentally. by Acoustic Emission, Key Engineering Materials 312:
305–310.
Carpinteri, A., Lacidogna, G. & Paggi, M. 2006b. Acoustic
Emission Monitoring and Numerical Modelling of FRP
ACKNOWLEDGEMENTS Delamination in RC Beams With Non-Rectangular Cross-
Section, Materials and Structures (RILEM) 40: 553–566.
The financial support for AE application provided by Carpinteri, A., Lacidogna, G. & Pugno, N. 2007b. Struc-
the European Union (EU) Leonardo da Vinci Pro- tural Damage Diagnosis and Life-Time Assessment by
gramme, ILTOF Project is gratefully acknowledged. Acoustic Emission Monitoring, Engineering Fracture
Special thanks for their collaboration go to the Arch. Mechanics 74: 273–289.
M. Muti, to the Eng. R. Meloni and to the Arch. L. Pollock, A.A. 1973. Acoustic Emission-2: Acoustic Emission
Amplitudes, Non-Destructive Testing 6: 264–269.
Regalbuto of the Soprintendenza of Syracuse. The
Privitera, S. 1863. Illustrazioni sull’AnticoTempio di Minerva
Soprintendenza of Siracusa supported the all the other Oggi il Duomo di Siracusa, Memoria del Parroco Ser-
NDT research. afino Privitera, Tipografia La Fenice Musumeci Catania
(in Italian).
Russo, S. 1991. La Cattedrale di Siracusa, Archivio Storico
REFERENCES Siracusano, III V (in Italian).
Russo, S. 1992, Siracusa Medievale e Moderna, Arnaldo
Agnello, G. 1996. Il Duomo di Siracusa e i Suoi Restauri, a Lombardi, Palermo (in Italian).
cura di S. L. Agnello, Ediprint, Siracusa (in Italian). Swindlehurst, W. 1973. Acoustic Emission-1: Introduction,
Agnello, S.L. 1950. La Rinascita Edilizia a Siracusa Dopo il Non-Destructive Testing 6: 152–158.
Terremoto del 1693, Archivio Storico Siciliano, serie III,
vol. IV, 1950–51, Palermo (in Italian).
382
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Shaking table tests on the large scale model of Mustafa Pasha Mosque
without and with FRP
ABSTRACT: Shaking table tests were carried out on a large scale model of Mustafa Pasha Mosque in Skopje
within the activities of Sixth Framework Program PROHITECH – “Earthquake protection of historical buildings
by reversible mixed technologies”. The main aim of experimental investigation was to evaluate the seismic
stability of the monument after applying a reversible technology for strengthening. Beside the experimental
shaking table testing of the model, three-dimensional finite element analyses were carried out using the general
purpose software package ANSYS. The numerical models were calibrated on the basis of mechanical properties
of masonry obtained from compression and shear tests on wall specimens built with same technique used for
the original prototype. The comparison between experimental and numerical results allowed the reliability of the
implemented models to be evaluated. Finally the effectiveness of strengthening and the evolution of crack patterns
observed during each phase of the testing programme has been analyzed on the basis of the obtained results.
383
Figure 2. Damage to the minaret (horizontal crack) and of
the mosque after phase 1.
384
Figure 5. Collapse of the upper part of the minaret.
385
Table 1. Performed seismic tests, phase 2.
386
Table 2. Performed seismic tests, phase 3.
Input
Scal. Input acc Top acc. displac. Top displ.
factor (g) (dome) (mm) (dome)
387
Table 3. Masonry material properties assumed in numerical
simulations of Minaret and Mosque.
γ E c ϕ δ
[kN/m3 ] [MPa] ν [MPa] [◦ ] [◦ ]
388
Figure 12. Distribution of first principal plastic strains on Figure 13. Distribution of first principal plastic strains on
the original large scale model of the Minaret at collapse load. the reinfroced large scale model of the Minaret at collapse
load.
389
1.8
1.6
original minaret strengthened minaret
1.4 Exp. Exp.
FE FE
1.2
0.8
0.6
0.4
0.2
0
0 1 2 3 4 5 6 7 8
Relative Displ. [mm]
Figure 15. Distribution of first principal plastic strains on Figure 16. Comparison between experimental and numer-
the reinforced large scale model of the Mosque at collapse ical response for the original and strengthened Minaret
load. model.
1.8
reinforced Mosque, allowed the effectiveness of FRP
reinforcement to be analyzed. The wraps around the 1.6
dome and the top of shear walls fully prevent the prop- 1.4
agation of cracks from the bottom part to the drum, as
shown in Figure 15. 1.2
Top Acc. [g]
390
than numerical ones. Considering that a non-linear in terms of ultimate seismic capacity and collapse
static FE analysis was performed, this discrepancy can mechanisms. Further studies are in progress, mainly
be ascribed to cyclic loading induced by seismic exci- devoted to refine the calibration of FE model by imple-
tations and to the rigid block-type behaviour of the menting non linear time history analysis with other
Minaret observed during testing. types of material models for masonry, such as smeared
crack. The calibrated non-linear numerical models will
represent a reliable tool both for the design optimiza-
5 CONCLUSIONS tion of FRPs and for the evaluation of the effectiveness
of other types of strengthening on the original Mosque.
In this study, a numerical and experimental investiga-
tion on the seismic strength of the masonry large scale
model of the Mustafa Pasha Mosque reinforced with ACKNOWLEDGMENTS
FRP has been presented.
The experimental testing of Mustafa pasha large It is significantly acknowledged the financial support
scale model was performed to investigate the seismic of the European Commission (grant No. INCO-CT-
stability of the monument after applying a reversible 2002-509119), for funding the research project PRO-
strengthening methodology, both for the minaret and HITECH (Earthquake PROtection of HIstorical Build-
for the mosque. ings by Reversible Mixed TECHnologies), which is the
The strengthening applied to the minaret enabled main framework of the experimental and numerical
stiffening and increasing of its bending resistance. activity presented in this paper.
The mosque model’s behaviour after strengthening
was evidently different in respect to that of the original
model. Under tests of moderate intensity, the existing REFERENCES
cracks were activated but during the subsequent more
intensive tests, the failure mechanism was transferred Gramatikov, K., Taskov, Lj., Krstevska, L. 2007. Final
report on “Shaking table testing of Mustafa-Pasha
to the lower zone of the bearing walls, in the direc- mosque model”. FP6-PROHITECH Project, Workpack-
tion of the excitation, where typical diagonal cracks age 7-Experimental analysis. University “Sts.Cyril and
occurred due to shear stress. Methodius”, Skopje, Macedonia.
In order to support the experimental test set-up, to Krstevska, L., Taskov, Lj., Gramatikov, K., Landolfo, R.,
design the FRP reinforcements and to analyze their Mammana, O., Portioli, F., Mazzolani, F.M. 2007. Exper-
effects on the prototype, a finite element analysis was imental and numerical investigations on the Mustafa
performed. Pasha Mosque large scale model. In Proceedings of Cost
In particular, the implemented finite element model Action C26 Workshop “Urban habitat constructions under
allowed the evolution of partial and global collapse catastrophic events”, Prague 30–31.3.2007, 158–169.
Landolfo, R., Portioli, F., Mammana, O., Mazzolani, F.M.
mechanisms observed during the tests to be analyzed. 2007. Finite element and limit analysis of the large scale
The results of numerical investigation have been com- model of Mustafa Pasha Mosque in Skopje strengthened
pared with the experimental outcome. Generally, a with FRP. In Proceedings of the First Asia-Pacific Con-
good agreement between the behaviour predicted by ference on FRP in Structures (APFIS 2007). Hong Kong,
numerical models and test results was observed, both China, 12–14 Dec. 2007.
391
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Jianli Yuan
Yangzhou Univ., Yangzhou, China
Donato Abruzzese
Department of Civil Engineering, Univ. of Rome, Tor Vergata, Rome, Italy
ABSTRACT: Dynamic behavior model is essential to the reliability evaluation and restoration scheme of
ancient pagodas. In this paper, the identification techniques and the main influence factors on the dynamic
behavior of ancient pagodas are discussed, and the modeling method integrated the predominance of parameter
forecast, sensitivity analysis, and model updating criteria is developed. The Huqiu Pagoda in Suzhou City,
a famous leaning pavilion-style masonry pagoda, was selected as a research case to present the application
characteristics of the method. A 3D finite element model of this pagoda has been constructed, and the main
structural parameters were updated according to the model updating criteria to match with the measured dynamic
characteristics of ambient vibration test to ensure the validity of the model.
Keywords: ancient pagoda; dynamic behavior; integrated modeling method; environmental random excitation;
finite element analysis; sensitivity system; model updating criteria.
393
the preliminary analysis model by ANSYS, (3) Esti- Table 1. Density of main materials.
mate the dynamic eigenvalues of the structure for the
selection of proper instrument parameters, (4) Acquire Material Density
the modal parameters by ambient vibration test,
(5) Establish sensitivity system and optimize structural Brick masonry 18 (kN/m3 )
parameters, (6) Select model updating criteria, update Granite masonry 26.4 (kN/m3 )
Limestone masonry 25.6 (kN/m3 )
the analysis model, (7) Evaluate the error between
Sandstone masonry 22.4 (kN/m3 )
the analysis value and test value by Errors Evaluation China fir 4∼5 (kN/m3 )
Program, (8) Select adjustment parameters for new Pine wood 5∼6 (kN/m3 )
updating, (9) Obtain the perfect model. Black tile roof 0.9∼1.1 (kN/m2 )
394
It is now clear that the identification of natural fre-
quencies of the structure can be deduced from the
auto-power spectrum of structural response.
For output-only modal analysis, there exist classi-
cal Peak Pick (PP) method (Bendat JS, Piersol AG.
1993) and more recent Enhanced Frequency Domain
Decomposition (EFDD) method (Brincker R., et al,
2001) in the frequency domain. When the PP method
is used in analyzing testing data of ancient pagodas,
the modal frequencies can be identified according
to following criteria: (a) Peak value of auto-power
spectrum of structure response occurs with the same
frequency at each test point; (b) Coherence function is
larger between test points at every modal frequency;
(c) At every modal frequency, each test point has
approximately identical or opposite phases.
Based on random vibration theory, components of
Figure 2. Sketch map of the parameters in Eq. (2). mode shape can be determined by the value of trans-
fer function located at characteristic frequency. As to
foundation (Fig. 2); and d = thickness of the wall at the ambient vibration testing, transfer function is the
foundation. ratio of response at the test point to that at reference
point. Taking reference point as input and test point as
output, the mode shape can be analyzed using transfer
function between both points, which can be expressed
3.3 Identification of modal parameters by ambient as follows:
vibration testing
Ambient vibration technique uses natural ground and
wind pulsations as excitation source, which is sim-
ple to conduct and does not damage the structure. where Gff and Gfy , respectively, are the auto-power
However, the excitation signal is generally faint for spectrum function and cross-power spectrum function
the ancient pagoda with large stiffness. To obtain the of the response signal.
required input signal, the high sensitivity sensors and The calculation of transfer function can be
low noise amplifiers should be selected. Furthermore, expressed in the form of complex numbers:
the pagoda is situated on the scenic spot usually, to
reduce the exterior vibration disturbance the better
sampling time is that with lesser noise from tourists
and vehicles. where |H (ω)| equals to the ratio of amplitude of the
The excitations from ground and wind usually occur signal at test point to that at reference point; and ϕ(ω)
at the same time, the output-only modal analysis is is the phase difference between the signal at test point
used in the ambient technique to avoid the difficulty and that at reference point. The signs of mode shape
of distinguishing the input signals. coordinate at each test point are correlated to the phase
According to the theory of random vibration, fre- relation of cross-power spectra: same signs for same
quency response function H (ω) can be expressed by phases, opposite signs for opposite phases.
the following formula: For ancient pagodas with small damping ratios and
well-separated modal frequencies, when ω ≈ ω̇ under
the arbitrary random excitation, the ratio between
peak value of auto-power spectrum and that of cross-
power spectrum, that is its transfer function, can be
where Gff (ω) and Gyy (ω) are the auto-power spectra of approximated by the ratio of mode shapes.
exciting force f (t) and structure response y(t), respec-
tively. When the frequency spectrum of input source
is smooth and approximately close to white noise with 3.4 Model updating and the reference criteria
a finite bandwidth, the power spectrum can be treated
as a constant C, which reduces Eq. (2) to: Due to the uncertainty of structural parameters of
the ancient pagoda, dynamic characteristics predicted
by the analytical model often differ from field mea-
surements. The proper reference criteria should be
395
provided for the structural parameters identification absolute value of SFpi , the more the sensitivity of model
in the model updating. characteristic to structural parameter pi . A little adjust-
From the field testing we can obtain the first ment to pi will often cause a big change in the dynamic
n order modal parameters of a structure with N characteristics.
degrees of freedom, i.e., natural frequencies [ωT2 ] = According to the structural dynamics, the γ-order
diag(ω12 , ω22 , . . . , ωn2 ) and mode shapes [φT ] = diag eigenvalue λγ and eigenvector φ(γ) , should satisfy:
[{φ1 }, {φ2 }, . . . , {φn }]. The structural mass matrix [MA ]
and stiffness matrix [KA ] can be determined through
structural analysis (finite element analysis for exam-
ple). Basically, there are two sets of reference criteria By solving the above formula’s partial derivative to
for structural parameters identification (Li Guoqian, the i−parameter, the sensitivity of the eigenvalue can
Li Jie, 2002): (1) [ω2 ] = [ωT2 ], [φ] = [φT ]; and (2) be obtained:
[ω2 ] = [ωT2 ], [M ] = [MA ].
Generally, [ωT2 ] measured from the field test is accu-
rate while [φT ] is less accurate. On the other hand,
[MA ] obtained from structural analysis is compara- And the sensitivity of the eigenvector is:
tively accurate but [KA ] is less close to the actual
values. For the ancient pagodas, because absence of
the original constructing data, the field test data is
very valuable for the structural parameters identifi-
cation. To take full advantage of the information from
field test, the reference criteria (1) is usually selected
as the reference for the analysis model updating. Taking the stiffness parameter of structure as main
Many model updating techniques have been devel- study object, the sensitivity of the γ-order eigenvalue
oped in the past (Smith M J, Hutton S G. 1992, and eigenvector to stiffness kij , respectively, are:
Farhat C, Hemez F M. 1993, Renken J A. 1995,
Alvin K F. 1996, Chung Y T. 1997), however, for a
pagoda with complicated architectural details, the 3D
F.E. model consists of tens of thousands of meshing
units and the degrees of freedom usually, it is still a bur-
densome task on structural parameters identification.
To improve the effect of the model updating, the sensi-
tivity system should be constructed for selection of the
structural parameters firstly. Besides, taking the fast
analysis advantage of ANSYS program, the conven- By assessing and analyzing the sensitivity of
tional trial-error method also can be used to simplify dynamic response to the adjustment of structural
the model updating procedure. parameters, the sensitivity system of the model can be
established, which will provide reference for optimiz-
ing the structural parameter adjustment, distinguish-
3.5 Sensitivity of dynamic behavior to structural ing the influence of parameters, and improving the
parameter adjustment efficiency of model updating.
The sensitivity-based model updating procedure
has been recognized as an effective approach for
4 INTEGRATED MODELING OF DYNAMIC
improving FE models and the correlative researches
BEHAVIOR OF HUQIU PAGODA
(Zhang Dewen, Zhang Lingmi, 1992, Jung H. 1992,
Friswelli MI, and Mottershead JE. 1995, Dascotte E,
4.1 Prominent characteristics of Huqiu pagoda
et al. 1995) are helpful for the construction of the
sensitivity system of ancient pagodas. Situated on the Huqiu hill of ancient city Suzhou,
Suppose the structural parameters such as mass, the Huqiu pagoda was built between A.D. 959 and
stiffness, geometric dimensions and material char- 961 and is considered as the one with the oldest
acteristics described as pi (i = 1, 2,. . . , n), and age and the greatest inclination among existing Chi-
the eigenvalues or eigenvectors are considered nese pavilion style masonry pagodas. The pagoda is
as derivative functions of structural parameters, a seven-storey tube structure, with 47.68 meters in
the dynamic characteristics of the pagoda can height and 6100 tons in weight. Ever since its con-
be expressed as F = F(p1 , p2 , . . . , pn ). Then the struction, the groundwork of the pagoda has started to
sensitivity of F to structural parameter pi is settle unevenly, causing the pagoda to incline north-
SFpi = ∂F(p1 , p2 , . . . , pn )/∂pi , and the bigger the ward. Under the erosion by winds and rains and with
396
Table 4. Material properties of the Huqiu pagoda.
397
Figure 6. Auto-power spectrum of test point 3.
398
Table 7. Comparison of the natural frequencies (Hz).
399
in the model as much as possible to consider the
influence to the dynamic behavior.
(2) Environmental random excitation test technology
uses natural ground and wind pulsations as the
excitation source and is non-destructive to ancient
pagodas. But the excitation signals from ground
and wind pulsations are generally faint, for the
ancient pagoda with large stiffness, only the lower
order frequencies can be obtained distinctly, and
the identification effect on mode shapes is reduce
obviously with the increase of the frequency. In
order to obtain the satisfactory data, besides the
a) The mode shapes from the test sensors and amplifiers with high sensitivity and
low noise should be selected, it is important to
ensure enough sampling time for each signal
record.
(3) The elastic modulus of the masonry has a remark-
able influence to the dynamic behavior of the
ancient pagodas, which can be selected as the main
modified parameter for model updating. It is con-
venient for model updating and analysis of the
computer by adjusting the elastic modulus to sim-
ulate the synthesis influence on the material vari-
ation, damage degree, and repaired conditions of
the masonry to the structural stiffness parameter.
(4) For the ancient pagodas, because absence of the
original constructing data, the field test data
is very valuable to identification of the model
parameters. To take full advantage of the informa-
tion from field test, the model updating criteria,
b) The mode shapes from F.E.A. [ω2 ] = [ωT2 ], [φ] = [φT ], can be selected as prefer-
ential reference for modification of the analysis
Figure 9. Comparison of the first four order mode shapes.
model. And the rule: “Predicted mode shapes
are in general agreement with the measurements
while the errors in natural frequencies are within
5 CONCLUSIONS allowable limit” will be helpful for obtaining the
appropriate analysis model.
The integrated modeling method for the dynamic (5) In order to enhance the effect of the integrated
behavior takes advantage of ambient vibration tech- modeling method, the link programs between the
nique and finite element analysis, which can obtain model updating, the parameter optimizing system,
not only the synthesis dynamic response of the whole as well as the error evaluation, should be improved
structure but also the contributions of individual fac- further.
tors such as connections, restraints, and damage con-
ditions. In the construction of the dynamic behavior
model of ancient pagodas, it is necessary to get hold ACKNOWLEDGMENTS
of the main characteristics of the method to improve
the application effect. This research was supported by the Cooperation
(1) Mastery of the main influence factors on dynamic Project granted by the 11th Sino-Italian S&T Joint
behavior of ancient pagodas and adoption of the Commission (2002, N. 47).
concise and proper finite element model, which Suzhou administration of cultural heritage is grate-
will redound to the parameter identification and fully acknowledged for the help in the investigation of
modification. The architectural details such as the Huqiu pagoda.
mast, eaves, and roofs of the pagoda can be simpli-
fied to reduce the number of elements and degrees REFERENCES
of freedom of the model. But the door holes, dif-
ferent structural materials, damaged positions, and Alvin KF. 1996, Finite element model update via Bayesian
complicated connections, etc. should be reflected estimation and minimization of dynamic residuals,
400
Proceedings of the 14th international modal analysis Li Guoqian, Li Jie. 2002, Dynamic Detection Theory and
conference, society for experimental mechanics, Inc, Application for Engineering Structure, Science Press, (in
561∼567. Chinese).
Bendat JS, Piersol AG. 1993, Engineering applications of Renken JA. 1995, Optimal solution selection for sensitivity-
correlation and spectral analysis. 2nd ed. USA, NewYork: based finite element model updating and structural dam-
Wiley Inter science. age detection. AD-A306053.
Brincker R, Ventura CE, Andersen P. 2001, Damping estima- Smith MJ, Hutton SG. 1992, Frequency modification using
tion by frequency domain decomposition, Proceedings of newton’s method and inverse eigenvector updating, AIAA
the 19th international modal analysis conference (IMAC), Journal, 30(7): 1866∼1891.
Florida, USA, 441∼46. Suzhou Huqiu pagoda archives, 1988, the archives of national
Code for design of masonry structures (GB 50003-2001), protection unit of historical and cultural heritages, (in
China Construction Industry Press, 2002 (in Chinese). Chinese).
Chung YT. 1997, Mass matrix updating for model validation, Yao Ling, 2003, Research and application on modeling tech-
Proceedings of the 15th international modal analysis con- nology of dynamic behavior of ancient masonry pagodas,
ference, society for experimental mechanics, Inc, Japan, Dissertation of Yangzhou University, April (in Chinese).
818∼824. Ying Huaiqiao and Liu Jingming, 2002, DASP – data auto
Dascotte E, Strobbe J, Hua H. 1995, Sensitivity-based sample and process system, China orient vibration and
model updating using multiple types of simultaneous state noise institution, Beijing (in Chinese).
variables. Proceedings of the 13th international modal Yuan Jianli, Li Shengcai, 2001, Analysis and investigation of
analysis conference, 1–5. seismic behavior for multistory pavilion ancient pagodas
Farhat C, Hemez F M. 1993, Updating finite element dynamic in China, Structural Studies, Repairs, and Maintenance of
models using an element-by-element sensitivity method- Historical Buildings, WIT Press, UK, 2001(5)129–137.
ology, AIAA Journal, 31(9): 1702∼1711. Yuan Jianli, et al., 2004, Inclination control and reinforce-
Friswelli MI, Mottershead JE. 1995, Finite element model ment technology applied to Huqiu Pagoda, China Civil
updating in structural dynamics, 2nd ed. The Netherlands: Engineering Journal, 37(5): 44∼49(in Chinese).
Kluwer Academic Publishers. Zhang Dewen, Zhang Lingmi, 1992, Matrix transforma-
Jung H. 1992, Structural dynamic model updating using tion method for updating dynamic model. AIAA Journal,
eigensensitivity analysis, Ph.D. thesis. London (UK): 30(5): 1440∼1443.
Imperial College of Science, Technology and Medicine.
401
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
G. Gaudini
Soprintendenza per i Beni Architettonici e per il Paesaggio di Verona, Vicenza e Rovigo, Verona, Italy
ABSTRACT: The “Arche Scaligere” is the gothic monumental funerary complex of the illustrious Veronese
“Scaligeri” family. Between these monuments, the “Arca” (stone tomb) of Cansignorio della Scala is the most
sumptuously decorated. In 2005 the “Soprintendenza per i Beni Architettonici e per il Paesaggio di Verona,
Vicenza e Rovigo” (the local administrative body in charge of the conservation of the landscape, monuments and
historical structures), with the aim of studying the stone tomb of Cansignorio and in the framework of a wider
research, gave the task of designing and installing a structural monitoring system, besides the implementation of
a detailed FE numerical model, to the University of Padova. In parallel with these studies, a light strengthening
intervention was carried out in order to stabilize some critic points of the structure and to intervene in deteriorated
parts or elements, to integrate the original materials. The paper describes the different activities carried out.
1 FOREWORD
403
1.2 Past restoration works
Throughout the centuries, several repair interventions
were necessary to preserve the delicate structure of
the stone tomb, such as those carried out in the XVII,
XIX and XX centuries. In 1676 the Verona municipal-
ity adopted a resolution to execute restoration works on
the tomb, comporting strengthening interventions and
substitutions on the upper part of the monument, with-
out however intervening on the supporting elements.
Between 1827 and 1829 other restoration works were
carried out, raising arguments on the type of marble
to be used in substitutions of the deteriorated parts.
Between 1838 and 1844 the fence was restored, and
on the 24th of July 1840 a portion of the southern
gablet fell down, being subsequently restored (1846)
and lodged back in the original position. Substitu-
tions comported the use of Candoglia marble elements,
secured with iron clamps fixed with melted lead. The
sealing of the cracks was performed with filler. Main
interventions carried out were: the reconstruction of
the spires of some tabernacles; the positioning of steel
reinforcing elements on two columns of a tabernacle;
the complete reconstruction of a column and capital of
a tabernacle, and of some gablets between the spires;
the reconstruction of the tail and the left rear leg of the
Figure 2. The upper part of the Cansignorio stone tomb: horse in the equestrian statue; the substitution of the
below the gablets with the statues of the virtues and the copper tie beams of the tabernacles with saint-warriors
tabernacles with angels; above the equestrian statue. with new ones in iron; the sealing of the vault’s groins.
Other interventions, similar to those executed at the
half of the XIX century, were carried out between 1910
and 1914. The monument was then protected against
sustaining gothic tabernacles, containing statues of the bombing during the two world wars. In 1919, after the
saint-warriors (St. George, St. Martin, St. Quirinus, removal of the shields, some light restorations were
St. Sigismund, St. Valentine and St. Louis, king of carried out. Then, during the positioning of the shields
France). The tomb starts with six columns sustaining a of the 2nd world war, an analysis of the conditions
red marble slab on which finds place the white marble of the tombs was carried out, with successive light
sarcophagus, sustained by eight pillars and decorated restoration works.
with bas-reliefs representing Gospel scenes. The cover
of the sarcophagus hosts a lying statue of Cansignorio,
watched over by angels.
2 THE INVESTIGATION ACTIVITIES
At the second level, six further spiral columns
sustain the canopy with polylobed arches. Above
2.1 Dynamic identification
these finds place a cornice sustaining six gablets
with allegorical figures representing the virtues. At Between different Non Destructive techniques that
the corners are positioned six further tabernacles may be profitably used for the achievement of an
with statues of angels. The roof, corresponding to advanced knowledge of the structural layout of a his-
an hexagonal pyramid made of white marble, finally toric masonry building, dynamic identification proved
supports the massive equestrian statue of Cansignorio to be a very effective tool (Modena et al., 2001; Gentile
(Fig. 2). et al., 2004; Ramos et al., 2006), being actually the
The stones used for the erection of the tomb are only method able to experimentally define parameters
the “Candoglia” white marble, the same employed related to the global structural behavior. Prior to the
in the Milan’s cathedral, and the “Rosso di Verona” installation of a Structural Health Monitoring (SHM)
(Verona’s red marble), besides the Pietra Gallina (a soft System, a dynamic investigation campaign took place
limestone from Vicenza). The inner part of the roof in August 2006. Tests were aimed at the definition of
(above the crossed vault and behind the stone fac- the optimal SHM system sensors’ positioning, and at
ing of the canopy) is composed by solid brickwork the characterization of the dynamic properties of the
masonry. monument for FE modelling calibration purposes.
404
Figure 3. Identified mode shapes: (a) 1st bending N-S, 3.19 Hz; (b) 1st bending E-W, 3.24 Hz; (c) 1st torsion, 5.88 Hz;
(d) 2nd bending N-S, 12.55 Hz; (e) 2nd bending E-W, 12.88 Hz; (f) 2nd torsion, 19.42 Hz; (g) FDD method, average of the
normalized singular values of spectral density matrices of all test setups.
Following the mode shapes emerged from the FE only identification techniques were used (Operational
numerical model, sensors were placed at the first level Modal Analysis). In particular, the recorded ambi-
(in the stone slab where the sarcophagus stands), at the ent vibrations were related to the wind excitation and
second level (on the cornice above the pointed arches) urban traffic.
and at the top of the monument (at the foot of the The modal parameter extraction method selected
equestrian statue). A total of six sensors was employed, was the FDD – Frequency Domain Decomposition –
considering three test setups for a total of 6 acquisition technique (Brincker et al. 2000) which estimates the
points, recording the acceleration in orthogonal (and modes, with the assumption that the excitation is rea-
parallel to the ground) directions. sonably random in time and in the physical space of
The acquisition system was composed by a com- the structure, using a Singular Value Decomposition
pact unit provided with 24-bit digital acquisition cards, (SVD) of each of the spectral density matrices. The
connected to piezoelectric mono axial acceleration data series acquired at 100 SPS were processed by a
transducers. Once fixed the transducers to the structure decimation of 2 (Nyquist frequency of 25 Hz), with
in the selected positions, tests consisted in acquiring segment length of 2048 points and 66.67% window
data over a predetermined period, at a determinate overlap. Several peaks related to structural frequen-
sample rate. Each test setup consisted in recording cies were detected in the frequency domain and the
the signal two times (65,536 points each) whit a sam- corresponding mode shapes defined (Fig. 3).
pling frequency of 100 SPS (samples per second), with
2.2 Monitoring
an overall setup signal recording duration of 21 51 .
For the identification of the modal parameters (natural The Structural Health Monitoring System (installed
frequencies and corresponding mode shapes), output in December 2006) is aimed at the control of static
405
corresponding to 4 daily readings. Dynamic data are
collected both at fixed time intervals (each 48 hours,
approximately 22’ of recording at a sample rate of 100
Hz) and on a trigger basis (shorter records, signals
are recorded when the vibration exceeds a predefined
threshold).
No meaningful variations in terms of displacements
were reported up to November 2007 (Fig. 5a). Varia-
tions remain limited and related to the environmental
parameters, presenting maximum differences (corre-
sponding to crack mouth opening) of about 1/10th of
millimeter. No seismic events were recorded in the
monitored period. Limited shifts (max 4%) were noted
in all of the identified frequencies (see also Ramos
et al., 2007), possibly related to environmental param-
eters, as reported in Figures 5b and c (slight decrease
with the relative humidity, seasonal variations).
3 STRUCTURAL MODELS
Figure 4. Positioning of the acceleration (left) and displace-
ment sensors (right).
3.1 Introduction
A detailed FE numerical model, based on a laser scan-
and dynamic parameters related to the structural func- ner geometrical survey of the monument previously
tioning of the monument. The system is composed carried out, was implemented in order to evaluate the
by an acquisition unit connected to six piezoelec- static and dynamic behaviour of the monument. The
tric accelerometers, two potentiometric displacement evaluation of the initial results of the numerical model
transducers and a temperature and relative humidity (linear static and natural frequency analyses) assisted
sensor. The central unit, located at the base of the the design phase of the strengthening intervention and
tomb, is provided with a WiFi router for remote data indicated the most suitable places for the sensors’posi-
transmission. tioning (dynamic identification and monitoring). The
The monitoring strategy is conceived both to collect first model was calibrated on the basis of the results
data at predetermined time-intervals (periodic moni- of the experimental activities, in order to be subse-
toring, i.e. cracks opening, changes in the dynamic quently used to simulate the response of the monument
response) and to automatically start to save data in case to different external actions.
of significant external events (such as seismic events).
Such controls will permit to appreciate possible vari-
ations in the assessed structural functioning with the
3.2 The FE model
passing of time and to have a record of the dynamic
behavior of the stone tomb during severe events. As a first step, linear elastic constitutive laws were
The acceleration transducers are placed in suitable assigned to all materials in order to define the static
positions in relation to the mode shapes of the struc- load pattern (self weight) and the dynamic properties
ture, as shown by the numerical modeling/dynamic of the monument. The model is composed by approxi-
identification (Fig. 4, left). Four sensors are placed mately 49,000 brick elements and 53,600 nodes. Finite
on two levels for the evaluation of the vibration in elements’ sides are comprised between 0.10–0.15 m.
the NS and EW direction (bending modes) and in the The mesh is more refined in the slender elements
horizontal planes (torsion modes). (columns) and in the junctions, rougher elsewhere. The
A couple of reference sensors is fixed at the base decorative elements and statues were modelled as the
for the record of the ground acceleration both in opera- structural parts: only areas too small to be considered
tional conditions (i.e. evaluation of the traffic induced significant were neglected (Fig. 6).
vibrations) and during seismic events. A tempera- The linear static analysis (self weight) indicates
ture/relative humidity sensor is fixed at the intrados that compressive stresses reach their maximum values
of the marble slab (first level). The displacement in the columns, where stresses of about 1.0–1.5 MPa
transducers are positioned across significant cracks (lower order) are found. In small areas of the upper
(Fig. 4, right, see also Fig. 14). The temperature, rela- order of columns compressive stress peaks of 2.0 MPa
tive humidity and displacement of the selected points are noted. Tensile stresses generally present very low
(crack mouth opening) are recorded each 6 hours, values or close to zero. However, non negligible tensile
406
Figure 5. Monitoring results: (a) displacement transducers PZ1/PZ2 and environmental parameters, recorded data plotted
vs. time; dynamic parameters, identified frequencies: (b) first two bending frequencies vs. time and (c) vs. relative humidity.
Frequency (Hz)
407
Figure 8. Strengthening of the stone tomb of Cansignorio
della Scala, detail of the interventions.
Figure 7. 1 to 6, first six mode shapes of the numerical supports (horse’s hooves) of the equestrian statue of
model. Cansignorio by means of CFRP strips; (2) strengthen-
ing a cracked capital with hoopings in high resistance
stainless steel cable.
4 THE STRENGTHENING INTERVENTION
408
Figure 9. Hooping A, details. Figure 11. Intervention C (a, b) and hooping D (c, d),
details.
409
The investigation activities carried out (dynamic
identification) proposed experimental evidences for
the calibration of behavioral models of the monument.
Finite Elements models of the Arca were imple-
mented on the basis of the laser scanner survey of
the monument. Models were used to predict the static
and dynamic behavior of the building and were succes-
sively tuned on the basis of the experimental activities.
In a successive stage the models will be considered
(with material non linear properties) to assess the
response of the monument to seismic events.
Figure 13. Strengthening of one of the stone capitals of the The installation of a static and dynamic Structural
spiral columns: (a) damage induced by the tie beam oxidation; Health Monitoring System gives the possibility to con-
(b) intervention layout. tinuously evaluate the conditions of the monument
by recording significant indicators (environmental
parameters, dynamic response, cracks opening). The
systems also allows to check the dynamic response of
the structure to traffic or seismic events. An impor-
tant aspect considered in the setup of the monitoring
system was the reduced impact on the monument,
given also the continuous attendance of tourists: sen-
sors were minimized and “camouflaged” as much as
possible (Fig. 14).
The design of the intervention was based on a almost
complete removability of the new structural elements,
positioned to compensate the lacks of the original
material. Interventions, except the consolidation of
heavily damaged parts (horse’s hooves), are based on
a mechanical assembly of metallic elements, avoid-
ing chemical connections with the original material.
New materials and structural elements were chosen in
order to maximally reduce their dimension (i.e. high
resistance stainless steel cables).
Figure 14. Above: a almost full-scale picture of the used dis-
placement transducers; below: displacement transducer PZ2
prior (left) and after (right) the restoration intervention.
ACKNOWLEDGEMENTS
The expansion of the iron tie beams due to mate- The authors would like to thanks student Manuel
rial oxidation caused the cracking of the capitals of the Marotto for his valid collaboration in the execution
upper order columns, in some cases with severe effects of the research activities.
(Fig. 13a). The strengthening intervention required
the sealing of the crack and the positioning of hoop- REFERENCES
ings on the capital, on 3 levels (1.6 mm diameter high
resistance stainless steel cable). Purposely shaped tita- Bertolini V. Da Nantes a Brou e a Ginevra: influssi possibili o
nium elements (Fig. 13b) were employed to allow the irrefutabili delle Arche Scaligere. 1993, Verona (in Italian)
clamping of the cable. Brincker, R., Zhang, L., Andersen P. Modal Identifica-
tion from Ambient Responses using Frequency Domain
Decomposition, Proc. 18th International Modal Analysis
5 CONCLUSIONS Conference, 2000, San Antonio, TX
De Maffei F. Le arche scaligere. 1955, Verona (in Italian)
Extensive studies were carried out on the Arca of Gentile C., Saisi A. Dynamic-based F.E. Model Updating to
Cansignorio della Scala, in Verona, in parallel with a Evaluate Damage in Masonry Towers. Proc. 4th Int. Sem-
light structural strengthening intervention and stone inar on Structural Analysis of Historical Constructions,
2004, Padova, Italy
restoration activities. The research involved several Modena C., Franchetti P., Zonta D., Menga R., Pizzigalli E.,
aspects, some of them not reported in the paper (e.g. Ravasio F., Muti M., Meloni R. and Bordone G. Static and
the laser scanner survey or stone characterization anal- Dynamic Analyses of Maniace Castle in Siracusa-Sicily.
yses), aimed at the achievement of a complete picture Proc. 3rd Int. Seminar on Structural Analysis of Historical
of the monument, for conservation purposes. Constructions, 2001, Guimarães, Portugal
410
Napione E. Le Arche Scaligere di Verona. Storia di un sepol- New Delhi, India. Proc. 5th Int. Seminar on Structural
cro dinastico, 2000–2003 tesi di dottorato di ricerca in Analysis of Historical Constructions, 2006, New Delhi,
Storia dell’arte, Università degli studi di Udine (in Italian) India
Napione E., Varanini G.M. «Gaspar recultor» e l’Arca di Can- Ramos L., Marques L., Lourenço P., De Roeck G.,
signorio della Scala, in Verona illustrata n.18. Museo di Campos-Costa A., Roque J. Monitoring historical
Castelvecchio, 2005, Verona (in Italian) masonry structures with operational modal analysis: two
Ramos L., Casarin F., Algeri C., Lourenço P.B., Modena C. case studies. 2nd International Operational Modal Analy-
Investigation techniques carried out on the Qutb Minar, sis Conference, 2007, Copenhagen, Denmark
411
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
P. Riva
University of Bergamo, Italy
E. Giuriani
University of Brescia, Italy
ABSTRACT: In this paper, the experimental study on the rocking behaviour of a full scale barrel vaulted
structure undergo cyclic horizontal loading is discussed. The study is the first part of an ongoing experimental
and theoretical research program, developed by the University of Brescia, concerning the seismic behaviour of
masonry buildings.
The scope of the paper is to provide some evidence of the rocking mechanism experienced by barrel vaulted
structures undergo horizontal loading. Understanding of the behaviour of such structural systems is fundamental
for their seismic vulnerability assessment, as well as for the correct design of possible strengthening techniques.
The structural behaviour is also investigated by means of non linear finite element analyses. Numerical results
are validated through comparison with experimental results. After validation, the FE model can be applied to
different case studies.
413
the phenomena involved in the mechanisms. It was
shown that in the at-rest condition, the arch lateral
thrust can be largely or entirely resisted by the buttress
action of the abutments, depending on their geome-
try and massiveness, thus possible ties must confine I II
only the part of the arch thrust exceeding the buttress
action (Giuriani & Gubana 1993, 1995). Conversely,
as rocking modifies the resisting mechanism of the A
abutments, increasing the arch thrust and consider- (a) (b) (c)
ably diminishing the confining buttress action, the arch
thrust must be entirely confined by the existing ties Figure 1. Definition of the experimental model geometry:
(Giuriani et al. 2007). A significant increase in the (a) reference structure, (b) unit stripe, (c) experimental model.
tie tensile force is expected during rocking. The same
conclusion was drawn by Lagomarsino et al. (2004)
0.25
by analysing the kinematics of the problem using the
principle of virtual work.
In this work, focus is paid to the experimental rock-
ing behaviour of a barrel vaulted structure undergoing
0.10
0.12
external lateral loading. The effect of seismic action
1.40
°
on the vault crown is beyond the scope of this paper 10
0.1
2
°
and will be studied in the future research.
10
The scope of the paper is to provide evidence of
1.61
the rocking mechanism experienced by barrel vaulted
3.40
51°
structures undergoing horizontal loading, which is
critical for their seismic vulnerability assessment, as
2.60
well as for the correct design of possible strengthening
2.00
techniques.
Focus is made on the modelling of the experimen-
0.99
tal test, on the choice of the geometry of the vault,
the technical and technological aspects concerning its
construction, as well as on the bench set up, the load
application system, and the instrument set up. 0.38 2.50 0.38
Furthermore, the experimental results are used to
validate a FE model. Validation of the FE model allows Figure 2. Geometry of the experimental model.
widening the field of its application to different case
studies.
The study is the first part of an ongoing research
modelling the barrel vault as a series of adjoining
program developed by the University of Brescia, which
arches of unit width (Giuriani E. et al. 1999). In this
is focused on the role of vaults in the seismic behaviour
scenario, according to this simplified analytical model,
of masonry vaulted structures, the assessment of the
a unit stripe A was selected and analysed (Fig.1 a,b).
effectiveness of some retrofitting techniques, and on
As a second step in the definition of the experimen-
the proposal of simplified analytical methods.
tal model, the upper masonry wall were removed and
their weight was introduced in the model by adding
point loads at top edges of the abutments. (Fig. 1c).
The geometry of the experimental model is detailed
2 EXPERIMENTAL TEST
and shown in Figures 2–3, respectively.
As for the construction technique, the wall layout
2.1 Modeling of the experimental specimen
and the layer by layer brick arrangement were derived
In this paper the behaviour of a barrel vaulted structure by historic construction treatises (Carbonara 1996,
undergoing rocking is analysed. As a reference, a real 2004, Giardina et al. 2007a; Fig. 4a). The resulting
two storey building having a barrel vault at the ground wall cross section is a single leave three head wall.
floor is considered (Fig. 1a). The thickness of the vault crown varies along the
Barrel vaults are 3D structures having a complex span: the vault ring is 0.12 m thick at the key and 0.25 m
three-dimensional stress distribution (O’Dwyer 1999). at the springing (Fig. 4b).
However, in the frequent case of long barrel vaults The structure is built on two reinforced concrete
lacking the connections to the head walls, an esti- support; each abutment is fixed to the support with
mate of the structural behaviour can be obtained by vertical rebars which allow rotations at the base,
414
Figure 3. Full scale model of the barrel vaulted structure.
Figure 5. Vertical load application system: (a) side view of
the loading apparatus, (b) load distribution at the wall top
edge, (c) cup-springs at the base reducing the vertical bar
stiffness.
415
Figure 8. Applied horizontal force versus abutment top
horizontal displacement curve (LVDT Bx , Fx ).
The tension in the vertical rebars and the tie rod were
also monitored during the test.
416
Figure 10. Crack pattern for: (a) positive and (b) nega-
Figure 9. Details of the (a) positive and (b) negative applied tive applied lateral loads; (c) detail of the cracks in crack
horizontal force versus abutment top horizontal displacement pattern (b).
curve (LVDT Bx , Fx ).
417
Figure 12. Vertical displacement at the vault intrados
springing and key sections (LVDT Cy , Ey , and Dy ) in case of
positive applied lateral loads (see Fig. 9a).
Figure 14. Tie traction force for a few cycles of: (a) positive
and (b) negative applied horizontal forces.
Figure 13. Sand backfill cracks due to internal tension
force. the tie length L = 3260 mm, the axial strain is equal
to ε = L/L = 1.23 × 10−5 , thus the increment in
The horizontal displacements of the abutments the axial force is F = εESAT = 0.29 kN (where
induced the bending of the vault crown. The vertical ES = 210000 MPa, AT = 113 mm2 ). The further tie
displacements at the intrados of the vault springing and increment indicates that following the development of
key, recorded by transducers Cy , Ey and Dy are shown the crack pattern, the ideal arch increases its span and
in Figure 12 for positive lateral load cycles. The maxi- the buttress action significantly reduces (Giuriani E.
mum uplift displacement of the vault intrados is equal et al. 2007).
to 5.6 mm at the right springing, corresponding to 1/3 When negative maximum displacements were
of the abutment top lateral displacement; whereas a applied, the mechanism changed, as shown in
2.5 mm maximum downward displacement, equal to Figure 10b. The crack pattern showed two wide cracks
1/8 of the abutment top displacement, was recorded at the abutment bases located at different heights, thus
at the left springing. The vault key also rose 1.9 mm. resulting in a negative relative displacement of the
Vault flexural deformations were emphasized by the abutments at the vault impost. In this case, the tie ten-
crack extending into backfill (Fig.13). sile force reduced to approximately 50% of its initial
The tie tensile force significantly changed for value (Fig. 14b).
increasing applied lateral loads (Fig. 14). Finally, the vertical tension in the rebars increased
When positive maximum displacements were during the test, as the lateral load and displacement
applied and the mechanism was that corresponding increased, despite the positioning of the cup-spring
to the crack pattern of Figure 10a, the tie tensile force packs at the bar bases. However, the maximum incre-
increased to 12% of its initial value (F = 0.82 kN, ment was at most equal to 15% of the initial value.
Fig.14a). Note that the test was stopped in this direction
well before the cracks had completely penetrated 4 NUMERIC ANALYSES
the element cross sections. Thus, for increasing lat-
eral displacement, a further significant increment Numeric analyses were carried out prior and after the
in the tie tension should be expected. Furthermore, experimental tests. Pilot analyses provided some use-
the abutments underwent a positive relative dis- ful information to guide the modelling and design
placement of L = 0.04 mm (Fig. 11a) at the vault of the experimental test. On the other hand, upon
springings, which only partially explains the incre- completion of the experimental test, the experimental
ment in the tie force. As a matter of fact, given results were used to validate the numeric model.
418
Table 1. Mechanical properties in Mesh B.
Masonry Backfill
419
action and the increase of the ideal arch span.
This result should be carefully taken into account
in seismic vulnerability assessment, as tie rods
might experience excess in tension during an earth-
quake, thus requiring additional strengthening. On
the other hand, depending on the triggered mech-
anism, tie tension might also decrease if rocking
induced displacements shorten the vault span.
The numerical study also confirmed that the crack
Figure 17. Mesh B deformed shape at failure (Magnifica- opening locations are a function of the geometry and
tion factor of 20). of the applied vertical loads. When large vertical loads
are applied, two cracks form in the vault ring close
opening is observed at the same locations where cracks to the springing where the thickness of the section
opened in the experimental test. It is worth noting that abruptly halves. On the other hand, for decreasing ver-
the chronological order of crack development is also tical loads, two cracks form in the abutments, close to
the same. the vault imposts, rather than in the vault crown.
Further numeric analyses showed that the crack Further experimental tests will be performed on
opening location is a function of the geometry and of unreinforced and reinforced masonry vaults, to eval-
the applied vertical loads. When large vertical load is uate the efficacy of some strengthening techniques,
applied to the structure (i.e. in the case of two or more such as: extrados spandrel walls, thin slab of clay
storey buildings), besides the crack at the abutment mortar, and FRP. Numerical and analytical models
bases, two further cracks form in the vault ring close will be improved to simulate cyclic behaviour of the
to the springing where the thickness of the section structures.
abruptly decreases. On the other hand, for decreasing
vertical loads, these two cracks form in the abutments,
close to the vault springings, rather than in the vault
ACKNOWLEDGEMENTS
crown.
Numeric and experimental results also compare
The authors gratefully acknowledge R. Squicciarini
well in terms of vertical displacements along the vault
for his contribution in carrying out the experimental
crown (Fig.16b), and tensile tie force.
test and L. Martinelli for his technical support. This
The deformed shape at failure is shown in Figure
work was developed within the research project DPC-
17. The crack locations correspond to the experimental
ReLUIS 2005–2008, Research line n. 1: “Vulnerability
mechanism surveyed during the experimental test.
assessment and antiseismic strengthening of masonry
buildings”. The experimental specimen was built by
the students of the Scuola Edile Bresciana, within a
5 CONCLUSIONS joined research agreement. Their work, as well as the
work of their supervisors, is gratefully acknowledged.
In this paper, the transverse rocking mechanism of
a barrel vaulted structure was investigated by means
of a full scale experimental tests and a non-linear FE REFERENCES
analyses.
The most significant results of the experimental Abruzzese D., Lanni G. 1999. Some developments on the lat-
study can be summarized as follows: eral strength of historical reinforced vaulted buildings. In
Structural Studies, Repairs and Mainteinance of histor-
• The structure is relatively ductile in terms of ical buildings, VI International Conference, STREMA,
displacements. However, little energy dissipation Brebbia and Jager Ed., Dresda.
capacity and a pronounced self centering behaviour Binda L., Baronio G., Gambarotta L., Lagomarsino S.,
are observed. Failure is reached with a four hinge Modena C. 1999. Masonry constructions in seismic areas
mechanism. Any possible retrofitting technique of central Italy: a multi-level approach to conservation,
should be respectful of this structural behaviour, 8NAMC 8th North American Masonry Conf., Austin,
by strengthening the structure without reducing its USA, pp. 44–55.
Borri A., Corradi M., Vignoli A. 2002. New materials
capacity to meet to displacement demand;
for strengthening and seismic upgrading interventions.
• The tie tension significantly changes during rock-
In International Workshop Ariadne 10, Arcchip, April
ing. The small abutment relative displacements 22–28 2002, Prague, Czech Republic, cd-rom: 1–24.
induced by rocking is not sufficient to explain this Carbonara G. 1996. Atlante del restauro. Torino: Utet.
increase in tie tension. The tension is likely to Como M., Lanni G., Sacco E. 1991. Sul calcolo delle catene
increase following the loss of the abutment buttress di rinforzo negli edifici in muratura soggetti ad azione
420
sismica. VII convegno nazionale L’ingegneria sismica in Giuriani E., Marini A., Porteri C., Preti M. 2007. Meccanismo
Italia, ANIDIS, Palermo. di rocking degli archi diaframma e vulnerabilità sismica
D’Ayala D., Speranza E. 2002. An integrated procedure for delle chiese. Technical Report n. 7. University of Brescia.
the assessment of seismic vulnerability of historic build- Giuriani E., Marini A., Porteri C., Preti M. 2007. Seismic
ings. In Proceedings of the 12th European Conference on vulnerability of churches associated to transverse arch
Earthquake Engineering, London, Elsevier Science, paper rocking. Submitted for possible publication to the Jour-
n. 561 (CD-ROM). nal of Architectural Heritage: Conservation, Analysis and
Giardina G., Marini A., Riva P., Squicciarini R. 2007a. Anal- Restoration.
isi dei meccanismi di collasso e delle tecniche di rinforzo Guerrieri (Ed.) 1999. Manuale per la riabilitazione e la
di archi e volte in muratura. Technical Report n. 13. ricostruzione post sismica degli edifici. Regione Umbria.
University of Brescia. DEI. Tipografia del Genio Civile.
Giardina G., Marini A., Riva P., Squicciarini R. 2007b. Housner G. W. 1963. The Behaviour Of inverted pendulum
Allestimento e collaudo del modello sperimentale e del structures during earthquakes. Bulletin of the seismologi-
banco di prova di una volta in muratura soggetta a carichi cal Society of America, 53 (2).
orizzontali. Technical Report n. 14. University of Brescia. Lagomarsino S., Podestà S., Risemini S., Curti E., Parodi S.
Giardina G., MariniA., Riva P. 2007c.Analisi numerica di una 2004. Mechanical models for the seismic vulnerability
volta in muratura soggetta a carichi orizzontali. Technical assessment of churches. In Structural analysis of histor-
Report n. 15. University of Brescia. ical construction (C. Modena, P. B. Lourenço, P. Roca
Giuriani, E., Gubana, A. 1993. Recupero e consolidamento Eds.), Proc. of IV Int. Seminar SAHC, Padova, Italy,
di volte in muratura. In Atti dei Colloqui Internazion- A.A. Balkema, London (UK), Vol. 2, pp. 1091–1101.
ali Castelli e Città fortificate, Storia Recupero, Valoriz- Modena C., Pineschi F., Valluzzi M. R. (Eds.) 2000. Val-
zazione, Palmanova, 24–25 settembre. utazione della vulnerabilità sismica di alcune classi di
Giuriani E., Gubana A. 1995. Extrados ties for structural strutture esistenti. Sviluppo e valutazione di metodi di rin-
restoration of Vault. In Proceedings of the Fourth Inter- forzo. Roma: CNR, Gruppo Nazionale per la Difesa dai
national Conference on Structural Studies of Historical Terremoti.
Buildings Strema 95, Architectural Studies, Materials and O’Dwyer D. 1999. Funicular analysis of masonry vaults. In
Analysis, Vol. 1, Ed. C. A. Brebbia, Wessex Institute of Computers & Structures, 73: 187–197.
Technology, UK.
421
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M. Skłodowski
Institute of Fundamental Technological Research Polish Academy of Sciences, Warszawa, Poland
ABSTRACT: The paper presents the results of testing of small microcore specimens using Compact Diagnostic
Test (CoDiT). Measurement of the evolution of strength and elastic properties of limestone is used to demonstrate
the applicability of microcores testing for the assessment of actual state of degradation of heritage construction
materials. The measurements are performed on artificially aged Portland Limestone, widely used as monuments
building material, e.g. in London. Microcore specimens are drilled out of samples which undergone the series of
cycles of accelerated Freeze and Thaw (F/T) ageing. Among the measured parameters there are the velocity
of ultrasonic waves and the flexural, tensile and compressive strengths of the material after various number
of F/T cycles. Results show that specimens drilled of the more aged sample are stronger than less weathered
material, which is in agreement with the results of the previous non-destructive testing of large samples done in
another laboratory. Based on the CoDiT image analysis, the effect is attributed to large material inhomogeneity
which is so important that it overpasses any possible material degradation introduced by the applied weathering
procedure.
423
that specimen which undergone 120 FT cycles had
better elastic properties than intact Portland specimens
(Morandini & Marini 2005).
Present research using microcore testing by the
Compact DiagnosticTest (CoDiT) procedure confirms
the previous results and shows that not only elastic
properties, but also strength properties of microcores
drilled out of these previously tested samples of arti-
ficially aged limestone are much higher. However, the
morphological analysis included in the CoDiT proce-
dure suggests that this “strengthening effect” should
rather be attributed to large material inhomogeneity
than to the FT ageing process.
424
Figure 4. Samples of Portland Whitbed with drilled micro-
core specimens.
Table 1. FT procedure.
Time
4 SPECIMENS
4.1 Weathering
Weathering of Portland Whitbed samples was done by
Figure 3. Unwrapped surface of the microcore specimen Politechnico di Torino, DITAG – Land, Environment
PW48 from the sample after 120 cycles of freezing & And Geo-Engineering Department (Bellopede & Man-
thawing.
fredotti 2006) according to the procedure presented in
Table 1.
material. Mechanical tests included in the CoDiT
4.2 Sampling of specimens
procedure shows the consequences of these various
material structures on elastic and strength properties In the present research the microcores were drilled
of the tested specimens. out of samples originally encoded as PW052 which
425
undergone 60 FT cycles and PW128 aged by 120 FT cut to flatten their bases and to obtain a short core
cycles. Microcores were drilled with their axis parallel of the length equal to the microcore diameter and a
to the limestone bedding plane. This allows to per- long core having the length equal to twice the diam-
form the measurements of the P-wave measurement, eter. After that, the propagation velocity of ultrasonic
the uniaxial compression strength and the indirect ten- P-wave along the longer sub-cores was recorded. As
sile strength with deformation localisation along the the last two steps, the destructive tests of both sub-
bedding plane. Flexural strength could be measured cores were done. Smaller cores were compressed along
perpendicular to the bedding. their diameter coinciding with the bedding plane to
As the first step, a set of lines parallel to bedding measure the indirect tensile strength (ITS) of the
planes were marked on PW052 and PW128 samples to material and longer cores were used for uniaxial
allow the unique identification of the direction of loads compressive strength measurement (UCS).
applied in measurement procedure. The second step Strength measurements are done using a specially
before sampling was the measurement of the velocity designed, hand operated, screw driven, mechanical
of propagation of ultrasonic surface wave (Rayleigh loading device. Recorded load-displacement curves
wave) along the marked vertical lines and perpendic- are based on strain gauge and LVDT sensor mea-
ularly to them using edge probes (Skłodowski 2005). surements, respectively. Analog signals are amplified,
After this measurement, seven long microcores were digitized online and stored in the connected PC.
drilled from each sample of limestone. Specimens
length was about 27 mm and their diameter between
6 RESULTS AND DISCUSSION
6 and 6.5 mm.
Reference samples of non-weathered limestone are
CoDiT results of limestone degradation assessment
the specimens reported in the previous author’s paper
presented here belong to three groups. The first group
(Skłodowski 2006c).
are the results of optical analysis included in the
Examples of each group of microcore specimens
CoDiT procedure. They are presented in the form of
are shown in Figures 1–3 revealing the various internal
unwrapped cylindrical surface images of the micro-
composition of the limestone. It has happed that these
core specimens given in Figures 1–3 and showing
various samples were used in previous FT weathering
specimens morphology.
test, each sample being frozen applying different num-
The analysis of post-critical deformation and crack
ber of cycles. Assuming that freezing process cannot
formation is also performed in each mechanical test to
change material morphology so much, it can be sad
make sure that deformation mechanisms are correct
that the weathering processes would have lower influ-
and hence that the strength measurements are reliable.
ence on material properties than differences between
Examples of these deformation modes are presented in
local properties of specimens used in each step of the
Figure 5 where cracks developed in the Brazilian Test,
FT test.
in the uniaxial compression test and in the three-point
bending test are shown for specimen PW29 after 60
5 MEASUREMENT PROCEDURE FT cycles.
It can be seen in Figure 5 that for fine grained
The microcore specimens prepared according to the Portland Limestone the deformation mechanisms in
above sampling method were subjected to the follow- all three tests are the same as in the case of testing
ing testing procedure. standard full-size specimens.
At first images of a side wall of the core were Forty two mechanical tests of microcores were
taken and combined, when necessary, into a single conducted including seven bending, seven diametral
image of the unwrapped side surface of the core. compression and seven uniaxial compression exper-
Examples of such images are shown in Figures 1–3. iments for each of the weathered Portland samples.
This allowed analysis of specimen’s morphology and Stress-displacement curves were recorded during each
material inhomogeneity. mechanical test. Their examples are presented below
Next, the line marking from the microcore base was in Figures 6–8. They are grouped in pairs to underline
elongated onto its cylindrical side wall to keep the track the differences recorded in the behaviour of the mate-
of the specimen’s orientation during the other tests. rial from the sample after 60 FT cycles of weathering
As the next step three-point bending test of the and that after 120 FT cycles.
core was performed to measure limestone flexural Load-displacement curves for bending of Portland
strength (FS). Direction of loading was the same as Limestone presented in Figure 6 show that the spec-
that of the line marked on the core. This allowed to imen drilled out of a more weathered sample can be
break the core perpendicularly to the material bedding. even two times more tough than the specimen from
Resulting two sub-cores, marked with additional let- a less aged material. Average difference in flexural
ters “a” and “b” (e.g bending of the specimen PW28 strength is also very pronounced (Fig. 9) but not as
results in obtaining specimens PW28a & PW28b) were high. It is worth to remind here that motivation for
426
Figure 5. Post-critical deformation modes developed in
microcore specimens tested according to CoDiT procedure.
Specimen diameter is approx. 6 mm.
427
Unloading-reloading curves can be used to get addi-
tional information about the stiffness of the specimens.
Let us observe that higher stiffness of specimen made
of more aged material shown at the bottom of Figure 7
can be seen even without thorough calculations. This
again confirms the previous results of non-destructive
tests.
Results from uniaxial compression tests of speci-
mens from the other sub-cores show the same tendency
which was observed in FS and ITS values. The strength
of specimens drilled out of the sample which under-
gone 120 FT cycles is much greater than the strength of
specimens from the sample weathered in 60 FT cycles.
Load-displacement curve in the upper part of the
Figure 8 is the example of the test which was discarded
due to an improper deformation mechanism of the
specimen and, in consequence, too low ultimate load.
This specimen was crushed in several places starting
from its upper base forming several short shear cracks
which reach about one third of the specimens height.
In the other uniaxial compression tests the pro-
portion between the recorded ultimate loads for both
weathered limestone specimens were close to 2:1
instead of 3:1 which could be suggested by the curves
presented in Figure 8.
The analysis is based on the measurement of five
parameters. Two of them are the indexes which char-
acterize the elastic properties of limestone. These are
the squared values of P-wave and R-wave velocities
VP2 and VR2 which are proportional the to modulus of
linear elasticity (Landau & Lifschitz 1986). The three
Figure 8. Force [N]-displacement [mm] for uniaxial com-
pression of specimens: PW28b (after 60 FT cycles) – upper, others are the measured values of FS, ITS and UCS.
and PW48b (after 120 FT cycles) – bottom. Average values of these parameters are presented in
Figure 9 where ordinate axis takes the dimension of
MPa or km2 /s2 depending on the index type.
Figure 9 illustrates in a single diagram the fact
that the trends in all the measured elastic and strength
properties are the same, thus showing that this is a pre-
dominant tendency. The presented results are average
values recorded for all the tested microcore speci-
mens from the weathered samples. It is easily seen
that one sample of the limestone has much better
properties than the other one. It is also clear that
the differences are so meaningful that they overpass
any possible material degradation introduced by the
applied weathering procedure. A possible explanation
of this phenomenon can be attributed to the material
morphology and internal composition of the speci-
mens.To discuss the problem let us look at Figures 1–3.
The specimen PW48 from Figure 3 is an example of
Figure 9. Average values of Portland Whitbed mechanical
specimens machined from the limestone sample which
properties measured for material samples artificially aged by
freezing and thawing. undergone ageing of 120 FT cycles. Its internal struc-
ture is dense with a sparse distribution of small and
hard inclusions of shells. It resembles a good concrete
material. This time the strength ratio between both material – of course in another scale.
specimens is less pronounced than in the case of bend- Specimen PW28 from Figure 2 is machined from
ing and the average values of diametral compression a sample which undergone medium ageing of 60 FT
ITS are even more close to each other. cycles. The unwrapped surface of the specimen shows
428
a large shell inclusion which acts as a strengthening ACKNOWLEDGEMENTS
element within the specimen on the one hand but also
as a stress concentrator on the other.As the result of this The measurements described in this paper were con-
inhomogeneity, a local increase of flexural strength is ducted using the same samples of Portland Whitbed
observed because the shell stripe is almost parallel to Limestone which were tested in Politecnico di Torino
stress trajectories in bending. Also the P-wave time of within the EU 6 Framework Program (MCDUR,
flight decreases as the wave travels the part of its dis- Contract G6RD-CT2000-00266).
tance choosing faster and easier path through the shell.
But in diametral compression and in uniaxial compres-
sion tests the specimen fracture is probably activated REFERENCES
along contact surfaces of oolite limestone and the shell
due to incompatibility of strains. Thus the strength Kasal, B. 2003. Semi-destructive method for in-situ evalua-
tion of compressive strength of wood structural member
increase is smaller than in the more homogeneous Forest products journal 53, no. 11–12, pp. 55–58.
PW48 specimen. Kasal, B., Drdácký, M. & Jirovský, I. 2003. Semi-destructive
Specimen PW6 shown in Figure 1 is machined from methods for evaluation of timber structures, in: Proceed-
an almost homogeneous sample of limestone com- ings STREMAH VIII, Series: Advances in Architecture.
posed mostly of calcite. This sample was not exposed Vol. 15, pp. 835–842, WIT Press, ISBN 1-85312-968-2.
to frost ageing but a relatively low strength of this Tiano, P., Filareto, C., Ponticelli, S. & Ferrari, M., and Valen-
intact material results from its mineralogical compo- tini, E. 2000. Drilling force measurement system, a new
sition with predominant weak calcite grains and no standardisable methodology to determine the “superficial
strengthening effect due to almost complete lack of hardness” of monument stones: prototype design and val-
idation. Int. Journal for the Restoration of Buildings and
shell inclusions. Monuments, 6(2), pp. 115–132.
Skłodowski, M. 2005. Application of ultrasonic Rayleigh
wave to testing of masonry materials. In: C. Modena,
7 CONCLUSIONS P. B. Lourenço & P. Roca (eds.), Structural Analysis of
Historical Constructions, pp. 395–400. London: Taylor &
The main conclusion is that the CoDiT method con- Francis Group.
firms the results of previous research showing that the Skłodowski, M. 2006a. Compact Diagnostic Test: outline of
sample which undergone 120 cycles of FT weathering historical monuments testing procedure. IFTR Reports,
has the best mechanical properties. However, the opti- 3/2006, Warszawa: IFTR.
Skłodowski, M., 2006b. Microcores technology for testing
cal analysis included in the CoDiT procedure suggests of historical masonry materials. In J. Radić, V. Rajčić,
that this effect should be attributed to morphological R. Žarnić (eds.), Proc. Int. Conf. HERITAGE PRO-
composition of the samples of limestone rather than to TECTION, constructional aspects, pp. 205–211, ISBN
the weathering processes. 953-95428-1-2
Degradation of Portland Limestone due to consec- Skłodowski, M. 2006c. Quasi-non-destructive testing of
utive freezing and thawing is much lower than the historical structural materials using microcores. In
material inhomogeneities. These inhomogeneities are P.B. Lourenço, P. Roca, C. Modena, S. Agrawal (Eds.),
large both in the sense of morphological inhomogene- Proc. 5th Int. Conf. Structural Analysis of Historical
ity and also local variations of elastic and strength Constructions, New Delhi 2006, 6–8 Nov., ISBN 972-
8692-27-7, pp. 859–865, MACMILLAN 2007.
material properties. Bellopede, R. & Manfredotti, L., 2006. In Fort, Alvarez
The Compact Diagnostic Test, unlike other research de Buergo, Gomez-Heras & Vazquez-Calvo (eds), Her-
methods used for assessment of material degradation, itage,Weathering and Conservatio,. pp. 539–546. London.
provides the information on the actual elastic and Taylor & Francis Group, ISBN 0-415-41272-2.
strength material properties and shows that they have Morandini, A. F. & Marini, P., 2005. unpubl. In Deliverable
usually a similar trend of qualitative correspondence. 10, Weathering Charts for AA Tests, Effect of the weath-
The advantage of the CoDiT method is that the small ering on stone materials: assessment of their mechanical
size of the specimen makes it possible to look at the durability, EU Research Project G6RD-CT2000-00266.
material structure from a relatively close perspective. Landau, L. D. & Lifschitz, E. M. 1986. Theory of Elasticity,
3rd rev. enl. ed. Oxford, England: Pergamon Press.
This makes it possible not only to state the mechan-
ical properties of the material, but also to see what
morphological structure gave rise to these properties.
429
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Raffaele Landolfo
University of Naples “Federico II”, Department of Construction and Mathematical Methods in
Architecture, Naples, Italy
ABSTRACT: Nowadays, in Italy historical metal structures represent a significant and important part of
national architectural heritage. In fact, their intrinsic architectural and technological value represents a milestone
in the history of constructions, being an attestation of great advance of technology and innovations reached more
than two centuries ago. These structures usually show a spread damaging mainly due to the corrosion and the
structural inadequacy. In particular, among the structural deficiencies the insufficiency of riveted connections
can be easily recognized. Hence, in order to characterize the mechanical behaviour of riveted connections, an
experimental and numerical research campaign has been planned. In this activity, riveted joint specimens have
been designed to be representative of the connections of the Italian existing structures, such as roofing structure
and steel railways bridges. This paper briefly summarizes the whole research activity and underlines the crucial
themes to be deeply investigated.
431
Figure 2. Replacement of a riveted connection of an Italian
Figure 1. The structure under examination: the roof struc- existing railway bridge.
ture of Gallery Umberto I (a); the railway bridge on Gesso
River (b).
432
internal tension results in the rivet. The direct conse-
quence of this phenomenon is its contribution to the
slip resistance of the joint just as do high-strength
bolts. However, the clamping force in the rivet is
difficult to control, is not as great as that developed by
high strength bolts, and cannot be relied upon. Upon
cooling, the rivets shrink diametrically as well as longi-
tudinally. The amount of shank-to-hole clearance that
results also depends on how well the rivet filled the
hole prior to shrinkage.
Having lightened the rivet process put in evidence
in which terms the Installation of hot-driven rivets
involves many variables, such as the initial or driv-
ing temperature, driving time, finishing temperature,
and driving method. Hence, it is of great importance
investigating in this field.
Moreover, a numerical study of the specimens to
be tested has been carried out. FE analyses have been
carried out involving a number of complexities, such
as material and geometric nonlinearities and contact–
separation modelling. In particular, in order to properly Figure 4. The plates sampled from Dome of Umberto I
plan the experimental tests the numerical activity has Gallery, before and after the superficial smoothing down:
been primarily concerned with predicting the ultimate 6 mm thick plate (a,b); 3 mm thick plate (c,d).
load-carrying capacity. After testing, the main objec-
tives of the FE analyses will be to closely examine
the performance of the specimens and to obtain, on
the basis of the static tests performed. In particular,
the load–deformation characteristics of the connec-
tions and the nature and magnitude of stresses resulting
in the different components are studied. In this phase,
the effect of clamping force in the rivets due to the
pre-stressing of the rivet processing on the response
of the connections has been also examined.
2 EXPERIMENTAL ACTIVITY
Figure 5. The stress-strain response of plates sampled from
2.1 Tests on materials Dome of Gallery Umberto I.
In order to characterize the material properties of
ancient riveted connections, several material spec- The chemical analyses revealed a carbon percentile
imens have been sampled from the Dome of the lesser than 0,1%. In fact, in the plate of 3 mm of thick-
“Umberto I” Gallery and from a dismantled railway ness the carbon content is about 0.085%, while it is
bridge dated from 1894 (coeval to the Dome of the 0.045%.in case of 6 mm thick plate.
“Umberto I” Gallery). Moreover, tests on steel plates The mechanical behaviour of these plates is illus-
and steel rivets having the same nominal properties of trated in Figure 5. In particular, the tests results
the riveted connection specimens have been performed revealed that the 6 mm thick plate does not show a clear
(D’Aniello et al. 2007). yielding, with a brittle behaviour (maximum strain of
In details, two plates have been sampled from the 6‰), while a more ductile behaviour was shown by
Dome: a 6 mm thick plate sample from the main arch the 3 mm thick plate (maximum strain of 10%).
of the Dome and 3 mm thick plate from the internal The plates and the rivets constituting the material
ring at the Dome basement (respectively Figures 4a,b of riveted specimens, that will be the core of the ongo-
and 4c,d). For these plates, the Brinnel hardness tests, ing study, have been obtained by materials stored by
chemical analyses and axial tensile tests have been car- RFI and they are dated from the ’60s of last century
ried out. The hardness tests showed an average value (hence, coeval to Bridge on Gesso River). In particu-
of 158 Brinnel index for 6 mm plates (with a standard lar, rivets were made from bar stock by hot- forming
deviation of 5,01) and 160 for 3 mm plates (with a the manufactured head. The heads were made of the
standard deviation of 3,62). high button-type.
433
Figure 6. Coupon sampled from a plate of riveted specimen
under testing (a); rivet coupon under testing (b).
434
Figure 10. The railway bridge (a) and the sampled riveted
joint (b, c, d).
435
3 PRELIMINARY NUMERICAL MODELLING
436
Figure 15. Simulation of the experimental load pattern.
437
Figure 21. Out-of vs. in-plane displacement of examined
riveted specimens.
4 CONCLUSION
438
it was recognized that a significant source of structural – Eng. Antonio D’Aniello, Department director of
deficiency could be found in the old rivet connec- the Italian railways (RFI), to the cooperation and
tions. Therefore, this paper deals with experimental assistance in the manufacturing of riveted connec-
and numerical works the objective of which aims at tion specimens and for giving the opportunity to
deepening the structural behaviour of riveted connec- study the railway bridges.
tions. In fact, a wide experimental activity on historical
riveted connections has been planned and, here, it
was summarized and presented. The first experimental
REFERENCES
results on materials have been summarized. Moreover,
the results of the predictive numerical analyses per- ABAQUS, Inc. User’s Manual. ABAQUS Standard V6.5.
formed on some representative specimens have been ABAQUS, Inc.; 2005.
shown. D’Aniello M., Fiorino L. & Landolfo R. (2007). “Riveted
connections in historical metal structures: analysis and
testing”. Proceedings of XXI C.T.A. Conference, Catania
ACKNOWLEDGMENTS (Italy) 1–3 October.
Fisher J. W. & Yoshida N. (1969) “Large Bolted and Riv-
eted Shingle Splices,” Journal of the Structural Division,
The following subjects are gratefully acknowledged:
ASCE, Vol. 96, ST9, September.
– PRIN prot. 2005087058_004 “Vulnerability and Guerrieri M.R., Di Lorenzo G. & Landolfo R. (2005). Influ-
reversible consolidation techniques for historical ence of atmospheric corrosion on the XIX century “iron”
metal structures” for having provided its financial structures: assessment of damage for Umberto I Gallery
in Naples. Proceedings of XX CTA Conference, Ischia
support;
(Italy) 26–28 September.
– PROHITEC project “Earthquake Protection of Shoukry Z. & Haisch W. T. (1970). “Bolted Connections with
Historical Buildings by Reversible Mixed Tech- Varied Hole Diameters,”Journal of the Structural Division,
nologies”, co-ordinated by Prof. F.M.Mazzolani; ASCE, Vol. 96, ST6, June.
439
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The study focused on the experimental investigations and numerical analyses regarding the
staircases in the interior of two historical buildings (named Caffa and Metellino) in the refurbished area of the
docks of Genoa harbour (Italy). Those stairways are made up of stone elements (pink granite of Sardinia Island)
and show a particular constructive technique that allows the monolithic steps to be linked together, providing an
overall behaviour among steps of a flight and among flights. This technique, suggested by the prescription (or
rule of thumb) of various ancient building manuals, allows the stress field to be distributed on each element and
it was modelled by means of FEM, in order to verify the effect of this technical measure relative to various load
cases. The numerical simulation, carried out through a detailed non-linear solid model of the flights of steps,
was preceded by an exhaustive diagnostic campaign (in situ and laboratory tests), useful in order to define, both
through direct and indirect tests, the mechanical parameters to be adopted. The obtained results were crucial
to preserve the staircases without performing any structural intervention that would have been useless (if not
harmful), modifying the architectonical and historical value of the original structure of the stairways.
441
Figure 1. The historical dock area in the Old Port of Genoa.
442
3 THE STAIRCASES OF CAFFA AND
METELLINO PALACES
443
Table 1. Geometry of the stone step.
Dimensions
444
Table 2. Results of compression tests.
445
1907 2907 3907 4907 5907 1586 2086 2586 3086 3586 4086 4586
0.0018 0.0014
0.0016 experimental data experimental data
0.0012
0.0014
lognormal distribution d = 140 cm lognormal distribution
normal distribution 0.0010 normal distribution
0.0012
0.0010 0.0008
0.0008 0.0006
0.0006 0.0004
0.0004
0.0002
0.0002
0.0000 0.0000
1586-1818
1818-2049
2049-2280
2280-2511
2511-2742
2742-2973
2973-3205
3205-3436
3436-3667
3667-3899
3899-4130
4130-4361
4361-4593
4593-4824
4824-5055
1907-2216
2216-2524
2524-2833
2833-3141
2833-3142
3449-3757
3757-4066
4066-4374
4374-4683
4683-4991
4991-5300
5300-5608
5608-5917
5917-6225
6225-6534
ultrasonic velocity (m/s)
ultrasonic velocity (m/s)
Figure 9. Probability density function of direct ultrasonic Figure 11. Probability density function of indirect ultra-
velocity of Metellino palace. sonic velocity, sensor distance 140 cm, of Caffa palace.
5 NUMERICAL ANALYSES
446
Table 4. Element types in the FEM model.
447
5.2 Scheme of the numerical simulations Table 5. Values of maximum (σI ) and minimum (σIII )
principal stresses in Models A and B.
Two analysis phases were carried out: in the first one,
the load cases prescribed by the technical rules (ulti- MODEL A MODEL B
mate and serviceability limit state) were imposed and,
in the second one, the behaviour related to incremental σI 1.03 MPa 1.76 MPa
loads was studied. σIII 1.88 MPa 2.67 MPa
Phase 1: on the two models (A and B), the following
analyses were performed:
LC1 (ULS): 1.4 Gk + 1.5 Wk1
LC2 (SLS): Gk + Wk1
where Gk is the dead load and Wk1 is the live load
(equal to 4 kN/m2 in case of common stairs);
– Phase 2: on the two models (A and B), the analyses
carried out concerned the monotonic increase of
two load typologies (uniform pressure on every step
and transversal line load on the central step), after
the dead load assignment.
It has to be noted that, in case of Phase 1, the
behaviour is very far from the inelastic range; so, it
was verified that the action assignment in two subse-
quent stages or the concurrent loading did not lead to
different results (i.e., the effect superposition is still
correct).
In both cases, live loads were assigned to the
walking surface of the steps.
448
Figure 14. Cracking pattern in case of Wk1 = 2 kN/m2 :
axonometric view from above (a) Model A; (b) Model B.
Figure 15. Model A, stress state in case of Wk1 = 72 kN/m2 :
(a) maximum principal stress σI [N/m2 ]; (b) minimum prin-
lateral section near the supports, the behaviour seems cipal stress σIII [N/m2 ].
instead to be more affected by torsional effects.
The strong interaction among the steps of one flight,
revelead by the formation of the strut in the numerical in this case, the stress peaks are located near the
simulations, confirms once more the static intuitions constraints (Fig. 15).
of the ancient builders and engineers. Observing the failure pattern, it can be noted that,
in Model A (as Rondelet wrote), the described con-
5.3.2 Phase 2: incremental analysis structive technique leads these elements to be more
After the dead load assignment, two load typologies effectively linked together.
(uniform pressure on every step and transversal line The force transfer through the moulding is more evi-
load on the central step) were monotonically increased. dent in respect to the other model, in which the rotation
of the ends of the step is allowed. In fact, the lower
5.3.2.1 Uniform pressure steps are more loaded in compression in correspon-
With reference to uniform live load, even the nominal dence of the mid-span point than the upper ones; this
action amplified more than 15 times does not cause force transfer lead the vertical planes in correspon-
significant damage pattern in the steps. In case of Wk1 dence of the fixed ends to be subjected to high tensile
equal to 72 kN/m2 , only slight cracking phenomena stresses and cracking develops in the lower part of the
develop near the supports (Fig. 14). flight.
In both the models, crushing states are not evident Further increase of the uniform pressure is not
and, in fact, during the loading phase, the principal investigated, because the hypothesis of fixed con-
stress σIII has not reached the surface of the fail- straints at the ends of the step (modelling the effect
ure domain. The higher obtained values for σIII are of the masonry walls) would have been not totally
12.6 MPa (Model A) and 22.9 MPa (Model B). Even correct. In fact, higher loads in the masonry could
449
6 CONCLUSIONS
450
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
S.P. Sparks
Sparks Engineering, Inc., Round Rock, Texas, USA
ABSTRACT: This paper presents a nondestructive approach for the evaluation of wrought-iron and steel in
nineteenth and early twentieth century bridges. Many historic bridges are non-redundant structures, and require
complete evaluation of both structural capacity and the condition of the members for continued service. The
conventional approach of material characterization in older structures has relied on tensile-testing of coupons
removed from a structure to determine strength values. That approach has limitations due to large statistical
corrections that are required when using only a few samples, and because the removal of coupons from critical
members is rarely feasible, and because each class of bridge member may have distinct mechanical properties.
An approach to in-place materials characterization is presented that is based on screening for low-ductility by
examining microstructure, chemistry, and hardness.
1 INTRODUCTION
451
The goal is to have an approach that in most cases
eliminates the need for physical sampling and testing
of bridge members, and where sampling is used can
help reduce the ambiguity that comes from a limited
sampling program.
2 DUCTILITY
452
Evaluation (AASHTO 1994) and other references,
based on date of construction. These strength values
are appropriate for use in preliminary analysis. They
are necessarily conservative, and higher values may be
justified where indicators of low-ductility are absent.
Where low-ductility is suspected or where fatigue is a
risk, lower values should be used.
3 HISTORICAL DATA
453
steels (below 0.84%C). In wrought iron, these same Table 1. Suggested screening values for chemical analysis
chemicals determine many of the characteristics, but of wrought iron and steel.
the quantity and distribution of slag inclusions have an
equally important role. C P S
In-place chemical analysis is now possible with spe-
Wrought Iron∗ 0.10 0.30 0.04
cialized optical emission spectroscopy instruments,
commonly called PMI (Positive Material Identifica- Steel
tion). One such instrument is the Arc-Met 8000 by Pins∗∗ 0.20 0.05 0.04
Eyebars∗∗∗ 0.30 0.08 0.05
Metorex. This instrument is routinely used for accu-
rate in-place analysis of steel and cast iron. Given ∗
Aston and Story 1936; and Gordon 2004.
proper access, it is possible to obtain dozens of ∗∗
Campbell, 1896.
field data of chemical analyses in a single day, pro- ∗∗∗
Gayler 1889; Oberg and Jones 1918.
viding an excellent statistical basis for judging the
materials.
For wrought iron, however, the use of the PMI investigation, including metallography and hardness
instrument is problematic because of the presence of testing.
slag, and the results may not be reliable. The operator In steels, manganese should normally be at least
must obtain multiple ‘burns’ at the same spot, which six times the sulfur content. In wrought iron, the man-
requires steady support for the operator and instru- ganese content is usually less than 0.10% by weight,
ment. Preliminary field results have been obtained though it does not seem to have an adverse effect on
using reference samples of known chemical content for ductility.
comparison. However, further evaluation of the pro- Using these criteria, the author screened historical
cess is required before the method can be considered data from 37 laboratory tests of wrought iron for which
useful for field analysis of wrought iron structures. chemical analysis and reduction of area were avail-
Laboratory chemical analysis can be done on very able (Holley 1877, Kaufman & Roberts 1990, Frank
small samples, such as drilling swarf, but obtaining 1974, Sparks & Badoux 1998). Of the ten samples with
even a small sample from a critical element is risky. reduction of area below 25% (low ductility), seven
Use of a ball endmill bit will produce a smooth, (70%) were caught by the screen. This was done on
concave depression, rather than a hole. chemistry alone without the aid of microstructure or
It is essential, whether in the field or in the lab, to hardness data. Taking reported high levels of slag as
use a type of instrument that can be calibrated to ana- a measure of poor microstructure, the screen would
lyze the lighter elements, especially phosphorus and have rejected 90% of the low ductility samples. Of
sulfur, which are the key determinants in screening the remaining samples having sufficient ductility, the
for low-ductility. For steels, it is possible to make a screen only rejected about 10%.
conservative estimate of fracture toughness based on
carbon content and comparison with chemically simi-
4.3 Hardness
lar modern metals. As carbon content increases, the
ductile-to-brittle transition becomes more dramatic, The field hardness survey is an inexpensive, rapid test
and occurs at higher temperatures. that serves several purposes in a structural evaluation:
One very important parameter for wrought iron is • To screen for insufficient ductility
the percentage of phosphorus contained in solid solu- • To indicate the variability of the material within an
tion. Phosphorus increases hardness, tensile strength,
individual member
and brittleness at low temperatures. In wrought iron, • To correlate the properties of one class of members
phosphorus is partitioned between both the ferrite and
with another
the slag (Gordon 2005). Chemical tests usually report • To estimate the tensile strength of the materials
the total phosphorus in the sample, and do not distin-
guish between the metal and the slag. Some historic Quoting Freudenthal (1950, p. 539):
tests of wrought iron show phosphorus content about
“The statistical information concerning the uni-
equally distributed between the ferrite matrix and the
formity of the ‘ultimate tensile strength’ of the
slag (Higgins 1934), though Aston & Story (1936)
material obtainable from a large number of
give a representative distribution of 0.10% in ferrite vs.
simple and rapidly performed hardness tests is
0.02% in the slag. The effects of phosphorus, and its
usually more relevant . . . than the limited infor-
real distribution, is an area of valuable further research.
mation supplied by a small number of tension
Table 1 suggests values for screening wrought iron
tests.”
and steel for ductility-reducing chemical concentra-
tions. Chemical contents above the maximum values Currently, several field hardness methods are avail-
shown in the table are not in themselves cause for able for field use on bridges including the Ultrasonic
rejection, but rather indicate the need for further Contact Impedance (UCI) method, used in instruments
454
Table 2. Suggested screening values of Brinell Hardness to an applied stress at a hypothetical crack is com-
Number (BHN) for different materials. pared with the fracture toughness KIc of the material.
This approach is accepted for evaluating the fracture
Typical range Maximum toughness and critical flaw sizes in iron and steel
truss bridges (ASCE 1979). Frank (1974) examined
Wrought Iron 95–120 130∗
the fracture toughness of wrought iron as compared
Steel to selected steels to estimate the approximate scale of
Pins 120–140 145 critical flaws and found that the critical flaw size for
Eyebars 100–120 140
wrought iron is often larger than that of older struc-
Average of three tests on an area of the same member.
tural steels. The same approach has been used for pins
∗
Local values in areas of phosphorus segregation may reach and eyebars using ultrasonic testing as the primary
160 BHN. inspection technique. (Sparks & Badoux 1998) and for
riveted connections (Keller 1995). Brühwiler (1990),
Keller (1995), Grundy (2004), and Schindler (1995)
such as the Krautkramer MIC10. Regardless of the examined fracture toughness and fatigue in riveted
instrument used, the results should be presented in wrought iron construction.
Brinell Hardness Number (BHN), which is closely cor- For both low-carbon steels and wrought iron
related with strength in carbon steels. As a rule, hard- with sufficient ductility, the critical flaw sizes in
ness values should be obtained at the same locations bridge members are usually large enough to permit
as the chemical testing and metallography. visual inspection and detection with ultrasonic flaw-
The following table suggests guidance for evalu- detection equipment (Sparks 1998 2004). Where a
ating hardness survey results, showing typical and material does not pass the screening for low ductility,
maximum values for wrought iron and steel. Hardness then the assumed critical flaw size must be adjusted
numbers somewhat above the maximum values shown downward.
in the table are not in themselves cause for rejection,
but rather indicate the need for further investigation,
including chemical analysis and metallography. For 5.2 Visual inspection
example, higher values may indicate that alloy or high- Visual inspection should be the primary method used
strength steels may have been used, which could be to assess the overall integrity of the bridge and to
verified by chemical analysis. It is important to repeat identify loss of section, physical damage, and previ-
that in the case of historic metals, high strength is not ous repairs. The visual inspection serves to integrate
necessarily an indicator of quality, per se, and is to be the findings of the detailed inspections and testing.
considered suspect. All elements should be inspected visually. However,
critical zones are often concealed due to the joint con-
figuration and the presence of spacing collars, bearing
5 FLAW DETECTION
plates, etc., making it necessary to use nondestructive
test methods. Knowing the extent to which the critical
5.1 Critical flaw size
zones will or will not be visible is essential to planning
Inspection and nondestructive flaw detection is neces- the assessment.
sary on critical bridge members for two reasons: the Depending on the type of bridge, the configuration
original materials were never tested to modern stan- of the connections, and the material characteristics, the
dards for flaws, and cracks (from corrosion, ductile following techniques may be appropriate.
failure, or fatigue) may have developed over time.
Knowledge of the critical flaw size (size of crack
5.3 Ultrasonic testing
which will propagate without an increase in the applied
stresses)is important when inspecting tensile-load car- 5.3.1 Eyebars
rying members. Critical flaw sizes can be derived for Because the faces of the eyebar ends are generally not
various defects, and this information can be used to set accessible for visual inspection, it is necessary to use
detection limits for inspection, and to develop a frac- a technique for circumferential scanning of the eye-
ture control plan. The available inspection methods bar head for cracks (Sparks & Badoux 1998). The
must be capable of detecting the critical flaw size. neck area of eyebars is a potential zone of failure due
A simplified fracture mechanics analysis can be to the possible presence of laps or welds, the stress
used to estimate the critical flaw size to establish detec- concentration at the neck-to-bar transition, and strains
tion limits for inspection. To estimate the fracture imposed during manufacture.
toughness, KIc , a correlation to relate CVN (Charpy An actual eyebar head of the same method of man-
V-notch impact toughness) to fracture toughness KIc ufacture and material type is the best way to calibrate
(Rolfe and Barsom 1977, ASCE 1979, Norris 1981). the testing technique. Known flaws (side-drilled holes
In such an analysis, the stress intensity factor, K, due and notches) can be introduced into these calibration
455
Table 3. Steel properties in 1896 Parker Truss bridge.
C Mn P S
Member BHN % % % %
Rolled Sections
Floor Beam 104 0.14 0.44 0.01 0.03
Chord Channel 105 0.15 0.48 0.01 0.04
Post Channel 94 0.12 0.44 0.01 0.03
Post Channel 100 0.14 0.47 0.02 0.02
Average 101 0.14 0.46 0.01 0.03
AISI 1012 .10–.15 .30–.60 0.04 0.05
Lower Chord
Eyebars
T10W 101 0.11 0.72 0.05 0.02
T4E 121 0.09 0.70 0.02 0.04
Figure 4. Circumferential ultrasonic testing of critical eye- T5E 108 0.13 0.74 0.04 0.04
bar on 1896 Steel Truss Bridge. Average 110 0.11 0.72 0.04 0.04
AISI 1010 .08–.13 .30–.60 0.04 0.05
pieces. In wrought iron, the sensitivity of the ultrasonic Pins
testing is reduced substantially due to inclusions. Steel L12W 146 0.19 0.50 0.02 0.03
can also contain laminations, inclusions, and coarse L1E 143 0.15 0.46 0.02 0.04
grain size that can disturb the ultrasonic signal. Prior L3E 140 0.18 0.45 0.03 0.04
L4E 135 0.18 0.48 0.03 0.04
knowledge of the microstructure is essential to obtain-
ing good results. Field trials are necessary to determine Average 141 0.18 0.47 0.02 0.04
the actual level of attenuation. The transducers should AISI 1020 .15–.20 .30–.60 0.04 0.05
generally have 2.25 MHz or higher frequency to be
Chemical analysis is from field metallurgy using an ArcMet
able to resolve the necessary discontinuities. 930. BHN is Brinnell Hardness Number. AISI numbered steel
standards are shown for comparison.
5.3.2 Pins
Whether pins are critical and require ultrasonic test- macro-etching, die penetrant, and by magnetic particle
ing should be determined by analysis. Nondestructive testing. This detailed examination will reveal signs of
testing of pins is best accomplished using longitudinal lap-weld or other anomaly in the eyebar.
wave ultrasonic transducers. In most steels and occa-
sionally in wrought iron, valid results can be obtained
by scanning from one end only. In wrought iron pins, or 6 EXAMPLE: 1896 STEEL TRUSS
those with shoulders or significant wear grooves, scan-
ning may be necessary from both ends. Test several Although it is ‘conventional wisdom’that older materi-
of the pins from both ends to determine whether valid als were highly variable, more so than modern metals,
results can be obtained from testing from one end only. the author has found very consistent data within
Calibration standards can ideally be made from member classes for an 1896, an 1887, and an 1881
actual salvaged bridge pins, if available. This is partic- bridge. Bridge builders were apparently selective in
ularly helpful for wrought iron, so that the attenuation material choices, which were selected based on the
properties can be included in the calibration. Also, a function of the member in the bridge. The following
standard having actual wear grooves or loss of section discussion is for a twelve-panel Parker through-truss
can assist in distinguishing these characteristics from bridge spanning 71m (234ft), originally constructed in
cracks. Alternatively, it is also sometimes possible to 1896.
field calibrate for pin inspection by introducing a side The initial material identification was based on
drilled hole or notch in the non-stressed area beyond visual observations, a simple spark test, and limited
the nut. field metallography. Using the spark test, the relative
carbon content of the bridge elements were estimated
in the following order, from lower to higher carbon
5.4 Dye penetrant and magnetic particle testing
content: eyebars, rolled sections, and pins. The spark
Dye penetrant and Magnetic Particle testing are estab- patterns for the eyebars were similar to wrought iron
lished test methods and are both well suited for or very low carbon steel (<0.08% C). The pins showed
revealing cracks in beam webs and flanges, around carbon indications similar to medium carbon steel.
rivet holes, and for qualifying eyebars. It is advisable The eyebars showed a visible weathering pattern on
to examine at least one representative eyebar head by the top edges resembling longitudinal ridges, similar
456
Table 4. Estimated allowable values for The aim of this paper has been to illustrate an
assessment: 1896 Steel Truss Bridge. approach that can be used to screen metals for
low ductility, based on the following key indica-
Yield stress tors: microstructure, chemical analysis, and hardness.
MPa (ksi)
New tools make obtaining this data much more effi-
cient. When informed by historical test data and
Eyebars: 220 (32)
Loop Rods 207 (30) in combination with detailed analysis, this approach
Pins: 317 (46) lends confidence to the evaluation. However, a good
Rolled Sections 220 (32) deal of work is needed to make such an approach
widely useful. A number of further research goals are
evident:
to what is often seen on wrought iron, where the
ridges are a result of the presence of slag filaments. 1. A program of trials to qualify the ductility screening
Under field metallographic examination, the eyebar method proposed here.
metal consisted mostly of equiaxed grains. Almost no 2. Improve portable metals identification equipment
slag fibers were visible on the wide face, but very and calibration procedures to accurately obtain
fine stringers could be seen on the edges of the eye- chemical analysis of wrought iron, in situ.
bars where there appeared to be approximately twelve 3. Develop an online catalogue of wrought iron, early
laminations. This laminated character is indicative of steels, and cast iron micrographs with associated
wrought iron; however, subsequent metallurgical test- tensile, ductility (reduction of area), hardness, and
ing, as described below, revealed that this was actually chemistry
laminated steel. 4. Expand the range of code-allowed notional strength
Based on their carbon contents, the example mate- values to include member classes (pins, eyebars,
rials should exhibit distinct ductile-to-brittle transition rivets).
as a function of temperature. For the eyebars, which 5. Develop a nondestructive in-place ductility mea-
are the most critical elements, this transition tem- surement
perature should be about −20-deg C (impact basis) 6. Research possible NDT methods for characterizing
(ASM 2004). The sulfur and phosphorus contents of slag distribution quantitatively.
all the steel elements were below the maximum values 7. Establish a library of typical bridge details, along
allowed for mild structural steels. As such, good duc- with proper methods of stress analysis, modes
tility is expected with no significant reduction in upper of failure, and guidance for determining stress
shelf fracture toughness. Furthermore, the Mn/S ratio concentration factors and critical flaw sizes.
is sufficient to fully convert the sulfur to manganese 8. Study the effect of phosphorus content in wrought
sulfide, a potential ductility reducer. iron and steel on ductility and fracture toughness at
The pin material generally meets the specifications low temperatures.
for pins proposed near the turn of the century by Cun- 9. Determine the typical range of phosphorus parti-
ningham (1896). These include limitations on sulfur tioning between ferrite and slag, and establish a
and phosphorus, and a tensile strength between 60- means of determining this in the field.
ksi and 70-ksi. The field hardness numbers indicate
that the pins have a tensile strength of approximately
REFERENCES
70-ksi. Furthermore, the pins showed a fine grain
structure without gross segregation. American Association of State Highway and Transportation
For structural analysis purposes, the following yield Officials, Manual for Condition Evaluation of Bridges,
strengths were used (Table 4). By comparison, the pre- 1994.
scribed AASHTO allowable for pre-1905 steel is 179 AASHTO Guide Specifications for Strength Evaluation of
MPa (26-ksi). Existing Steel and Concrete Bridges.
The eyebar, plate, and rolled sections are easily American Society of Civil Engineers, Repair and Strength-
weldable due to the low carbon content and absence ening of Old Steel Truss Bridges, ASCE, 1979.
American Society for Testing Materials, “Quality Standards
of contaminants.
for Wrought Iron,” Report of Special Subcommittee X on
Research, 1934.
7 CONCLUSIONS American Society for Testing and Materials, ASTM Standard
E 140 Hardness Conversion Tables for Metals.
ASM International 2004, ASM Metals Handbook, Tenth Edi-
There is a clear need for a nondestructive approach for tion, Volume 9. Materials Park, Ohio: ASM International.
historic iron and steel bridge evaluation. In most cases, p. 739.
it is not feasible or appropriate to remove sufficient Aston, James and Edward B. Story, Wrought Iron: Its Man-
material to rigorously quantify the yield stress, tensile ufacture, Characteristics and Applications, A. M. Beyers
strength, and ductility of the material. Company, Pittsburgh, 1936.
457
Brockenbrough, Roger L., Design Guide 15: AISC Reha- Kunz, F. C., Design of Steel Bridges Theory and Practice for
bilitation and Retrofit Guide: A Reference for Historic the use of Civil Engineers and Students, McGraw-Hill,
Shapes and Specifications, American Institute of Steel 1915.
Construction, 2002. Larsson, Tobias, Material and Fatigue Properties of Old
Brühwiler, E. I.F.C. Smith, and M. A. Hirt 1990. “Fatigue Metal Bridges, Licentiate Thesis 2006:26, Luleå Univer-
and Fracture of Riveted Bridge Members.” Journal of sity of Technology, Department of Civil and Environ-
Structural Engineering, Vol. 116(1): 198–214. mental Engineering, Division of Structural Engineering,
Buonopane, Stephen & Sean Kelton 2007. “Mechanical Sweden, 2006.
Properties of Wrought Iron from Penns Creek Bridge Maniar, D.R., M.D. Englehardt, and D.E. Leary 2003. Evalua-
(1886)” Dept. of Civil and Environmental Engineering, tion and Rehabilitation of Historic Metal Truss Bridges: A
Bucknell University Case Study of an Off-System Historic Metal Truss Bridge
Cunningham, A. C. 1896. “The Condition of Steel in Bridge in Shackleford, County, Texas, Research Report 1741-3.
Pins,” Transactions of the American Society of Civil Austin: Center for Transportation Research, University of
Engineers, Vol. XXXVI (December). Texas.
Federal Highway Administration, Comprehensive Bridge Norris D.M., J.E. Reaugh, and W.L. Server 1981. “A
Safety Inspection Program, “Section 15: Inspection of Fracture-Toughness Correlation Based on Charpy Ini-
Fracture Critical Bridge Members.” tiation Energy” STP743-EB. West Conshohocken, PA:
Frank, K. H., “Mechanical and Chemical Properties of ASTM International
Selected Steels Used in Bridge Structures,” Report Oberg, Erik and Franklin D. Jones 1918. Iron and Steel, First
No. FHWA-RD-75-79, Federal Highway Administration, Edition, New York: The Industrial Press.
Washington, DC, 1974. O’Sullivan, M. 2007 “Mechanical properties of wrought-iron
Freudenthal, Alfred M. 1950 The Inelastic Behavior of Engi- published at various times” End of first year PhD report,
neering Materials and Structures. New York: John Wiley University of Manchester.
& Sons. Pilkey, Walter D., Peterson’s Stress Concentration Factors,
Gayler, Carl, “Wrought Iron and Steel Eye-bars,” Transac- John Wiley & Sons, New York, Second Edition.
tions of the Association of Engineering Societies, read Rolfe, Stanley T., and John M. Barsom, Fracture and Fatigue
January 16, 1889. Control in Structures, Prentice-Hall, Inc., 1977.
Gordon, Robert B., American Iron 1607-1900, The Johns Sauveur, Albert, The Metallography of Iron and Steel,
Hopkins University Press, Baltimore, 1996, pp. 208–211. McGraw-Hill Book Company, New York, 1912.
Gordon, Robert B., “Strength and Structure of Wrought Iron,” Schindler, Hans-Jakob, “Toughness Evaluation and Assess-
Archeomaterials 2:109–137, 1988. ment of Old Bridge Steel,” Proceedings of the Interna-
Gordon, Robert B., “Evaluation of Wrought Iron for Con- tional Association for Bridge and Structural Engineering
tinued Service in Historic Bridges,” ASCE Journal of Symposium (1995): 1313.
Materials in Civil Engineering, July/August 2005, pp. Sparks, Patrick 2004. Guide to Evaluating Historic Iron &
393–399. Steel Bridges. Austin: Texas Department of Transportation
Grundy, Paul, “Fatigue Assessment of Wrought Iron Struc- Sparks, Patrick, “Wrought Iron: Properties, Evaluation, and
tures,” Steel Construction, Volume 38, Number 1, Aus- Reliability,” presented at the Association for Preservation
tralia, March 2004. Technology Annual Conference, Banff, Alberta, Canada,
Hertzberg, Richard W., Deformation and Fracture Mechanics October 1999.
of Engineering Materials, Second Edition, John Wiley & Sparks, Patrick & Marc Badoux 1998. “Non-Destructive
Sons, New York, 1983. Evaluation of a Historic Wrought-Iron Truss Bridge in
Hartle, Raymond A. et al., Bridge Inspector’s Training Man- New Braunfels, Texas”, APT Bulletin: Journal of the
ual (Washington, D.C.: Federal Highway Administration, Association for Preservation Technology, Vol. XXIX (1).
1990). Steel Construction Institute, Appraisal of Existing Iron and
Holley, A. L. 1877 “The Strength of Wrought Iron as Affected Steel Structures, SCI Publication 138, 1997.
by Its Composition and By Its Reduction in Rolling,” Tilley, Graham 2002. Conservation of Bridges. London: Spon
Trans. Am. Inst. Min. Eng., 6, pp. 101–124. Press.
Institution of Structural Engineers, Appraisal of Existing Tschorn, Gerhart 1963. Spark Atlas of Steels: Cast-Iron, Pig-
Structures, 2nd Edition, 1996. Iron, Ferro-Alloys and Metals, New York: The MacMillan
Kaufmann & Roberts 1990, “Strength and Weldability Eval- Company.
uation of the Wrought Iron Beams of the US Capitol Twelvetrees, W. N., Structural Iron and Steel. A Text Book
Terraces,” ATLSS Report No. 90-10, Lehigh University. for Architects, Engineers, Builders, and Science Students,
Keller, Andreas, et al., “Assessment of a 135Year Old Riveted Whittaker & Co., London, 1900.
Railway Bridge,” Extending the Life Span of Structures, Walton, Charles F. (ed.) 1971 Gray and Ductile Iron Castings
Proceedings of the International Association for Bridge Handbook. Cleveland: Gray and Ductile Iron Founders
and Structural Engineering Symposium, San Francisco, Society, Inc.
1995.
Kirkaldy, David 1862. Results of an Experimental Inquiry
into the Tensile Strength and Other Properties of Various
Kinds of Wrought-iron and Steel. Glasgow.
458
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M. Piazza
University of Trento, Trento, Italy
ABSTRACT: An accurate geometric and mechanical evaluation of two king-post trusses, based in grading
results with data gathered from NDT as well as from mechanical evaluation, followed by full-scale tests were
performed. The trusses were reassembled in laboratory and submitted to a series of cyclic tests under symmetric
and asymmetric loading. Strengthening techniques evaluated in precedents research steps were used in a second
phase of the carrying tests.
459
193 70 50 73 evaluations methods were conducted: Sylvatest, using
ultrasonic waves, and the mechanical evaluation of the
62 60
modulus of elasticity.
212 210 75 109
3.3.1 Hardness tests
75 41 In the Turrini-Piazza tests, hardness is estimated by
195 65 80 50
measuring the load force R required to embed a 10 mm
φ steel hemispherical bit to 5 mm (Turrini & Piazza,
(a) (b) 1983). The value of R was obtained by averaging the
tests results made in the four longitudinal faces of the
Figure 2. Comparison between the geometric properties of element, in numbered sections every 80 cm wide. Each
the cross-sections: (a) supposing rectangular and (b) with test consisted in five measures taken in each tested
detailed measurements (dimensions in millimetres).
area. The result of each test was obtained by averaging
the three middle values among the five measures. It
Table 1. Visual strength grading according to UNI 11035.
means that, for each considered section: 9 for the tie
Truss Element Strength class E0,mean (MPa) beam, 5 for the rafters and 2 for the struts and king
post; twenty tests were done (four by surface). At the
1 K10 S1 12000 end, nearly 1000 hardness tests were performed. The
T10 S2 10500 correlation between R and elastic modulus in longi-
S11 S1 12000 tudinal tension E0 is expressed by the experimental
S12 S2 10500 equation:
R11 S2 10500
R12 S1 12000
2 K20 S1 12000
T20 S1 12000 where δ indicates a reduction factor, 0.5 ≤ δ ≤ 0.8, that
S21 S2 10500 depends on the defectiveness of the element.
S22 S3 9500
Because the Equation 1 is experimentally founded,
R21 S1 12000
R22 S1 12000 its reliability depends on the limits of the experi-
mentation itself: namely on the species of the tested
samples that are Silver fir (Abies Alba) and Larch
(Larix decidua), on loading conditions (700 ÷ 3000 N)
Each element, every 40 cm, was marked, named and and moisture content in the range 12%–14%. However,
the cross-section measured. Figure 2 presents the due to the large amount of hardness test performed and
cross-section A of the tie beam of the Truss 1, as since mechanical determination of the elastic modulus
example of the difference that can be obtained if an of some elements (tie-beams and rafters) is available,
inadequate measurement of the cross-section is done. it was possible to calibrate the equation for the case o
Spruce (Picea Abies Karst.). The value of δ = 0.6 was
founded. After the calibration of the δ value, it was pos-
3.2 Visual grading sible to calculate the elastic modulus of all elements
The grading criteria according to the European (EN that compose the two king post trusses.
518:1995) and the Italian national standards (UNI
3.3.2 Pilodyn
11035-1-2:2003) have been applied. The species of the
The Pilodyn method uses a steel pin of a fixed diam-
two king-post trusses analyzed is Spruce (Picea Abies
eter driven into the material by a dynamic force. The
Krast.), from North Italy, accordingly, grading criteria
depth of penetration is correlated with material density.
referred to the group ‘softwood 1’ (conifere 1) in UNI
Görlacher (1987) and Feio (2005) developed relation-
11035 §5.3 have been adopted. For visual grading, it
ships between the depth of penetration of a standard
was necessary to record the moisture content and to
pin and the density of the wood. Those correlations
calculate density.
depended on the number of measurements, species
and are affected by moisture content. For all elements
3.3 Non-destructive tests (NDT) that compose both trusses, in each section used for the
geometric assessment, the penetration depth obtained
Non-destructive tests based on local and global evalua-
by the Pilodyn in the 4 faces, was measured. Table 2
tion of the material were performed. Resisting drilling
summarized the Pilodyn tests results obtained.
obtained by means of the Resistograph® device, the
penetration depth achieved with the Pilodyn® and the 3.3.3 Resistograph
hardness Turrini-Piazza tests results are examples of The use of a small-diameter needle-like drill was intro-
techniques based on local evaluation. After, global duced by Rinn (1992). The cutting resistance of a
460
Table 2. Depth penetration means values for Pilodyn. Table 3. Sylvatest results using the direct method.
461
Table 4. Elastic modulus (MoE) obtained by 13500
MoE = 0.624MoE(Turrini-Piazza) + 4655.4
means of four-point bending tests. 13000
2
R = 0.6197
462
60
Load target
50
25%
50%
Load (kN)
75% 40
Load
FDx
30 FSx
20
Self-weight
10
Time (a) Symmetric loading history
Displacement (mm)
FSx
R7 R8 FDx 10
Ai - Displacement
R5 R6 Ri - Rotation
8 A7
A6 R3 R4 A7
A1
R1 6 A6
A5 R2
A2 A3 A4 4
2
Figure 5. Instrumentation layout. 0
10
frame fixed to the reaction slab of the laboratory, were 8 A3
used for applying the loads. As boundary conditions, 6 A5
two supports restraining the vertical displacement 4
and the right support restrained also the horizontal 2
movement were applied. 0
4.2 Evaluation of the trusses behavior Figure 7. Behaviour of the connection between the
king-post and the tie beam under symmetric loading
The response of the timber trusses under the carrying conditions.
tests performed is highly depending of the variabil-
ity, previously detected and reported, along the truss 2
members in terms of cross sections area and princi- A1
Displacement (mm)
1.6
pal moments. Moreover, it is recognized the difficulty
associated with tests over existing structures, in which 1.2
the members can present residual deformations and
signs of decay and/or the connections are not well 0.8
tight. The tests results shows that even under symmet- 0.4
ric loading conditions, the behavior of both trusses
is asymmetric. This conclusion is more evident in 0
the response of Truss 1 under symmetric loading
corresponding to SLS, Figure 6. Figure 8. Horizontal displacement recorded in the support
The connection between the king-post and the tie under symmetric loading.
beam, made by a steel rod of 20 mm diameter fixed
in the king-post, aligned in the longitudinal direction, Under asymmetric loading conditions, the asym-
passing through the tie beam, with a nut at one end, metric behavior of the trusses, already pointed out, is
fastening the two elements, performs adequately. The emphasized. Channels A6 and A7 shows asymmetric
king post suspends the tie beam, Figure 7. displacement, Figure 9.
As a consequence of the truss deformations under This asymmetric behavior of the trusses introduces
the vertical loads applied, horizontal displacement at bending stresses in the rafters and tie beam. Because
the free support was recorded, Figure 8. of the rotational stiffness offered by the king-post/tie
463
40
35 A7
Displacement (mm)
30 A6
25
20
15
10
5
0
-4
A2
-6
A4
-8
-10
-12
464
Figure 14. Strengthening of Truss 2. Screws and binding
strip applied in the frontal part of the step in the rafter/tie
beam connection.
465
the surface resisting to shear stresses is insufficient.
The double step connections of the right side (over the
fixed support) failed as a consequence of a first shear
failure in the rear step.
Once the rear step fails, the stresses concentrated in
the frontal step and the failure is reached, Figure 16.
5 CONCLUSIONS
466
has been carried out with a partial financial contri- Görlacher V.R. 1987. Zerstörungsfreie Prüfung von Holz: ein
bution of the Italian Earthquake Engineering Labora- in Situ Verfahren zur Bestimmung der Rohdichte. Holz als
tory Network (RELUIS), within the research program Roh- und Werkstoff, Vol.45, No.7, 273–278.
carried out for the Italian Agency for Emergency Kotlínová M., Kloiber M., Vasconcelos G., Lourenço P.B. &
Branco J.M. 2007. Non-destrcutive testing of wood
Management. structures. Report E-12/07. DECivil, University of
Minho, 63 pp.
Lear G.C. 2005. Improving the assessment of in situ
REFERENCES timber members with the use of non-destructive and
semi-destructive testing techniques. M.S. Thesis – North
Branco J.M., Cruz P.J.S., Piazza M. & Varum H. 2005. Exper- Carolina State University, Raleigh, 137 pp.
imental Analysis of Birdsmouth Joints. Report E-11/05. Rinn F. 1992. Chancen und Grenzen bei der Untersuchung
DECivil, University of Minho, 49 pp. von Konstruktionshölzern mit der Bohrwiderstandsmeth-
EN 408 2004. Timber structures. Structural timber and glued ode. Bauen mit Holz, (9).
laminated timber. Determination of some physical and Sylvatest 1991. Sylvates user’s manual. Cableries &
mechanical properties. CEN European Committee for Trefileries de Cossonay SA. Switzerland. 19 p.
Standardization. Turrini G. & Piazza M. 1983. Il recupero dei solai in legno.
Feio A.O. 2005. Inspection and Diagnosis of Historical Esperienze e realizzazioni. In Recuperare, Vol. 7.
Timber Structures: NDT Correlations and Structural UNI 11035-1 2003. Structural timber – Visual strength
Behaviour. PhD Thesis, University of Minho. grading for Italian structural timbers: terminology and
Frankl J., Kloiber M. & Bryscejn J. 2006. Non-destructive measurement of features. UNI – Milano, Italy.
inspection of a historical wooden structure damaged UNI 11035-2 2003. Structural timber – Visual strength grad-
by fire. In Engineering mechanics 2006, Svratka, zech ing rules and characteristics values for Italian structural
Republic, 2006, pp. 62–63. timber population. UNI – Milano, Italy.
467
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M. Madhoushi
Department of Wood & Paper Sci. and Indus., Gorgan Uni. of Agri. Sci. & Nat. Res., Gorgan, Iran
J. Eimanian
Iranian Cultural Heritage Organization, Gloestan Head Office, Gorgan, Iran
M.P. Ansell
Department of Mechanical Engineering, University of Bath, Bath, UK
ABSTRACT: The House of Bagheri is one of the most important buildings in Gorgan city (North of Iran)
which is a tourist site and the House is a Cultural Heritage Construction. Solid wood was utilized extensively
for structural members in association with other traditional materials. It has been exposed to natural environ-
mental conditions for a long time, so some faults have developed including decay of the wooden members.
The mechanical properties of decayed members have been measured in laboratory tests and the buildings have
been non-destructively evaluated. Samples from the buildings were prepared which are representative of the
whole structure. Flexure and compression tests were then performed on these samples. The results of NDE
(nondestructive evaluation) tests on the samples were correlated with those on the buildings.
1 INTRODUCTION
469
Figure 2. A plan of one building of the House of Bagheri.
Figure 3. Measuring the transmission time of stress waves.
470
Table 1. The measured transmission time of samples and Table 3. Comparison between stress wave NDT, laboratory
their condition. examination and visual inspection.
No 1∗
2 3 Baseline (µs) Condition Stress wave NDT 50 12.5 12.5 12.5 12.5
Lab Examination 50 25 – – 25
1-h∗∗ 333 357 341 650 TD∗∗∗ Visual Inspection 25 – – 50 25
1-s 863 829 888 143 TD
2-s 634 678 618 158 TD
3-h 567 599 584 650 SD
4-h 3891 4587 3125 52 UU of investigation.Table 3 proves that there is a difference
5-h 432 489 473 585 NS between visual inspection and scientific examination.
6-s 364 313 403 520 S
7-s 741 783 718 156 TD 3.3 Comparison of the data
8-s 282 311 272 234 SD
A comparison of the data is shown in Table 3. There
∗
Place of measurements on sample. were not any unusable samples in laboratory exam-
∗∗ ination and visual inspection, because they wouldn’t
h for hardwood, s for softwood.
∗∗∗ be used again in restoration of the building. It can be
TD: total decayed, SD: slight decayed, UU: unusable,
NS: nearly sound, S: sound. seen that if we ignore the unusable samples tested using
NDT, both the stress wave NDT and lab examination
Table 2. The mechanical properties of old wood members results are very close to each other. Both methods show
measured in laboratory. that 50% of the samples are totally decayed and 25% of
samples are nearly sound. However, it is more difficult
Condition∗ to distinguish the slightly decayed by laboratory exam-
MOR MOE E in C|| ination and these results depend on the extent of decay.
∗∗
Sample 1 2 (MPa) (GPa) (GPa) This outcome is discussed in more detail by Madhoushi
et al. (2006). Visual inspection based on personal expe-
Rafter 1 NS SD 79.12 10.46 3.16 rience produces is inherently variable and results are
Rafter 2 S SD 88.73 8.9 3.43 not to be trusted.
∗∗∗ ∗∗∗
Column 1 S TD 2.44 In general, since there is good accordance between
∗∗∗ ∗∗∗
Column 2 NS TD 3.35
Door Frame 1 NS S 79.77 9.09 10.2
the results of stress wave analysis and laboratory test
Door Frame 2 TD TD 12.67 2.12 2.15 results, both methods can be used to assess the integrity
Beam 1 TD TD 22.56 3.48 2.58 of timber during repair and restoration activities. How-
Beam 2 NS S 74.8 9.15 10.68 ever, NDT is a more economic and cost-effective
method.
∗
Evaluated (1) visually by expert carpenter based on expe-
rience, (2) after examination.
∗∗
Modulus of elasticity in compression parallel to grain. 4 CONCLUSIONS
∗∗∗
was not measured.
• The House of Bagheri is a heritage wood con-
original condition without any repair or treatment, and struction building in Gorgan in which the timber
they are not in a good condition. members contain some damage and decay.
• Stress wave nondestructive method has been used in
the evaluation of the building and it showed that 50%
3.2 Static mechanical properties
of the structural elements tested are totally decayed
The static mechanical properties measured in the lab- and decay is initiated in 25% of the structural
oratory show (Table 2) that some elements do not elements tested.
possess acceptable strength in spite of the visual eval- • Laboratory testing demonstrated that some old
uation to the contrary. For example samples from members are not of acceptable strength in spite of
Rafter 1 were considered to be nearly sound elements, their good appearance and visual inspection.
whereas in fact the MOR (Modulus of Rupture or • As much of the building as possible will be retained
bending strength) and MOE and the modulus of elas- for restoration.
ticity in compression parallel to grain (E in C|| ) have • There is good correspondence between stress wave
all been reduced by decay. In contrast, Door Frame nondestructive evaluation and laboratory mechani-
1 is sound and has acceptable mechanical properties cal tests.
in spite of its grading as nearly sound. It is clear that • It is advised that the NDT method should be
visual evaluation, even by an expert carpenter, is not used in association with visual inspections in the
accurate and using NDE methods is advised in this type assessment of this heritage building.
471
REFERENCES 11DBMC Inter. Conf. on Durability of Building Materials
and Components, Istanbul, 11–14 May, 2008.
Ceccotti, A., Faccio, P., Nart, M., Samghaas, C. and Pasanen, A.L., Kasanena J.P., Rautialaa S., Ikaheimoa M.,
Simeone, P. 2006. Seismic behaviour of historic timber Rantamakib, J., Kaariainenb H. & Kalliokoskia, P.
frame buildings. ICOMOS Intern. Wood Committee; Proc. 2000. Fungal growth and survival in building materials
15th intern. Symp., Istanbul and Rize, 18–23 September under fluctuating moisture and temperature conditions.
2006. International Biodeterioration & Biodegradation 46 (2):
Eimanian, J. 2001. Restoration suggestions on House of 117–127.
Bagheri’, MSc, Thesis, Islamic Azad University. Pellerin, R.F. & Ross, R.J. 2002. Inspection of timber struc-
Haines, D.W., Leban, J.M. & Herbe, Ch. 1996: Determination tures using stress wave timing non-destructive evaluation
of Young’s modulus for spruce, fir and isotropic materi- tools. In R.F. Pellerin & R.J. Ross (eds), Nondestructive
als by the resonance flexure method with comparisons to Evaluation of Wood: 135–148. Forest Products Society.
static flexure and other dynamic methods. Wood Science Ronca, P. & Gubana, A. 1998. Mechanical characterisation of
and Technology 30 (4):253–263. wooden structures by means of an in situ penetration test.
Lourenco, P.B., Luso E. & Almeida, M.G. 2006. Defects and Construction and Building Materials 12 (4): 233–243.
moisture problems in buildings from historical city cen- Ross, R.J. & Pellerin, R.F. 1994. Non-destructive testing for
tres: a case study in Portugal. Building and Environment assessing structures: A review. Gen. Tech. Rep. FPL-GTR-
41 (2): 223–234. 70 (Rev.). Madison, WI, U.S. Department of Agriculture,
Machek, L., Edlund, M., Sierra Alvarez, R. & Militz, H. Forest Service, Forest Products Laboratory. 40 p.
2004. A non-destructive approach for assessing decay in Ross, R.J., De Groot, R.C., Nelson W.J. & Lebow, P.K. 1996.
preservative treated wood.Wood Science and Technology Assessment of the strength of biologically degraded wood
37 (5):411–417. by stress wave NDE. In C. Sjostrom (ed.), Durability of
Madhoushi, M., Ansell, M.P. & Hashemi, M. 2006. Appli- Building Materials and Components 7 (V1): 637–644.
cation of NDT stress wave method for assessing property E & FN Spon.
changes in decayed wood of Iranian beech (Fagus orien- Tampone, G. & Messeri, B. 2006. Compliance of the practice
talis). Integrated Approach to Wood Structure, Behaviour of strengthening ancient timber structure in seismic areas
and Applications; Proc. ESWM and COST Action E35 with the official documents on conservation. ICOMOS
meeting, Florence, 14–17 May, 2006. intern. wood committee, Proc. 15th intern. Symp., Istanbul
Madhoushi, M. & Eimainan, J. (in press). Faults and Repairs and Rize, 18–23 September 2006.
in House of Bagheri: A Cultural Heritage Construction
in Gorgan (North of Iran) – A Case Study. Submitted to
472
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: In spite of the fact that there are plenty of examples of historic spatial roof structures – in spires,
castle towers, etc. – in the technical literature information about this kind of structure can rarely be found,
especially about their structural behaviour. The development of computerised 3D programs of analysis and
modelling helped us to improve our understanding of this. This paper offers a guide to assist in the understanding
of the structural behaviour of the load bearing structure of spires, offering a typology, followed by a detailed
presentation of these structural analyses.
473
Figure 2. The horizontal framing system.
474
Figure 4. The horizontal framing system.
Preventing the spire from overturning is provided 2.2.4 Other elements affecting the structural
by a similar system to that described previously, using behaviour of a spires
metal bars at the corners. Additional elements affecting the performance of
The main difference is in the bracing system acting spires are: the pinnacles at the four corners, the
against wind-loads. Supported by the framing systems, tympanum-windows at the centre of the facades,
there are lower and upper purlins, connected by scissor clock-ledges, other ornaments (especially in the case
bracings, set out at two or more levels (Figure 5). If of baroque structures). All of these elements can dis-
the spire is the typical eight sided model, these scissor turb the direction and intensity of loads and so the
bracings above the second level are set out alternately structural behaviour of the spire’s main structure.
in four-four sides. If the spire is four-sided, these braces
are set out on each face, but they can be missing at the
top level. 2.3 Numerical (computerised) modelling
Common rafters are connected at the top to the short The main aspects of the structural behaviour of the spa-
Kingpost, which is supported by the last (or last two) tial roof structures which can be resolved by modelling
framing systems. are as follows:
Regarding the construction of spires, we have found
a description in a nineteenth century carpenter’s book, a) supporting arrangements – whether the structure is
written in the period when a lot of spires of this type supported mainly at the top of the wall, or lower
were constructed: “The spires should not be very rigid; down and by the fixings or “building in” systems;
they should allow a small amount of movement during b) effectiveness of joints – whether the timber con-
storms. For this reason the beams of the framing sys- nections allow the parts of the members to rotate
tems should be screw fixed to the common rafters, but freely or provide restraints against rotation;
the roof bracing system’s beams should not be screw c) the role of the different components of the structure
fixed to the common rafters, rather they should finish (especially the Kingpost).
a few centimetres from the rafters. The basic framing
system should be well fixed into the sidewalls of the Other considerations which are not subject of this
tower. The main common rafters (spanning from the paper:
corners) are connected to the short Kingpost, which a) consideration of the applied loads, especially the
is secured firmly at the bottom of the framing system. wind-loadings – whether the loads act as the simpli-
The other common rafters are connected either into the fied European Standards states, or whether account
main common rafters, or to header beams”. [2] should be taken of the distortion effect of the
wind-loading;
2.2.3 A third type of spires can be defined as a b) the role of the additional elements: pinnacles, win-
combined structure dows, galleries, other ornaments (baroque roofs);
As the spire of Bela tower of the Mathias Church, c) the effect of irregularities and missing elements to
from Budapest (from 1896) with a long Kingpost, and the structural behaviour of the spires.
475
Figure 7. Modelling the spires with long kingpost. Figure 8. Modelling the spires with short kingpost – contact
elements.
476
Figure 10. Stresses in the Kingpost – Nx (kN) – for
Dead-loads (1)/Wind-loads (2) for fixed joints (a)/joints with
articulations (b).
2.4.2 Results
a) Spires with long Kingpost – the roof structure of
the tower of Reformed church from Barabas.
The joints have been considered in two ways:
with free rotation and with restraint against rota-
tion, but this – as you can see it below – had
only a minor effect in the stresses of the different
elements.
The other question was the role of the different Figure 11. Bending moments in the Kingpost – My (kN m) –
elements (especially the Kingpost) in the structural for Dead-loads (1)/Wind-loads (2) for fixed joints (a)/joints
behaviour of the spire. with articulations (b).
The role of the Kingpost:
In spite of the fact that the Kingpost is an
extremely slim element, with it’s cross-section very the common rafter, permitting a little movement
much weakened by the joints with other elements, during rugged wind.
the Kingpost is very important to the structure, The other elements, such as common rafters,
working for both vertical and horizontal actions compound rafters, collar beams etc. are work-
(such as roofing and wind-loads). ing – as usual – in both compression or tension
For dead-loads the Kingpost is working as and sometimes common rafters also for bending
usual – resulting only axial stresses. What is moments.
unusual is that wind-loads producing bending The spires with long Kingpost are not very sen-
moments in the Kingpost especially in the joints sitive to the rigidity of the joints. The distribution
with the compound rafters, but also at joints with of loads are very much depends of the rigidity of
knee-braces and collars. This is why the bracing the Kingpost, and the possibility of movements of
system of the spire should not be screw fixed to the bracing elements during storms.
477
Figure 15. Stresses in the scissor bracings – Nx (kN) – for
Dead-loads (1)/Wind-loads (2) for fixed joints (a)/joints with
Figure 12. Stresses in the compound rafters – Nx (kN) – for articulations (b).
Dead-loads (1)/Wind-loads (2) for fixed joints (a)/joints with
articulations (b).
The Kingpost in this case is not presenting very
much stresses neither for dead-loads, nor for wind-
loads. The role of the Kingpost is much more
constructive, than supporting the structure of the
spire – so that it is an assembling element helping
to join together the common rafters.
More important role have the horizontal framing
systems and also the scissors bracings, what are
also very sensitive to the movements:
The spires with short Kingpost are very sensitive
to the rigidity of the contact elements, and rotations
in the joints between elements, producing impor-
tant redistribution of stresses. The Kingpost has a
secondary importance; in this case primary ele-
ments are the scissor bracings and the horizontal
Figure 13. Stresses in the short Kingpost – Nx (kN) – for railing systems.
Dead-loads (1)/Wind-loads (2) for fixed joints (a)/joints with
articulations (b).
3 CONCLUSIONS
478
REFERENCES 5. Schunck, E., Oster, H.J., Barthel, R., Kiessl, K., 2003,
Roof Construction Manual. Pitched Roofs, Birkhäuser –
1. Bruzsa, L., Toth, E. 2007, Roof structures from A to Z Publishers for Architecture, Basel-Boston-Berlin, Edi-
(Hungarian), Verlag Dashöfer, Budapest, Hungary tion Detail Munich, Germany
2. Diesener, H., 1897, Carpenter works (Hungarian), Keil 6. Szabo, B. Gy. 2005, Illustrated Dictionary of His-
József Publisher, Budapest, Hungary toric load-bearing structures, Kriterion/Utilitas Pub-
3. Gabor, Laszlo, 1976, Doctrines of the construction ele- lisher, Cluj-N., Romania
ments (Hungarian), Tankonyvkiado, Budapest, Hungary 7. Yeomans, D., 1992, The Architect and the Carpenter,
4. Mohrmann, K. 1901, Gothic Structures’ Handbook RIBA Heinz Gallery, London, UK
(German), Leipzig, Germany 8. http://en.wikipedia.org/
479
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The “Circo Romano” was built by the Emperor Massimiano in the last years of the 3rd century.
It was erected with the attempt to give Milan a building capable to represent the city as a capital of the Empire.
Nowadays the only witness of such an imposing group of buildings is the Tower of the “carceres”. The intervention
planned for the “Torre quadrata del circo romano”, includes the project for static improvement and the more
complex preservation project of the upper part of the building (the loggia).
In order to prepare the project it was necessary to carry out a diagnostic campaign to verify the effective condi-
tions of the building. All the retrieved information were represented on drawings identifying geometrical survey,
image rectification and stratigraphic readings referring to the historical-archival and the brickwork discontinu-
ities. In addition, there was also representation of the readings of crack networks and of out-of-plumbs; structures
and materials non-intrusive or light-intrusive diagnostics performed by means of sonic method; checking of the
status of sleepers, tie beams and chains in order to verify their functioning; mortars and bricks sampling for the
identification of the base elements; the results of petrographical and mineralogical analysis.
1 INTRODUCTION
481
purpose of verifying the Torre’s geometrics, the status 2 SURVEY, MAPPING OF CONSTRUCTION
of the works already implemented and of providing PHASES AND VERTICALITY CONTROL
indications on the conservation status of the stone
materials and of the connecting grout in the wall In order to provide better support in interpreting the
section relating to the loggia. results obtained and, at the same time, to provide an
additional way of looking at the situation, the stratigra-
phy survey carried out in the Nineties by the architect
Gianfranco Pertot was re-examined. The stratigra-
phy, in fact, correlates quantitative and qualitative
information, utilizing the chronological succession
of the structure construction episodes as the guiding
parameter.
The geometric findings, together with the photo-
graphic straightening, was the first operation under-
taken, by which it traced accurate maps on which,
subsequently, information acquired as a result of the
other diagnostic surveys would be mapped.
A closed topographic polygon was realized (with
laser total station) composed of seven stations in which
more than seven hundred points required for tracing
the structure geometrics were measured. Thereafter,
a detailed profile was executed, performed with the
help of direct measuring instruments such as laser dis-
tance meters, rigid meters and ribs to integrate the
missing information and to study the level of detail
of certain more significant portions of the structure.
Figure 2. The topographic polygon. In addition, a self-leveling laser was used to determine
482
the difference in level between the stairwell landings In order to acquire the photographic material, a dig-
in the internal staircase. ital reflex camera equipped with standard optics (focal
Seven different outlines were realized, each one cor- length 50 mm) was used and it was also equipped
responding to a landing, in addition to that for the with tele-optics (210 mm) based on the camera sta-
ground floor and for the loggia, with five sections and tion; wide angle optics (20 mm) were used for the
four external prospects. The considerable amount of loggia shots, and the relevant deformations were
details determined made it possible to correctly high- subsequently corrected by software.
light the entity of the out-of-lines, measured on the The procedure for photo straightening and placing
four corners of the structure, as well as the opportu- the single images in mosaic order was realized with
nity to coherently represent the deformations and the dedicated software. The coordinates provided by the
bulging of the walls. topographic shots were used, as specified.
In order to correctly restore the plumb lines, also The geometric tracing and the control of verticality
taking into consideration the particular deterioration made it clear that the front presenting the most dis-
of the bricks with which the structure was realized, placement is that facing west, on the contrary to that
it was deemed advisable to execute the readings on facing east, which is nearly perfectly plumb.
the corners from very close up (approximately every The information determined on the external corners
30–40 cm for the entire height of the Torre). Suffi- of the structure is also confirmed by the measurements
cient information thus became available to identify and carried out inside the loggia. These were effectuated
isolate any mistakes in the reading or in collimation. by using a high accuracy laser plumb and once again
The topographic bearings, used for the geometric it is the west wall that shows the most displacement
restitution, were also employed as references with real from the vertical. In this case the values are more
coordinates for the photographic straightening pro- modest but, in fact, even the portion of the wall
cedure for the four external fronts and for the four analyzed is considerably less extensive than on the
internal prospects on the loggia level. external walls, where the out-of-plumb was calculated,
483
Figure 6. Stratigraphic analysis.
taking into consideration nearly the entire height of the With respect to the Torre, seven different construc-
building. tion phases were identified and eight different periods
The analysis of the situation concerning crevices of interventions were classified, to which the three
evidenced a modest presence of lesions and crevices, restoration phases, carried out from the second half
at the same time showing the existence of no-anchoring of the 19th century, must be added.
areas among the angles of the Torre and the loggia’s
supporting fill walls.
It is believed that the origin of this phenomena is not 3 DIAGNOSTIC ANALYSIS
attributable to structural settlement or to yielding, but
rather to the particular genesis of the entire structure. In addition to providing priceless information as to the
It was constructed in many phases and over a very long evolution dynamics of the structure, the stratigraphy
period of time and presents construction stratifications readings also constitute a fundamental basis for the
at times not properly solved and generated by an urgent choice of homogeneous samples to analyze the char-
need rather than by a single project. acteristics of the materials. At the same time these
On the basis of the plans realized by the photo- surveys were carried out, an analysis campaign was
graphic straightening, a stratigraphic analysis of the organized that aimed at recognizing the stone materi-
structure was outlined for the four external façades. als present in the structure and at the characterization
Thanks to a high definition photographic support, sub- of the connecting grout used for the walls in the loggia.
ject to verifying the stratigraphy unit readings, this This in-depth study is justified by the evident decay
analysis made it possible to restore the development affecting the Torre’s stone elements, especially the log-
and interpretation of the numerous construction activ- gia, where decorations can no longer be read and where
ities that followed each other during the course of time. entire sections of material have disappeared.
In fact, the stratigraphy analysis tables describe 14 samples were taken from the four façades of the
the perimeter of the stratigraphy units (S.U.) and loggia to analyse the column, plinth and stone materi-
of the architectural elements (A.E.). Through appro- als present in the masonry in the area below the loggia’s
priate background information, they communicate floor.
the construction activities in chronological sequence, Samples were analysed in a laboratory with polar-
highlighting wall irregularities and the architectural ized light microscopy (MO-LP) using traditional
structure’s critical points. methods of mineralogical analysis in “thin sections”
484
Figure 7. Localization of the different stone materials.
that entails the use of a very thin (about 0.03 mm) analyzed are in fact one different from the other: the
layer of stone glued between two slides. The material re-used stone materials originate from different urban
is observed through a polarizing microscope and the “quarries”.
analysis is conducted under different types of polarized Laboratory analyses continued, as previously men-
light in order to clearly observe all the characteris- tioned, with the characterization of the mortar beds
tics of that particular sample. These characteristics, used in the loggia and were conducted with X-rays
unique to each mineral, allow specialized technicians diffraction (XRD).
to recognize the materials on sight. Diffraction allows the crystalline planes present in
The results obtained confirm the visual analysis, the samples analyzed to be recognised and, if com-
in that they showed the great variety of the materials pared with appropriate reference tables, they identify
sampled, including: Pietra di Angera, Rosso di Verona, the elementary components of the material. In brief,
Trachite dei Colli Euganei, Pietra di Saltrio, Pietra the procedure uses X-rays that are directed towards
Aurisina and Marmo Apuano. a rotating sample and then diffracted with varying
The absence of consistent placement inside the angles according to the materials that make up the sam-
building for each litote is moreover confirmed in the ple. At the same time, a detector measures the width
context of the elements themselves. The four columns of the diffraction angles and records them on file for
485
Figure 9. Ultrasound analysis carried out on the loggia’s columns.
subsequent analysis. This method provides qualitative situation that is worse than that seen by mere observa-
and in some cases semi-quantitative results. tion. The column on the left especially shows clear
The results highlighted a rather uniform situation signs of chipping and lamination of the crystalline
both in terms of the composition (high concentrations planes of the material with a progressive reduction of
of quartz and calcite with a significant presence of pot- the resistant section.
sherds) and of the state of conservation that is not good The Torre has undergone a number of maintenance
no matter the location and morphological aspect of the and restoration works over the past two centuries that
sample under study. In particular, nearly all the mor- implemented a series of measures aimed at improving
tars are poorly consistent with pulverization in some the structure’s static equilibrium.
sporadic cases, though, on the average they show a The surveys conducted therefore focused on con-
uniform and compact structure. trolling the condition of these measures, and in partic-
It is thus difficult to identify deterioration problems ular the tension of the metal chains within the loggia
with respect to the exposition of the interior fronts as that were not located inside the walls and the presence
well as with respect to other factors external to the of steel strands within the masonry built (it is assumed)
material. The results show a general state of conserva- during the restoration works conducted in 1981.
tion that is slightly worse as regards the mortars used to The check on the chain tension was conducted
build the loggia parapet and those used for the masonry through the dynamic analyses of free vibration.
located over the arches and in the normal parameters Only the north and south chains (only two of the
relative to angle irons. four chains in the loggia) were assessed, inasmuch as
Analyses on the stone materials continued with they were the only ones detached from the masonry. In
ultrasound tests to control the compactness of the fact, this circumstance greatly compromises the reli-
elements. ability of the analysis that was consequently not been
Ultrasound tests are based on the relationships that conducted on the metal elements facing east and west.
link the propagation speed of elastic waves in a given The analysis was conducted by disturbing the chain,
material to the elastic properties of the material itself. applying an impulse and recording the accelerations
The propagation speed of these waves can be directly produced with an accelerometer placed at the center of
linked to the physical and mechanical characteristics the chain.The results acquired through a personal com-
of that material and, more to the point, a reduction puter showed the poor tension of the metal elements
of the propagation speed indicates irregular (cracks, that will undergo a tension enhancement procedure or
empty spaces, detachments) or deteriorated material. will be replaced during the restoration works.
Ultrasound tests can be carried out in three ways The last analysis consisted in checking, (using a
(by transparency, on the surface and by radial trans- magnetometer), the presence of metal consolidation
mission) according to the position of the transmitter strands which according to the projects that we possess
and receiver on the surfaces under study. were installed in 1981.
These tests were performed on the small stone The theory that the documentation on paper referred
columns in the loggia and on the huge columns in to was confirmed by the presence of markers on the
cipolin marble at the Torre entrance. They revealed building’s façades that were painted with colored paint
that the loggia elements are in good condition, despite which according to the other documents, represents the
the external surface attacked by corrosive elements actual position in which the building firm that carried
and sulphatation, and that the material is compact. On out the consolidation work had placed the strands.
the other hand, the columns at the entrance revealed Six readings were made on the external façades
low speed values and in some points the total absence (on the north and east fronts) and two on the internal
of propagation of the ultrasound waves, indicating a fronts (west façade), in order to verify the presence
486
stratigraphic facts that suggested the “critical areas” in
which the conservation project will have to focus on.
Material data and geometrical aspects that have
been surveyed provide considerable information and
attest the building’s past history, giving, at the same
time, useful hints and indications on the methods to
apply for a correct approach to the intervention project.
As a matter of fact, all the information resulting
from this campaign, correlated with the direct mea-
surements, with the monitoring of verticality and with
the diagnostic analysis, let us achieve detailed and
exhaustive answers to the complexity of the prob-
lems that affect the building. The clearness and com-
pleteness of the preventive diagnostic outline, further
strengthened by petrographical, chemical and physi-
cal analysis of materials, has allowed for the correct
development of a specific conservation and structure
Figure 10. Localization of the analysis carried out with improvement project.
magnetometer.
487
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
C. Stammers
Department of Mechanical Engineering, University of Bath, Bath UK
ABSTRACT: This paper analyses a series of shaking-table tests on three 1/10-scale 3 dimensional dry masonry
models of same plan area and different height and total mass. Sinusoidal waves varying in frequencies based
on both constant amplitudes and constant accelerations are used as input. The aim is to gain better insight
into the behaviour of masonry historic structures subjected to seismic action by using the approach of multi
rigid body dynamics. The models are relatively simple. Yet a sufficient number of blocks is used to ensure that
the fundamental interaction among parts of masonry walls connected in 3D to form buildings is replicated to
reasonable accuracy. Results are presented in terms of relative displacement, frequency and energy content of
the motion of selected instrumented block. Global results in terms of deformed shape and crack pattern of the
walls are also discussed.
489
dry work. The aim is to gain better insight into the need not to be addressed, the second is that results can
behaviour of masonry historic structures subjected to be more easily generalised to masonry structures with
seismic action by using the approach of multi rigid a variety of materials.
body dynamics. The models are relatively simple. Yet
a sufficient number of blocks is used to ensure that 2.2 Experimental setup
the fundamental interaction among parts of masonry
walls connected in 3D to form buildings is replicated Three series of models were built and tested on the
to reasonable accuracy. In the following sections the Instron Structural Testing Multi-Axis Shaking Table
background to this set of tests and the test set up (IST-MAST) system with 6 degrees of freedom, sited
are first presented, then the results discussed and a sim- at the Mechanical Engineering laboratory of the Uni-
ple control approach is proposed to provide a simple versity of Bath, UK. The control system used is
and effective modelling. the Labtronic 8800 Control Electronics with a Basic
8800 Console, control cards for each axis position
and SCM data acquisition boards-table accelerometer
2 SHAKING-TABLE TESTS monitoring and customer feedback. Models tested
have rectangular plan shape so as to collect informa-
2.1 Background tion on façades with different height to length ratio.
Panels perpendicular to y axis have 7 bricks in length,
The series of shaking-table tests presented here is while the facades perpendicular to the x axis have 5
part of an experimental program to identify dynamic bricks. They are built without windows and at corners
behaviour of masonry structures. Previous theoretic the superposition of bricks is equal to half the length
and experimental studies conducted by D’Ayala & resulting in competent connections. The bricks have
Speranza; 2003 Restrepo et al., 2004 and Shi & dimensions (L × W × H) 100 × 50 × 35 mm, and they
D’Ayala, 2006; Shi et al., 2008, have shown how using are regularly staggered to provide s/h = 50/35 = 1.4,
limit state analysis and pseudo-static tests is possible which from previous experimental and analytical stud-
to derive a consistent model of behaviour of cracking ies (D’Ayala, 2005, Shi et al., 2008) has shown to
and damage of historic masonry subjected to lateral provide good equivalent shear strength to the masonry.
action and how the behaviour can be correlated to a Model dimensions are shown in Figure 1. The first
relatively small numbers of geometric and structural and second series of models were built with 9 courses
parameters, without relying on stress analysis. While
results of these studies are also confirmed by in situ
observation of damage to buildings subjected to earth- x
quakes, their static nature fails to provide insight in
the damaging process and hence fails to accurately
quantify the strength and “ductility” resources that are
available during the hysteretic behaviour. It is argued y
here that notwithstanding the fact that the constituent
materials of masonry, bricks or blocks and mortar, are
not ductile, substantial dissipation of energy can take
place during the damaging process at the cracks inter-
faces due to sliding and rocking of portions relative to
each other, and hence via friction and impact.
As the processes outlined above are post material
capacity, both in tension and compression or shear, it (a)
is not essential to replicate the continuum nature of the
masonry. This is the main reason for adopting models
in dry masonry so that the emphasis is directly on the
post-elastic behaviour. It may be argued that the lack of
mortared joint would increase the dissipating capacity
of the scale model with respect to a real masonry struc-
ture, because of the possibility of each unit to randomly
move with respect to the adjacent one.This indeed does
not occur, as verified by the tests, because the grav-
ity load distribution is proportional to the real case
and hence actual relative movement does indeed occur
only along main lines of crack pattern. Neglecting the (b)
elastic phase has two advantages: the first is that in
scaling down the model the considerations on stresses Figure 1. Plan (a) and elevation (b) of the three models.
490
of bricks, while the third series had 15 courses to inves- each direction for Model III. Details of the transducers
tigate the difference in height to length ratio further layout are shown in Figure 3.
and to study the effect of different area to mass ratio. Using a simple sinusoidal input with 50-cycle dura-
The roof structure is simulated by five timber beams tion, each of the models is subjected to different group
set a regular spacing on the top course with weights of tests where only the amplitude to the accelera-
bolted at each end to simulate the actual roof mass and tion of the signal was increased, as summarised in
to ensure proper simulation of the constraining effect Table 1. One of the objectives of the tests is to clar-
of floor structures onto the wall. Of course the issue of ify whether for a same energy input greater levels
rigid or flexible diaphragm action is not addressed at of damage are triggered by acceleration or increased
all here. However, to obtain a more even distribution of amplitude of the motion. A second objective is to iden-
mass and constraint, especially over the corners, series tify whether the response is frequency sensitive, in
II models have extra weight applied to the corners other words, notwithstanding the fact that the speci-
for a total 20% more with respect to series I models men is not a continuum, whether something akin to
(Figure 2). natural frequency or range of frequency for which the
Besides the accelerogrammes registering the six response is enhanced, can be identified. Hence the
component of motion of the shaking table, each speci- input frequency varies following two principles: (1)
men is instrumented with 6 displacement transducers, under constant amplitude. This is used on Model I
3 for each of the x and y alignment, set on the cen- to identify resonant frequencies. (2) Under constant
tral brick and the corner bricks on each side. As the peak acceleration. To ensure this, as the input wave
bricks of the top layer are less stable without mortar, is defined in terms of frequency and amplitude, the
in order to represent the global shaking, displacements relationship between frequency and amplitude of two
of the 2nd top layer are measured. Additionally, there successive input series is (fa )2 /(fb )2 = Ab /Aa , where
is one more transducer in the middle of the 8th layer in f is the input frequency and A is the corresponding
amplitude. For Model II Series, the effect of input
1 8 y Amplitude I
2 5 Acceleration
3 5 x Amplitude
4 5 Acceleration
5 4 y Acceleration II
6 5 x Acceleration
7 7 x, y Acceleration
8 9 y Amplitude III
Figure 2. (a) Series I and II and (b) Series III set up model 9 3 y Acceleration
setup.
491
non parallel to the structure’s walls was also consid- (a) Constant input amplitude
ered by providing sinewaves of equal amplitude and 2
displacement in the two coordinate directions. 1.8 x direction
1.6 y direction
Amplification
3 EXPERIMENTAL RESULTS ANALYSIS 1.4
1.2
Both pseudo-static tests. (Shi & D’Ayala, 2006; Shi 1
et al., 2008) and extensive on-site post earthquake 0.8
observation have shown that under lateral action,
0.6
masonry walls are more vulnerable to out of plane
than in-plane failure, even for shape ratio H /L < 1, 0.4
and slenderness ratio H /t < 11, as it is the case for 0.2
this series of tests. This was confirmed by the obser- 0
vation of both the recorded videos of the shaking and 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2
the comparison of overall displacement of in-plane Input frequency
and out-of-plane instrument for any given motion test.
However differently from the monotonic static tests (b) Constant input acceleration
carried out on single walls with wings, the inversion 300
of motion caused by the sinusoidal wave results in a x direction
much stronger interaction at the corner between par- Dissipated energy (J) 250 y direction
allel and orthogonal walls with effects on the two sets
of walls which will be further discussed in detail in 200
the following subsection where results are presented
separately for each series of tests. 150
100
3.1 Analysis of results on Series I models
In Figure 4 maximum amplification of motion and 50
dissipated energy for series of test with increased fre- 0
quency and constant amplitude of 12 mm and constant 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6
acceleration of 1.53 m/s2 are shown for the two coordi-
nate directions of motion. The y direction relates to the Input frequency
out of plane motion of the longer façade. This shows
a peak amplification of 2 for 2.0 Hz, with a distribu- Figure 4. Amplification and energy response for constant
amplitude and acceleration for Series I model.
tion very similar to an amplification spectrum. For the
shaking in x, peak response is associated to an input
frequency of 1.9 Hz.and a value of 1.5, with a second
peak at 3.6 Hz. Assuming a simple 1 degree of free-
dom oscillator as basic model for the behaviour of the
walls would yield values of stiffness 29.5 KN/m and
15.7 KN/m respectively, contrary to expectations.
This can be explained by the fact that the floor struc-
ture afforded better friction restraint in the y direction
shaking, resulting in overall greater stiffness.
In the case of constant input acceleration, and hence
constant input energy, it is observed that for increas-
ing input frequency damage and collapse occur for a
smaller number of cycles. This phenomenon is sum-
marised clearly in Figure 4b, where the total energy
dissipated in each test is measured against the value of
input frequency. Figure 5. Damage on side panel of Series I models with
Figures 5 and 6 show that similar portion of the walls excitation in y as 2.2 Hz, 12 mm.
are excited in both directions for each peak amplifica-
tion case. In the photos the panel parallel to the motion line shows the relative sliding of blocks in the direc-
are shown. The thick lines at the edge of the panels tion of motion. It should be noted that these two crack
shows the portion of this that participate to the out of patterns relates closely to the theoretical assumptions
plane motion of the orthogonal walls while the thinner taken in D’Ayala & Speranza, 2003 to derive out of
492
x-1.9Hz 7th cycle
2
positive peak
middle point
negative peak 1
positive peak 2
0
-18 -12 -6 0 6 12 18
Panel // y
-1
-2
-3
493
Table 3. Difference in input – output phase angles of When the maximum response amplitude exceeds the
Model I shaking in y direction. input amplitude the top starts drifting more and more
while the oscillations reduce in amplitude. When the
Input Input frequency is closer to the resonance the first oscilla-
Frequency Amplitude Lagging Phase difference
tion has the greatest amplitude and then the subsequent
(Hz) (mm) (s) (◦ )
ones are slightly smaller but associated with substan-
2.0 9.7 0.16 18.33◦ tial drifting, although overall slightly smaller than the
2.0 12 0.2 22.92◦ previous case.
2.2 12 0.2 25.21◦
2.5 6 0.16 22.92◦
2.5 12 0.18 25.78◦ 3.2 Analysis of results of Series II models
In order to further study the effect of corner restraint
and decoupling of the two orthogonal walls, extra
(a) Excitation of 1.8Hz, 12mm weights were placed directly at the edge of each wall.
20 This addition resulted in a slight increase in natural
10 frequency from 2.0 Hz to about 2.1 Hz in y direction
and from 1.9 Hz to 2.0 Hz in x direction with respect
0 to the Series I models with a substantially smaller
Displacement
494
2.7 (a) x-2.0Hz 28th cycle
(a) in x in y 4
2.4 positive peak
negative peak
2.1 zero point 2
positive peak 2
Amplification
1.8 0
1.5 -15 -12 -9 -6 -3 0 3 6 9 12
-2
1.2
Panel // y
0.9 -4
0.6
-6
0.3
-8
0
1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6
-10
Input frequency
-12
240 Panel // x
(b) in x
210
in y (b) y-2.0Hz 16th cycle
Dissipated energy (J)
180 diagonal 4
positive peak
150 negative peak
zero point 2
120 positive peak 2
0
90
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
60 -2
Panel // y
30
-4
0
1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 -6
Input frequency
-8
Figure 10. (a) amplification and (b) energy dissipation
response with different input frequencies for Series II models -10
for constant acceleration. Panel // x
495
0.024 ACKNOWLEDGEMENTS
0.021 Model II Model III
This project is developed with support from the
0.018 Dorothy Hodgkin Award scheme and from the Royal
Unit Top Drift
496
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
C.L. Brookes
Gifford, Southampton, UK
ABSTRACT: This paper describes how during the last 10 years Gifford has been applying a novel numerical
technique to accurately assess the strength of masonry arch bridges, and for weak bridges, to design repairs and
strengthening. This has involved a major development programme including full-scale laboratory tests, supple-
mentary load tests on bridges in the field, monitoring programmes and verification of the arch bridge structural
analysis which has been based on the Finite/Discrete Element technique. The advantages this technique provides
over conventional masonry arch bridge analyses, both mechanism and traditional Finite Element modelling, are
described and how through partnering an innovative assessment and strengthening service is being delivered. The
relevance of this approach to emerging serviceability limit state arch bridge assessment, which is seen as being
particularly important for Railways, is also discussed. Some practical issues relating to bridge strengthening and
working within national codes of practice are also covered.
1 INTRODUCTION the arch bridge form, the number of spans, rings and
piers that can be modelled. Also, unlike many simpler
It is likely that there are at least half a million masonry strength assessment methods, there is no adherence to
arch bridges in use throughout the world today, princi- predetermined failure mechanisms, for instance, a set
pally carrying road and rail. European railways alone number and pattern of hinges.
account for 200,000 bridges (Orbán 2004). These The application of the Finite/Discrete Element
bridges form a vital asset. Their replacement cost is method has marked a step change in the rigour that can
almost incalculable yet a world wide insatiable appetite now be applied to the structural analysis of masonry
for economic growth is in some cases pushing their use arch bridges. Not only can it be used to accurately
to the limit. assess strength but also to determine bridge defor-
Despite being ancient in form, masonry arches mation, including all significant non-linear effects,
are notoriously difficult to accurately assess. At all making it possible to assess behaviour at both strength
limit states their behaviour is complex, deriving their and serviceability limit states. Also, being a gener-
overall behaviour from the interaction of individual alised approach the behaviour of complex bridges
parts, blocks, bricks, mortar and fill. Several meth- can be assessed where for example a concrete sad-
ods for assessing the strength of arch bridges have dle may exist, or the bridge is propped and in the
become well established, but their generalised use is case of strengthening, retrofitted reinforcement is
limited and their application for designing strength- introduced.
ening difficult. Finite Element analysis, which has to Through partnering with Cintec International who
be non-linear to predict strength, has also been suc- manufacture and install a masonry anchor system,
cessfully applied but the choice of tensile material Rockfield Software who produce the ELFEN (Rock-
properties can be problematic as this can artificially field 2003) structural analysis software, Gifford have
influence the predicted strength. completed over 170 bridge assessments and bridge
The Finite/Discrete Element method, which strengthening designs, mainly in the UK but also in
involves the automatic computation of interacting the USA, Australia and India. Known as Archtec, this
bodies is, therefore, a natural choice for representing service was originally conceived for efficient, eco-
masonry and this type of non-homogenised structure. nomic and sympathetic strengthening of arches, but
Like the conventional Finite Element method, being the method of structural analysis can also provide
a generalised approach also means that, subject to accurate strength assessment of existing bridges and
verification, any geometric form of masonry can be on many occasions has been used to show that bridges
simulated. Consequently, there are no restrictions to do not need to be strengthened.
497
2 CONVENTIONAL ASSESSMENT operational load levels they quite often become dif-
ficult to use to predict ultimate strength and damage.
Methods of strength assessment have been categorised This is generally because of the type of solver that
(McKibbins & Melbourne 2006) as semi-empirical, is used, normally an implicit solver, and the effort
limit analysis and solid mechanics methods. required to ensure internal forces are in equilib-
rium with external loads, as brittle materials such as
2.1 Semi-empirical methods masonry soften and redistribute load. The solution to
the equilibrium problem is normally to use a hypothet-
Most semi-empirical methods are based on the MEXE ical masonry tensile strength but choosing a suitable
(Military Engineering Experimental Establishment) value, large enough to achieve equilibrium conditions
method which evolved from work undertaken in the are met but small enough not to influence the result,
1930s for the military to rapidly assess arch bridges. can be a challenge.
It is often still used as a first pass strength assessment
but its use is highly subjective and there are many lim-
itations. It is of little value for any detailed work such 3 THE FINITE/DISCRETE ELEMENT METHOD
as the design of strengthening.
3.1 Description
2.2 Limit analysis methods Numerical techniques have been devised to represent
Most conventional bridge assessments are now carried discontinua where body or particle interaction defines
out using computerised versions of limit analysis also overall behaviour (Cundall 1971). Perhaps, the most
known as mechanism analysis. In its simplest form advanced technique that describes this behaviour is the
these methods consider a 2D arch comprising a series Discrete Element Method (DEM). The relatively new
of blocks of infinite compressive strength, which can- Finite/Discrete Element Method (FDEM) (Munjiza
not slide against each other and cannot carry tension. 2004) is a combination of FEM and DEM and pro-
A routine is used to establish the locations of hinges in vides a more natural approach to the simulation of
the span followed by calculations of reaction and then many materials and structures. It has been applied to
vector algebra to position the resultant line of thrust. a diverse range of engineering and scientific prob-
The method produces a lower bound solution. In other lems from food processing to rock blasting. Through
words, if a load path can be found that lays entirely automated adaptive modelling, even the transition
within the masonry then the modelled arch is capable from continua to discontinua and the fracturing and
of sustaining that load even if it is not the true load path. fragmentation process can be represented.
Limit analysis techniques have proved to be excel- FDEM is aimed at problems involving transient
lent tools for first phase strength assessments but dynamic systems comprising large numbers of
several restrictions exist that are important in the deformable bodies that interact with each other. Mod-
design of strengthening. The most important of these els involve typically thousands, but in extreme cases
is the inability to calculate strain and displacement. millions, of separate Finite Element meshes auto-
Consequently, it is not possible to determine the distri- matically interacting with each other using DEM
bution of stress at operational load levels, it is difficult contact algorithms. The solution of the continuum
to assess the serviceability of bridges and in the case equations associated with FEM is well established, the
of strengthening, it is not possible to determine the algorithms within DEM less so.
share of load between the existing bridge and the Contact detection and contact interaction lay at the
strengthening. heart of DEM. Contact detection is aimed at iden-
tifying discrete elements that can potentially come
into contact with each other and eliminating those
2.3 Solid mechanics methods
far away from subsequent contact interaction algo-
The established technique used to model continuum rithms. Different algorithms have been developed for
based phenomena in solid mechanics such as deforma- different packing densities for example, sparse and
bility is the Finite Element Method (FEM). Not sur- moving or dense and static. The chief aim here is to
prisingly this has also become the most popular solid reduce computing effort. Contact interaction applied
mechanics method used for arch bridge analysis, and to the surfaces of discrete elements coupled through
there are numerous well developed industry quality the detection process is where interface behaviour is
computer programs available. calculated. Here interface laws are applied according
Like limit analysis most work is carried out using to the surface characteristics of the contacting discrete
2D representations, generally plane strain, but 3D shell elements for example, frictionless no-tension contact.
and solid models are used for special assessments. During the solution of transient dynamic problems of
Although these techniques can be good for even quite modest size millions of contacts will be
determining displacements, strains and stresses at detected and resolved.
498
Another key aspect of FDEM is that the analysis
involves all equations of motion, is therefore dynamic
and uses an explicit central difference solution scheme
(Owen 1980). This involves a time stepping procedure
that is conditionally stable, but unlike many conven-
tional FE solvers that use an implicit solution scheme,
does not involve computationally intensive matrix fac-
torisation. Solutions are achieved only through the use
of very small time steps. The critical time step size Figure 1. Part of a modelled masonry arch barrel showing
FE mesh (left) and DE mesh (right).
below which steps must remain for stability and accu-
racy is given by the time taken for a stress wave to travel
across the smallest finite element. The efficiency of
DEM contact detection and the avoidance of equilib-
rium calculations allows FDEM simulations to predict
failure, collapse and post-failure kinematic behaviour.
499
Figure 4. FDEM simulation showing use of temporary
formwork to support the barrel whist dead load is applied.
500
of practice. Unfortunately, outside of the UK, there
are few rules to help Engineers assess arch bridges.
For example live loading is almost always developed
for beam arrangements of bridges where load support
is primarily through bending, masonry strength assess-
ment is often permissible stress based, and bridge
specific earthquake rules are geared towards steel and
concrete construction. In India the railways have a code
of practice for masonry arch bridges which impose
almost arbitrary performance limits on deflection.
The use of FDEM to simulate arch bridges is
a performance based method, useful for limit state
assessment and design, but cannot be directly used for
Figure 6. Installation from below in a multi-span bridge
using precisely aligned diamond drilling coring equipment. rules that have been developed for linear, often inac-
curate, working stress approaches. In these instances
as to provide bending strength at the critical locations to satisfy bridge technical authorities hybrid analyses
thereby resisting the development of the hinges. By are run along side the more realistic and reliable limit
providing bending resistance the arch barrel is able to state work. The results allow additional checks to be
resist the critical loading conditions more efficiently made with local codes of practice.
and the peak compressive stresses in the masonry
are reduced. A similar procedure is applied to more 5 VERIFICATION
complex arrangements including multi-span arches
although failure mechanisms and anchor positioning The process which has been undertaken to verify the
requires anchors to be placed in different positions. FDEM analytical methods employed by Archtec have
included a number of key strands and evaluations as
4.2 Benefits follows:
Compared with conventional arch bridge strengthen- 1 Against conventional methods of arch assessment.
ing such as concrete saddling and lining, the Archtec 2 Against published data from full-scale tests of
service has several practical benefits: unstrengthened arches carried out by others.
1 Through the use of FDEM a good assessment of 3 Against full-scale tests by TRL of bridges strength-
existing strength and bridge behaviour is obtained. ened by the Archtec method which were specif-
This allows accurate matching of strengthening to ically commissioned as part of the verification
the loading requirements if the bridge is under process.
strength, thus minimising any intervention. Alter- 4 Against the results obtained by monitoring bridges
natively, strengthening may be avoided. in the field including before and after strengthening
2 Strengthening is invisible which is particularly comparisons.
important for historic and heritage bridges. Additionally, a philosophy of fixing material param-
3 Construction is small scale and fast to implement. eters for whole series of tests where similar masonry
4 Disruption to bridge users during strengthening is construction has been employed (compressive strength
much less than saddling. of bricks, mortar type etc.) has been adopted. This
5 A more sustainable solution with lower envi- makes it impossible to adjust an individual arch anal-
ronmental impact, embodied energy and carbon ysis to gain better correlation with tests within a series
emissions. without influencing all others. Similarly, the analysis
6 Because displacement and strain is predictable of Archtec strengthening follows on from verified and
assessments and strengthening designs can be fixed unstrengthened analyses.
based on limits states other than purely ultimate A selection of the verification work (Brookes 2003)
strength. and recent field trials illustrating the accuracy and
7 Each anchor installation provides a core of infor- flexibility of FDEM arch simulation follows.
mation that can be used to confirm the materials
and internal arrangement of the bridge.
8 In many instances all these factors equate to 6 FULL-SCALE ARCH TESTS
reduced cost.
6.1 Strengthened arches
4.3 Working with codes of practice
In order to test the practical implementation of
Archtec services have to be provided within a frame- Archtec, to validate the FDEM method of structural
work which embraces as far as possible national codes analysis, to help quantify key strength parameters and
501
Figure 8. Comparison of load versa deflection charac-
teristics between the first Archtec test results and FDEM
predictions. Deflection is measured in the barrel at the span
position of the load.
502
separated barrel restoring it to that of the fully bonded
case (as-built condition). The implication is that strains
in the intrados have been reduced and the risk of
bricks loosening is thereby also reduced. Provided an
arch is maintained in reasonable condition the risk of
bricks loosening should be reduced compared to an
unstrengthened arch. There is also no reason to doubt
that similar trends in behaviour will occur if the inner
ring itself is in a deteriorated condition.
Bridge owners and experts in the field recognise the
desirability of further research with respect to the ser-
viceability limit state and phenomena such as masonry
fatigue. However, at the current time no specific guid-
Figure 10. Comparison of maximum predicted compressive ance or criteria exist with respect to explicit evaluation
stresses between strengthened and unstrengthened bonded of the serviceability state in arches.
rings. To provide increased confidence that the service-
ability of a bridge is being improved by Archtec
strengthening additional checks have been introduced
Both observations of monotonic loading and cyclic
into the design process. As a precautionary measure
loading have led to the recommendation of a 50% rule
in the absence of other guidance, the following addi-
and are in effect stress limit based. The current Archtec
tional serviceability criteria are included in the design
design method, which uses load factors based on the
process:
UK Highways Agency standards, embraces the ser-
viceability limit state implicitly within the load and 1 Either check that stresses under the required live
material factors used at the ultimate limit state. Whilst loading do not exceed those in the unstrengthened
this method is consistent with current practice, FDEM bridge under existing live loading, or alternatively
analysis used in the design of Archtec strengthening check that stresses in the strengthened bridge are
enables the behaviour of the arch under serviceabil- below an agreed serviceability limit state value.
ity loading to be investigated in ways never before 2 To be sure that existing defects are not made worse,
possible. or for that matter introduced into arch barrels by
Comparison of results from the unstrengthened Archtec strengthening, strains along the intrados
and Archtec tests show that under identical loads, under the required live loading are checked to
displacements are very similar. Corresponding struc- ensure they do not exceed those in the unstrength-
tural analysis of the test arches predicts compressive ened bridge under existing live loading. Strains are
stresses in the Archtec strengthened arch that are lower calculated over a reasonable length so that an esti-
than the unstrengthened arch under the same load- mate of radial joint cracking, critical to loosening
ing. For example, using the bridge proportions of the of bricks, is included.
Archtec tests and UK highway 40/44 tonne vehicle
These criteria are considered very conservative and
axle loading, under the maximum service load the
stresses and strains beyond these limits may be quite
maximum compressive stress in the masonry barrel
safe and have no adverse serviceability effects. How-
was reduced by approximately 15%. The reduction in
ever, further fundamental research is required to
stress is due to the fact that the strengthening intro-
establish appropriate limiting criteria.
duces bending capacity into the arch barrel, which can
therefore resist the applied loading at the critical points
more effectively. Hence, on the basis that serviceabil- 7 FIELD MONITORING
ity can be defined by a stress limit, the reduction of
stress levels in the masonry in Archtec strengthened Several bridge monitoring programmes have been
bridges has a beneficial effect on serviceability. undertaken during the last decade to help verify FDEM
Other aspects of bridge serviceability might be con- arch simulations, and for strengthened bridges, to
cerned with specific deteriorated conditions in arch make before-and-after behaviour comparisons. The
barrels, such as loose bricks and ring separation. The most recent of these was for Indian Railways with the
risk here is that debris falling from a bridge would first part of the programme completed in 2007 and
represent an unacceptable hazard. An example of an which involved two unstrengthened multi-span brick
arch barrel in a weakened condition that could develop arch bridges.
loose bricks as a result of partial ring separation has The principal aims of the programme were to
been tested and used in comparison with Archtec. compare measured barrel vertical displacements and
Displacement results show that Archtec strengthen- intrados circumferential strains with those predicted
ing significantly increases the stiffness of the ring using FDEM simulations. Load tests were carried out
503
10, 40 and 65 km/h. During the dynamic tests vertical
accelerations were also recorded.
Vertical displacement results were found to be
sensitive to the stiffness of the foundations, which is
essentially unknown, and also to the transverse distri-
bution of the live loading. A process involving linear
regression was used to take account of these effects.
Overall, measured and predicted results compared well
for all static load cases considered.
Vertical displacements measured during the
dynamic tests were compared with predictions and
good correlation obtained. For predictions, the engine
and two wagons traversed the bridge. Good correlation
was obtained for the first eight axles of the train.
504
McKibbins, L.D. & Melbourne, C. et al. 2006. C656 – Hendry, A.W. 1990. Masonry properties for assessing arch
Masonry arch bridges: condition appraisal and remedial bridges, Contractor Report 244. Crowthorne, England:
treatment. London: CIRIA. Transport and Road Research Laboratory.
Cundall, P.A. 1971. A computer model for simulating BD 21/01. 2001. The Assessment of highway bridges and
progressive, large scale movement in blocky systems. structures. England: The Highways Agency: Design Man-
Proceedings Symposium ISRM, Nancy, France, Vol. 2, ual for Roads and Bridges.
129–136. BS 5628-1. 2002, Code of practice for use of masonry, Part 1:
Munjiza, A. 2004. The combined finite-discrete element Structural use of unreinforced masonry. England: British
method, Chichester: John Wiley & Sons Ltd. Standards Institution.
Owen, D.R.J. & Hinton, E. 1980. Finite elements in plasticity, Brookes, C.L. 2003. Archtec, Verification of Structural
theory and practice. Swansea: Pineridge Press Limited. Analysis, Rep B1660A/V10/R02 Rev C. Southamp-
Brookes, C.L. & Mehrkar-Asl, S. 1998, Numerical Modelling ton: Gifford http://www.bridgeforum.org/dir/weblink/
of Masonry Using Discrete Elements – Seismic Design B1660AV10.html
Practice into the Next Century, Society for Earthquake
and Civil Engineering Dynamics, London pp 131–138.
Rotterdam: Balkema.
505
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M. Piazza
University of Trento, Trento, Italy
H. Varum
University of Aveiro, Aveiro, Portugal
ABSTRACT: In-situ cyclic tests on an existing traditional timber truss were performed. The main goal of
the tests was to evaluate the overall behavior of the timber truss under symmetric and non-symmetric loading.
Moreover, the influence of the location of point loads application, without and with eccentricity relatively to
joints, was assessed. The carrying tests were preceded by a visual and non-destructive inspection aiming to
collect geometric data and to assess the decay level. The field tests results of a queen-post truss are presented
and analyzed. A numerical model was developed to reproduce the test results.
507
Ridge (80x180) Rafter (80x200) 600
500
Drill resistance
Purlin (80x170) Purlin (80x170)
400
King-post Struts Purlin
Purlin (80x190) (80x145) (80x170)
300
(80x195)
Queen-post Queen-post 200
(80x150) (80x150) 27°
100 decay
Tie-beam (80x220) 0
11,8m 0 15 30 45 60 75 90 105 120 135 150 165 180 195 210
Measurement length (mm)
Figure 1. Truss geometry.
Figure 3. Calculation of the cross-section reduction based
in a Resistograph test profile.
Decay
0-5%
5 - 10 %
10 - 20 %
20 - 25 %
Figure 2. Emergence holes over the surface of sapwood. Figure 4. Map of the decay extension in the truss.
located below the lower purlin. Point loads out of the 3 TEST SETUP, INSTRUMENTATION AND
joints, causing bending moments in the rafters, are the PROCEDURE
most common error detected in the preliminary survey
performed in previous steps of the research program The main goal of the tests was to evaluate the overall
(Branco et al. 2006). behavior of the timber truss selected under symmetric
The truss is made of maritime pine (Pinus pinaster, and non-symmetric loading. Moreover, the influence
Ait.). The timber members of the truss are slender, of the location of point loads application, without and
as characteristic of traditional Portuguese roofs struc- with eccentricity relatively to joints, was assessed.
tures, with cross-sections varying from 80 × 145 mm2 This is the consequence of the roof geometry, which
for the struts to 80 × 220 mm2 for the tie-beam. The exhibits two purlins located with an eccentricity rela-
tie-beam is suspended to the posts by iron straps nailed tively to the intermediate joint of the rafter. Therefore,
into the posts. Between the tie-beam and the king-post firstly, joints loads were applied in the joints (F1 , F2
there is a gap of 5 cm while queen-posts are in con- and F3 ) and, in a second step, loads were applied over
tact with the tie-beam. Connections between the others the purlins and ridge (F1 , F4 , F5 , F6 and F7 ).
timber members are made by single step joints, in some Loading and unloading were recorded and an
cases nailed, and the queen-posts/rafters connections attempt to measure the creep of the structure under
have a heel strap nailed (25 mm wide and 5 mm tick), symmetric loading was made. The behavior of the truss
Figure 1. under non-symmetric loading was evaluated, in the
Despite the apparent good condition of the timber first scheme, only by one test and in the second with
members of the truss, the visual inspection revealed two tests (one in each pitch side). Table 1 resumes the
insects attack in the tie-beam, queen-posts and struts. tests performed.
In these timber members, emergence holes over the Wood pallets suspended to the truss by four steel
surface of sapwood are visible however, without signs cables (φ 6 mm) supported the 35 kg cement bags
of active infestation, Figure 2. used as loads. Each loading and unloading proce-
To evaluate the extension of decay in the timber dure was divided in steps of 175 kg (5 bags). A total
truss, Pilodyn® and Resistograph® non-destructive load of 2625 kg (3 × 875 kg) and 2975 kg (5 × 595 kg)
tests were performed. The Pilodyn 6J was used with the was used in the first (three point loads) and second
aim to assess the surface hardness through the depth (five point loads) schemes, respectively. The differ-
penetration of the pin steel (2.5 mm) measured in each ence in the maximum load value applied between
test performed. Resistograph permits to plot profiles both schemes, 350 kg, is due to the difficulty to
(drill resistance versus penetration depth) that can be increase the number of bags over the pallets in the
used to determine the location and extent of voids, first case. To record the deformation of the truss
allowing the calculation of the residual cross sec- during the tests, eight LVDTs (Linear Variable Dif-
tion (since decayed wood presents lower penetration ferential Transformer) and six dial gauges (DG) were
resistance), and variation in material density. used. LVDTs were responsible for measuring the
508
Table 1. Summary of the tests performed.
875
F1
F2 F3 700
Load (kg)
525
Procedure 350
Load (kg) Load (kg)
LVDT-1
875 875 175
LVDT-3
161 LVDT-2
0
Time (min.) Time (min.) 0 5 10 15 20 25
Test Point loads Test Point loads Displacement (mm)
3F-C F1, F2 and F3 3F-N F3
Figure 6. Displacement recorded by LVDTs 1, 2 and 3
F1
during 3F-C test.
F5 F6
F4 F7
In every loading and unloading step the displace-
ment values of the LVDTs were recorded; however,
Procedure
in the case of the dial gauges, only some steps were
acquired, as results of the little variation verified.
Load (kg) Load (kg)
595 595
509
875 875
Load (kg)
700
525
700 350
DG-7
175
Load (kg)
0
525 0 2 4 6 8 10 12 14
Displacement (mm)
350
(a) Damage (b) load-displacement curve of DG-7
Load (kg)
525
875
350
700
LVDT-1
(3F) DG-4 LVDT-3
175
Load (kg)
510
3150 E90 orthogonal to it, plus a shear modulus, G, and a
2800 Poisson’s ratio, ν. For the material properties values
2450 the LNEC (1997) was followed.
Total Load (kg)
2100
1750
5.2 Geometric and mechanical models
1400 (5F) LVDT-4
1050 The geometry and loading of the truss permits the
(5F) LVDT-6
700 assumption of a state of plane stress for the model. The
(3F) LVDT-4
350
truss is analyzed as a frame structure, having the mem-
(3F) LVDT-6 ber an elastic behavior while a semi-rigid behavior
0
0 0.25 0.5 0.75 1 1.25 1.5
has been assumed for the connections. The semi-rigid
behavior of the connections is introduced in the model
Rotation (10-3 rad)
with the Nonlinear link elements (Nlink).
Figure 11. Comparison between rotations calculated based
on values recorded by LVDTs 4 and 6 during 3F-C and 5F-S 5.3 Loads
tests.
During the tests two types of loads were applied in
3150
the structure. Uniformly distributed loads represent-
2800
ing the self-weight of the truss member, automatically
2450
computed, and joint loads as result of the self-weight
(5F) LVDT-1 of the roof structure, transmitted to the truss by the
Total Load (kg)
2100
(5F) LVDT-2
1750 (5F) LVDT-3
purlins, and the ones applied during the different load
1400 (3F) LVDT-1 procedures of the tests performed.
1050 (3F) LVDT-2
(3F) LVDT-3
700
5.4 Semi-rigid modeling of connections
350
0 Traditional timber joints, even without any strength-
0 5 10 15 20 25 ening device, usually have a significant moment
Displacement (mm)
capacity. Common constraint models, like hinges or
full restraint connections, indeed, cannot satisfacto-
Figure 12. Comparison between the displacements
rily describe the real behavior of these joints. The joint
recorded by LVDTs 1, 2 and 3 during 3F-C and 5F-C tests.
behavior may be classified as semi-rigid and, being
based on friction, is influenced by the time-varying
three point loads (3F-C) is significantly greater but, level of compression between the joined members
in this case, the influence of the damages observed (Parisi & Piazza 2000). In order to properly describe
in the rafter/tie-beam connection must be taken into this behavior, the elastic stiffness of each Nlink must
account. be defined, according to the different geometric and
mechanical features of the elements at each joint.
Candelpergher & Piazza (2001) have proposed
5 NUMERICAL ANALYSIS some expressions to define the rotation stiffness of
traditional timber connections, however, these rules
The structural analysis program SAP 2000 has been should be verified for the Portuguese traditional tim-
used to model the tests. It has been developed a ber connections case. Because this calibration process
model involving variable dimensions beam elements, is not yet finished, only the symmetric tests will be
and semi-rigid joints between the different elements, numerically analyzed. In the case of the symmetric
assuming anyway a linear behavior of the material. tests performed the rotation stiffness of connections
has a trivial influence in the overall behavior of the
tested truss. However, the axial stiffness of the con-
5.1 Material model nections is crucial in the truss response (deformation
Timber is assumed as an orthotropic material in and stress distribution).
the system so-called anatomic cylindrical coordinates The axial stiffness (kax ) depends from the lum-
corresponding to the longitudinal, L, radial, R, and ber mechanical properties, the geometric proportions
transversal, T , directions of the tree trunk. Cylindri- of the connected elements and the connection angle
cal coordinates may be approximated as orthogonal, (skew angle):
for the material extracted from the outer region of
the trunk. The elastic modulus, for an asymmetric
model, are E0 in the direction along the fibers, and
511
where, applying the Hankinson Equation (2):
512
Table 2. Comparison between numerical values (Nu.) and Table 3. Axial stiffness values (kN/m) used in the numerical
tests results (Exp.). Error (Er.) express in (%). model.
15.4 15.5 0.21 12.0 13.8 14.6 10.9 10.0 8.14 particular, the connections between the tie-beam and
the posts, recorded by DG-3, DG-4 and LVDT-5 are
Relative displacements – tie-beam/posts connections (mm) sensible to the connections rotational stiffness.
513
REFERENCES and Maintenance of Heritage Architecture VIII, May,
Halkidiki, Greece.
Branco J.M., Cruz P.J.S., Piazza M. & Varum H. 2006. LNEC 1997. Timber for structures – Maritime pine for struc-
Portuguese Traditional Timber Roof Structures, In WCTE tures. LNEC (eds), Ficha M2. ISSN 0873-6472, Lisbon,
2006 – World Conference on Timber Engineering, 6–10 12 p. (only available in Portuguese).
August, Portland, Oregon, USA. Parisi M.A. & Piazza M. 2000. Mechanics of plain and
Branco J.M., Cruz P.J.S. & Piazza M. 2008. Diagnosis and retrofitted traditional timber connections. Journal of
analysis of two king-post trusses, In SAHC 2008: Struc- Structural Engineering, ASCE; 126(12): 1395–403.
tural Analysis of Historical Constructions, Bath, UK, Piazza M., Brentari G. & Riggio M.P. 2004. Strength-
02–04 July. ening and control methods for old timber trusses: the
Candelpergher L. & Piazza M. 2001. Mechanics of traditional queen-post truss of the Trento theatre. In SAHC 2004:
connections with metal devices in timber roof structures. Structural Analysis of Historical Constructions, Padova,
In Proceedings of the Seventh International Conference IT, II: 957–965.
STREMAH 2001, Bologna, Italy. SAP 2000. Static and Dynamic Finite Element Analysis of
Del Senno M. & Piazza M. 2003. Behaviour and Structures. Structural Analysis Program. Computers and
rehabilitation of queen post timber trusses. A case Structures. Inc., Advanced 9.03. California. USA.
study. In STREMAH 2003: Structural Studies, Repairs
514
Assessment and analytical techniques
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The role of horizontal diaphragms in the seismic response of buildings is significant both in terms
of stiffening and transfer of horizontal loads to vertical elements. Their behaviour in 3D modelling is frequently
assumed (with a rough approximation) as rigid: this hypothesis may be not completely acceptable in case of exist-
ing buildings (with reference to historical masonry structures), where various archaic constructive technologies
can be found (e.g. structural brick or stone vaults). In the framework of equivalent-frame models, the adequate
definition of the equivalent elastic stiffness of various ancient floor technologies is noteworthy. In this paper, the
attention is focused on the structural brick or stone vaults, frequently built in historical heritage. Detailed FEM
elastic analyses are performed on different typologies of vaults (barrel vault, cross vault, cloister vault), in order
to correlate their axial and shear stiffness to that of an equivalent membrane. Various geometrical parameters and
boundary conditions are analysed. On the other hand, stiffness degradation, due to the non-linear behaviour of
masonry, and failure mechanisms have to be studied. In fact, damage in the vaults may substantially alter their stiff-
ening effect and the distribution of horizontal loads to vertical elements. Also in this case, detailed FEM analyses,
using a non-linear constitutive model are performed, focusing the attention on the effect of masonry pattern.
517
order to correlate their axial and shear stiffness to The presence of loose fill, structural backing and
that of an equivalent membrane. Various geometrical constraint boundary conditions may be considered
parameters and boundary conditions are analysed. further factors affecting the vault behaviour.
On the other hand, damage in the vaults may In particular, the constraint boundary conditions are
substantially alter their stiffening effect and the dis- a crucial point if the attention is focused on the vault
tribution of horizontal loads to vertical elements. All role in the context of the seismic overall behaviour of
elastic and geometrical conditions being equal, other a complex entity as a building is (see §3 The vault
masonry characteristics (e.g. masonry pattern and modelling as an equivalent membrane).
strength parameters) may strongly affect this feature. From this point of view, the constraint conditions
In this case, detailed FEM analyses, using a non-linear to be modelled are function of the interlocking pro-
constitutive model (Gambarotta & Lagomarsino 1997) vided by the walls in correspondence of the vault
are performed, focusing the attention on selected case springing points. The constraint degree may derive
studies. by a precise constructive purpose or by cracking or
degradation situations, which the structure may have
2 MORPHOLOGY AND SEISMIC ROLE OF undergone during the ages. As an example of the first
VAULTS case, lunettes in barrel vaults are often built without
connection to lateral walls: in this condition, when the
The building technique of masonry vaults continued vault and the supporting walls perform out-of-phase
throughout the ages until recent times, influencing the oscillations, they act as struts (or counterforts) if com-
history of architecture and originating a multitude of pressed, but they may undergo no damage when the
typologies and shapes (simple and complex: barrel wall tends to separate. The second situation may be
vault, cross vault, cloister vault, etc.). represented by the loss of continuity between the arch
Manifold factors affect their structural behaviour, and the end wall, e.g. in case of barrel or cross vaults,
such as geometry, stiffness and mass distribution, due to crack developing near the wall.
chronological succession of the building works, sub- In presence of seismic loads, another interaction
sequent alterations. effect between vaults and supporting walls, besides
As an example, different masonry pattern can be the constraint degree, is constituted by the activation
adopted on equal terms of geometry: this aspect plays a and developing of out-of-plane collapse mechanisms
fundamental role in a structural system where the equi- of the walls. Even if the analysis of these mechanisms
librium is ensured by the mutual thrust of the blocks is certainly very interesting, this study focuses on the
making up the vault. In Figure 1, some examples of structural role of the vaults in transferring the inertial
pattern are shown. loads to vertical walls, in the framework of the overall
As a consequence, this aspect represents a distinc- seismic behaviour of the building.
tive building feature, strongly affecting the overall There is no intention to provide information about
behaviour: paying attention on the barrel vault, vary- the detailed seismic behaviour of the vault itself, but
ing the brickwork bond, its structural response may be the aim is obtaining some knowledge about the realis-
interpreted as that of a three-dimensional shell (par- tic stiffening effect that a vault may produce in case of
allel pattern) or a set of adjacent arches (orthogonal an earthquake and comparing it to that of a plane floor.
pattern). The brick (or stone) bond influences not only In fact, no dynamic analyses are carried out on the
the static behaviour, but also the damage pattern: in vault system, but only simplified load configurations
fact, this latter is obviously affected by the potential (see next paragraphs) are studied.
sliding planes, identified by the scheme of bed and
head joints.
3 THE VAULT MODELLING AS AN
EQUIVALENT MEMBRANE
518
The moduli of elasticity Ex and Ey represent the L × L), thickness (s) and material (characterized by
normal stiffness of the membrane and each of them the elastic moduli Ex , Ey and Gx,y ). An isotropic
structurally describes the connection degree between plane element is accounted for if the vault typology
the floor and the vertical wall parallel to its reference is symmetrical (e.g. cloister and cross vault) and an
direction, both in linear and non-linear phases. In a few orthotropic one if unsymmetrical (e.g. barrel vault).
words, the normal stiffness of the floor along X-axis In the first case, aiming to evaluate the vault equiva-
provide a link between the piers of a wall parallel to lentYoung’s modulus EV , a uniform displacement state
X-axis, influencing the axial force the spandrels, but it along x-axis (u) is imposed to one of the orthogonal
does not control the loads transferred to a wall parallel sides. The opposite one is fixed, as in Figure 2.
to Y-axis. From the isotropic linear elastic relationships, the
The term Gx,y represents instead the shear stiffness numerical FEM results may be interpreted as follows.
of the floor and the horizontal force transfer among The Young’s modulus EV is given by eq. (1).
the walls, both in linear and non-linear phases.
The evaluation of these quantities may be rather
simply identified in case of recurrent floor typologies,
ascribing it to the structural role shown by some spe-
cific elements. For example, the shear stiffness of an
r.c. floor with beams and slab is mainly ascribable where: E is the Young’s modulus of the plane element,
to the slab. In this example, the beam axial stiffness σ is the normal stress, ε the axial strain, u is the
leads to the definition of the Young’s modulus of the applied displacement to one side, n is the number of
equivalent plane element. nodes on this side, rk is the reaction nodal force in the
On the contrary, in case of vaults, beside thickness k-th node along the analysed direction.
and material properties, the stiffening contribution In order to evaluate the vault equivalent shear mod-
strongly depends on shape and geometrical proportion ulus GV , an auto-equilibrated system of forces (having
(e.g., rise-to-span ratio). So, the definition of the elas- a total value F on each side) is applied. The horizontal
ticity moduli to be attributed to the equivalent plane displacements of one edge of the model are adequately
element may be highly arbitrary. constrained. By means of the isotropic linear elastic
The proposed modelling strategy starts from the relationships, eq. (2) provides the ratio between GV
elastic numerical simulation of vault response, in case and G (the shear modulus of the plane element).
of pre-defined load configurations, aiming to schema-
tize the axial-only and pure shear behaviour by impos-
ing special force or displacement conditions (Fig. 2).
The scope is to establish a functional relationship
among the elastic parameters of the equivalent plane
where: τ is the tangential stress, γ is the shear strain, ūx
element and the significant quantities of the vaults.
and ūy are the displacement components of the oppo-
This obviously excludes the rigorous theoretical inves-
site corner A (Fig. 2), derived by the FEM numerical
tigation of the vault behaviour as a membrane and
simulations.
flexural shell.
Obviously, EV and GV are representative of the 3D
In this way, an ideal equivalence is set between the
behaviour of each vault typology, summing up the
typology of vault examined and the plane element,
shape effect and the mechanical characterization of
having the same planimetric dimension (squared plan
the material. Further considerations about the bound-
Lx Lx+∆u
ary condition are pointed out in §3.1 Description of
the analysed configurations.
y y
The equivalent parameters, describing the vault
∆u rk
Ly
519
Figure 4. Ev /E ratio in case of different vault typologies.
520
Table 1. Numerical values of the equation coefficients ai .
Vault typology
SS F SS F SS
Starting from the results of the numerical simu- Figure 7. Gv /G ratio in case of different vault typologies.
lations, by means of error minimization, analytical
relationships are defined. These equations are func- Even in case of shear stiffness, from the results
tions of thickness-to-span (s/L) and rise-to-span (f/L) of the numerical simulations, by means of error
ratios. In eq. (3), the relation in case of barrel and minimization, analytical relationships are put forward.
cloister vaults is reported. In eq. (5), the relation in case of barrel and cloister
vaults is proposed.
521
Table 2. Numerical values of the equation coefficients bi .
Vault typology
522
the strain-softening regime. Since the adopted shell
element has different Gauss points in thickness, the
evolution of this parameter can be monitored on both
extrados and intrados. With reference to the intrados
results, as expected, in case of the pattern b, dam-
age turns out to be more pronounced near the corners,
because of cracks that propagate orthogonally to the
displacement direction. On the contrary, in case of
pattern a, damage propagates starting from the vault
keystone (less compressed and hence more vulner-
able to shear failure). The described damage scalar
parameter αm , combined with the information from
the inelastic strain components, allow to evaluate that,
in case of pattern b, the predominant failure mode is
the tensile one (crack opening), whereas, in case of
pattern a, is the shear one (shear sliding).
Further remarks, regarding the role of these
masonry bonds, can be obtained from the direction of
the tensile principal stress σI in various analysis steps
(Fig. 11).
The extent of the direction modification of the
tensile principal stress σI is influenced by the basic
hypothesis of the adopted constitutive law. In fact,
because of the masonry idealisation as a layered con-
tinuum, this component tends to be placed normally
to the head joint, which are indefinitely resistant. As
a consequence, it is fair to expect a variation of this
result if the head joints are kept into account.
Finally, in case of the cross vault (f/L = 0.5), the
results corresponding to three different patterns are
shown in Figure 12. As an example, for τmr /σmr ratio
equal to 1, similar remarks to the ones of the cloister
vault can be obtained.
Figure 13 shows the distribution of αm parameter
on the vault intrados: again the damage follows the
possible paths mapped by the masonry pattern.
Figure 10. Cloister vault (f /L = 0.5; s/L = 0.045): distri- 4 FINAL REMARKS AND FUTURE
bution of αm parameter on the vault intrados, corresponding DEVELOPMENTS
to three analysis steps.
In the paper, for some typical vault typologies, the
In case of herringbone pattern, a much more marked evaluation of an equivalent elastic stiffness is pro-
decay can be noted; it suffers to a higher degree from posed in order to idealize these structural elements
the variation of τmr /σmr strength ratio (see also Fig. 10). as isotropic or orthotropic membranes. The study is
The bond affects not only the crack path, but also the developed in the context of the“equivalent frame”
main failure mechanism that involves the bed joints (no modelling approach for masonry buildings.
block crushing is revealed from the analyses). Some Even if through simplified assumptions, the elastic
considerations may be carried out thanks to the review FEM simulations allow to evaluate the minima reli-
of the obtained inelastic strain (excised in this paper for able contributions, varying the vault typologies and the
brevity) and of the damage pattern shown in Figure 10. geometrical and constraint boundary conditions. On
In particular, Figure 10 shows the value distribution the other hand, the non-linear FEM simulations, even
of the dimensionless damage scalar parameter αm that if undoubtedly no exhaustive, provided some prelimi-
describes the evolution of the mortar joint damage. nary knowledge about the sensitiveness of the stiffness
Values lower than 1 correspond to the un-cracked state, degradation in the early non-linear response.The effect
while a value equal to 1 corresponds to the attainment of masonry pattern and strength characterisation of the
of the peak strength of the joint (both in shear or in bed joints is analysed. This work represents the first
tension). Lastly values greater than 1 correspond to step of further investigations.
523
Figure 11. Cloister vault (f /L = 0.5; s/L = 0.045;
τmr /σmr = 2): direction of the tensile principal stress σI on
the extrados vault.
0.15
Pattern a’
Pattern b’ Figure 13. Cross vault (f /L = 0.5; s/L = 0.045;
Pattern c’ τmr /σmr = 1): distribution of αm parameter on the vault
0.1 intrados corresponding to different steps of the analysis.
Gv/G
ACKNOWLEDGMENTS
0.05
524
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
E. Verstrynge
Department of Civil Engineering, KULeuven, Heverlee, Belgium
S. Ignoul
Triconsult NV, Lummen, Belgium
ABSTRACT: A rheological model is used to describe the long-term behaviour of masonry. Special attention is
paid to the evolution of damage and the description of the damage parameters, based on the continuum damage
theory. The parameters of the model are derived from a test program, including three types of mortar compositions
and three kinds of tests: monotonic compression, short-term and long-term creep tests. Furthermore, a range of
stress evolutions is simulated to evaluate the model and the results are compared with experimental test data.
2 MODELLING OF DAMAGE
ACCUMULATION
525
failure times to describe the growth of ‘hot’ crack
arrays up to a critical density. In the model presented
here, a continuum damage mechanism will be fol-
lowed, describing the overall damage throughout the
material as a continuous phenomenon.
The theory of Kachanov (Kachanov 1985, Kra-
jcinovic & Lemaitre 1987), which lies at the base
of continuum damage mechanics, poses to describe
the absence of material deterioration by means of a
field variable, the continuity. The more frequently used
damage parameter (D) can be seen as the comple-
ment of this continuity. For a completely undamaged
material D equals 0, whereas D = 1 corresponds to a
completely deteriorated material. This model has been
extended in terms of the introduction of the effective
stresses, which act upon a decreasing surface area as Figure 2. Decrease of Young’s modulus due to damage
a result of growing micro voids. increase (from Hult 1987, p. 29).
526
2.2 Viscous damage
As the aim of the proposed model is to simulate long-
term deformations of masonry, viscous damage, which
governs the strain rate at constant stress levels and ini-
tiates the tertiary creep phase, has a highly important
role. The viscous damage parameter (Dv ) decreases
the relaxation time of the Maxwell dashpot, thereby
increasing the strain rate during the steady-state creep
and describing the accumulation of damage up to fail-
ure. In order to describe the evolution of Dv in time,
a twofold definition is considered: firstly, the viscous
damage parameter is a function of the stress level, as a
relation between the strain rate and the stress level has
been observed during the current test program, as well
as in literature (Challamel 2005, Anzani 2000). Sec-
ondly, a damage-rate formulation is adopted to enable
Figure 3. Evolution of Young’s modulus in function of the simulation of damage increase in time when the
stress at each step during a short-term creep test on a wallet stress level remains constant.
with lime-cement mortar. Stress is formulated dimensionless To establish a relation between the amount of vis-
(division by the maximum stress acquired during the test).
cous damage and the stress level, Dv was calculated for
each load step of the aforementioned short-term creep
tests on masonry wallets, according to the equation for
a Maxwell dashpot:
527
presented in a simple constitutive formulation, to
describe the vertical deformation, as follows:
528
Table 1. Overview of the parameters used during the sim- step of the short-term creep tests, revealed a power
ulations with the rheological model. Presentation of results: law relation between the Maxwell relaxation time and
average value (± standard deviation). the time-interval during which the stress remained
constant;
Parameters of
rheological Hydraulic Cement Hybrid
model lime mortar mortar
529
85%
80% 85%
75%
80%
75%
85%
80%
75%
530
Figure 10. Experimental results and simulation of Figure 11. Experimental results and simulation of
short-term creep test on hydraulic lime masonry, column 3. short-term creep test on hydraulic lime masonry, column 5.
Evolution of horizontal en vertical strain in time (above) and Evolution of horizontal en vertical strain in time (above) and
volumetric strain (below). volumetric strain (below).
531
The proposed straightforward model is able to Binda, L. & Anzani, A. 1993. The time-dependent behaviour
describe isotropic behaviour of masonry, includ- of masonry prisms: an interpretation. The masonry society
ing monotonic stress increases and high sustained journal. Vol. 11 (2).
stresses. As other phenomena influence the stress- Bodner, S.R. & Chan K.S. 1986. Modeling of continuum
damage for application in elastic-viscoplastic constitutive
strain behaviour during short-term, monotonic load- equations. Eng. Fracture Mech. Vol. 25 (5/6): 705–712.
ing, this loading condition is not considered here. Boukhavrov, G.N., Chanda M.W. & Boukharov N.G. 1995.
A comparison with experimental data has shown a The three processes of brittle crystalline rock creep. Int. J.
satisfying fit between the model and the experimen- Rock Mech. Min. Sci. & Geomech. Vol. 32 (4): 325–335.
tal results, but more experimental testing is necessary Challamel, N., Lanos C. & Casandjian C. 2005. Creep
in order to enlarge the amount of available data and damage modelling for quasi-brittle materials. European
improve the fitted parameter values. Journal of Mechanics A/solids 24: 593–613.
Hult, J. Introduction and general overview. 1987. In Con-
tinuum Damage Mechanics Theory and Applications,
D. Krajcinovic and J. Lemaitre (eds.): Springer–Verlag,
ACKNOWLEDGEMENTS New York.
Ignoul, S., Schueremans L., Binda L., et al. 2006. Creep
The authors express their thanks to the Flemish Fund behavior of masonry structures – failure prediction
for Scientific Research (FWO) for the doctoral grant, based on a rheological model and laboratory tests. In
offered to Els Verstrynge. P.B. Lourenço, P. Roca, C. Modena, S. Agrawal (Eds.)
The authors highly appreciate the collaboration with Proc. Of the 5th int. seminar on structural analysis of
the Politecnico di Milano and the University of Minho historical constructions.Vol. 2: 913–920. New Delhi 2006.
in this research area. Kachanov, L.M. 1985. On creep rupture time. Proc. Acad.
Sci. USSR Div. Eng. Sci. 8: 26–31
Krajcinovic, D. & Lemaitre, J. (eds). 1987 Continuum damage
mechanics: Theory and applications: Springer–Verlag,
REFERENCES New York.
Papa E. & Talierco A. 2005. A visco-damage model for brittle
Anzani, A., Binda, L. & Roberti G.M. 2000. The effect of
materials under monotonic and sustained stresses. Inter-
heavy persistent actions into the behaviour of ancient
national journal for numerical and analytical methods in
masonry. Materials and Structures. Vol. 33 (228):
geomechanics, Vol. 29 (3):287–310.
251–261.
Verstrynge, E., Ignoul, S., Schueremans, L.,Van Gemert, D. &
Anzani, A., Binda, L. & Taliercio A. 2005. Application of a
Wevers, M. 2008. Damage accumulation in masonry under
damage model tot the study of the long term behaviour
persistent loading evaluated by acoustic emission tech-
of ancient towers. Proc. 1st Canadian Conf. on effective
nique. Proc. 14th Int. Brick & Block Masonry Conference,
design of structures, Ontario, 10–13 July 2005.
Sydney, 17–20 February 2008.
Bazant, Z.P. (ed.) 1988. Mathematical modelling of creep and
shrinkage of concrete: Wiley & Sons Ltd.
532
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Giuseppe Gariup
Departamento de Ingeniería de la Construccíon, U.P.C., Barcelona, Spain
Igino Pitacco
Dipartimento di Ingegneria Civile, Università degli Studi di Udine, Udine, Italy
ABSTRACT: A multi-body approach to solid dynamics is presented. The numerical method implemented
is a discrete element method (d.e.m.) specifically intended for the analysis of masonry structures. Con-
tacts between different blocks are treated using interacting forces. Contact forces are modelled using penalty
method for the normal component and an elasto-plastic behaviour for the shear. The model considers 3
different types of damping. A first quantitative validation phase is presented together with possible future
developments.
1 INTRODUCTION
533
brick with a single block. Applicability of this method 2 CONCEPTUAL MODEL
to complex structures has been considered in the last
section of this paper. Discrete element method analyzes the interactions
The model is characterized by the following aspects: between a finite number of distinct blocks. Each block,
in the implementation used in this work, is elas-
– Formulation of the physical background for the
tically deformable, while non-linear deformation is
development of the model (Simo 1998). Choice of
concentrated in the contacts.
finite strain theory and hyper-elastic behaviour of
All the equilibrium and constitutive equations have
block material.
been obtained under the hypothesis of finite defor-
– Geometrical and finite element definition of the
mations. The deformation of a single block has been
blocks (Belytschko 2000).
considered infinitesimal but the one of the whole sys-
– Definition of the contact algorithm including the
tem, due to possible roto-translation of its parts, can be
geometrical definition of contact, contact updating
large compared to the usual finite elements problems.
through the analysis, the description of the normal
Each block or body is composed by one or more
response using a penalty method approach and the
8-node hexahedrons depending on the complexity of
description of the shear response considering stick-
the geometry. In the contact routine each body is iden-
ing and sliding and plastic behaviour of the shear
tified and interact with the others with its external
reaction (Wriggers 2002).
surface.
– Definition of damping, considering mass, stiffness
The external surface of a block is given by the
and contact damping (Zhong 1993).
external faces of the finite elements forming it (Fig. 2).
– Numerical method used to solve the system of dif-
At the beginning of the analysis all the possible pairs
ferential equations representing the f.e. problem.
of contacting faces are determined and stored in an
Choice of an explicit integration algorithm (Bathe
array. The uncertainty is introduced by the roughness
1996).
of this first check: there can be surfaces included in a
The phase of validation is focussed on two-block contact pair which are not actually touching each other
dynamic analysis. The two blocks are superposed, the but are simply close enough to trick the contact crite-
lower one serves as “ground” for the motion of the rion. During the calculation, at each time step, each
upper one and displacements are constrained on its contact pair is considered separately and is marked as
base. Different cases are modelled and analyzed with active or not-active. A contact pair is considered active
the numerical code. The results are confronted with only if one surface is penetrating into the other and
analytical ones or with the expected behaviour. Each some other criteria established to avoid singular situ-
case has been specifically chosen to test a particular ations are satisfied. This procedure has to be repeated
characteristic of the model: during the calculation every n time-steps to follow the
physical evolution of the system.
– Stability: the upper block is leaning on the lower Each active contact pair represents a contact
one at the beginning of the analysis. Gravity is the between two blocks. The procedure to calculate the
only force considered.
– Impact: the upper block is placed at a vertical dis-
tance d from the lower one at the beginning of the
analysis. Gravity is the only force considered.
– Friction: the upper block leans on the lower one
at the beginning of the analysis. Gravity is applied
together with a horizontal force. Friction is studied
varying the static friction coefficient of the contact.
– Updating of contact: the upper block leans in a
instable position on the lower one whose upper face
is inclined. The evolution of motion is given by
the overturning and successive rolling of the upper
block. Gravity is the only force considered.
– Damping tests: the upper block is inclined with
respect to the vertical and touches the lower one
with an edge at the beginning of the analysis.
Varying the damping coefficients different rocking
responses are simulated. Gravity is the only force
considered.
Figure 2. 2D representation of blocks: (a) single finite ele-
In the next sections the theoretical introduction and the ment; (b) elements forming a block; (c) block composition;
results of the numerical tests are presented. (d) block boundary.
534
forces belonging to each contact consists in the fol- by the shear stiffness K S . The value obtained is con-
lowing: fronted with the absolute value of the normal penalty
force Fn multiplied by the static friction coefficient µ.
– The larger surface is identified as the target-surface
If d·KS ≥ µ. · Fn , the shear force is still in the elastic
and the other as the hitting-surface.
range, its absolute value is given by d·KS , the direction
– If the distance between the 4 nodes of the hitting-
and verse is given by the vector connecting X H 1 and
surface is higher than a tolerance, a regular grid of
X H 0.
virtual nodes is created on it.
If d · KS < µ. · Fn , the shear force, calculated as an
– The position of each node, real or virtual, of the hit-
elastic reaction to the displacement, exceeds the static
ting surface, relative to the target one, is calculated.
friction force. In this case the direction and verse of
If the node is not penetrating, the pair hitting-
the shear force are the ones calculated in the elastic
node/target-surface is considered non-active.
range while the absolute value is scaled to µ. · Fn .
– If the node is penetrating the target surface, normal
The contact forces, together with the external ones,
and shear forces are calculated and added to the
are used to calculate the displacements and the defor-
forces acting on the two surfaces.
mations of the blocks at the successive time-step.
The normal force at contact is calculated using the Central difference method has been implemented to
penalty method. In this approach a small penetration solve the dynamic differential system given by
is accepted during the impact of two blocks. After the
penetration, a reaction force, proportional to the pen-
etration and the penalty stiffness, orthogonal to the
target surface, starts to counteract the superposition where u is the displacement vector, ü its second time
of the two bodies. Choosing an adequate penalty stiff- derivative, M the mass matrix, F ext is the vector of
ness the penetration can be limited to an acceptable external force, F c is the vector of contact forces,
threshold. F int = K · u is the vector of internal forces and K is
The shear force has been modelled with elastoplas- the stiffness matrix.
tic behaviour. Once a hitting point penetrates for the The reasons leading to the choice of an explicit
first time in the target surface, the natural coordinates method in the integration of Equation 1 are mainly its
X H 0 of the normal projection of the point onto the sur- robustness and its simplicity, suitable characteristics
face are stored together with the point-surface contact to deal with problems with many non-linearities.
pair. At the successive time-steps has to be checked if In performing dynamic analysis with any code, it
the point is still penetrating the surface. In this case is usually necessary to account for energy losses in
the new coordinates of the hitting-point X H 1 have the physical system which are not considered in the
to be calculated. The shear force has been modelled numerical algorithm. In this formulation the elasto-
with elasto-plastic behaviour. At first the distance d plastic behaviour of contact shear is the only possibility
between X H 0 and X H 1 is calculated and multiplied the system has to dissipate energy. Other forms of
damping have to be introduced in order to get a better
modelling of real structural dynamic behaviour.
The damped system of equations becomes
535
The values of constants α, β and γ have to be
determined to adjust the model to experimental
observations.
The third form of damping seems the one with
stronger connection to the dissipative nature of the
impact. The first one has no physical meaning and it is
purely a numerical tool to dissipate energy. Stiffness
damping is related to the dissipation due to anelastic
phenomenons in the deformation of the blocks which
are normally neglectable in masonry analysis.
In the next section some examples of discrete
element analysis are presented and validated.
3 EXAMPLES
3.1 Stability
a certain value it has a repulsive effect and pushes A
Two cubic blocks A and B with 1.0 m side are super- upward.
posed. The equilibrium is obtained when the magnitude of
The lower one B has its base constrained. The grav- the penalty reaction d · K s equals the weight W A of A.
itational force is the only force acting. Due to the d is therefore the penetration that has to be tolerated
deformability of the blocks this is not an equilibrate in the analysis and depends on the penalty stiffness K c .
configuration. If no damping is considered in the analysis, A will
The geometrical configuration, forces and con- oscillates indefinitely around the equilibrium position
straints and node numbering are summarize in with more or less the same amplitude (Fig. 5).
Figure 4. The analysis was performed without damping to test
A tends to press B which will react deforming and the numerical stability of the method.
lowering its upper face. The entity of this displacement
is neglectable compared to the residual penetration
3.2 Impact
imputable to the penalty approach. During the impact
A does not encounter a physical obstacle in B, but Two cubic blocks A and B, with side equal to 1.0 m,
experiences a force representing the contact. This force are considered. The blocks are aligned in the horizontal
tends to contrast the interpenetration, when it reaches directions. The lower one B has its base constrained.
536
Figure 7. Z displacement of node 10: comparison between
a model with contact damping equal to 0.1 and an undamped
model.
537
Figure 9. X displacement of node 10: comparison between
the analytical and the numerical solution.
538
4 ANALYSIS OF RESULTS
539
Calculating a consistent stiffness matrix would Lourenço, P. (1996). Computational strategies for masonry
increase the computational cost of the method. structures. Ph. D. thesis, Delft University of Technology,
Rigid-blocks analysis seems to be a viable option Delft, The Netherlands.
considering that the deformation in bricks and stones Mamaghani, I.H.P. & Aydan, O. K. Y. (1999). Analysis
of masonry structures under static and dynamic load-
is negligible compared to block displacement. Never- ing by discrete finite element method. Structural Eng.
theless considering bodies deformable could be used Earthquake Eng., JSCE 16(2), 75–86.
to take into account the deformations of the mortar Oliveira, D. (2003). Experimental and numerical analy-
joints or possible damages of a block. sis of blocky masonry structures under cyclic loading.
Future developments should explore these possi- Ph. D. thesis, Universidade do Minho, Guimares, Portugal.
bilities, confronting results and choosing the most Orduña, A. (2003). Seismic assessment of ancient masonry
suitable one. structures by rigid blocks limit analysis. Ph. D. thesis,
A validation phase with experimental results (Peña Universidade do Minho, Guimares, Portugal.
2005) is also needed and will be taken into account. Peña, F. (2005). Dinámica de bloques de piedra. Experi-
mentación y simulación numérica. Laboratorio Nacional
de Engenharia Civil, Proc. 0305/17/14483.
Simo, J.C. & Hughes, T. (1998). Computational inelasticity.
REFERENCES Springer.
Wriggers, P. (2002). Computational contact mechanics. John
Bathe, K. (1996). Finite element procedures. Prentice Hall. Wiley & Sons.
Belytschko, T. & Liu, W. M. B. (2000). Nonlinear finite Zhong, Z. (1993). Finite element procedures for contact-
elements for continua and structures. John Wiley & Sons. impact problems. Oxford University Press.
Cundall, P. (1992). Universal distinct element code, User’s
manual. USA: ITASCA Consulting Group Inc.
540
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
G. de Felice
Department of Structures, University Roma Tre, Rome, Italy
ABSTRACT: The assessment of load-carrying capacity of masonry bridges is usually based on limit anal-
ysis methods, which implicitly assume that masonry has unlimited ductility capacity in compression; aiming
at taking into account the effective ductility capacity of masonry, an approach that makes use of beams with
fibre cross is developed. The approach is able to take into account the non-linear behaviour of masonry in
compression, including the post-peak softening branch and therefore provides a reliable estimate of the ultimate
load. A numerical investigation of one to three span masonry arch bridges under travelling concentrated load,
shows that, by parity of constitutive assumptions, the method provides the same estimate of ultimate load given
by limit analysis, as well as the same kinematics at collapse. The influence of limited ductility is then investi-
gated and the overestimate provided by classical limit analysis is highlight for the previously examined bridge
typologies.
541
especially for multi-span bridges, which pertain to a is subdivided into longitudinal fibres, whose stress-
more refined analysis and is not justified for initial or strain behaviour is defined according to the material
intermediate-level assessment. properties that can be easily obtained through experi-
The aim of this paper is to investigate the influence ments. The non-linear constitutive relation of the beam
of a limited ductility capacity of masonry on the load- section is not given explicitly but is derived by integra-
carrying capacity of arch bridges. To this end, a non- tion of the stress-strain relation of the fibres as follows
linear incremental analysis tool is developed, along a (Spacone et al. 1996). According to the assumption
research line that dates back to the well-known study of that plane section remain plain and normal to the
Castigliano (1894) for the bridge over the Dora close beam axis x during the deformation history, referring
to Turin, and includes some recent contributions by to figure 1, for a given section generalized defor-
(Boothby 1997; Molins and Roca 1998; Brencich and mations d(x) = {χz (x) χy (x) ε(x)}, comprising the two
De Francesco 2004). In the present paper, aiming at curvatures and the axial strain at the reference
taking into account the effective behaviour of masonry axis, the strain distribution over the section writes:
in compression, the bridge is modelled using beams ε(x, y, z) = a(y, z)d(x), where a(y, z) = {−y z 1}. Once
with fibre cross section. The use of beams with fibre the material constitutive relation is specified, the
cross section has become current for the analysis of strain distribution yields the tangent material modu-
reinforced concrete structures but, at the knowledge of lus E(x, y, z) and the corresponding stress σ(x, y, z).
the author, has not yet been used for structural analysis Therefore, according to the virtual work equality, the
of masonry arch bridges. section stiffness matrix k(x) and the resisting forces
The paper is organized as follows. In section 2 D(x) = {Mz (x) My (x) N (x)}, comprising the two bend-
the non-linear beam model with fibre cross section ing moment and the axial force, that correspond to
is briefly recalled, then, in section 3, a comparison of deformation d(x) are obtained by integrating over the
the results provided by the present approach and those cross section A(x) as:
given by limit analysis is performed for one to three-
span railway arch bridges under travelling concen-
trated load; for the comparison, masonry is considered
either with infinite compressive strength or finite com-
pressive strength with unlimited ductility. Finally, in
section 4, the influence of a limited ductility capacity is
examined, using the Kent & Park model for describing The evaluation of the integrals (1,2) is performed
the uniaxial constitutive behaviour of masonry, when by subdividing the generic section x into n(x) fibres
varying the slope of the post-peak descending linear and applying the mid-point integration rule:
branch. The numerical results, which provide the ulti-
mate travelling load of the bridges, highlight the effect
of material post-peak behaviour on the load carrying
capacity, showing to what extent the assumption of
unlimited ductility would lead to an overestimate of
loading capacity.
A flexibility-based formulation is adopted for the
FE analysis, which ensures that equilibrium is strictly
2 FIBRE BEAM MODEL satisfied within the element, and allows to overcome
the numerical instabilities arising after the onset of
Some recent experiments on masonry prisms sub- strain softening; the latter point is crucial in masonry
jected to eccentric compression (Brencich and Gam- elements, much more than in reinforced concrete ele-
barotta 2005; de Felice, 2006) have shown that the ments, since due to the lack of steel, the masonry
classical hypothesis that plane section remains plane section displays a rapidly descending branch after
after deformation, holds true for brick masonry even in the peak load. According to flexibility approach the
the non-linear range, when stress concentrations and element stiffness matrix is obtained by inverting the
cracks start to develop; besides, the intrinsic difficulty flexibility matrix: K = F −1 , where
in defining the non-linear constitutive behaviour of the
beam, as for instance the dependence of flexural hys-
teretic behaviour on axial load, also due to the lack of
experimental data, suggests deriving the section force-
deformation relation from the uniaxial stress-strain L being the element length, and b(x) the force interpo-
behaviour of the material. Therefore, looking for a lation functions matrix, which is derived form the equi-
compromise between simplicity and accuracy, the arch librium of axial force and bending moments within the
is discretised into rectilinear beams and each beam element.
542
Figure 3. Three span bridges.
Shallow Deep
Geometrical and Mechanical arch arch
properties bridges bridges
3 LOAD-CARRYING CAPACITY OF
MULTI-SPAN MASONRY ARCH BRIDGES: Figure 4. Load diffusion through the fill.
COMPARISON WITH LIMIT ANALYSIS
In this section, the load carrying capacity of one to (rise/span ratio equal to 0.35); their geometrical and
three-span arch bridges representative of European mechanical characteristics are reported in table 1.
railway network (figures 2–3) has been estimated The ultimate value of a concentrated travelling load
according to the present approach that makes use of has been evaluated for the classical hypothesis of
fibre beam elements, and the results compared with no-tensile resistance and either infinite compressive
those provided by limit analysis. The same bridge strength, or 5 MPa compressive strength with unlim-
typologies studied in (Brencich and De Francesco ited ductility. As regards the application of travelling
2004) have been considered, having 15 mt span and load, a diffusion angle trough the fill 2θ = 40◦ + 40◦
0.75 mt arch depth; the bridges are made either with was assumed, and the effective loading on the arch was
shallow (rise/span ratio equal to 0.25) or deep arches schematised as uniformly distributed (figure 4).
543
Figure 7. Load carrying capacity of a two-span shallow arch
Figure 5. Load carrying capacity of a single-span shallow
bridge under concentrated travelling load.
arch bridge, under concentrated travelling load.
Figure 6. Load carrying capacity of a two-span deep arch Figure 8. Load carrying capacity of a three-span shallow
bridge, under concentrated travelling load. arch bridge, under concentrated travelling load in the central
span.
Each arch was modelled by 100 non-linear beams
with cross section discretised in 100 fibres. The results about 25% and 40% is obtained for deep and shallow
obtained with the Fibre Beam (FB) model have been arch bridges examined typologies respectively.
compared to those obtained with Limit Analysis (LA) As regards the kinematics of the bridge close to
making use of the program Ring (Gilbert 2005). The collapse, the formation of sufficient number of plastic
figures 5–8, give the load carrying capacity of the hinges that turn the bridge into a mechanism takes
bridge (one meter depth) as a function of load posi- place.
tion for the different bridge typologies under study. As shown in figure 9, the collapse mechanism of the
All the analyses show an excellent agreement between bridges provided by FB approach is similar to the one
the results provided by the two approaches, by parity of provided by LA in terms of “hinges” location, apart
constitutive assumptions, with differences that never from a different treatment of the kinematics of joint’s
exceed 4%. When neglecting the compressive strength opening. In fact, while in LA approach no overlapping
of masonry, an overestimate of the ultimate load of of the blocks in contact across a joint is allowed, in
544
Figure 10. Load carrying capacity of three-span deep arch
bridges, under concentrated travelling load, when varying
backfill height.
545
material behaviour turns from perfectly plastic to post- Brencich, A., Gambarotta, L. 2005. Mechanical response
peak brittle, in the case of deep and shallow arches of solid clay brickwork under eccentric loading. Part I:
respectively. An even stronger dependence is found Unreinforced masonry, Materials and Structures, 38:
for three-span shallow arch bridges, with a reduc- 257–266.
Boothby, T.E. 2001. Load rating of masonry arch bridges,
tion in load capacity of about 32%, from 385 kN/m ASCE Journal of Bridge Engineering, 6(2): 79–86.
to 260 KN/m. Boothby, T.E. 1997. Elastic plastic stability of jointed
These results can be compared with those presented masonry arches, Engineering Structures, 19(5): 345–351.
in (Brencich & de Francesco 2004) since at section Boothby, T.E., Brown, C.B. 1992. Stability of masonry piers
level, the post-peak slope of softening branch can be and arches, ASCE Journal of Engineering Mechanics
associated with an equivalent available ductility; how- 118(2): 367–383.
ever, as far as the overall behaviour of the bridge close Boothby, T.E., Domalik, D., Dalal, V. 1998. Service load
to collapse is concerned, the differences in the constitu- response of masonry arch bridges, ASCE Journal of
tive assumptions would imply a different redistribution Structural Engineering. 124(1): 17–23.
Castigliano, C.A.P. 1879. Theorie de l’equilibre des sys-
of the normal force and bending moment through the temes elastiques et ses applications, Augusto Federico
arches. It is worth noting that, while in (Brencich & de Negro, Torino, Italy; Andrews E.S., Elastic stress in struc-
Francesco 2004) the hinge mechanism does not takes tures. Transation by Andrews, Scott Greenwood & Sons,
place since the arch collapses when the ultimate duc- London, 1919.
tility capacity in the critical section is reached, in the Cavicchi, A., Gambarotta, L. 2005. Collapse Analysis of
present approach, thanks to a more accurate descrip- Masonry bridges taking into account arch-fill interaction,
tion of the behaviour of masonry in compression, the Engineering Structures, 27: 605–615.
analysis can follows the progressive flexural strength Clemente, P., Occhiuzzi, A., Raithel, A. 1995. Limit
decrease until the hinge mechanism forms. behaviour of stone arch bridges, ASCE Journal of Stru-
cural Engineering, 121: 1045–1050.
de Felice, G. 2006. Experimental investigation on his-
toric brickwork subjected to eccentric axial load, Proc.
5 CONCLUSIONS SAHC’06 Conference, New Delhi, India.
Fanning, P.J., Boothby, T.E., Ronert, B.J. 2001. Longitudi-
An approach for estimating the load carrying capac- nal and transverse effects in masonry arch assessment,
ity of multi-span masonry arch bridges is proposed Construction and Building Materials, 15: 51–60.
that makes use of a beam model with fibre cross sec- Gilbert, M. 2005. Ring, Theory and Modelling Guide, Uni-
tion. A numerical investigation of one to three span versity of Sheffield, Sheffield, UK.
Gilbert, M., Melbourne, C. 1994. Rigid-block analysis
masonry arch bridges under travelling concentrated
of masonry structures. The Structural Engineer, 72:
load, shows that, by parity of constitutive assumptions 356–361.
(infinite strength in compression, or finite strength Heyman, J. 1966. The stone Skeleton. International Journal
with unlimited ductility) the method provides the same of Solids and Structures, 2: 249–279.
estimate of the ultimate load given by limit analysis, Heyman, J. 1982. The masonry arch, Ellis Horwood.
as well as a similar kinematics at collapse. How- Melbourne, C., Gilbert, M., Wagstaff, M. 1995. The
ever, while limit analysis method implicitly assumes behaviour of multi-span masonry arch bridges. Proc.
a perfectly ductile behaviour, the proposed approach 1st International. Conference on Arch Bridges, London,
is able to take into account the effective behaviour of Thomas Telford.
Molins, C. Roca, P. 1998. Capacity of masonry arches and
masonry in compression, including the post-peak soft-
spatial structures, ASCE Journal of Structural Engineer-
ening branch. Once the limited ductility capacity is ing, 124 (6): 653–663.
taken into account, a significant decrease in load car- Page, J. 1987. Load test to collapse on two arch bridges at
rying capacity is found, which depends on the slope of Preston, Shropshire and Prestwood, Staffordshire, Depart-
softening branch: a reduction between 15% and 30% ment of Transport, TRRL Research Report 110, TRL,
has been estimated for the bridges under study, when Crowthorne, England.
the material behavior turns from perfectly plastic to Prentice, D.J., Ponniah, D.A. 1994. Testing of multi-span
post-peak brittle. Now, since masonry cannot be con- model of masonry arch bridges. Proc. Centenary year
sidered as perfectly ductile in compression, it appears bridge conference, Cardiff, Elsevier Science: 169–174.
Robinson, J.I., Ponniah, D.A., Prentice, D.J. 1997. Soil pres-
that the use of fibre beam method provides a signif-
sure measurements on a multi-span brick arch. Proc. 7th
icant improvement in the evaluation of load carrying Int. Conf. On Structural Faults and Repair, Eng. Technics
capacity of masonry arch bridges. Press, Edinburgh: 111–119.
Salençon, J. 1985. Calcul à la Rupture et Analyse Limite,
Presses de l’ENPC, Paris, FR.
REFERENCES Spacone, E., Filippou, F.C., Taucer, F. 1996. Fibre beam-
column model for non-linear analysis of R/C frames:
Brencich, A., De Francesco, U. 2004. Assessment of multi- Part I. Formulation, Earthquake Engineering and Struc-
span masonry arch bridges. Part I: a simplified approach. tural Dynamics, 25: 711–725.
ASCE Journal of Bridge Engineering, 9(6): 582–590.
546
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: In masonry building, while the behaviour of regular walls (vertically and horizontally aligned
opens) was exhaustively assessed in the past, irregular walls (not aligned openings and geometry varying both
along the horizontal and vertical direction), although representing the greatest part of existing constructions,
are still object of study by many researchers. In this paper, some well-known design methods were used and
compared through applications to masonry irregular walls. Non linear static analyses on four cases-study of
bearing wall (1: regular; 2: horizontally irregular; 3: vertically irregular; 4: horizontally and vertically irregular)
allowed investigating and assessing the influence of irregularity in masonry structures. Finally, the procedure was
applied to a real wall still damaged. The following results are discussed: (i) collapse mechanisms of masonry
panels composing the wall; (ii) initial stiffness, peak load and ductile branch in the base shear versus top
displacement diagram.
1 INTRODUCTION
547
function; (ii) Spandrel panels (included between two 1 2 n
openings in the horizontal direction) bearing the slab 1 H1
548
H2d H3s
H1d H3d H1s H2s
(a) (b)
549
1 2 3 4 5 1 2 3 4 5
1.1 1.2 1.3 1.4 1.5
1.1 1.2 1.3 1.4 1.5
1 1
2.1 2.3 2.5
2.1 2.3 2.5
2 2
3.1 3.2 3.3 3.4 3.5 3.1 3.2 3.3 3.4 3.5
3 3
4.1 4.3 4.5 4.1 4.3 4.5
4 4
(a) (b) Figure 8. Characteristic curves of pier panel 2.1: (a) positive
1 2 3 4 5 1 2 3 4 5 seismic verse; (b) negative seismic verse.
1.1 1.2 1.3 1.4 1.5 1.1 1.2 1.3 1.4 1.5
1 1
2.1 2.3 2.5 2.1 2.3 2.5
2 2
3.4 3.5 3.4 3.5
3 3
4.1 4.3 4.5 4.1 4.3 4.5
4 4
(c) (d)
550
(a) (b)
(c) (d)
551
Figure 15. Capacity curves of walls: (a) regular; (b) irreg-
ular height; (c) irregular length; (d) irregular length and
Figure 14. 3MURI failure mechanisms: (a) regular; height.
(b) irregular height; (c) irregular length; (d) irregular length
and height.
allows the application through the use of spread-
sheets. This implies a simpler use of the methodology
5 COMPARISON OF RAN, SAM AND 3MURI and the possibility of adapting it to specific needs,
although it is more time consuming.
Once the three analyses were completed, a comparison Finally, a quantitative comparison of the three
of methodologies has been carried out and discussed methodologies has been performed. Some non-linear
as follows. static analyses of the whole structure were carried out.
Obviously, the first difference among procedures For regular walls (Fig. 15.a) the three curves show
is the geometric modelling of the panels. Extensively nearly the same value of shear force at the base. The
reported in §2, it is opinion of the authors that a pair- RAN curve is stiffer than the 3MURI and SAM one.
ing of the three methodologies is definitely necessary. The 3MURI ductile branch is longer than SAM’s. Once
RAN method rightly considers the updating of the geo- the vertical irregularity is introduced, RAN method
metric configuration according to the seismic verse, provides an higher shear force and stiffness value than
whilst SAM and 3MURI rightly evaluate the external 3MURI and SAM. These latter show the same initial
cross panels dimension. stiffness but can not be compared in terms of peak load
Regarding mechanical characteristics, the parame- and displacement (Fig. 15.b). The same considerations
ters required by 3MURI are practically the same of on the regular wall can be made for the horizontal irreg-
RAN whilst SAM, being mainly targeted to artifi- ular wall (Fig. 15.c). The last irregular wall, which
cial units, requires a more refined calibration in the considers both the horizontal and vertical irregularity,
perpendicular and parallel directions of load. shows the same behaviour of the vertical irregular wall
Another difference of methods is represented by (Fig. 15.d).
the collapse mechanisms of pier and spandrel panels. In all these cases the RAN method does not provide
The 3MURI and SAM methods did not detect shear the same ductility value as the other two methods, since
collapse mechanisms in any of the panels, but only the non linear analysis was conducted under force but
mechanisms of flexural-axial load failure or uncracked not displacement control (as SAM and 3MURI).
panels. Conversely, in the application of the RAN
method, shear collapse mechanisms influenced the
panel pier 2.3 for horizontal irregular wall. The pres- 6 ANALYSIS OF A REAL DAMAGED WALL
ence of this kind of collapse mechanism revealed by
the RAN method for stocky panel appears more consis- Having examined the four kinds of irregularity, a real
tent with the geometry and load condition of the panel. wall in a building placed in Avellino County and hit by
It has to be admitted that a general disagreement about the 1980 earthquake (Magnitude 6.9) was analyzed.
the failure modes is detected. Since no retrofit interventions were made after the
Moreover, it can be observed that, whilst SAM seismic event, it was possible to carry out a detailed
and 3MURI, being implemented in computer codes, geometrical, mechanical and load survey. In Figure 16
request the use of a “close” calculus, the RAN method the irregular wall and the crack pattern are shown.
552
Figure 17. Panel discretization: (a1) RAN method for pos-
itive forces; (a2) RAN method for negative forces; (b) SAM
method; (c) 3MURI method.
Value
kN/m3
553
7 CONCLUSIONS Augenti, N. & Romano, A. 2008. Seismic design of masonry
buildings through macro-elements. XIV International
This paper deeply investigates the behaviour of Brick and Block Masonry Conference; Proc. intern. conf.,
masonry irregular walls through some well-known Sydney, 17–20 February 2008.
Benedetti, D. & Tomazevic, M. 1984. Sulla verifica sismica
analysis methods. The hypothesis at the base of each,
di costruzioni in muratura. Ingegneria Sismica I: 9–16.
the procedure adopted, their criteria and applica- Cattari, S., Curti, E., Galasco,A. & Resemini, S. 2005.Analisi
tive capabilities have been listed. Modelling through sismica lineare e non lineare degli edfici in muratura.
macro-elements allows a remarkable reduction of the Napoli: Esselibri – Simone.
degrees of freedom in respect to Finite Elements. For Dolce, M. 1991. Schematizzazione e modellazione degli edi-
the applicative simplicity and the plurality of offered fici in muratura soggetti ad azioni sismiche. L’industria
possibilities, macro-element methodologies are cer- delle Costruzioni 242: 44–57.
tainly a valid solution for the practitioner analysing Gambarotta, L. & Lagomarsino, S. 1997. Damage models for
a simple masonry residential building. the seismic response of brick masonry shear walls. Part I:
the mortar joint model and its applications. Earthquake
Different kinds of simple irregularities in masonry
Engineering and Structural Dynamics 26: 423–439.
walls were individuated and analyzed starting from Gambarotta, L. & Lagomarsino, S. 1997. Damage models for
a regular wall and then introducing the irregularities the seismic response of brick masonry shear walls. Part II:
increasing the grade of complexity. Great attention the continuum model and its applications. Earthquake
has been paid to the geometrical discretization of Engineering and Structural Dynamics 26: 441–462.
panels, according to damages observed after earth- Magenes, G. & Calvi, G.M. 1997. In-plane seismic response
quakes. Then, in order to compare the numerical of brick masonry walls. Earthquake Engineering and
analyses performed with reality, a wall damaged by Structural Dynamics 26: 1091–1112.
a severe earthquake in the 80’s has been considered. Magenes, G. & Della Fontana, A. 1998. Simplified Non-
linear Seismic Analysis of Masonry Buildings. Proc. of
Non linear analyses were performed on both the four
the British Masonry Society 8: 190–195.
case-studies and the real wall: the collapse mecha- Magenes, G., Bolognini, D., & Braggio, C. 2000. Metodi
nisms of the panels and the capacity curves have been semplificati per l’analisi sismica non lineare di edifici in
plotted. muratura. Roma: CNR – Gruppo Nazionale per la Difesa
Unfortunately, due to the sensitiveness of the dai Terremoti.
response depending on the used method, irregular Penna, A. 2001. Una procedura a macroelementi per l’analisi
walls are still characterized by a great uncertainty as dinamica non lineare di edifici in muratura. PhD The-
regards a correct modelling. sis, Politecnico di Milano, Dipartimento di Ingegneria
Further researches are necessary and it is opinion Strutturale, Milano.
Presidenza del Consiglio dei Ministri, O.P.C.M. n. 3431
of the authors that a pairing of the procedures relying
del 3 Maggio 2005, Ulteriori modifiche ed integrazioni
on RAN, SAM and 3MURI methods could be a good all’O.P.C.M. n. 3274 del 20 marzo 2003, Roma.
start. Raithel, A. & Augenti, N. 1984. La verifica dei pannelli
murari. Second Conference ASS.I.R.C.O.; Proc. nat. conf.,
Ferrara, 1984.
ACKNOWLEDGMENTS Raithel, A. & Augenti, N. 1984. La deformazione dei pannelli
murari. Second Conference ASS.I.R.C.O.; Proc. nat. conf.,
Ferrara, 1984.
This research has been supported by ReLUIS “Rete
Raithel, A. & Augenti, N. 1984. Il pannello murario vinco-
di Laboratori Universitari Ingegneria Sismica” in the lato. Second Conference ASS.I.R.C.O.; Proc. nat. conf.,
context of the activities of “Linea 1 – Valutazione e Ferrara, 1984.
Riduzione della Vulnerabilita’ di Edifici in Muratura”. Raithel, A. & Augenti, N. 1984. Il legame T-D nei pannelli
murari. Third Conference Seismic Engineering in Italy;
Proc. nat. conf., Roma, 1987.
REFERENCES Norme Tecniche per le costruzioni del 14. settembre 2005,
Roma.
Augenti, N. 2004. Il calcolo sismico degli edifici in muratura, Tomazevic, M. 1978. The computer program POR. Report
Torino: Utet. ZRMK, Institute for testing and research in materials and
Augenti, N. 2006. Seismic behaviour of irregular masonry structures, Ljubljana, (in sloveno).
walls. First European Conference on Earthquake Engi- Tomazevic, M. 1982. Lateral load distribution in masonry
neering and Seismology; Proc. intern. conf., Geneva, 3–8 buildings. VI International Brick and Block Masonry
September 2006. Conference; Proc. intern. conf., Rome, 16–19 may 1982.
554
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Geometrical and static aspects of the Cupola of Santa Maria del Fiore,
Florence (Italy)
A. Passerini
Leonardo Società di Ingegneria S.r.l., Florence, Italy
ABSTRACT: The purpose of this research is to clarify, in the language of differential geometry, the geometry
of the internal surface of Brunelleschi’s dome, in the Cathedral of Santa Maria del Fiore, Florence; the statics
of a Brunelleschi-like dome have also been taken into consideration. The masonry, and, in particular, the “lisca
pesce” one, together with the construction and layout technologies, have been main topics of interest for many
researchers: they will be the subjects of further research.
555
perchè nel murare la praticha insegnera quello ches-
sara a seguire” (Brunelleschi’s specifications), nor
about the mechanical apparatus he would have used
later to raise the heavy weights “tirare i pesi per via
di contrappesi e ruote, che un sol bue tirava quanto
avrebbero appena tirato sei paia” (Vasari 1550).
556
Figure 4. Surveyed points of the western web of the dome.
557
Figure 6. The estimated straight line at 64◦ degree on the
Figure 5. The estimated straight line at 64◦ degree on the vertical plane projection.
horizontal plane projection. The values in the figure represent
the difference between the calculated points and the surveyed
ones.
558
Table 1. Surveyed dimensions. Notice: curves φ=cost are parallel horizontal lines,
whose projections on i1 , i2 are inclined β with respect
Dimensions m. to line AC; curves v=cost are translations of the ruling
circle in direction v.
AB 17,26 Deriving the vector x with respect to each Gaussian
AA 36,11
coordinate (φ , v):
BB 36,23
AF 44,94
BG 44,96
AA /AF 0,803
BB /BG 0,805
Zenithal angle AIV (lantern) 36.21◦
Through the cross vector product the unit vector
normal to the cylinder is readily obtained:
559
Then:
560
Figure 9. Equilibrium of a cylinder surface with the lines
of curvature. q is the vertical force for unit area.
Figure 10. Equilibrium along the web ribs.
561
Figure 12. The Brunelleschi-like meshed dome. Figure 14. The diagram of the 1ST principal stress in Pa.
Nonlinear solution.
562
DAVIDSOHN, R. 1956. Storie di Firenze. Sansoni
DI PASQUALE, S. 1977. Primo rapporto sulla Cupola di
Santa Maria del Fiore, CLUSF, Firenze
DO CARMO, M. P. 1976. Differential Geometry of Curves
and Surfaces, Prentice Hall, New Jersey
FITCHEN, J. 1961. The Construction of Gothic Cathedrals,
Oxford
FONDELLI, M. 2004. La Cupola di Santa Maria del Fiore, in
Giuseppe Rocchi Coopmans de Yoldi, S. Maria del Fiore,
ALINEA, Firenze
GIOVANNI DI GHERARDO DA PRATO, 1421. Document
conserved in the Museum of Opera del Duomo. Firenze
GUASTI, C. 1887. Santa Maria del Fiore: la costruzione
della chiesa e del campanile secondo i documenti tratti
dall’archivio dell’Opera Secolare e da quello di Stato, Tip.
M. Ricci, Firenze
Figure 16. The diagram of the 1ST principal stress in Pa with
HEYMAN, J. 1977. Equilibrium of Shell Structures (Oxford
the lantern. Nonlinear solution
Engineering Science), Oxford University Press
HEYMAN, J. 1966. The Stone Skeleton, Int. Journ. Solids
Then, the equations of the differential geometry of and Struct., 2
the dome are useful to understand and quantify the KRAUS, K. 1998. Fotogrammetria, Levrotto & Bella, Torino
geometrical properties of the dome. MANETTI, A. 1480?. Vita di Filippo Brunelleschi
Finally, the linear solution of the finite element NAGHDI, P.M. 1972. The Theory of Shells and Plates.
analysis fits the considerations of Timoshenko and Handbuch der Physik VI, Springer-Verlag
Heyman, but shows positive stresses, unsuitable for OPERA DEL DUOMO, 1691. Archivio di Stato. Filza 366.
the brick masonry. Firenze
The equilibrium of the dome is possible with a ROCCHI COOPMANS DE YOLDI, G. 1996. S. Maria del
diffuse cracking, especially in the lower part. Fiore, Università degli Studi di Firenze, Dipartimento
di Storia dell’Architettura e del Restauro delle strutture
architettoniche, Firenze
SANPAOLESI, P. 1962. Brunelleschi, G. Barbera, Firenze
REFERENCES SOKOLNIKOFF, I. S. 1951. Tensor Analysis – Theory and
applications, John Wiley and Sons
ANSYS® Inc. Southpointe 275 Technology Drive Canons- TIMOSHENKO, S. P. 1959. Theory of Plates and Shells, Mc
burg, PA Graw Hill
BARTOLI, L. 1994. Il disegno della cupola del Brunelleschi VASARI, G. 1550. Le vite de’più eccellenti pittori scultori e
Firenze Leo Olschki Editore architetti Firenze
CECCHI, A. & PASSERINI, A. 2006. Survey, digital recon- WITTKOWER, R. 1962. Architectural Principles in the Age
struction, finite element model of the Augustus Bridge in of Umanism, Alec Tiranti Ltd., London
Narni (Italy). 5th International Conference on Structural
Analysis of Historical Constructions, New Delhi 2006
563
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: In this paper the problem of out-of-plane collapse of masonry walls under vertical and horizontal
loads is addressed through the review of ancient building rules and the study of the European (Eurocode 6 and 8)
and new Italian seismic codes on unreinforced masonry buildings.
The limitations of the analysed seismic codes and the empirical geometrical rules of ancient constructors are
compared with the results of some structural analyses carried out on rectangular masonry panels with different
slenderness h/s, aspect ratio h/b (height/width) and boundary conditions. In particular, the analyses have been
carried out both with the limit analyses and non linear FEM analyses with Abaqus computer code.
The results of these parametric analyses show that the respect of the seismic codes limitations on the geomet-
rical slenderness h/s allows to implicitly have a minimum resistance to out-of-plane collapses of the masonry
walls.
A proposal is made to overcome the unjustified provisions of some codes which do not allow to have distance
among orthogonal walls larger then 7 m.
565
Nf/Npl
N
h 1L 2L 3L 4L
1.0 b
s (a) (b) (c) (d)
)
.75
4L
=0
80
(ρ2
Figure 1. Effect of slenderness on resistance of a masonry 3L
'03
5)
TC 1)
0. 7
2=
C6
wall under axial load. (ρ2=
'07
(8) '03 (ρ
6'03
:E
60 = 1)
C
E (ρ2
(5)
(3):
C6
:N
'03
C6
:E
: E
(6)
(in the chart Nf is normalized to the squash load Npl 40 (4)
(1): EC6'03 (ρ2=0.75)
of the wall). It can be observed that the diagram can (2): EC6'03 (ρ2=1) 2L
be divided in three fields, characterized by different 20 (7): NTC'07
where: hef is the effective height of the wall; sef is In the EC6 the slenderness λc should not be greater
the effective thickness of the wall, that is usually than 27 when the wall is subjected mainly to vertical
taken equal to the actual thickness s (for cavity walls loading. This limitation on λc , together with the sug-
the EC6’03 suggests the following formula: sef = gested limitations on ρ, may be used to establish the
(t31 + t32 )1/3 where t1 and t2 are the actual thickness of maximum h/s ratio allowed by the EC6.
the leaves; the Italian code NTC’07, instead, in every In Figure 3 the relationships between h/s and h/b
case assumes sef = s)1 ; ρ is a reduction factor depend- are shown in graphical form. In particular, the curves
ing on the edge restraint or stiffening of the wall. (1)–(2) are associated to the limits of the wall 2L,
In the examined codes (EC6’03 and NTC’07) the while the curves (3)–(4) and curves (5)–(6) are asso-
suggested values of the coefficient of restraint ρ are ciated to the limitations (3) and (4), for wall 3L and
related to the aspect ratio h/b and to the degree of the 4L, respectively.
edge restraints. Particularly, with reference to the walls The comparison among the curves (1) ÷ (6) of
of Figure 2, the EC6’03 makes the difference among Figure 3 underlines the effect of the lateral edge
the following three cases: restraints on the required h/s ratio. For example, in
– wall 2L (Fig. 2b): the value of ρ (appointed as ρ2 ) the case of wall with aspect ratio h/b = 1, the maxi-
is depending on the degree of restraint (ρ2 = 0.75 mum values of geometrical slenderness h/s is equal to
in presence of r. c. floors at the top and bottom of 36 (ρ2 = 0.75) or 27 (ρ2 = 1.0), for the wall 2L; it is
the wall; ρ2 = 1.0 for walls restrained at the top and equal to 38 (ρ2 = 0.75) or 30 (ρ2 = 1.0), for the wall
bottom by timber floors); 3L; it is equal to 56 (ρ2 = 0.75) or 54 (ρ2 = 1.0), for
the wall 4L.
Instead, the new Italian code NTC’07 does not con-
1
In the following part of this paper, we will assume sef = s. sider the case of wall 3L. In fact, in NTC’07 are
566
recommended reduction factors ρ only in the case of 12·β=λ/(h/s)
wall 2L (ρ2 = 1.0) and wall 4L, for which ρ depends 1.2
only by the aspect ratio: ρ4 = f(h/b).
(2): EC6'03 (ρ2=1)
The maximum slenderness λc permitted in the 1.0
NTC’07 is 20; this limit, together with the recom- (1): EC6'03 (ρ2=0.75)
mended values of ρ, allows to plot in Figure 3 the
curves (7) and (8) associated to the walls 2L and 4L, 0.8
respectively.
The comparison among the curves (1)÷(6) associ- 0.6 (3): NTC'07
ated to the limits of EC6’03 and to the curves (7) and
(8) derived by the NTC’07, shows that the Italian code 0.4
is more conservative than the EC6 both for wall 2L and 4L
wall 4L (Fig. 3). For the case of wall 4L, in Figure 4 it is
proposed the comparison between the coefficients of 0.2
restraint β and ρ, that define the relationships between ρ=λc/(h/s)
the slendernesses λ and λc and the h/s ratio (Eqs. (1) 0
and (2), respectively). Therefore, the diagram allows to 0 0.2 0.4 0.6 0.8 1.0 1.2
compare the differences between the theoretical slen-
derness λ (Eq. 1) and the conventional slenderness λc Figure 4. Comparison between the coefficients of restraint
(Eq. 2). β and ρ.
To define the reduction factors β of the wall 4L, it
is utilized the elastic theory of the buckling of the rect- NTC’07, only for small values of reduction factors
angular plates under compressive loads. If the wall’s (lower than 0.5) there is coincidence between β and ρ;
edges may be considered pinned, the critical stress σcr for higher values (greater than 0.5), in EC6 (ρ2 = 0.75)
of the wall 4L becomes: the factor β is larger than ρ, while in the NTC’07 β is
lower than ρ.
By observing the curves plotted in Figure 4, it is
possible to found an approximate relationship between
λ and λc . In fact the curves (1), (2) and (3) suggest the
where: m is the number of half waves in witch the following identities:
plate buckles in the vertical direction (m = 1, 2, . . ., n);
E·s3
D = 12·(1−ν 2 ) ; E is the modulus of elasticity; ν is the
567
Nf/Npl 3.0 α (4) h/s=10 (3)
1 λc,max
e/s=0.05
2.5 4L 3L
0.8
e/s=0.10 EC6'03 α4=2.4
e/s=0.15
2.0
e/s=0.20
0.6
e/s=0.25 1.5
0.4 e/s=0.30 α3=1.2 2L
1.0 α2=0.8
e/s=0.33 (2)
0.2 0.5 1L
α1=0.1 (1)
0 λc 0.0 h/b
27
0 5 10 15 20 25 30 0 0.5 1 1.5 2 2.5 3
λ
108
0 ~ 20 ~ 40 ~ 60 ~ 80 ~ 100 ~120
(a) Figure 6. Comparison among the multipliers α of the walls
Nf/Npl 1L, 2L, 3L and 4L.
1 λc,max
When the wall is sufficiently retrained at the top
0.8 NTC'07
and bottom by the floor diaphragms (wall 2L) the
multiplier α becomes (Hendry et al. 1997):
0.6
m=6e/s=0
0.4 m=6e/s=0.5
m=6e/s=1
0.2
m=6e/s=2 m=6e/s=1.5 Instead, if the wall is stiffened on one or on two
0 λc
vertical edges (wall 3L and 4L, respectively), in the
0 5 10 15 20 25 30 hypothesis that the yield lines are slanted at 45◦ , the
λ collapse multipliers α are, respectively:
0 ~ 17 ~ 34 ~ 51 ~ 68 ~ 85 ~102
(b)
568
30 h/s (d) h/s=12 1+(h/b)2
(a)); h/s = 10, medium stability (curve (b)); h/s = 12,
low stability (curve (c)).
4L Particularly, the curve (d) shows that the required
h/s=21 h/s ratio to guarantee the stability of the wall increases
20
with the aspect ratio h/b of the masonry panel; for
h/s=17 1L example, if h = 3.5 m, for h/b = 1 it is necessary to
(c) h/s=12 have a minimum value of h/s = 17, i.e. a wall thick-
10 (b) h/s=10 ness s = 21 cm; instead, for h/b = 1.5 it is necessary to
(a) h/s=8 have a minimum value of h/s = 21, i.e. a wall thickness
s = 17 cm.
The above Rondelet’s rules, in terms of collapse
0 h/b
1.5 multiplier α (Eqs. (10) and (13)), lead to the curves
0 1 2 3
(a)÷(d) reported in Figure 8. The comparison among
these curves shows the great increase of resistance to
Figure 7. Comparison among the rules of J. B. Rondelet.
out-of-plane loads of wall 4L (curve (d)) with respect
to wall 1L (curves (a)÷(c)). Further, the trend of curve
3.0 α 4
(d): α4= (d) shows that the multiplier α4 increases with h/b
2.5 4L 3· 1+(h/b)2 · (1-2/3·h/b) ratio; indeed, the curve (d) is asymptotic to the vertical
line h/b = 1.5. In such meaning, Rondelet writes that
2.0 the collapse of wall is impossible when the distance b
1.5 between the transversal stiffening walls is low.
569
3.0 α
(5): EC8'03 (ρ2=0.75)
(9): NTC'07 (seismic zones 3 and 4)
(8): NTC'07 (seismic zones 1 and 2)
30 h/s 2.5 4L
)
(q): seismic zones 3 and 4 (λc =12)
=1
ρ2
(p): seismic zones 1 and 2 (λc =10)
3(
2=
1) 2.0
(ρ
8'0
5) '03
EC
20
): '0 3 ( ρ2=0.7 ): EC8 1.5 (n): seismic zones 3 and 4 (λc =12)
C8 (4
(6
(3): E (1): EC8'03 (ρ2=0.75);
(m): seismic zones 1 and 2 (λc =10)
NTC (seismic zones 3 and 4) 1.0
10 2L
(2): EC8'03 (ρ2=1)
0.5
(7):NTC (seismic zones 1 and 2) 2L h/b
0.0
0 h/b 0 0.5 1 1.5 2 2.5 3
0 1 2 3
Figure 10. Collapse multipliers for EC8’03. Figure 12. Comparison among the limitations of Ron-
delet, EC8’03 and NTC’07 (in seismic zones) on collapse
In terms of collapse multiplier α (Eqs. (11)÷(12)), multiplier α.
the condition (16) recommended in EC8’03 allows to
obtain the curves plotted in the diagram h/b-α of Fig- walls 2L and 4L it is possible to note a certain coin-
ure 10. Particularly in figure are reported: the curves cidence among the plotted curves; particularly, it can
(e) and (f) for the wall 2L; the curves (g) and (h) for the be observed that the Rondelet’s curve (d) is overlaid to
wall 3L; the curves (i) and (l) for the wall 4L. Each cou- the curve (q) (associated to NTC’07) and to the curve
ple of curves is associated to the condition ρ2 = 0.75 (g) of EC8’03 (wall 3L, ρ2 = 0.75) when h/b > 1 and
or ρ2 = 1 that the EC6’03 considers in the definition h/b < 0.8, respectively.
of hef .
The limitations (17) and (18) of Italian code
NTC’07 (seismic part) are plotted in the h/s-α dia- 4 F.E.M. PARAMETRIC ANALYSIS
grams of Figure 11 for wall 2L (curves (m) and (n))
and for wall 4L (curves (p) and (q)). The diagram In this paragraph, non linear analyses on rectangular
shows that in high seismicity zones (curves (m) and walls are carried out through the F.E.M. computer code
(p)), the Italian code recommends a minimum value of Abaqus 6.7-1 [Simulia, 2007]. On the basis of these
collapse multiplier α higher than the case of low seis- analysis results, some considerations about the col-
micity (curves (n) and (q)). Moreover, the curves (p) lapse multiplier α of walls 1L, 2L, 3L and 4L subjected
and (q) associated to wall 4L show a constant trend in to horizontal loads, are derived.
the range h/b = 0.5÷1, i.e. in the most common range
of aspect ratio h/b for the masonry walls. The pres-
4.1 Modelling of the masonry walls
ence of the horizontal plateau in the curves (p) and (q)
depends on the fact that the Italian code uses the same In Table 1 are reported the characteristics of the 42
expressions of the collapse mechanisms as the ones examined models. The specimens have been labelled
examined in §3.1 to define the reduction factor ρ. according to the following criteria: h-b-s Ln, where the
In Figure 12 the comparison among the limitations h, b and s are the geometrical dimensions of the wall
of Rondelet, EC8’03 and NTC’07 (in seismic zones) (in cm) and Ln (n = 1÷4) defines the edge restraint
on collapse multiplier α, is provided. In the cases of of the wall as schematized in Figure 2. Regarding the
570
Table 1. Examined models. 3.0 α
4L
Specimen 350-350-35
2.5 h/b=1; h/s=10
Specimen h/b h/s α4=2.4
2.0 3L F.E.M. analysis
350-350-45 L1 1 8 1.5 α3=1.2 Limit analysis
350-350-45 L2 1 8 1.0 α2=0.8 2L
1L
350-1400-45 L3 0.25 8 0.5
α1=0.1
350-700-45 L3 0.5 8 0.0 d [mm]
350-470-45 L3 0.75 8 0 5 10 15 20 25
350-350-45 L3 1 8 (a)
350-280-45 L3 1.25 8
350-233-45 L3 1.5 8
350-1400-45 L4 0.25 8
350-700-45 L4 0.5 8
350-470-45 L4 0.75 8
(b)
350-350-45 L4 1 8
350-280-45 L4 1.25 8
350-233-45 L4 1.5 8
350-350-35 L1 1 10
350-350-35 L2 1 10
350-1400-35 L3 0.25 10 (c)
350-700-35 L3 0.5 10
350-470-35 L3 0.75 10 Figure 13. Specimens 350-350-35: (a) comparison
350-350-35 L3 1 10 among the push-over curves and the collapse multipliers;
350-280-35 L3 1.25 10 (b) deformed shapes with stress tensor vectors of the F.E.M.
350-233-35 L3 1.5 10 models; (c) collapse mechanisms considered for limit
350-1400-35 L4 0.25 10 analyses.
350-700-35 L4 0.5 10
350-470-35 L4 0.75 10
350-350-35 L4 1 10 In order to correctly calibrate the model parameters,
350-280-35 L4 1.25 10 reference has been made to the curve fitting procedure
350-233-35 L4 1.5 10 was made by Giordano A. (2002), which utilizes the
350-350-30 L1 1 12 results of the experimental tests on masonry tuff walls.
350-350-30 L2 1 12 The applied loads are the self weight and the hor-
350-1400-30 L3 0.25 12 izontal load, which increases with monotonic low up
350-700-30 L3 0.5 12 to the end of the analysis.
350-470-30 L3 0.75 12
350-350-30 L3 1 12
350-280-30 L3 1.25 12 4.2 Results and comparisons
350-233-30 L3 1.5 12
350-1400-30 L4 0.25 12 In Figure 13a, a summary of the non linear analyses
350-700-30 L4 0.5 12 for the walls 350-350-35 is reported; particularly, the
350-470-30 L4 0.75 12 pushover curves are depicted. In this diagram, the col-
350-350-30 L4 1 12 lapse multipliers calculated through the limits analyses
350-280-30 L4 1.25 12 and already provided in Figure 6 are also reported.
350-233-30 L4 1.5 12
The comparison among the nonlinear static and limit
analyses shows that the F.E.M. results provide collapse
multipliers higher than limit analyses, because the ten-
geometry, all the walls have height h = 350 cm; the sile strength of masonry is considered in the Abaqus
width b varies between 233 and 1400 cm in order to models.
have walls with aspect ratio between 1.5 and 0.25. The Indeed, in Figure 13b and c the visualization of
thickness s is set equal to 30, 35, 45 cm, in order to have the F.E.M. deformed shapes, with stress tensor vec-
walls with h/s ratio equal to 12, 10 and 8, respectively. tors, and the hypothesized collapse mechanisms are
Four-nodes shell elements (S4R5 elements) are reported. The comparison shows a good agreement
used to model the masonry tuff walls; reduced inte- between the Abaqus and limit analyses.
gration is used for the shell elements; the number In Figure 14 the comparison among the collapse
of integration points through the thickness of shell multipliers computed by means of F.E.M. analysis and
element is equal to five. of limit analyses (Eqs. (10)÷(13)) is provided. Par-
All the examined walls have been subjected to non ticularly, the diagrams refers to walls characterized
linear analyses using a smeared cracking approach as by h/s ratio equal to 10. In almost all analyzed spec-
implemented in the computer code Abaqus. imens (Table 1) the diagrams confirm the previous
571
6 α (4) (3) h/s=10 (Fig. 15a) and NTC’07 (Fig. 15b). The thicknesses s of
4L 3L the analyzed specimens have been obtained by Figure
5 9 curves (4) (wall 3L) and (6) (wall 4L) for EC8’03,
Abaqus 4L
4 and by Figure 9 curves (8) and (9) for NTC’07. In the
case of EC8’03, the results of F.E.M. analyses show
3 an increasing trend of the curves (Fig. 15b), while in
Abaqus 3L the case NTC’07 it possible to note a sub-horizontal
2
2L trend of the F.E.M. curves.
1 (2) Moreover, when h/b ratio is larger than one, the
0 h/b NTC’07 curves show a slight decrease of collapse mul-
0 0.5 1 1.5 2 2.5 3 tipliers; this seems to underline that the limitations of
NTC’07 give very large values of maximum h/s ratio
Figure 14. Comparison among the collapse multipliers when the wall is characterized by high values of aspect
of walls with h/s = 10 evaluated through F.E.M. and limit ratio h/b.
analyses. Finally, the comparison among the F.E.M. results
reported in the curves (a) of Figure 15 shows that the
6 α (l) (ρ2=1) (h) (ρ2=1) EC8’03 is on the safe side with respect to NTC’07; in
4L 3L fact, it can be observed that the curve “Abaqus 4L” of
5
Figure 15a is characterized by higher values of collapse
4 Abaqus 4L multipliers α than the curves “Abaqus 4L” plotted in
Figure 15b.
3
2 Abaqus 3L
5 CONCLUSIONS
1
The study of the out-of-plane collapses of masonry
0 h/b
(a) 0 0.5 1 1.5 2 2.5 3 walls subjected to vertical loads has suggested the
relationship between the slendernesses λ and λc or,
equivalently, between the coefficients of restraint β
6 α
and ρ. In term of slenderness λ, it has been observed
4L
5 that a wall characterized by λ < 20 can be defined as
stocky wall, having a low reduction of axial strength
4 Nf with respect to the squash resistance Npl of the wall;
(p): seismic zones (q): seismic zones
3 1 and 2 (λc=10) 3 and 4 (λc=12) when λ > 50÷60, instead, the wall can be considered
as slender wall. Finally, if the slenderness λ is com-
2 Abaqus 4L prised between 20 and 50÷60, the wall is characterized
1 by an intermediate behaviour, with interaction between
the crushing and the elastic buckling of the wall.
0 h/b In the second part of this paper, the issue of
0 0.5 1 1.5 2 2.5 3
(b) out-of-plane collapses of masonry walls subjected to
horizontal loads has been studied through (i) the anal-
Figure 15. Comparison among the collapse multipliers yses of the collapse mechanisms; (ii) the review of
evaluated through nonlinear static and limit analyses on ancient rules of the art; (iii) the study of the European
walls designed according to the geometrical requirements of (EC8’03) and new Italian seismic codes (NTC’07) for
EC8’03 (a) and NTC’07 (in seismic zones) (b).
unreinforced masonry buildings.
In the case of walls 3L and 4L, the comparison
observation regarding the multipliers α; in fact, the among the geometrical limitations underlines that the
values of α computed through pushover analyses are recommended h/s ratio becomes greater than the val-
generally higher than the values calculated with the ues h/s = 8, 10 or 12, that Rondelet suggests for the
limit analyses. Finally, also the curves associated to the wall 1L. For example, in the range h/b = 0.5÷1.5, the
results of nonlinear static analyses carried out on walls required geometrical slenderness h/s for the wall 4L
3L and 4L show an increasing trend with the aspect is equal to or smaller than: 13÷21 according to Ron-
ratio h/b; this confirms the beneficial effect of the ver- delet’s rule (Eq. 15); 11÷27, according to EC8’03;
tical edge restraints on the resistance of the walls to 10÷32, according to NTC’07.
out-of-plane collapses. These observations are also confirmed in the prac-
The Figure 15 shows the comparison among the tice for a wider range of aspect ratio h/b, as shown
F.E.M. results and the curves relative to collapse mul- in Figure 16 for the 8 classes of macro-elements of 10
tipliers α of the walls 3L and 4L associated to EC8’03 masonry churches studied in Brandonisio et al. (2008).
572
50 h/s masonry walls has been presented. In the F.E.M. anal-
yses both h/s and h/b ratios and the edge restraints have
40 been varied. The results confirm the observation done
through the application of the limit analyses, i.e. the
respect of the seismic codes limitations on the geo-
30
metrical slenderness h/s allows to implicitly have a
minimum resistance to the out-of-plane collapses of
20 the masonry walls subjected to horizontal loads.
h/s=12 In the light of the previous consideration, it could
10 h/s=10 be suggested to overcome the indications concerning
h/s=8
h/b
the maximum spacing between the cross walls equal
0 to 7 m often recurrent both in ancient treatises and in
0 1 2 3 4 5 6 the modern seismic codes. In particular, it seem that
a more appropriate limitation on orthogonal elements
Figure 16. h/s vs. h/b relationships for the masonry walls of should be given in non dimensional forms, i.e. in terms
10 Italian masonry churches analyzed in Brandonisio et al. of aspect ratio of the masonry panel: h/b ≥ 0.5.
(2008).
573
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: Environmental degradation, inevitable aging, negligence and extreme events are some of the
most frequent factors that can threaten the structural integrity of historical constructions. Hence, architectural
heritage owners and administrators urgently require cheap and effective tools for structural health assessment.
Restoration works on the Holy Shroud Chapel in Turin, Guarino Guarini’s masterpiece which was extensively
damaged by a fire in 1997, offered the authors the opportunity to test some innovative techniques that are able
to provide an exhaustive picture of the Chapel’s structural health. Dynamic tests were carried out by applying
several exciting actions and the signals were acquired to perform a modal identification of the dome in the
time-frequency domain. A stochastic multiple model approach, used to update a FE model of the Chapel, is
also presented. The PGSL algorithm, implemented in the model updating procedure, is able to define different
groups of models, each one referring to a particular damage scenario that the building is probably undergoing. The
improved knowledge about the structure is the starting point for the development of symptom-based structural
health assessments.
575
have been applied as the starting point for the develop- obtained cannot be considered as the only possible
ment of symptom-based structural health assessments: one. Furthermore, there are many error and uncertainty
the Holy Shroud Chapel is presented as an example. sources that condition the final result. These should be
considered in order to obtain a consistent estimation
of the model parameters and consequently a reliable
2 STRUCTURAL HEALTH ASSESSMENT
structural health assessment.
TECHNIQUES
On the basis of the remarks made so far, the authors
focused attention on the development of a model
The monitoring of common structures represents a
updating method which could meet monumental build-
well established and widespread practice, whereas the
ing requirements:
situation of monumental buildings is rather different.
Only a few cases can be found referring to research – the treatment of errors obtained from the acquisi-
due to the strict constrains necessary to preserve the tion of experimental measures;
cultural and artistic value of the buildings. Tools which – the treatment of uncertainties due to construction
combine a suitable level of reliability with less invasive complexity (mechanical properties, constructive
interventions are preferred. Dynamic system identifi- methodologies, unknown design details, etc.);
cation in this sense constitutes an optimal compromise – the management of information redundancy
which is finding widespread approval among experts. obtained from many distributed sensors;
Dynamic tests are very powerful identification tech- – extrapolation ability in the definition of numerical
niques which are able to reveal global quantities such models.
as natural frequencies, modal shapes and damping
Solution accuracy is the main problem which the
features of the structure under examination. These
authors tried to solve in a convincing way. As a unique
parameters are closely related to the variations in
optimal solution for this problem cannot be found,
stiffness of the system and are therefore extremely sen-
the authors concentrated on the possibility of generat-
sitive to damage and deterioration phenomena. Many
ing a set of different but sufficiently reliable models.
modal identification techniques are now available but
The adopted approach is generally referred to as the
the complexity and uncertainty that affect ancient
“multiple-model” due to the fact that it generates
masonry constructions drastically reduce their appli-
a huge number of models to solve the problem of
cation possibilities. Time-frequency domain methods
the solution uniqueness. Application of the adopted
have proved to be accurate in treating non-stationary
method to the case study has allowed some models
signals and they are robust against noise. Further-
to be defined and has made them able to accurately
more, some instantaneous estimators, obtained as auto
predict its structural behaviour. An interesting devel-
and cross-correlation of bi-linear time-frequency sig-
opment the authors intend putting into practice regards
nal transforms, allow the modal features of systems to
the identification of possible damage scenarios for
be reliably detected.
each model obtained using the updating procedure.
In the structural health assessment field FE mod-
The natural consequence is the estimation of the resid-
elling also plays a very important role, even thought in
ual performance a building is still able to supply, i.e.,
many cases the lack of numerical modelling accuracy
its structural health.
and a deficient correspondence to reality represent
The case of the Holy Shroud Chapel in Turin is pre-
decisive limitations. There is the risk of choosing
sented in order to clarify the concepts introduced so far.
model parameters which do not fit real structure prop-
erties well. Model Updating (MU) provides a valuable
solution to the problem. It can be considered as the
3 CASE STUDY: THE HOLY SHROUD CHAPEL
missing link between structural identification and FE
IN TURIN
modelling since it gives a model calibration crite-
rion on the basis of the experimental results. Model
3.1 The experimental campaign
updating techniques have undergone an exponential
growth during the last few years due to their latent The Holy Shroud Chapel in Turin (Fig. 1) is univer-
potentialities. Recently, stochastic approaches have sally recognized as a outstanding example of Italian
become more and more popular and can be distin- Baroque architecture. Emanuale Filiberto di Savoia
guished from traditional deterministic methods due entrusted its design to the famous Italian architect
to their capacity to deal with behaviour uncertainties Guarino Guarini, who built the Chapel from 1667 to
and problem complexity in a robust way. Deterministic 1694. Since the very beginning, the monarch’s intent
methods, whether direct or parametric, focus on model was that of housing the precious relic of Christianity
calibration through the minimization of a target func- in a more prestigious seat. The architectural results
tion obtained as the difference between experimental obtained by Guarini are extraordinary. The whole
results and analytical data. Since Model Updating building conveys the Guarini’s obsession for architec-
represents an inverse problem, the solution that is tonic originality and a sense of mystery which are well
576
previously been covered over. As a result of these
investigations, a 3D computer geometrical model of
the entire building was set up and more detailed
knowledge in the structural morphology was acquired.
The mechanical properties of the materials were
measured both in laboratory and in situ tests, in the
latter case using cross-hole ultrasonic measurements.
The tests revealed the chemical degradation induced
by the heat on the surface of the elements and a reduc-
tion in the material load carrying capacity in the zones
affected by stresses caused by the effect of constrained
thermal deformations.
Local inspections provided useful information con-
cerning about the mechanical properties of the mate-
rials and structural configuration and allowed a first
preliminary estimation of extent of the damage to be
obtained. However, due to their spatial limitation, these
tests were not able to completely identify the global
behaviour of the construction. They therefore had to
be integrated with other techniques that were capa-
ble of predicting the residual structural capacity of the
building considered as a whole.
In order to achieve a complete overall image of the
Chapel’s structural health, the Research Unit to which
the authors belong, in association with the University
Figure 1. Section of the dome.
of Kassel (Prof. M. Link), designed a dynamic test-
ing programme, prepared a FE model and adjusted it
according to the acquired experimental knowledge.
expressed by the structural complexity and the rich-
In order to perform the dynamic experiments, 25
ness of perfect shapes and theological, astronomic and
accelerometers were positioned on six different lev-
mathematical symbols. The chapel is composed of a
els to measure the response of the structure along
tambour bored by six large windows and surmounted
three different orthogonal directions: radial, vertical
by three big arches which sustain the dome. On the
and horizontal normal to the radius. The dynamic tests
inside, a series of small arches overlapped and dis-
were realized adopting four different exciting actions:
posed on six levels, creates an hexagonal geometry
which diminishes towards the top where it becomes the – environmental excitation (traffic, wind, micro-
circular base of a lantern. On the outside, another series quakes);
of small arches creates a complex plaiting effect and – impulsive excitation produced by hammering;
the alternation of black marble and grey stone grants – impulsive excitation caused by a sphere dropped to
a particular sense of dynamicity. the ground near the foot of the building;
The recent history of the Holy Shroud Chapel has – wind turbulence produced by a Fire-Brigade heli-
been marked by a tragic event. A fire broke out in 1997 copter flying around the top of the dome.
during some restoration works and it seriously dam-
The obtained signals were used to identify the struc-
aged the structure, producing incalculable economic
ture of the dome using the TFIE (Time Frequency
and artistic loss.
Instantaneous Estimators) method.
In the following months, the Sovraintendenza
dei Beni Architettonici encharged the Politecnico di
Torino with the project to carry out a general experi-
3.2 Modal identification
mental campaign on the materials and structure (Prof.
P. Napoli) and a dynamic test program (Prof. A. De The results obtained from the dynamic tests performed
Stefano). on the chapel provided the starting point for the subse-
Mainly in situ investigations were performed in quent structural identification phase. A well-designed
order to obtain detailed knowledge of the structural structural health monitoring programme cannot disre-
morphology in term of geometry, marble and masonry gard this fundamental step which supplies information
organization, position of metal ties, etc. The investi- concerning the overall behaviour of the system and it
gations, conducted through topographic and camera is the basis for the FE model calibration.
surveys and deep extraction of samples, allowed the Many effective techniques are now available for
experts to detect some structural elements that had structural identification, but in the case of complex
577
monumental buildings not all are suitable to robustly
face the several uncertainties that can affect this kind
of structure. Generally, output-only methods are pre-
ferred to input-output ones since they are less invasive.
Identification techniques can also be distinguished on
the basis of the parameters that need to be identified. It
is therefore possible to choose between direct methods,
which try to determine the [M], [C] and [K] matrices
of motion equation (1):
578
Table 1. Subdivision of the elements in the FE model.
579
All these parameters are found in the central and lower 2. the Probability Updating Cycle, in which all the
part of the structure. These parts in fact have a greater probability distributions are updated as follows:
influence on the global behaviour of the structure, the probability values of the sub-interval containing
mainly for the first modal shapes. the Best Sample selected in the preceding cycle and
All the selected parameters are the Young’s Mod- its neighbours are increased, while, the probabilities
ula of different sub-structures into which the model in the other regions are decreased.
was previously divided. Each substructure is an axial- 3. the Focusing Cycle, which modifies the structure
symmetric structure; this fact influences the behaviour of the variation interval of each parameter in order
of the structure, making it more regular in every to focus the search near the actual best solution.
direction. To avoid this, a further subdivision of each 4. the SubDomain Cycle, which works on the solution
sub-structure into four different ones was made. This space and increase the resolution in order to ease
subdivision leads to an increase in the number of the confluence.
selected parameters from 5 to 20 and is necessary
Each operation performed by the algorithm is man-
because the data obtained in the identification stage
aged by specific parameters which have to be cho-
highlight a non axial-symmetric behaviour of the
sen carefully because some of them present a high
structure.
dependence on the considered problem.
In the case study, the Root Mean Square Error
3.4 The stochastic model updating between the identified and calculated frequencies is
utilized as the objective function which has to be min-
In order to solve the intrinsic problem of the solution
imized. At each step, a new parameter distribution is
uniqueness and to treat error and uncertainty sources
created by the algorithm and new model outputs are
a stochastic “multiple-model” approach has been
obtained. In order to match the numerical and exper-
adopted to update model parameters. This method can
imental modal shapes, the MAC function, that gives
be divided into the three following phases:
a measure of similarity between two different modal
1. generation of a huge number of models varied on shapes, is calculated.
the basis of some structural parameters through the The model generation procedure has been divided
minimization of objective functions that are able to into two different phases, each one referred to a partic-
explore a consistent part of the solutions space; ular objective functions. Firstly, the PGSL algorithm
2. selection, among all the generated models, of generates a huge number of models through the mini-
those which best fit the system behaviour with a mization of a function set on the first modal frequency,
prescribed level of accuracy; which is the best-identified in the dynamic tests. The
3. analysis of the detected model properties and their second objective function instead utilizes the first
classification through clustering techniques. three model frequencies and mode shapes matched by
means of the MAC function. This second step allows to
The adopted method implements the PGSL algo-
select only some of all generated models, while mod-
rithm introduced by Smith (1998) which creates the
els that have a higher value than a certain threshold are
models and assigns them parameter values derived
neglected.
from probability distributions. The principal assump-
All these selected models can be considered as rea-
tion is that better points are likely to be found in the
sonable candidates to represent the real building, but
neighbourhood of families of good points. Hence, the
their large number makes necessary the application of
search is intensified in regions containing good solu-
some data mining techniques that can be applied to
tions. The search space is sampled by means of a
discover different types of patterns in the model data.
PDF defined over the entire search space. Each axis
The PCA (Principal Component Analysis) is per-
is divided into a fixed number of intervals and a uni-
formed to generate a new set of variables called
form probability distribution is initially assumed. As
principal components that are linear combinations of
the search progresses, the intervals and probabilities
the original ones. The goal of the PCA is to find a
are dynamically updated so that sets of points are gen-
system of principal components that are sorted so that
erated with a higher probability in regions containing
the first components can explain most of the data vari-
good solutions.The search space is gradually narrowed
ance. In this way the number of dimensions can be
down so that convergence is achieved.
reduced by choosing only the first two or three princi-
The algorithm includes four nested cycles:
pal components. Thus the original data are represented
1. the Sampling Cycle, where a certain number of by a linear combination of the original parameters in a
models is generated and analysed through the new and lower dimensional space. The application of
target function, using the actual probability distri- the PCA and the examination of the first three prin-
bution. The Best Sample is selected, memorized cipal components supplied an important result. The
in the database and then used to recalculate the first three principal components are not sufficient to
probability distribution; explain most of the data variance. For this reason it
580
its subdivision into sub-structures and the sensitiv-
ity analysis, performed in order to reduce the num-
ber of the subsequently updated parameters, are also
shown.
The last part of this study is focused on the devel-
opment of a stochastic model updating technique. The
adopted approach is known as “multiple-model” due
to the fact that it generates a huge number of models
to solve the problem of the impossibility to find an
optimal unique solution and to treat error and uncer-
tainty sources. The PGSL algorithm is implemented
in order to create these models and to assign parame-
ter values derived from their probability distributions,
progressively recalculated around those values which
minimize a target function. The application of this kind
Figure 5. Percentage of the selected models which belong of model updating procedure raises the problem of
to each group. information redundancy, which can be faced through
the adoption of some data mining techniques such
as the PCA and clustering. The final obtained results
are some models that can be considered as reasonable
is not possible to represent all data in a 3D reference candidates to represent the real structure.
system. The models obtained from the updating procedure
Finally, some clustering techniques have been can be considered as the starting point for the following
applied to identify and collect models into groups. The phase of damage detection and structural health assess-
adoption of the k-means algorithm, which sub-divides ment. It is now important to underline that these results
the data into k subsets minimizing the distances within are obtained from a linear analysis while damage and
each group and maximizing the distances between the degradation effects are strictly non-linear phenom-
k different groups, allowed the definition of 9 models. ena. Therefore, the subsequent numerical simulations,
These models are the centroids of the 9 groups identi- applied to identify symptoms which can reveal pos-
fied by the k-means algorithm and can be considered sible damage scenarios on the construction, cannot
as the best candidates to represent the real structure. disregard these non-linear implications. The updated
The obtained results are reported in Figure 5, where models supplied by the multi-model approach repre-
the percentage of the selected models which belong to sent a useful resource in order to estimate the health
each group is shown. condition of the building. Non-linear analyses per-
formed on the obtained models can reveal the effective
residual performance the construction is still able to
provide for each damage scenario.
4 CONCLUSIONS
581
Mottershead, J.E. & Friswell, M.I. 1993. Model Updating in Saitta, S., Raphael, B. & Smith, I.F.C. 2005. Data Mining
Structural Dynamics: a Survey. In Journal of Sound and Techniques for Improving the Reliability of System Iden-
Vibration, 1993, 162: pp. 347–375. tification. In Advanced Engineering Informatics, 2005,
Raphael, B. & Smith, I.F.C. 2003. A Direct Stochas- 19(4): pp. 289–298.
tic Algorithm for Global Search. In Journal of Smith, I.F.C. 2005. Multi-Model Interpretation of Mea-
Applied Mathematics and Computation, 2003, 146: surement Data with Errors. CANSMART 2005, RMC,
pp. 729–758 Canada, 2005: pp 13–22.
582
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The Paderno d’Adda Bridge, Lombardia, northern Italy, is one of the very first great iron con-
structions designed through the practical application of the theory of the ellipse of elasticity, a graphical-analytical
method of structural analysis that was developed in the 19th century. It embeds a natural discretisation of the
structure into a series of elastic elements, treated then with standard tools of geometry of masses. In this work,
the application of such theory to the calculation of the parabolic arch of the bridge is inquired, attempting to
breathe, at the same time, the beauty of the architectonic and structural conception directly linked to that; later,
results are compared with much modern approaches that also consider now-available numerical discretisation
methods. A further, definite aim of this work is also that of trying to promote interest on the bridge, on its actual
state of conservation and future destinations. Not only it represents a true industrial monument and a living tes-
timony of the scientific and technological developments of the time but also a beautiful, effective achievement
of architecture and engineering through the methods of Strength of Materials.
583
This theory was originally conceived by Karl Culmann ±8.63◦ to the vertical. The parabolic axis of the arch
(1821–1881) and then systematically developed and has a span of 150.00 m and rise of 37.50 m in the
applied by his pupil Wilhelm Ritter (1847–1906). It inclined plane; the transverse arch’s cross section is
represents a very elegant method for the analysis of 4.00 m high at the keystone and 8.00 m high at the
the flexural elastic response of a structure and is based abutments (i.e. in the same 1:2 ratio between rise and
on an intrinsic discretisation of a continuous beam in half span). The two mean inclined planes of the arches
a series of elements, each with a proper elastic weight, are located at a distance of 5.096 m at the keystone and
directly proportional to its length and inversely pro- 16.346 m at the shoulders. The wall of each composing
portional to its bending stiffness. The theory of the arch is also a truss structure with two main T-ribs con-
ellipse of elasticity is based on the concepts of projec- nected by vertical and inclined bars. The two couples
tive geometry, which lead to a correspondence between of twin arches are gathered together by two transverse
the ellipse of elasticity of the structure and the central brace systems located at the extrados and at the intra-
ellipse of inertia of the distribution of the elastic weight dos of the arch’s body. In essence, the resulting cross
of the structure. This correspondence brings back the section of the main parabolic arch supporting the hor-
problem of the determination of the flexural elastic izontal beam is trapezoidal, with variable, increasing
deformation of a beam to a problem of pure geome- cross section from the crown to the shoulders. This,
try of masses, of more convenient solution and direct and specifically the inclination of the twin arches, is a
interpretation in terms of the design of the structure. beautiful key feature of Röthlisberger’s conception of
the bridge, in view of counteracting effectively wind
1.2 Main technical features of the bridge and transverse horizontal actions in spite of the con-
siderable slenderness of the structure. The arch cross
The main technical features of the bridge are reported
section at the impost is inclined of 45◦ to the hori-
in details in Nascè et al. (1984), which is, to our knowl-
zontal, so as the local tangent to the parabolic axis
edge, the most comprehensive publication, and one of
of the arch to the vertical. The vertical bridge piers
the very few, concerning the bridge. We rely very much
that sustain the upper continuous beam are made by
on this very valuable contribution and on the Techni-
eight T-section columns, linked to each other by a brace
cal Report (SNOS 1889) that was originally issued
system with horizontal bars and St. Andrew’s crosses
at the time of the first try-out. Here, the essential
and, on top, by transverse beams that directly serve as
characteristics are reported.
supports for the bearing devices of the upper beam.
The 266 m long upper flyover is made by a continu-
For inspection and maintenance purposes a 1 m large
ous box girder with nine equally-distributed supports,
boardwalk is provided into the body of the arch and,
at 33.25 m distance from each other. Four of the sup-
inside the bridge piers, a system of ladders along their
ports are sustained by a big parabolic metallic arch; two
height. The bridge is a riveted wrought iron structure of
of them bear directly on the same arch’s masonry abut-
about 2600 t of metals, with near 100000 rivets just in
ments (made with Moltrasio masonry, with Baveno
the arch.
granite coverings); a seventh, on the Calusco bank,
rests on a smaller masonry foundation placed between
1.3 Aim of this work
the arch shoulder and the higher bridge supports; the
last two, in masonry work as well, are the two direct In this work, which in its main part largely refers to
beam bearings at its two ends, on top of the two river a study developed in a Laurea Thesis (Ferrari 2006),
banks. The four piers resting on the arch are placed a detailed analysis of the SNOS Report (1889) is pre-
symmetrically, in between keystone, haunches and sented. The point of view here is the following: inquire
shoulders of the arch. The inner side of the beam girder, the application of the theory of the ellipse of elastic-
on which the railway is located, runs at about 255.00 m ity to the calculation of the bridge, breathe the beauty
on the sea level (osl); the rails are placed at 255.45 m of the architectonic and structural conception directly
osl, the upper road at 261.75 m osl. The main verti- linked to that, compare results with modern structural
cal longitudinal trussed beams of the upper continuous approaches that also consider now-available numerical
girder are 6.25 m high and placed at a respective trans- discretisation methods.
verse distance of 5.00 m, leaving a free passage for After a careful review of the Report by the SNOS,
the trains of 4.60 m. They are composed of two main a full 3D truss Finite Element model of the arch of
T-ribs connected by a metallic truss. The upper-level the bridge has been elaborated, based also on direct
road is 5.00 m wide and includes also two additional inspections of the bridge and on the screening of
cantilever sidewalks, each 1 m long, with iron parapets the marvellous original drawings that are guarded
1.50 m high. at the Archivio Storico Nazionale di Torino. Dif-
The big arch is composed by two couples of sec- ferent loading conditions have been considered and
ondary inclined arches. Each couple is formed by two results compared with those reported in the SNOS
arches posed at a respective distance of 1 m and lay- Report, showing the remarkable accuracy of the
ing symmetrically to a mean plane inclined of about adopted graphical-analytical methods and allowing to
584
experience the unrepeated beauty of the original anal-
ysis with respect to rather impersonal computer struc-
tural analysis. Moreover, the model that has been put
in place shows promise for possible further analyses
that could inspect other behaviours of the bridge, as for
example dynamical and inelastic, also connected to the
present and future state of conservation of the struc-
ture. These aspects are left for further developments
of the present study.
The paper is organized as follows: Section 2 pro-
vides a short account on the theory of the ellipse Figure 2. Representation of a force R applied along line of
of elasticity; Section 3 reports its application to the action r to a beam section A and the centre of rotation C of
structural analysis of the arch of the bridge; Sec- A. C is the symmetric of C with respect to the centre S of the
tion 4 presents an independent validation of the orig- ellipse of elasticity of beam AB referred to section A.
inal design results with present analytical-numerical
methods.
in view of its conception and design. Furthermore, the
ellipse itself may actually play the role of an hidden,
underlying, graphical construct. Indeed, the proper-
2 ON THE THEORY OF THE ELLIPSE OF
ties of projective geometry that are attached to that
ELASTICITY
allow the elastic solution of the structure even without
the explicit drawing of the ellipse itself. The methods
2.1 Fundamentals
are also said graphical-analytical because, in practice,
The theory of the ellipse of elasticity can be consid- main technical steps that are framed on the graphi-
ered as a main icon of the so-called Graphical Statics, cal constructions may be carried-out analytically, by
the discipline which often characterised the resolving working-out formulas that arise from the inspection
approach of practical design problems during the 2nd of the drawings (Belluzzi 1942).
half of the 19th century. It represents a very elegant The concept of the ellipse of elasticity referred to a
and practical method for the analysis of the flexural section of an elastic structure is achieved by inspect-
response of an elastic structure. It is based on an intrin- ing the correspondence existing between the line of
sic discretisation of a continuous elastic problem. This action r of a force R applied to a section A of a gen-
theory is basically associated to the two outstanding eral, curvilinear, elastic beam (with little curvature and
figures of Culmann and Ritter, but also of people, like continuously-varying cross section) and the centre of
Giulio Röthlisberger, that were formed at the time at rotation C of the same section (Fig. 2).
the Polytechnical Schools in Europe and that became In particular, refer to a planar beam acted upon
later structural engineers and designers and largely by forces laying in the same plane and cross sections
contributed in the practical and effective application of the beam that, during the beam’s deformation, are
of the method. assumed to remain plane and perpendicular to the geo-
The theory is based on the following main hypothe- metric axis, also deforming in its original plane. The
ses (we refer here to the Italian text by Belluzzi 1942, theory states that there exists an involutory relationship
which reports results from the technical literature between the line of action r and the centre of rotation
of the time, basically ascribed to the two names of C of the section. Moreover, the ellipse of elasticity
Culmann and Ritter): (a) linear elastic behavior of is the fundamental real conic of the polarity existing
the material and the structure, which leads to the between the line of action r and the point C , which is
proportionality between acting forces and (reversible) the symmetric of C with respect to the centre S of the
displacements provoked by them (property that in turn ellipse. In other words, the ellipse of elasticity can be
implies the validity of the principle of superposition defined as the fundamental conic with respect to which
of effects); (b) existence of the ellipse of elasticity, the lines of action r and the respective centres of rota-
referred to a section of a structure; (c) correspondence tion C correspond to each other through an antipolarity
between the latter and the central ellipse of inertia of relationship.
the distribution of the so-called elastic weight of the The determination of the central ellipse of inertia of
structure.This correspondence transforms the problem the distribution of the elastic weight of the structure,
of the determination of the elastic response of a con- which coincides with the ellipse of elasticity above,
tinuous structure to a task of pure geometry of masses. is linked first to the definition of the general concept
The latter can be feasibly handled by taking advantage of elastic weight and then to the quantification of its
of the assumed discrete character of the distribution of distribution for the structure under consideration. The
the elastic weight and is endowed with a visible inter- concept of elastic weight goes as follows. If on a sec-
pretation of the elastic performance of the structure, tion A of a beam, a moment M acts in the plane which
585
contains the geometric axis, it causes a rotation φ of
A, around the centre S of the ellipse of elasticity. The
rotation is proportional to the applied moment M as:
586
Figure 4. Calculation of line of action LO of left reaction Figure 6. Equilibrium requires that the left and right reac-
A; the segments FL and VO that the left reaction A locates on tions A and B form with load P a closed polygon of forces.
the vertical lines from A and on load P, below and above the Segments V = OQ and V = QP represent the vertical com-
horizontal line from centre S are determined. ponents of A and B, with V + V = P; segment H = TQ their
horizontal component.
graphical-analytical procedure. Figs. 4 and 5 represent where quantities a, a , f, f are represented in Fig. 6.
two ways to solve this problem (SNOS 1889, Ferrari Following similar arguments, it is possible to derive
2006). Underlying to these constructions lay the com- the in-plane arch’s deflections dy at any position x of
patibility conditions φ = 0, dx = 0, dy = 0 for the left the arch, for the different locations a of the load P
built-in constraint at terminal section A. These can be acting on the arch (SNOS 1989). The equation that
worked-out from Eq. (4), leading to: gives the mean parabolic line at the arch is taken as:
587
determined without the constant proportionality fac-
tor E = 17000000 t/m2 , Young’s modulus of the iron,
i.e. G = s/J . The coordinates xS , yS , xY , yX and
total elastic weight G = EG are finally found as:
588
Table 1. Arch’s vertical deflections calculated at design
stage for four different loading tests (SNOS 1889, p. 71).
Negative values indicate downward displacements.
589
Table 3. Bending deflections listed in SNOS (1889, p. 59). 4.2 Structural analysis of the arch by the FEM
Load Pier I Pier II Pier III Pier IV Currently, an attempt is made of building a full FE
Point (m) (m) (m) (m) model of the bridge. So far, a truss mesh of the arch of
the bridge has been developed and appropriate loading
Pier I −0.0089 −0.0029 +0.0071 +0.0042 conditions has been considered for validation of the
Pier II −0.0011 −0.0042 +0.0026 +0.0030 previous results. The FE analysis has been run with
Pier III −0.0004 −0.0003 +0.0005 +0.0001 the commercial code ABAQUS® .
Pier IV +0.0001 +0.0001 −0.0000 −0.0001
Total −0.0103 −0.0073 +0.0102 +0.0071
4.2.1 Structural model
The FE model consists of a 3D truss frame, reproduc-
ing as much as possible the actual arch geometry. It
Table 4. Bending deflections calculated by the VWP. consists of two planar parabolic trusses referring to
the in-plane geometry of the arch (see Fig. 7), placed
Load Pier I Pier II Pier III Pier IV in two inclined planes (of ±8.63◦ to the vertical). The
Point (m) (m) (m) (m) inclined planes are placed at a distance of 5.096 m from
each other at the axis of the arches at the keystone.
Pier I −0.00893 −0.00298 +0.00703 +0.00422
Pier II −0.00126 −0.00425 +0.00255 +0.00297 The truss nodes are linked to each other through a
Pier III −0.00038 −0.00033 +0.00055 +0.00016 reticular system that corresponds to the actual brac-
Pier IV +0.00006 +0.00010 −0.00004 −0.00013 ing of the arch. To each bar of the model, a cross
Total −0.01051 −0.00746 +0.01009 +0.00722 section with equivalent geometrical characteristics is
attributed (area, principal moments of inertia, torsional
stiffness). Some approximations were made as regards
to the section’s attribution to the superior and inferior
Table 5. Total vertical deflections (1st distribution) reported arch ribs, which are made with a variable number of
by SNOS (1889, p. 60) and present results by the FE model. longitudinal plates. Also, at the node junctions, there
are additional reinforcing plates, for local stiffening.
Vertical loads SNOS deflections FE deflections
Point (t) (m) (m)
In spite of this, the model has been simplified with
bars of constant average cross section. The model is
Pier I +340.6 −0.0120 −0.015903 comprised of 752 beam elements and 266 nodes. Built-
Pier II +144.0 −0.0113 −0.013847 in constraints are imposed at the nodes of the arch
Pier III −18.5 +0.0062 +0.010322 shoulders.
Pier IV +4.9 +0.0054 +0.007222
4.2.2 Obtained results
First trial loading cases considered a unitary load (1 t)
regard the vertical deflections (due to pure bending) at applied at the piers and at the keystone, in view of
the four piers due to forces acting at their locations. verifying the order-of-magnitude agreement with the
The reason of the discrepancies above may be due deflections that were listed in the Report and calculated
to the fact that the SNOS might have used in the final here by the VWP. After these preliminary checks, the
Report also data concerning a preliminary project. deflections were evaluated for five given load distribu-
Indeed, the bridge’s layout was slightly modified in tions with vertical loads acting at the four piers (SNOS
the executive project, due to new requirements on the 1889, p. 58–62).
railway trace that were posed by the Strade Ferrate A sample of these outcomes is given in Figure 9
Meridionali, after checks on the Adda’s banks (Nascè for the 1st load distribution already considered in
et al. 1984). It is therefore possible that some specific Tables 3–4, with main results summarized in Table 5.
data were referring to a previous project, while impor- Such distribution considers spans 2–3 charged by a
tant global quantities, such as the deflections caused uniformly distributed load of 9 t/m, leads to a max-
by external actions, were indeed corresponding to the imum pressure on Pier I and is somehow similar to
final one. As a matter of fact, even if this Report has Test IV considered in the try-outs (Fig. 8). The maxi-
been probably conceived to present to a general audi- mum compressive axial force at the intrados of the left
ence the main steps of the calculations, including a shoulder is found in 377.0 t, in good agreement with
very valuable account on the theory of the ellipse of the value 751.6/2 = 375.8 t calculated by the SNOS
elasticity and on its explicit application to the analysis (1889, p. 62). Agreement is also found for the hor-
of the arch, it is doubtlessly very concise and it obvi- izontal thrust H and vertical reactions V , V at the
ously presents concepts and practical considerations shoulders, with H = 185.8 t, V = 417.2 t, V = 53.8 t.
that may not be directly apparent to the contemporary The FE outcomes confirm, to a quite good degree of
reader. accuracy, the SNOS results, showing the true potential
590
bridge. All this shows the ingenuous, effective and
beautiful approach to the design of the bridge. One
might think at this once contemplating the giant still
standing there silently, serving since almost 120 years
of duty, with an actual state of conservation that actu-
ally poses serious questions about its future survival.
We should take care of it.
REFERENCES
Figure 9. FE deformed configuration of the arch for the
Belluzzi, O. 1942. Scienza delle Costruzioni. Bologna:
1st load distribution reported in Table 5 (amplification
Zanichelli.
factor = 500).
Benvenuto, E. 1981. La Scienza delle Costruzioni e il suo
Sviluppo Storico. Firenze: Sansoni.
of the theory of the ellipse of elasticity as applied to Culmann, K. 1880. Traitè de Statique Graphique. Paris:
the elastic analysis of the arch of the bridge. Dunod.
Ferrari, R. 2006. Sulla concezione strutturale ottocentesca
del ponte in ferro di Paderno d’Adda secondo la teoria
5 CLOSING REMARKS dell’ellisse d’elasticità. Laurea Thesis in Building Engi-
neering. Advisor E. Rizzi, Università di Bergamo, 228
An attempt to scrutinise in details the SNOS pages.
Nascè, V., Zorgno, A.M., Bertolini, C., Carbone, V. I.,
Report (1889) has been made in view of breathing the
Pistone, G., Roccati, R. 1984. Il ponte di Paderno: sto-
original conception of the design of the bridge and its ria e struttura – Conservazione dell’architettura in ferro.
specific calculation through the elegant method of the Restauro, Anno XIII, n. 73–74, 215 pages.
theory of the ellipse of elasticity. The intrinsic discreti- Società Nazionale delle Officine di Savigliano 1889.Viadotto
sation in elastic elements is remarkable and makes a di Paderno sull’Adda (Ferrovia Ponte S. Pietro-Seregno).
natural connection with discretisations that can now Torino: Tip. e Lit. Camilla e Bertolero.
be provided by FE codes. The beauty of the executive Timoshenko, S.P. 1953. History of Strength of Materials. New
project can be totally appreciated only by the comple- York: McGraw-Hill.
mentary screening of the 147 drawing tables of the
591
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: This paper applies Thrust-Network Analysis, a three-dimensional computational method for
obtaining lower-bound solutions of masonry vaults with complex geometries. The method extends thrust-line
analysis to three-dimensional problems by finding equilibrium force networks within the vault’s geometry, repre-
senting possible paths of the compression forces. Through two case studies, this paper demonstrates the potential
of the method as a powerful tool for understanding, visualizing and exploring the equilibrium of compression-
only structures. First, an analysis of a series of groin and quadripartite rib vaults investigates the interrelationship
between different parameters and the range of possible equilibrium solutions of these vaults. A second case study
analyzes the fan vaults of King’s College Chapel in Cambridge, England.
593
to construct a hanging string model, Gaudi had to
decide on a suitable force pattern topology to represent
the structural action of the vaults. Then, after choos-
ing the structural logic, it is still challenging to control
or even predict the final shape, since the equilibrium
of each string influences the equilibrium of the entire
network. It is a tedious, iterative process of adjusting
and refining.
New form-finding programs which explore hanging
models in the virtual world, based on dynamic relax-
ation, such as Kilian’s CADenary tool have to deal with
the same issues (Kilian and Ochsendorf 2005). It is
very hard to control and predict how the final shape of
the compression network will look like if local changes
are being made or a string model is being assembled
and hung under gravity.
This is true for both physical and virtual string
models as for graphical methods.
Figure 1. A pseudo-3D analysis of a gothic rib vault using In order to analyze a three-dimensional indetermi-
graphic statics (Wolfe 1921). The web of the vault is cut into nate system these unknowns need to be understood
strips which are analyzed as 2D arches. The main ribs bring and controlled. This can be achieved by describing the
the forces from those arches down to the supports. problem as an optimization problem. The equilibrium
requirements then are formulated as a set of constraints
which have to be satisfied while optimizing a certain
analyst typically must slice the structure, reducing it objective function.
to a combination of two-dimensional problems. In this O’Dwyer (1999) implemented optimization meth-
way, structural behavior is reduced to a combination of ods to investigate masonry vaults by finding possible
arch actions. This process obviously does not capture compression-only force networks which are entirely
the full three-dimensional behavior of the structure and contained within the boundaries of the vault. Recently,
must rely heavily on the chosen discretization. These the problem of controlling a virtual hanging strings
limitations have been the main reasons why thrust- network has been elegantly and efficiently been imple-
line analysis has not been used extensively for the mented within a similar optimization framework by
assessment of complex 3D structures. Andreu et al. (2007).A hanging string network is found
Wolfe (1921) demonstrated how by slicing up a which fits within the inverted geometry of the vault to
structure a global pseudo-3D analysis can be done by be analyzed.
combining local thrust lines (Fig. 1). This methodol- The following section briefly summarizes Thrust-
ogy was entirely manual and quickly becomes tedious Network Analysis, a new fully three-dimensional
(Boothby 2001). Block et al. (2006b) proposed a extension to thrust-line analysis based on projec-
method to produce models which contain the graphical tive geometry, duality theory and linear optimization
construction but which are parametric and interac- (Block and Ochsendorf 2007). Examples in Sections
tive, reducing the tedious iterative nature of traditional 4 and 5 will demonstrate its value for the analysis of
graphic statics. The models demonstrate the range of vaulted structures in unreinforced masonry.
possible solutions and can be used to investigate the
effects on the global stability of the vault of varying
different parameters such as buttress thickness, arch 3 THRUST NETWORK ANALYSIS
thickness, level of fill, etc. Smars developed in his PhD
thesis (2000) computational tools to perform such a 3.1 Motivation
pseudo-3D analysis in an automated fashion, starting
from the actual measured geometry of the vaults. Our goal was to develop a three-dimensional version
of thrust-line analysis similar to previous applications
using interactive graphic statics. This means that the
2.2 Fully 3D equilibrium analysis methods following features should be preserved:
The main problem with three-dimensional equilib- – a graphical and intuitive representation of the forces
rium analysis for masonry vaults is that they are in the system; and
highly indeterminate structures. Antoni Gaudí’s physi- – an interactive exploration of the range of equi-
cal form-finding process for the church of the Colonia librium solutions bounded by a minimum and
Guell can be used to explain this. First, before starting maximum thrust.
594
In order to cope with the challenges of the high degree is represented by the geometry of the other grid and
of indeterminacy of three-dimensional problems, as vice versa.
discussed in section 2.2, we want to be able to: Thrust-network analysis is developed for loading
conditions where all forces are parallel to each other,
– identify and control the many unknowns (degrees
such as gravitational loading. Note that this method
of freedom); and
is therefore appropriate for historic structures in unre-
– negotiate between the unknowns by formulating
inforced masonry since the dominant loading is self-
an optimization problem with different objective
weight. It is important to note that in this case the
functions.
external forces do not appear in the projection of the
Therefore we want to explore the impact of differ- system on the plane perpendicular to the direction
ent assumptions about the force patterns, internal of the forces. This means that a plane force diagram
force distributions, boundary conditions or loading can be produced which represents the equilibrium in
conditions. that plane of the system independent of the externally
applied loads. As a result the force diagram is scale-
3.2 Methodology less since the external forces which typically give scale
to the force diagram are missing.
Thrust-network analysis extends O’Dwyer’s (1999)
work on funicular analysis of vaulted masonry struc- 3.3 Overview of the main steps
tures by adding the reciprocal relationship between
the geometry and the in-plane internal forces of net- The set-up of the program is summarized below. Block
works (Williams 1986), which was first described and Ochsendorf (2007) presents details on the problem
by Maxwell (1864). Figure 2 demonstrates this formulation and solving procedures.
relationship: the internal force equilibrium of one grid (a) Defining a solution envelope:
The compression-only solutions must lie within
given boundaries defined by an intrados and an
extrados (Fig. 3a). These put height constraints
on the nodes of the solution. These limits are
obtained from a three-dimensional model of the
actual vault.
(b) Choosing a force pattern :
In plan, a possible force pattern topology is con-
structed (Fig. 3b). The branches represent possible
load paths throughout the structure. The loaded
nodes represent the horizontal projections of cen-
troids (cf. step d). This pattern is the horizontal
Figure 2. The two plane grids have a reciprocal relationship projection of the final solution.
as defined by Maxwell. The equilibrium of a node in one of (c) Generating the reciprocal force diagram ∗ :
them is guaranteed by a closed polygon in the other and vice The reciprocal force diagram (Fig. 3c) is produced
versa. The labeling uses Bow’s notation (Bow 1873). from the force pattern such that corresponding
Figure 3. The input for the Thrust Network Analysis method: (a) the boundaries, the intrados and extrados of the vault; (b) a
possible force pattern defined on the horizontal plane; (c) the reciprocal force diagram ∗ automatically produced from ;
and the weight associated per node coming from the vault’s self-weight and other imposed loads.
595
Figure 5. Possible objectives for the LO problem: (a) maxi-
mizing the load factor of an applied load; (b) maximizing the
geometric safety factor; and (c) finding the range of thrust
values, defined by a minimum and maximum thrust value.
596
The first force pattern and diagram (Fig. 7a) repre-
sent the minimum thrust state shown in Figure 6. The
second pattern (Fig. 7b) shows a different assumption
on how the forces could travel through the structure: all
force lines go directly to the corner supports. The force
diagram represents in a clear fashion the equilibrium of
the corner nodes and center node. A similar force pat-
tern is used in Section 5 to analyze the fan vaults. The
last pattern (Fig. 7c) shows a fully three-dimensional
Figure 6. Possible thrust values at the corners for this groin force pattern. An important difference with the previ-
vault range from 21% to 32% of its total weight. ous two assumptions is that forces no longer only go
to the corner supports. This network assumes that a
we want to minimize and maximize the scale of the part of the vault is carried along the edges. As a result
force diagram, resulting in globally the smallest ver- the thrust at the corner supports will be reduced. The
sus the largest horizontal forces in the system or also main force lines (heavier lines in the force pattern)
the deepest versus the shallowest solution which fits are distinguished from a continuous, regular grid in
within the boundaries of the structure. between. As can be seen in the force diagram, more
The viability of the proposed method is demon- force is attracted to the main force lines compared to
strated through two case studies, highlighting its mul- the grid lines resulting in primary structural action over
tiple applications. First, an analysis of a series of groin these lines and three-dimensional vault action between
and quadripartite rib vaults investigates the interrela- them. The location of these force lines can for example
tionship between different parameters and the range of be inspired by the location of ribs in the actual vault.
possible equilibrium solutions of these vaults. A sec- Figure 7d shows a three-dimensional compression-
ond case study looks specifically at the fan vaults of only thrust network which fits within the groin vault’s
King’s College Chapel in Cambridge, England. geometry.
Variables influencing the range of vault thrust include The fan vaults of King’s College Chapel in Cambridge,
the influence of the web geometry, the boundary condi- England were constructed between 1512 and 1515
tions, the role and effect of cross-ribs, the assumptions (Fig. 8a). These double-curvature vaults were first
for the internal force patterns, the presence of cracks analyzed by Heyman (1977) using membrane theory.
and other pathologies, and the impact of fill above the These vaults with complex geometries can be analyzed
haunches. using thrust-network analysis.
The groin vault shown in Figure 6 is found to have a From available documentation (Leedy 1980), a
range of possible horizontal thrust values at the corners detailed three-dimensional model is constructed
which vary from 21% to 32% of the total weight of (Fig. 8b, c). This model is used for obtaining precise
the vault. This results from the choice of a pseudo- nodal height constraints and good approximations of
3D force pattern inspired by Wolfe’s analysis (Fig. 1), the weights applied at each node.
where arches are assumed to span between the ribs, Since there are cracks between the transverse arches
which carry the loads to the supports. The presence of and the conoid fan vaults (Leedy 1980), we can assume
Sabouret cracks which run parallel to the edges of the that no compressive forces can be transferred between
vault could justify such a pattern. them. The transverse arches and the fan vaults work
Figure 7 illustrates the relation between the cho- independently of each other. The chosen force pat-
sen force patterns (i.e. the horizontal projection of tern should reflect this, i.e. no branches should cross
the thrust network solution), the reciprocal force dia- the interfaces between the fan vault and the trans-
grams, and the shape of the 3D equilibrium solu- verse arch. Figure 8d shows a thrust network which
tions. The image demonstrates how the force diagrams fits within the vault’s section demonstrating that it is
clearly visualize the internal force distribution of the stable. The force pattern follows the radial rib pattern.
different force path assumptions (e.g. how much more The fill adds weight and alters the thrust-network
force goes into the diagonals compared to the arches but also adds more depth to the section for the thrust-
spanning in between them). Not only the proportions network to travel through. This is very clear if we look
inside the force diagrams are useful, but the global at the equilibrium of the main transverse arch (Fig. 9).
scale allows for a direct comparison of the overall mag- Without the level of fill, this arch would be too thin
nitude of the forces in the system between the different to stand under its own weight. The level of fill causes
force pattern assumptions. an increase of thrust, i.e. the horizontal component, of
597
Figure 7. Different possible force patterns for a groin or rib vault: (a) the diagonal ribs bring down the forces to the corner
supports, arches in the web span in between the ribs; (b) all force lines go directly to the corner supports; and, (c) primary
force lines accumulate towards the supports and a continuous 3-D mesh distributes the loads to these force lines. (d) shows
the three-dimensional thrust network, resulting from the assumptions in (c), which fit within the vault’s geometry.
Figure 8. (a) Inside view of the fan vaults of King’s Chapel in Cambridge, (b,c) A detailed model of the geometry of the
vaults and (d) a 3D thrust-network solution fitting inside the section of the vaults.
598
6 DISCUSSION
599
For the input of the vault geometry and the loading Heyman, J. 1966. The stone skeleton. International Journal
conditions and for the output of the three-dimensional of Solids and Structures 2: 249–279.
results, implementations are written in existing archi- Heyman, J. 1977. Equilibrium of Shell Structures. Oxford:
tectural software. This allows for clear visualizations Clarendon Press.
Heyman, J. 1995. The Stone Skeleton: Structural engi-
of the results, and the smooth integration of the neering of masonry architecture. Cambridge: Cambridge
form-finding process in the analysis process. University Press.
Through several examples, this paper demonstrated Huerta, S. 2001. Mechanics of masonry vaults: The equilib-
the potential of thrust-network analysis as a power- rium approach. In P.B. Lourenço, P. Roca (eds.), Historical
ful tool for understanding, visualizing and exploring Constructions: 47–70, Guimarães.
the equilibrium of compression-only structures such Huerta, S. 2004. Arcos bóvedas y cúpulas. Geometría y equi-
as historic vaults in unreinforced masonry. librio en el cálculo tradicional de estructuras de fábrica.
Madrid: Instituto Juan de Herrera.
Leedy, W.C. 1980. FanVaulting: A Study of Form, Technology,
and Meaning. Santa Monica: Arts and Architecture Press.
REFERENCES Maxwell, J.C. 1864. On reciprocal figures and diagrams of
forces. Phil. Mag. Series 4(27): 250–261.
Andreu, A., Gil, L. & Roca, P. 2007. Computational Analysis Ochsendorf, J.A. 2002. Collapse of masonry structures.
of Masonry Structures with a Funicular Model. Journal PhD. dissertation, Department of Engineering, Cambridge
of Engineering Mechanics 133(4): 473–480. University.
Boothby, T.E. 2001. Analysis of masonry arches and vaults. O’Dwyer, D.W. 1999. Funicular analysis of masonry vaults.
Progress in Structural Engineering and Materials 3: Computers and Structures 73(1–5): 187–197.
246–256. Schek, H.-J. 1974.The Force Density Method for formfinding
Bow, R.H. 1873. Economics of construction in relation to and computation of general networks. Computer Methods
frames structures. London: Spon. in Applied Mechanics and Engineering 3: 115–134.
Block, P. Dejong, M. & Ochsendorf, J. 2006a. As Hangs the Smars, P. 2000. Etudes sur la stabilité des arcs et voûtes.
Flexible Line: Equilibrium of Masonry Arches. The Nexus PhD Thesis, Department of Civil Engineering, Katholieke
Network Journal 8(2): 13–24. Universiteit Leuven.
Block, P., Ciblac, T. & Ochsendorf, J. 2006b. Real-time limit Ungewitter, G. 1890. Lehrbuch der gotischen Konstruktio-
analysis of vaulted masonry buildings. Computers and nen. Leipzig: Weigel Nachfolger.
Structures 84(29–30): 1841–1852. Williams, C.J.K. 1986. Defining and designing curved flex-
Block, P. & Ochsendorf, J. 2007. Thrust Network Analysis: ible tensile surface structures. In J.A. Gregory (ed.), The
A new methodology for three-dimensional equilibrium. mathematics of surfaces: 143–177. Oxford: Clarendon
Journal of the International Association for Shell and Press.
Spatial Structures 48(3). Williams, C.J.K. 1990. The generation of a class of structural
Cremona, L. 1879. Le Figure Reciproche nella Statica forms for vaults and sails. The Structural Engineer 68(12):
Grafica. Milan: Ulrico Hoepli. 231–235.
Föppl, A. 1892. Das Fachwerk im Raume. Leipzig: Verlag Wolfe, W.S. 1921. Graphical analysis: a handbook on graphic
von B.G. Teubner. statics. New Cork: McGraw-Hill.
600
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: In this paper main results on the study of the seismic vulnerability of a basilica-type church are
presented. Static behavior and dynamic properties are evaluated using the finite element method. To this end a 3D
non-linear analysis is done that takes into account the non-linear behavior of masonry: constitutive assumptions,
characterized by elasticity, damage and friction, are made. The standard FEM modeling strategy, based on the
concepts of homogenizated material and smeared cracking constitutive law, is used. Seismic vulnerability is
evaluated using a pushover approach. Results obtained with the FEM technique are then compared with the
simplified schemes of limit analysis, and the capacity of the church to withstand lateral loads is evaluated
together with the expected demands from seismic actions. The comparison demand vs. capacity confirms the
susceptibility of this type of buildings to extensive damage and to collapse, as frequently observed. Starting
from a single case study, a contribution to the issue of modeling and analysis of monumental masonry buildings
under seismic action is provided. Eventually a retrofitting proposal is presented.
601
7m
8m
2160
Figure 1. Aerial view of the Church case study.
41 m
analyses. In particular an in-situ survey on the geom-
etry (and on the structural details) aimed to identify
any irregularities of the structure was made.
This is necessary also to localize possible col-
lapse mechanisms that might be activated during an
earthquake. The investigation consisted in a geomet-
rical survey to identify the actual damage, a quality
check up of connections wall to wall and wall to roof,
and masonry texture characterization. This step was
accompanied by an historical survey of the building
over centuries in order to determine the original shape
and to consider the church modifications over the cen-
turies. The paper reports mainly, for brevity, results on
6m
602
Table 1. Yield criterion (main elements).
603
Figure 6. Vertical stresses σz [N/mm2 ]: facade walls detail.
The maximum compression value is found in the Figure 7. First mode on the nave T1 = 1.704 sec.
columns on the facades (Fig. 6) where the stress state
reaches the value of 2.5 N/mm2 . Even if it is high, this transversal and torsional stiffness, and significant out-
value is lower than the crushing limit of the column’s of-plane deformations of the elements. Furthermore
material (stone). These are the critical elements of the the plan deformed configurations of the structure con-
church as is demonstrated by the fact that they are firm that the seismic loads acting along the church
reinforced by a steel collar. transversal direction involve remarkable out of plane
deformations of the orthogonal structural elements.
2.1.2 Modal analysis
The first two modes of the “as is” building involves the
2.2 Pushover analysis
translation in the two principal direction of the bell-
tower; since the paper is focused on the analysis of the The study of the seismic behavior has been made using
church building, these results are not reported here. a non-linear static analysis method. Seismic loads are
The third modal shape (Fig. 7) is the first translational evaluated with respect to the New Italian rule (DM
mode for the Church, and it acts on the transversal 14/09/05 & OPCM 3274) using a pushover analysis
direction of the nave. Following modal shapes (not (Galasco et al. 2006, Kim & D’Amore 1999). Based
represented here for brevity) are a combination of on this analysis method, the effects of the seismic loads
transversal vibration mode and torsional mode on the are evaluated through the application of two systems of
church. horizontal forces perpendicular to one another. These
The distribution of the modal shapes shows that forces, not acting simultaneously, are evaluated taking
the “as is” building, though characterised by very into account two load distributions. The first load dis-
stiff structural elements on the perimeter, displays low tribution is directly proportional to building’s masses
604
(uniform); the second load distribution is proportional
to the product of the masses for the displacements of
the corresponding first buildings modal shape. These
load distributions could be considered as two limit
states for the building’s capacity. The first distribu-
tions assume that the horizontal loads representative of
inertia seismic forces are constant with respect to the
building’s height. This means that the displacements
on the lower level of the building are overestimated,
while the opposite happens on the displacement on
the top level. On the contrary the second distributions
overvalue the displacements on top level. In this study
a conventional pushover is assumed, i.e. loads applied
on the building don’t change with progressive degra-
dation of the buildings during loading. This means
that the conventional pushover does not account for Figure 8. Displacement (mm).
the progressive changes in modal frequencies due to
yielding and cracking on the structure during loading. The peak ground acceleration (PGA) for the reference
This is a critical point for the application of conven- return period is denoted by ag and for the case study
tional pushover to the analysis of historic masonry is assumed equal to 0.35 g. S is a factor depending on
buildings, because it is predictable that the progressive the ground type that in this case (ground type A) is
damage of the building may also lead to period elonga- equal to 1.0. Next, results concerning church behavior
tion, and therefore to different spectral amplifications are detailed. The critical load distribution is the one
(Antoniou & Pinho 2004). Hence the hypothesis of acting on the y-direction (direction perpendicular to
invariance of static loads could cause an overestima- main nave direction). This is quite expected due to
tion in the analysis of masonry building especially the fact that the transversal direction of the church
when a non uniform damage on the buildings or a high involves remarkable out of plane deformations of the
level of cracking are expected. Further studies will be orthogonal structural elements.
devoted to the investigation of this point. However, also Figure 8 reports the displacements on the transver-
in its conventional form, the pushover provides an effi- sal direction at the end of the analysis; Figure 9 reports
cient alternative to expensive computational inelastic the corresponding crack pattern. The capacity curve is
dynamic analyses. reported in Figure 10 with the corresponding behavior
The seismic loads to apply to the building have of the equivalent bilinear systems. For this system it
been evaluated through modal analysis with the elas- has been obtained, by eq. (2) (see DM 14/09/05 for
tic response spectrum. For the case study the Class more details), that the seismic displacement demand
1 spectrum as reported in the New Italian Rule (DM is dmax = 89 mm against the building displacement
14/09/05), with a ground type A corresponding to rock capacity that is equal to 88 mm.
or other rock-like geological formation, is assumed.
Following expressions are derived:
605
Figure 9. Cracking pattern (pushover, y-direction).
kN Capacity Curve
20000
18000
16000
14000
12000
10000 Figure 11. Reference case.
8000
Original MDOF system
6000
4000
Bilinear SDOF system bell-tower and the clock-tower, they are not reported
2000 here for brevity’s sake.
0
0 10 20 30 40 50 60 70 80 90
displacements (mm)
simplified schemes based on the limit analysis. The After the evaluation of the collapse multiplier, the
structure is considered as the assemblage of a certain corresponding seismic spectral acceleration is evalu-
number of components depending on the structural ated by (where M∗ is the mass of the idealized SODF
compound geometry and shape (e.g. the whole façade) system and g is the acceleration of gravity):
and on the details (e.g. quality of existing connec-
tions) whose behavior is similar to the behavior of
analogous macro-elements in other buildings, where
the collapse mechanisms can be identified. In this
study, several possible mechanisms have been taken
into account (D’Ayala & Speranza 2003), mostly out-
of-plane mechanisms. Next the results are reported The respect of the analyzed limit state is assured by
with respect to the most diffused collapse mechanisms. the respect of the following inequality (where q is the
To this end, several elementary macro-elements behaviour factor assumed equal to 2.0, Z is the height
have been identified in the church; applying the The- of the mass above the level of application of the seismic
orem of Virtual Work (TVW) the collapse multipliers action and H is the total height of the macro-element):
(α0 ) have been evaluated by eq. (3) that refers to
the simple scheme illustrated in Figure 11. After the
identification of the collapse mechanism a rigid-body
system is considered in application of the TVW. The
seismic load is assumed as an overturning force while The step where the church is subdivided into macro-
the gravity load is assumed as a stabilization load. elements is a crucial one. It consists in identifying the
Analyses of past events have permitted to select architectonic elements where the seismic behavior acts
a group of mechanisms to analyze. These mecha- quite independent by the rest of the structure. Major
nism concerns partial of local overturning of the main macro-elements could be searched on the façade,
façade, of the lateral bearing walls, of the apse. Other on the apse or on the bell tower. For each macro-
mechanisms take into account the behavior of the element must be considered its typology and their
606
Figure 15. Mechanism 4: lateral wall overturning.
Figure 12. Mechanism 1: pronaos overturning (local).
connection with the remaining parts of the building. reports results on the mechanisms that could be acti-
All these informations must be collected during the vated on the pronaos (Figs 12–14), in the lateral wall
in-situ survey of the church, and every structural detail of the church (Figs 15–17), in the lateral wall of the
need to be recorded (Brencich et al. 1998). This paper nave (Fig. 18) and on the apse (Fig. 19).
607
Table 3. Collapse multiplier.
608
macro-elements have been identified and their vul- D’Ayala, D., Spence, R., Oliveira, C. & Pomonis A. 1997.
nerability has been assessed. Taking into account the Earthquake Loss Estimation for Europe’s Historic Town
results obtained with the two methods a retrofitting Centres. Earthquake Spectra 13(4): 773–793.
proposal that offer an increase to the seismic resis- D’Ayala, D. & Speranza, E. 2003. Definition of Col-
lapse Mechanisms and Seismic Vulnerability of Historic
tance of the building has been proposed. The proposed Masonry Buildings. Earthquake Spectra 19(3): 479–509.
strengthening use traditional material and techniques D.M. del Ministero delle Infrastrutture e dei Trasporti
in order to propose a reversible intervention. 14/09/2005. Norme Tecniche per le Costruzioni. G.U.
It is believed that the results and the conclusions 23/9/2005, n. 222, S.O. n. 159.
obtained with respect to the static assessment and Galasco, A., Lagomarsino, S. & Penna, A. 2006. On the use
the operative model identification procedure herein of pushover analysis for existing masonry buildings. First
proposed could be extrapolated for a wide variety of European Conference on Earthquake Engineering and
masonry cultural heritage structural compound. Seismology. Geneva, Switzerland, 3–8 September 2006
(paper 1080).
Kim, S. & D’Amore, E. 1999. Push-over analysis procedures
in earthquake engineering. Earthquake Spectra 15(3):
REFERENCES 417–434.
Lagomarsino, S. 1999. Damage survey of ancient churches:
Antoniou, S. & Pinho, R. 2004. Advantages and limita- the Umbria-Marche experience. In A. Bernardini (ed.),
tions of adaptive and non adaptive force-based pushover Seismic Damage to Masonry Buildings, Proceeding of
procedures. Journal of Earthquake Engineering 8(4): the International Workshop on “Measures of seismic dam-
497–522. age to masonry buildings”, Monselice, Italy, Rotterdam:
Binda, L., Saisi, A. & Tiraboschi, C. 2000. Investigation Balkema.
procedures for the diagnosis of historic masonries. Con- Lagomarsino, S. & Podestà, S. 2004. Seismic Vulnerability of
struction and Building Materials 14(4): 199–233 Ancient Churches: I. Damage Assessment and Emergency
Brencich, G., Gambarotta, G. & Lagomarsino, S. 1998. Planning. Earthquake Spectra 20(2): 377–394
A macroelement approach to the three-dimensional seis- Lourenço, P.B. & Orduña, A. 2003. Seismic Analysis and
mic analysis of masonry buildings. Proceeding. of the XI Strengthening of a 17th Century Church in Azores.
European Conference on Earthquake Engineering, Paris, Proceeding 9th North American Masonry Conference,
A.A. Balkema (Abstract Volume & CD-ROM): 602. Clemson, USA: 932–941.
Chiostrini, S., Galano, L. & Vignoli, A. 1998. In situ tests and Lourenço, P.B. & Oliveira D.V. 2007. Improving the seismic
numerical simulations on structural behaviour of ancient resistance of masonry buildings: Concepts for cultural
masonry. Proceedings Workshop on Seismic Performance heritage and recent developments in structural analysis.
of Monuments: 197–206. XII Convegno Nazionale ANIDIS “L’Ingegneria Sismica
Chiostrini, S., Foraboschi, P. & Vignoli, A. 1992. Struc- in Italia”, Pisa.
tural analysis and damage evaluation of existing masonry O.P.C.M. 3274 del 20/3/2003. Primi elementi in materia di
building by dynamic experimentation and numerical mod- criteri generali per la classificazione sismica del territorio
elling damage. Proceedings of Tenth World Conference on nazionale e di normative tecniche per le costruzioni in
Earthquake Engineering, Madrid: 3481–3486. zona sismica. G.U. 8/5/2003, n. 105, S.O. n. 72 e O.P.C.M.
Chiostrini S., Galano L. & Vignoli, A. 1997. Mechanical 3431 del 3/5/2005.
characterization of stone masonry panels and effective- Willam. K.J. & Warnke, E.D. 1975. Constitutive model for
ness of strengthening techniques. Proceeding Fourth Int. the triaxial behaviour of concrete. Proceeding Interna-
Symposium on Computer Methods in Structural Masonry: tional Association for Bridge and Structural Engineering,
128–35. Bergamo, Italy: 174–186.
609
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: This paper aims at developing the tools and strategy for assessing the dynamic and seismic
performance of the Byzantine Walls of the city of Thessaloniki in order to evaluate the seismic history of the city
as a whole. The particular Walls have been constructed at the end of the 4th century A.D. in the reign of Theodosius
the Great and as such, their structural integrity and record of damage reflects to a certain degree the level of
seismic forces that has developed during the centuries. Moreover, the fact that they are extending in kilometers
within the civil grid of the modern city allows the study of the role played by the local soil conditions for a
given earthquake scenario. It is worth noting that despite their relatively simple structural system, their foreseen
seismic behavior as a 3D body has not been thoroughly studied so far, primarily due to the lack of efficient
numerical tools and the high computational cost related, especially towards the study of their response in the
time domain. Along these lines, a refined dynamic analysis approach is proposed and the structural performance
of particular parts of the Walls complex is examined for a number of realistic earthquake scenarios, accounting
for the site specific soil conditions, the spatially variable nature of the incident seismic waves, as well as the
overall geotechnical/geotectonic environment of the area. Through this advanced simulation scheme, an upper
bound of the historical level of seismic forces for the city of Thessaloniki is traced (through back analysis), while
the overall refined approach can be also used as a guide for the direct assessment of the existing seismic capacity
of monuments as a whole.
1 INTRODUCTION
from the seismic study of other historical structures
of the Byzantine Era throughout the city, in order to
1.1 Scope
attempt to shed some light not only on the structural
The impact of earthquakes on monumental heritage history of the Walls through the centuries, but if possi-
is a critical issue that has attracted growing scientific ble, to back analyze and evaluate the reliability of the
interest during the last decades. Monuments however, seismic scenarios developed for the Metropolitan City
are most often complex structures, whose preservation of Thessaloniki as a whole.
and/or seismic strengthening heavily relies on the clear The Walls, still surrounding partially the old town
understanding of all factors affecting their vulnerabil- of Thessaloniki were initially built in 315 B.C. by
ity as well as on the accurate study of the effects of past the king Kassandros and were completed at the time
earthquakes. Until recently, the investigation of the of Great Theodosius (379–395 A.D.). Nowadays, the
seismic performance of important and extended mon- Walls extend in kilometres within the civil grid of
uments was restricted by the inherent limitations of the modern city but their continuity has been dis-
numerical analysis thus preventing the engineers from rupted due to partial or complete collapse at numerous
the study of their dynamic response in the time domain locations. Historically, due to their dynamic nature,
under realistic (recorded or artificial) ground motion in the sense that they could be adapted in accor-
scenarios. This problem was further stressed in the dance with civilian needs, thus being repaired after
case of (Classical, Roman, Byzantine or Medieval) city sieges and following rules of economy and function-
(Sea or Land) Walls whose dimensions were normally ality where the art of war was concerned, the Walls of
significantly large, hence the complexity of the analyt- Thessaloniki (as those of Constantinople, Nicaea and
ical or numerical procedures to be followed, (as well as others), did in fact changed considerably over the cen-
the subsequent computational cost) were considerably turies (Bouras, 2002) following the heavy fortification
high. Along these lines, it was deemed interesting to requirements that arose. From the overall Byzantine
focus on the city Walls of Thessaloniki, utilizing state- Walls complex which extends in kilometres within the
of-the-art numerical tools and the experience gained city, two sections (Figure 1) were chosen to be studied,
611
Eptapygriou street, near the Trigonion Tower, inclu-
sive of the two twin gates at the East section (widened
and named thereafter by Anna Palaiologina) up to the
circle tower (Figures 2–5) constructed later. The par-
ticular Walls Section was encircling the Byzantine
acropolis (Velenis, 1998) thus separating it from the
Ancient Acropolis and it consists of numerous rectan-
gular (primarily) and triangle defensive towers. This
is essentially a monolithic and straight complex with
its main axis being parallel to the East-West direc-
tion. The superstructure was constructed of masonry
made by alternate bands of stones and bricks (Velenis,
1998). The masonry displays a mixture of construction
methods as can be seen in Figure 3.
From the two locations selected to be studied,
the small, free-standing masonry wall illustrated at
the bottom of Figure 1 was chosen on the basis
of the well controlled study environment which was
ensured by its clear geometry and structural system,
its proximity to the unique location where the 1978
earthquake was recorded and the good knowledge of
the underlying soil.
On the other hand, the extended Land Walls Sec-
tion in the upper part of the city (illustrated at the top
of Figure 1) was selected to be studied primarily for
three reasons: (a) due to its structural integrity which
has remained almost intact within the centuries thus
allowing to assume that its seismic capacity must have
not been exceeded for the last 1600 years, (b) due to
its overall length (reaching approximately 500 m) that
forms one of the most extended preserved parts of the
fortification and (c) due to the additional opportunity
that this length provides, to further investigate (in con-
trast to what is commonly performed) the potential
effect of the (typically neglected) traveling nature of
the incoming seismic waves to the dynamic response
of the system.
The scope of this research effort is therefore, to
utilize state-of-the-art knowledge and numerical tools
for the pilot study of the seismic performance of the
selected Thessaloniki city Land Walls in order to:
a) verify the good seismic performance of the specific
Walparts during the 20/6/1978 Thessaloniki severe
earthquake (Ms = 6.5).
b) assess the level of safety during future seismic
Figure 1. Contemporary photograph of the 4th Century
events (especially for the long circuit) through
A.D. Byzantine Land Walls ofThessaloniki (North side view). appropriate earthquake scenarios that account for
a realistic representation of the overall seismotec-
tonic and local site conditions.
particularly: (a) a simple relatively small, essentially c) investigate the potential effect of ignoring spa-
stand-alone part of the Walls located approximately tially variability of earthquake ground motion in
in the center of the modern city and (b) the Walls the analysis of long and monolithic structures.
circuit in the northern part of the Byzantine fortifi- d) if possible, to compare through back numerical
cation. This part, which due to its size is studied more analysis, the level of the highest earthquake excita-
thoroughly, forms a statically independent structural tion (that could have damaged the particular Wall
system, extending from the beginning of the West Gate section but apparently has not occurred yet) with
(namely “Pyrros Gate”) to the main East Gate along the the Maximum Credible Earthquake (MCE) defined
612
Figure 2. Contemporary photograph of the 4th Century
A.D. Byzantine Land Walls of Thessaloniki (North side
view).
Figure 4. Axonometric view of rectangular wall between
along the East-West axis (North side view). After Velenis,
1998.
613
1998, Galasco et al., 2004, Rota et al., 2005), with
emphasis on the non-linear response of the walls, the
particular structure was studied herein as purely lin-
ear elastic, in order to permit primarily focus on the
macroscopic dynamic response of the system under
simultaneous excitation along the three principal axes.
Subsequently both local material and geometric non-
linearities (inclusive of cracking and sliding at critical
locations) were not considered.
The structure was assumed as fully fixed at its base
since geotechnical and geophysical studies have shown
that the soil conditions at the particular city loca-
tion are very stiff (Anastasiadis et al., 2001). Based
on experimental results for monuments of the Byzan-
tine Era (Manos et al., 2004, Stylianidis and Sextos,
2006) an average uniform compression strength of
fmc = 2.0 MPa was adopted for the construction mate-
rials of the whole system for simplicity, based on
the weaker brick masonry. The corresponding ten-
sile strength was set equal to fmt = 0.15 MPa. The
Young’s modulus of the masonry was taken equal
to E = 3.5 × 106 kPa (corresponding to the compos-
ite material as a whole) and the self weight equal
to γ = 22 KN/m3 while 6% Rayleigh damping was
assumed for the circular frequency range ω = 40–
200 rad/sec. It is noted that the two twin gates of
the Trigonion Tower (i.e. Annis Palaiologinis Gate)
as well as the Pyrros Gate (Velenis, 1998), have
not been modeled due to their relatively smaller
dimensions compared to the overall Wall Section
length.
614
2.3 Spatially variable earthquake ground motion
along the Land Walls under study
Development of spatially variable seismic ground
motion scenarios for the study of the dynamic response
of extended structures (primarily bridges but also dams
and pipelines) is attracting increasing scientific atten-
tion after strong evidence that not only seismic motion
may differ substantially in terms of amplitude and fre-
quency content but also that this difference may have
a detrimental effect on the displacements and stresses
of the extended structure under certain circumstances
(Sextos et al., 2003). The sources of this spatial and
temporal variations of seismic motion have been well
identified: (a) waves travel at a finite velocity, hence
they arrive at consecutive support points with a time Figure 7. Characteristic eigenmodes of the system.
delay; (b) wave coherency loss in terms of gradual
reduction of the waves’ statistical dependence with
distance and frequency, due to multiple reflections, 3 SEISMIC PERFORMANCE OF THE LAND
refractions and superpositioning of the incident seis- WALLS UNDER STUDY
mic waves during propagation and (c) variation in local
site conditions that strongly affects wave amplification 3.1 Dynamic characteristics of the long Wall circuit
and frequency content. under study
For the particular Wall Section, and since the soil
conditions are generally stiff, it can be assumed that The dynamic characteristics of the system were ini-
from the above sources of spatial variability, the effect tially investigated, and as anticipated despite the height
of local site conditions and the loss of wave coherency of the Walls, the overall complex is rather stiff (i.e.
is rather small; as a result, the asynchronous excita- its fundamental period is equal to 0.12 sec). It is also
tion is estimated solely on the phase lag as is can be notable that due to the presence of the intermediate
defined by the distance from the reference point and towers, there is no clear transverse mode; on the con-
the apparent velocity of seismic waves (taken equal to trary the system vibrates on a large number of similar
Vapp = 2000 m/sec). In order to account for the afore- local modes having similar modal contribution to the
mentioned time delay at all base points (nodes) of the overall response.
500 m long Wall complex, the Finite Element model Clearly, the fundamental mode is also local as it
was internally programmed to permit the subsequent corresponds to the east main wall which is not only
excitation of each one of the 5830 supporting points i unsupported laterally (i.e. no intermediate defensive
with a distinct 3-D ground displacement Ui,x , Ui,y , Ui,z towers exist) but also is on average 2 m higher than
which is a function of space and time t and is equal to: other parts of the Wall circuit. The most characteristic
from these modes are illustrated in Figure 7 where it is
seen that it is only after the 25th mode that the structure
vibrates (asymmetrically) as a whole.
615
sections instead of local points. Nevertheless, the over-
all assessment is that seismic demand developed is
clearly inferior to the estimated capacity of the Walls.
In terms of shear stresses, the most critical loca-
tion was identified to be the North side of the East
Gate as depicted in Figure 9. Notably, both the shear
stresses developed and the shear strength are a func-
tion of time as the latter depend on the compression
stresses which also vary in time. For the assessment of
the safety factor against shear, the following equation
was used according to the provisions of Eurocode 6:
616
Figure 10. Variation of normal stresses with time under
asynchronous excitation for three characteristic locations.
the response can be considered as rather similar, the 3.4 Seismic response of simple stand-alone
extremes of the stresses are different. In particular, the walls at different city locations
extreme tensile stress at the East Gate has increased by
50% (from 0.12 MPa to 0.19 MPa) while the extreme Following the numerical investigation of the seismic
tensile stress at the East Main Wall has decreased performance of the long upper city Wall circuit, case
by 40% (from 0.13 MPa to 0.08 MPa). Especially for A scenario (1978 earthquake excitation) was applied
the middle point (on the East Gate) the maximum for the free standing wall of the second location under
stresses occur at a slightly different time (7.15 sec study (Stylianidis and Sextos, 2006). Similarly to the
instead of 7.02 sec). Such distinct response under asyn- long Wall complex, good performance was also ver-
chronous earthquake ground motion has also been ified as neither loss of stability due to rocking nor
observed in a large number of bridges (Sextos et al., exceedance of its tensile strength was observed dur-
2003) and is pronounced in the case of long and ing the application of the 1978 ground excitation.
stiff (monolithical) structures (Tubino et al., 2003) Moreover, it was estimated that the particular sys-
due to the excitation of higher (primarily antisymmet- tem studied could resist at least twice the earthquake
ric) modes (Sextos et al., 2003) and the subsequent load imposed during the 1978 earthquake without
development of additional pseudo-static forces. How- significant damage.
ever, it is also noted that the ‘effective’ tensile and
compressive stresses are again lower than the esti-
mated corresponding strength, hence the comments 4 CONCLUSIONS
made regarding the satisfactory historical and future
performance of the Land Walls under study remain This paper is an attempt to propose a refined approach
valid. towards the assessment of the seismic performance
In can be generally stated though, that the dif- of a part of the Land Byzantine Walls of the city
ferences observed between the synchronous and the of Thessaloniki. Utilizing advanced numerical tools
asynchronous excitation of the particular Wall Section, and generating various earthquake ground motion sce-
even if they do not impose a threat for the structural narios, both a 500 m long section of the Walls and
integrity of the system, are not negligible if one con- a small free-standing part of the Walls at a second
siders that due to the very stiff (rock) formations at the location were studied in the time domain under using
617
synchronous and asynchronous excitation. The main
conclusion drawn can be summarized as follows:
REFERENCES
Figure 11. Variation of normal stresses (σz ) with time for Ambraseys N., Smit, P., Berardi, R., Rinaldis, D., Cot-
Case A scenario (excitation with 1978 earthquake record). ton, F. and Berge. C. 2000. Dissemination of European
618
Strong-Motion Data. CD-ROM collection. Council of Galasco, A., Lagomarsino, S., Penna, A. and Resemini.
European Communities. Environment and Climate Pro- S. 2004. Non-linear seismic analysis of masonry struc-
gramme. ENV4-CT97-0397. Bruxelles. Belgium. tures, 13th World Conference on Earthquake Engineering,
Anastasiadis A., Raptakis D. and Pitilakis K. 2001. Thes- Vancouver, Canada, Paper No. 843.
saloniki’s Detailed Microzoning: Subsurface Structure Rota, M., Pecker, A., Bolognini, D. and Pinho, R. 2005.
as basis for Site Response Analysis, Pure and Applied A methodology for seismic vulnerability of masonry
Geophysics – Pageoph, Vol. 158(12), 2597–2633. bridge walls, Journal of Earthquake Engineering, Vol. 9,
ANSYS Engineering Analysis System 2005. User and Theo- Special Issue 2, 331–353.
retical Manual. ANSYS, Inc., Southpointe, Canonsburg, Sextos, A., Pitilakis, K. and Kappos, A. 2003. A global
Pennsylvania, Version 10.0. approach for dealing with spatial variability, site effects
Bouras, C. 2002. The Economic History of Byzantium: From and soil-structure-interaction for non-linear bridges: Part
the Seventh through the Fifteenth Century, Laiou, A. (Edi- 1: Methodology and analytical tools, Earthquake Engi-
tor in-Chief), Dumbarton Oaks, Trustees for Harvard neering Structural Dynamics, 32, 607–629.
University, Washington,D.C. Stylianidis, K. and Sextos, A. 2006. Seismic response of
Bommer, J. 2002. Deterministic vs. Probabilistic Seismic Byzantine Walls as a means of assessment the seismic his-
Hazard Assessment. An exaggerated and obstructive tory of the city of Thessaloniki, 1st Hellenic Conference
dichotomy, Journal of Earthquake Engineering, 6 (special on rehabilitation works, Thessaloniki, CD-ROM volume,
issue no. 1), 43–73. Paper No. 084 (in Greek).
Casolo, S. 1998. A three-dimensional model for vulnerability Tubino, F. Carassale, L. and Giovanni, S. 2003. Seismic
analysis of slender medieval masonry towers, Journal of response of multi-supported structures by proper orthogo-
Earthquake Engineering, Vol. 2, No. 4, 487–512. nal decomposition, Earthquake Engineering & Structural
CEN. 2004. Eurocode 8: Design of structures for earthquake Dynamics, 32, 1639–1654.
resistance. EN 1998-1 General rules, seismic actions and Velenis, G. 1998. The Walls of Thessaloniki: From Kasandrus
rules for buildings, Brussels. to Heraklios, University Studio Press, Thessaloniki, (in
CEN. 2004. Eurocode 6. Design of masonry structures. ENV Greek).
1996-1 General Rules’, Brussels.
619
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
H.P. Mouzakis
National Technical University of Athens, Greece
ABSTRACT: The Katholikon of Dafni suffered throughout its history a large number of earthquakes that
caused many structural problems and damage to it. After the catastrophic earthquake that affected the region
of Attica during the September 1999, and in the frame of a strategic plan undertaken by the Hellenic Ministry
of Culture for the conservation and restoration of the monument including its mosaics, a seismic monitoring
system has been installed. This paper presents the monitoring system installed and the analysed data recorder
during four earthquakes. Two of them took place before the grouting injections and the other two some months
after the implementation of grouting to almost the half of the height of the perimeter walls. The comparison of
digitized data during weak earthquakes, occurred before and after grouting application show the effect to the
overall behaviour of the structure.
1 INTRODUCTION
621
and supporting urgent measures were subsequently
undertaken (Miltiadou-Fezans et al. 2003a), together
with all the necessary investigations and studies,
including also a first series of finite element analyses
for the numerical verification of the existing damage of
the structure (Miltiadou-Fezans et al. 2003b & 2004).
This analysis helped for a better understanding of the
structural behaviour of the monument and guided the
design of the monitoring system. The monitoring sys-
tem was installed in 2003 after the implementation of
the aforementioned emergency measures and before
the application to the masonry structure of any repair
Figure 2. Past interventions to the monument: North façade. intervention. Thus, when the hydraulic lime grouting
The different types of construction, mainly of the upper and application started (Miltiadou et al. 2008), in June
lower part of the monument, are shown. 2006, the system was functioning and had already
recorded some weak earthquakes.
(Vintzileou et al. 2004), it was shown that in both cases
This paper presents the monitoring system that
a three-leaf masonry is used. As expected the width of
was installed to record the seismic response of the
each of the three leaves at the lower and the upper
Katholikon of Daphni Monastery and the analysed
part is not the same due to the bigger stones used in
data recorder during four earthquakes. Two of them
the external leaf of the lower part. The existence of
took place before the grouting injections and the rest
three-leaf masonry for the construction of the internal
some months after the first phase of grouting. During
vertical walls was also ascertained (Côte et al 2008),
this first phase, grouting was implemented to almost
while plain brick masonry was used in the construc-
the half of the height of the perimetric walls. The dig-
tion of all the monolobe, bilobe and trilobe windows,
itized data during weak earthquakes which occurred
and all the vaulted roof.
after the application of the injections confirm the effi-
The various materials and construction techniques,
ciency of grouting interventions and their effect toe
the foundation and the soil condition, as well as the
the overall behaviour of the structure.
past modifications and local reconstructions taken
place to remediate damages provoked due to strong
earthquakes, affect significantly the mechanical and 2 MAJOR 19TH AND 20TH CENTURY
dynamic characteristics of the structure. The varia- INTERVENTIONS
tions of temperature, settlement, creep and ambient
vibrations have also a great influence on the seismic The monument, during its life (from 11th century
response of the monument. As a result, cracks occur to our days) has been subjected to various modifi-
and divide the whole structure in interacting parts with cations, it suffered numerous earthquakes, whereas
nonlinear behaviour. major interventions for its preservation were applied
Due to these aforementioned uncertainties in addi- in the 19th and 20th centuries (Delinikolas et al. 2003,
tion to the variability of the seismic excitation, even if Miltiadou-Fezans et al. 2004). During the last two
detailed accurate finite element models and sophis- centuries strong earthquakes have severely affected
ticated analysis is developed, the estimation of the the church. The 1889 earthquake (M = 6.7 R), the
seismic safety of the structure is not enhanced. 1894 (M = 7.0 R), the 1914 (M = 6.0 R) the 1981
In order to increase our understanding of the seismic (M = 6.7 R) and the 1999 (M = 5.9 R) have to be men-
structural response and to decrease the uncertainty of tioned. The major intervention (1890–1897) include
the seismic action, the use of an earthquake monitor- the following: Demolition and reconstruction of the
ing system is necessary. Moreover, the implementation (heavily damaged) central dome, repairs in the lower
of the recorded and analyzed data will be the basis parts of the church, removal and partial replacement
for the further development and calibration of detailed of mosaics that were heavily damaged. Installation of
numerical models, which will be used for the para- three concentric iron rings (I-beams) was carried out at
metric study of the seismic behaviour before and after the base of the drum of the dome. After, the 1894 earth-
interventions. quake the narthex, its collapsed groined vaults and
Such monitoring systems have already installed the NE chapel were reconstructed. Between 1897 to
in Hagia Sophia and Suleimaniye in Constantinople 1907, the two stone buttresses to both sides of the north
(Durukal, E. et al. 2003) and recorded the response entrance of the church were constructed and metallic
during strong earthquakes, as part of the efforts for devices were installed around the two south masonry
the determination of their earthquake worthiness. piers aiming to their confinement. For the same pur-
The Katholikon of Dafni Monastery suffered severe pose, in 1954, an iron ring was installed in the external
damage during the strong Athens’ earthquake of 1999 face of the upper part of the drum over the windows.
622
Table 1. Grouting periods. 0 80
PEAK ACCELLERATION IN G
Phase Start End
0 60
I 26/6/2006 4/8/2006
II 15/12/2006 4/4/2007 0 40
0 20
The 1981 earthquake caused numerous hair cracks to
the building and damages to the mosaics. Finally, the 0 00
1999 earthquake caused severe damages to the mon- 10 100 1000 10000
ument. Extensive cracking of arches was recorded, RETURN PERIOD
the N-E corner was severely damaged, and horizontal
cracks appeared along the perimeter of the drum (both Figure 3. Peak ground acceleration and return period.
at its base and top). Severe damages were also recorded
at the 16 pillars of the drum (Miltiadou-Fezans A. et al. the urgent supporting measures and before the begin-
2003b and Miltiadou-Fezans A. et al. 2004). ning of masonry repair, while the second one occurred
during the preparation of masonry for grouting by
stitching of cracks and local deep re-pointing. After
3 EMERGENCY SUPPORTING AND PHASES three and four months from the end of the first grout-
OF STRUCTURAL RESTORATION SCHEME ing period, during which grouting was implemented up
to the springing level of groin vaults, two more weak
Taking into account the damage of the structure and earthquakes were recorded.
the danger of eventual aftershocks temporary mea-
sures were urgently taken (Miltiadou-Fezans A. et al. 4 SEISMIC HAZARD EVALUATION
2003). Independent new steel elements were designed
and constructed to provide vertical support to various The selection of the technical specifications of the
arches. A system of three steel buttresses is used, at accelerometers that are installed on the monument and
the NE corner of the monument, where a pronounced on soil surface has to be based on the quantitative esti-
tilting of the external walls was noticed. This system mation of the acceleration of the ground motion, which
is able to carry horizontal seismic loads to especially is assessed by seismic hazard analysis. It was evalu-
constructed concrete footings. ated that the annual mean peak horizontal acceleration
A first phase of structural restoration interven- with mean frequency of two events per year is 0.059 g,
tions has been decided and implemented. It comprised while the maximum expected peak acceleration on soil
all the measures considered necessary for the repair surface is 0.80 g, as shown in Figure 3. Considering the
and strengthening of masonry elements (stitching and standard deviation of the used sample earthquake data
deep re-pointing where necessary, systematic grouting and in order to have a high signal to noise ratio, the
injection to fill voids and cracks, local reconstruc- noise level was selected to be 0.001 g.
tions, etc). The design of a second phase of inter-
ventions (installation of ties, metallic confinements,
diaphragms, etc) is under elaboration, and will take 5 SEISMIC MONITORING SYSTEM
into consideration all the data collected from research
programs, in situ surveys during the first phase of A digital dynamic monitoring system has been
interventions and the monitoring system. installed to record the response of the core of the mon-
In the framework of the first phase of interven- ument. Accelerometers and displacement meters are
tions, a hydraulic lime based grout was selected and the used to measure acceleration and relative displace-
application of injections to the monument was imple- ment at characteristic locations. These locations were
mented in two time periods (Miltiadou-Fezans A. et al. selected in order to avoid local vibrations of indi-
2007 & 2008), as shown in Table 1. vidual masonry units during earthquakes, taking into
The seismic monitoring system was installed on consideration the existing cracking patterns.
the monument before the beginning of restoration The accelerometers were placed at five character-
works, in order to collect data before, during and istic levels (a) on ground level free field; (b) on the
after the completion of the interventions, with the cornice where the central arches and the squinches
aim to estimate the effect of the intervention mea- are springing; (c) on the base of the drum; (d) on
sures to the overall behaviour of the structure. In fact the base of dome (top of the drum). The displacement
the results presented in this paper are based on the transducers were placed to record the relative displace-
elaboration of data collected from four earthquakes. ment between north and south as well as east and west
The first earthquake took place after the application of internal walls of the central part of the church.
623
Figure 4. Control room. Figure 6. Accelerometers located at the Drum’s Base.
624
Table 2. Seismic events. accelerometers that have a frequency of 1.38 kHz and
damping coefficient ζ = 70%.
Recorded The acceleration time histories of the earthquake of
event 1 2 3 4
08/01/06 for channels 1–14 are presented in Figure 10,
Location South Kythira Penteli Fthiotis respectively. A lowpass digital filter with cutoff 25 Hz
Euboea (Attica) and a highpass with cutoff 0.7 Hz were used for data
Date 26/09/04 08/01/06 05/11/06 12/12/06 processing.
Time GMT 03:06 11:34 22:55 05:50 Total acceleration response spectrum (Nigam, N.C.
Latitude 38.31 36:21 38:05 38:62 & Jennings, P.C. 1969) on soil surface for the same
Longitude 24.18 23:41 23:87 23:31 earthquake for damping coefficient 5% are shown in
Epicenter Euboea Off the 15 km 80 km Figure 11.
eastern east of N.N.W The amplification factor between the maximum
coast of Athens of Athens.
acceleration at various locations on the structure and
the island (Penteli Malesina
of Kythira Area) Fthiotis the acceleration on the soil surface of the same
Magnitude 4.2 6.4 3.7 4.5 orientation is given in Table 3.
(Richter) The damping coefficient of the first mode is given
Epicentral 20 km 69 km 21 km 20 km in Figure 12. The natural period of the first mode was
Depth changed during earthquakes as shown in Figure 13.
Epicentral 58 km 402 km 21 km 73 km
Distance
Characteristics Weak Strong Weak Weak 8 PRESENTATION OF MAIN RESULTS
Data Source NOAIG NOAIG OASP- OASP-
NOAIG NOAIG
8.1 Earthquake 16/09/2004
The acceleration of the northern region of the base of
the dome along E – W direction is higher than the
acceleration of the southern region. The acceleration
of the west region of the base of the drum is higher
than that of the east region. Period of the longitudinal
and transversal translational modes were not changed
during the earthquake. On the cornice the acceleration
of the south – west corner along E – W direction is
higher than the acceleration of the north – east one.
The drum was mainly deformed and consequently the
acceleration on the base of the dome was increased.
625
0.2
m/sec2
0.1
0
-0.1
-0.2
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
0.2
m/sec2
0.1
0
-0.1
-0.2
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
0.3
m/sec2
0.1
-0.1
-0.3
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
1.5
m/sec2
0.5
-0.5
-1.5
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
1.5
m/sec2
0.5
-0.5
-1.5
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
2
m/sec2
1
0
-1
-2
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
1.5
m/sec2
0.5
-0.5
-1.5
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
1
0.5
m/sec2
0
-0.5
-1
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
0.6
m/sec2
0.2
-0.2
-0.6
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
1.5
m/sec2
0.5
-0.5
-1.5
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
0.6
m/sec2
0.2
-0.2
-0.6
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
1
m/sec2
0.5
0
-0.5
-1
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
0.3
m/sec2
0.1
-0.1
-0.3
15 20 25 30 35 40 45 50 55 60 65 70
Seconds
of grouting was clearly distinguished, as the dynamic characteristics of the monument except the evolution
characteristics of the structure have changed. of grout hardening. The period of the structure was the
same along the N-S direction, while along E-W direc-
8.4 Earthquake 12/12/2006 tion a very slight increase was recorded during the
This specific earthquake occurred almost one month earthquake. The damping of the structure was higher
after the previous one, four months after the than that of the previous event but it remained lower
completion of the first period of grouting. There was compared to the damping before the first period of
no other new element that could have affected the injections.
626
0.7 Table 3. Amplification factor.
Damp. 5.0%
0.65
0.5 Direction /
0.45 channel Position 26/09/04 08/01/06 05/11/06 12/12/06
0.4
0 40
0.65 0 32 0 39
0 35
Period Ti (sec)
0.6 0 30
0 30
0 27 0 26 0 26
0.55 0 25
0 20
0.5 0 20
0 22
0.45 0 15
0.4 0 10
0.35 0 05
0.3 0 00
26/9/2004 8/1/2006 5/11/2006 12/12/2006
0.25
West - East South - North
0.2
0.15
0.1
Figure 13. Variation of the period of the first mode.
0.05
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Period [sec]
were taking place, together with the stitching and deep
re-pointing of the cracks on the extrados of the vaults
Figure 11. Total acceleration Response Spectrum and this could have an effect on the dynamic character-
(a) Vertical; (b) South – North; (c) West – East direction. istics of the overall structure. It will be very interesting
to compare the existing records with the ones of future
The differences presented between the last two weak earthquakes, as from April 2007 the first phase
earthquakes can be probably attributed to the fact that of interventions has been accomplished, and injections
at that period local reconstructions of external face have been implemented until the level of the base of
of masonry of the S and W aisle of the central cross the drum.
627
9 CONCLUSIONS Côte, Ph., Dérobert, X., Miltiadou-Fezans, A.,
Delinikolas, N., Durand, O., Alexandre, J., Kalagri, A.,
The monitoring system installed to the Katholikon of Savvidou, M., Chrysopoylos, D.., Anamaterou, L. and
Dafni Monastery to record the seismic response of the Georganis, F. 2008. Application of non-destructive tech-
niques at the Katholikon of Daphni Monastery for
structure during weak earthquakes has been proved
mapping the mosaics substrata and grouting monitor-
very useful, as the analysed data collected until now ing. Submitted to the 6th International Conference on
during four earthquakes give important information StructuralAnalysis of Historical Constructions, Bath, UK.
for its behaviour before interventions, as well as after Delinikolas, N., Miltiadou-Fezans, A., Chorafa, E. and
application of grouting until half height of the walls. Zaroyianni, E. 2003. Study on restoration of the Katho-
After the urgent measures and before any interven- likon of Dafni Monastery: Phase A-Architectural and
tion the stiffness of the structure was decreased during historical Survey. Hellenic Ministry of Culture (in Greek,
the earthquake 08/01/2006. This is totally expected, with an extensive bibliography).
as no remedial measures had been applied, while Durukal, E. et al. 2003. Dynamic Response of two Histori-
cal Monuments in Istanbul Deduced from the Recordings
the emergency measures have not been designed to
of Kocaeli and Düzce Earthquakes. Bull.Seism.Soc.Am.,
undertake seismic actions. v.93-2, pp. 694–712.
As a result of the implementation of grouting injec- Millet, G. 1899. Le monastère de Daphni, Histoire,
tions to the half height of the walls the period of Mosaiques. Monuments de l’Art Byzantin, I, Paris.
the first mode was changed. Specifically, the periods Miltiadou-Fezans, A., Tassios, T. P., Delinicolas N.,
were decreased. The damping of the monument before Chorafa, E., Zarogianni, E. and Candrinos, I. 2003a.
restoration measures was 7.75% and 9.30% along Earthquake structural problems and urgent measures
E – W and N – S directions respectively.After the appli- undertaken to support the Katholikon of Dafni Monastery
cation of injections of hydraulic lime based grout to in Athens, Greece, Proceedings of the 8th International
Conference STREMAH 2003 “Structural Studies, Repairs
the half height of the structure, damping was decreased
and Maintenance of Heritage Architecture”, Halkidiki,
to 4.8% for E – W and 7.6% for N – S direction. Thus, Greece, 7–9 May 2003.
it was proved that the effect of grouting on the overall Miltiadou-Fezans, A., Delinikolas, N., Chorafa, E. and
behaviour of the structure can be evaluated by a seismic Zaroyianni, E. 2003b. Study on restoration of the Katho-
monitoring system, under the condition to have earth- likon of Dafni Monastery, Phase A-Structural Survey,
quakes recorded both before and after the application Analysis and remedial measures. Hellenic Ministry of
of grouting. Culture (in Greek).
All these data will be used for the further analyses Miltiadou-Fezans, A., Vintzileou, E., Delinikolas, N.,
of the structural behaviour of the Katholikon using Zaroyianni, E. and Chorafa, E. 2004. Pathology of Dafni
Monastery: survey, monitoring of cracks, interpretation
adequate methods and models, while complementary
and numerical verification. Proceedings of the 4th Inter-
instrumentation has now been installed to cover the national seminar on StructuralAnalysis of Historical Con-
whole monument, apart from the central core already structions,Vol.2, Padova, 10–13 November 2004, Mod-
monitored since 2003. ena C., Lourenco P.B., Roca P. (ed.), Balkema Publishers,
All the instruments are going to stay on the struc- London.
ture in an almost permanent way, to record the seis- Miltiadou-Fezans, A. Kalagri, A. and Delinikolas, N. 2007.
mic behaviour of this important monument of world Design of hydraulic grout and application methodology
heritage. for stone masonry structures bearing mosaics and mural
paintings: the case of the Katholikon of Dafni Monastery,
Proceedings of the International Symposium: SHH07-
ACKNOWLEDGEMENTS
Studies on Historical Heritage, Edited by Görün Arun,
Antalya, Turkey, 17–21 September 2007.
The authors would like to acknowledge the support Miltiadou-Fezans, A., Kalagri, A., Delinikolas, N.,
of the Electronic Engineer S. Asimakopoulos who has Kakkinou, S. and Ziagrou, A. 2008. Methodology for in
been responsible for the installation and the opera- situ application of hydraulic grouts on historic masonry
tion of the system. The project has been undertaken structures. The case of the Katholikon of Dafni Monastery.
by the Hellenic Ministry of Culture in collaboration Submitted to the 6th International Conference on Struc-
with the National Technical University of Athens and tural Analysis of Historical Constructions, Bath, UK.
co-funded by the European Regional Development Nigam, N.C. & Jennings, P.C. 1969. Calculation of Response
Spectra from Strong Motion Earthquake Records. Bul-
Fund (ERDF-75%) and by National Funds (25%).
letin of the Seismological Society of America, Vol.59,
pp. 909–922.
REFERENCES Vintzileou, E., Miltiadou-Fezans, A., Palieraki, V. and
Delinikolas, N. 2004. The use of radar techniques and
Bernhard, B. 2002. Measuring with Crystals: Principles and
endoscopy in investigating old masonry: the case of
applications of the piezoelectric measuring technique,
Dafni Monastery. Proceedings of 4th International Sem-
pages 70.
inar on Structural Analysis of Historical Constructions.
Bouras, Ch. 1998. The Daphni monastic complex recon-
Modena C., Lourenco P.B., Roca P. (ed.), Rotterdam,
sidered, AETOS studies in Honor of C. Mango, B.G.,
Balkema, Vol. 2, pp. 351–360.
Teubner, Stutgart & Leipzig.
628
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
T. Hirai
Obayashi Corporation, Tokyo, Japan
ABSTRACT: Bayon temple, Angkor, consists of dry-masonry structures made of sandstone. Strong wind in
the rainy season is suspected as one of the causes for its progressive deterioration. Since 2003, micro-tremor has
been measured at most of the tower-type structures and the continuum-equivalent elastic modulus was found
to be 1/9 to 1/27 of that of the sandstone specimen in laboratory tests. In 2006, micro-tremor was measured at
libraries in Bayon temple and the similar reduced equivalent modulus was also obtained for these frame-like
sandstone-dry-masonry structures. This extremely low elastic modulus implies the inadequacy of continuum
modeling for bearing capacity of dry-masonry structures. Then the Discontinuous Deformation Analysis was
applied to safety evaluation of the library against wind load. It was an illustrative example; however, the evaluated
safety margin for the wind velocity of 40 m/s is about two.
1 INTRODUCTION
629
obtained base-fixed natural frequencies of horizontal
vibration modes with adjusted elastic constants. The
simulated equivalent Young’s modulus ranged from
1/27 to 1/9 of the laboratory test. This extremely
low modulus may be attributable to spring effects
of some material existed between the interfaces of
sandstone blocks, or to reduced contact areas at the
interfaces.
In 2006, we had a chance to measure micro-
tremor at the libraries. While the main tower and
sub-towers are basically stonewalls, the libraries are
relatively small frame-like structures. Comparing the
equivalent elastic modulus of these different types of
structures, insight on the underlying common phys-
ical phenomenon of sandstone-dry-masonry may be
found.
Our final objectives are the evaluation of bearing
capacity for these dry-masonry structures. Since the
extremely low continuum-equivalent elastic modulus
may reveal incapableness of continuum mechanics in
simulating rupture processes of dry-masonry struc-
tures, we have applied the Discontinuous Deformation
Analysis (DDA) (Shi 1993) to the evaluation of the
safety margin of the library against the wind load as
an illustrative example.
630
Transfer Function
25
20 L4WY
15 L4NY
10 L4SY
5 L4EY
0
0 2 4 6 8 10 12 14 16
180
Phase Lag
90 L4WY
L4NY
0
L4SY
- 90 L4EY
- 180
0 2 4 6 8 10 12 14 16
Frequency (Hz)
(a) NS direction
Transfer Function
15
L4WX
10
L4NX
5
L4SX
L4EX
0
0 2 4 6 8 10 12 14 16
Frequency (Hz)
180
Phase Lag
90 L4WX
0 L4NX
L4SX
- 90
L4EX
- 180
0 2 4 6 8 10 12 14 16
Frequency (Hz)
(b) EW direction
Figure 4. Transfer functions of the northern library in NS Figure 5. FEM model for the superstructure of the northern
and EW directions. library.
631
Table 3. Equivalent Young’s modulus of the structures.
632
Figure 9. Modeled portion of the northern library by DDA.
633
regardless of sizes and types of structures, based on
micro-tremor measurements. Since the extremely low
modulus is not yet clearly explained, we cannot rely
on a continuum model to evaluate bearing capacity of
dry-masonry structures. DDA was selected to replace
FEM and tested for fundamental features by simple
problems. Finally DDA was applied to the northern
library with plane strain model. The bearing capacity
for the first slide occurrence is evaluated as about twice
larger than the wind load corresponding to 40 m/s.
ACKNOWLEDGEMENT
REFERENCES
Ikeuchi, K., Hasegawa, K., Nakazawa A., Takamatsu, J.,
Oishi, T. & Masuda, T., Bayon Digital Archival Project,
10th International Conference on Virtual Systems and
Multimedia (VSMM2004), November 2004.
JSA (Japanese Government Team for Safeguarding Angkor),
1995. T. Nakagawa (ed.), Annual Report on the Technical
Survey of Angkor Monument 1995.
JSA, 2000. T. Nakagawa (ed.), Report on the Conservation
and Restoration Work of the Northern Library of Bayon,
Angkor Thom, Kingdom of Cambodia, 2000.
Maeda, T., Sugiura,Y. & Hirai, T. 2005. Vibration characteris-
tics of the Bayon temple main tower, Angkor, Cambodia,
In C. A. Brebbia & A. Torpiano (eds), Structural Stud-
ies, Repairs and Maintenance of Heritage Architecture 9,
Malta 2005: 255–264, Southarnpton: WIT PRESS.
Maeda, T., Yamamoto, T., Wako, Y. & Hirai, T. 2007. Vibra-
tion characteristics and equivalent Young’s modulus of the
Northern Library and the main tower, Bayon, Cambodia,
In C. A. Brebbia (ed.), Structural Studies, Repairs and
Maintenance of Heritage Architecture 10, Prague 2007:
Figure 11. Deformation and principal stress in the northern 493–502, Southarnpton: WIT PRESS.
library DDA model for horizontal load. Sugiura, Y., Fukumoto, Y. & Maeda, T. 2004. Vibration char-
acteristics of the main tower, the Bayon temple, 21st Inter-
national congress of theoretical and applied mechanics,
4 DISCUSSION 2004.
Shi, G-H. 1993. Block system modeling by discontinuous
The continuum-equivalent Young’s modulus for deformation analysis, In C. A. Brebbia & J. J. Connor
sandstone-dry-masonry structures in Bayon temple (eds), Topics in Engineering Volume 11, Southampton:
ranges from 1/27 to 1/9 of the laboratory value Computational Mechanics Publications.
634
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The present work illustrates the results of structural analyses carried out on one of the temples
located in the archaeological area of Mỹ So,n in Central Viet Nam. The temple is a relatively well-preserved
building, hosting two square chambers covered by false vaults. A corner of the temple was destroyed by a bomb,
and some passing-through cracks were likely induced by the vegetation. The temple was built using multi-leaf
brick masonry in most of the walls, the outer leaves being more regular in texture than the inner one; also, the
bricks in the outer leaves are stuck by natural resin, which provides an excellent bonding. Finite element models
of the temple were set up, according to an accurate recently performed survey, and to drawings dating back to
the beginning of the 20th century, as far as the geometry of the nearly intact building is concerned. The models
take the layered nature of masonry into account. The masonry leaves were supposed to be either perfectly or
partially connected, allowing for a Mohr-Coulomb’s type failure condition at the interface between adjacent
leaves. Numerical analyses were carried out to evaluate the presumable stress conditions of the still intact temple
and to determine whether or not the self-weight alone might be the cause for the collapse of parts of the temple.
Also, the damages induced by bombs and vegetation were taken into account, and the current stress conditions
were determined. Some conclusions on the safety of the temple in its past and current conditions are drawn; also,
the influence of the models employed to allow for the contact between the different masonry layers is discussed.
635
Figure 3. Evolution of the plan and northern prospect from
1909 to 2001.
636
Figure 5. View of temple E7. Figure 7. Three-dimensional model of the structure:
transversal cross-sections.
637
Figure 9. Section of the leaves of the wall.
reduced to few centimeters, implies a weak toothing Figure 10. Elevation of the north side of temple E7.
between the external leaves and the central leaf, and
does not guarantee an adequate monolithicity to the
masonry, whose behavior tends to be similar to that of
three independents leaves (Condoleo, 2007).
Differently from the external leaf, which is homo-
geneous from the base to the roof, the internal leaf has
different characteristics from the base to the principal
body. The main difference between the two parts is rep-
resented by the dimensions. The thickness of the base
is approximately the same as that of the foundations,
with a value of approximately 110 cm. The principal
body, which rests on the base, has a smaller thickness,
with a value close to 75 cm. The internal leaf of the
base of the annexes is constituted by layers of entire
and half bricks, with irregular arrangement, and a fill-
ing of variable thickness made of clay, chamotte and
quartzitic temper (Fig. 9).
Before surveying the external and internal surfaces
of the walls it became necessary to remove the vege-
tation layers. This cleaning process made the survey
of all surface damages possible, together with a clear
identification of the missing parts. The results of the Figure 11. Crossing cracks pattern of the northern side.
survey process were reported on drawings, in order to
facilitate the damage interpretation and the detection
of its causes. For this reason, it was decided to put a timber provi-
The overall structural stability is mainly endangered sional structure. In order to avoid any modification of
by several cracks, passing through the entire wall thick- the structural behavior, the scaffolding was designed
ness. Moreover, the loss of the lintel and part of the to be passive, which means that its bearing capacity is
vault of the main entrance jeopardized its stability activated only in case of collapse or movements of the
(Figs 10–11). building itself (Fig. 12).
638
Figure 13. Finite element mesh of the intact temple.
639
neglecting any connection, thus considering the three
leaves as fully independent (case C). As a matter of
fact, the leaves are likely to be partially connected,
according to the construction technique (see Sec. 2),
so that stresses can be transmitted from a leaf to the
neighbouring ones to a certain extent. To model this
situation, the interface between each pair of adjacent
leaves was supposed to comply with a Coulomb-type
interface (case B). The friction angle was given a value
of 30◦ , whereas the tensile strength of the interface
was neglected: the latter is likely to be a conservative
assumption.
Only the effects of the self-weight of the materi-
als were investigated in the analyses shown in Sec. 4,
failing any evidence of significant ground settlements.
4 NUMERICAL ANALYSES
The main results of the numerical analyses carried out Figure 15. Case A: contour plots of the minimum principal
on the two models of the temple are presented in the stress in the intact temple.
form of contour lines of the extreme principal stresses
plotted on the deformed FE mesh. The stress values
are expressed in Pa units. To appreciate the effect of
the different assumptions made about the degree of
connection of the wall leaves, the same magnification
factor for the displacements (=1650) was employed
in all the figures shown in the continuation of this
Section. Only the results obtained in the cases A and B
are shown: the results pertinent to case C were found
to be unrealistic, as illustrated in the continuation.
640
Figure 17. Case B: contour plots of the minimum principal Figure 19. Case A: contour plots of the minimum principal
stress in the intact temple. stress in the damaged temple.
641
was found to be compatible with the material strength
properties when the temple was still intact, unless
the excessively severe assumption of fully indepen-
dent wall leaves is made. Thus, the possibility that the
collapse of the temple might be partially due to the
self-weight alone seems to be excluded.
Nevertheless, its current state is the result of the
damage caused by the bombings during the war and the
lack of repair and maintenance during the last decades.
Therefore, the analysis was repeated to address the
damaged situation.
As very high tensile stresses were computed in the
damaged temple accounting for a limited interface
frictional bonding (case B), the building in its current
condition is not found to be safe. Provisional scaffold-
ings have already been put up, to avoid the collapse
of the upper part of the building (see Fig. 12). The
safety assessment of the damaged building is worth
Figure 21. Case B: contour plots of the minimum principal being further analyzed. To this end, it might be appro-
stress in the damaged temple. priate employing more reliable constitutive laws, both
for the brick leaves (accounting for their brittleness)
and the interface between the leaves (accounting for its
non negligible tensile strength, coming from the partial
degree of connection of the leaves). Also, an attempt
at evaluating the local stress concentrations due to the
vegetation that took root in the temple is planned, by
introduction of a suitable pressure into some of the
main cracks.
ACKNOWLEDGMENTS
REFERENCES
Figure 22. Case B: contour plots of the maximum principal Binda L., Condoleo P., Cucarzi M., Lê Thành V., Pichard P.,
stress in the damaged temple. Hoàng Ða.o K. 2007. Preservation of temples in Mỹ
So,n (Vietnam). In P.B. Lourenço, P. Roca, C. Mod-
ena, S. Agrawal (Eds.), Structural Analysis of Historical
values, however, are of the order of −0.5 MPa, that Constructions. New Delhi: McMillan India Ltd.
is, compatible with the material compressive strength. Binda L., Tedeschi C., Condoleo P. 2006. Characterisation
Wide regions where the maximum principal tensile of materials sampled from some Mỹ So,n temples. Proc.
stress exceeds 0.2 MPa are found (Fig. 22). They are ICCE 7, Tehran, 8-10/5/2006, CD ROM.
mostly located at the base of the deck of the tem- Condoleo P. 2007. I templi di Mỹ So,n: indagini e progetto di
ple, where the lower walls collapsed and the deck conservazione del Gruppo G, PhD thesis in Preservation
behaviour resembles that of a cantilever. As these high of Culture Heritage of Architecture Politecnico of Milan.
Kwiatkowski K. 1985. Recherches sur les Monuments du
stresses could not be borne by the material, a scaffold- Champa. Rapport de la Mission Polono-Vietnamienne
ing was duly placed in the opening caused by bombing 1981–1982. Warsaw.
to prevent the temple deck from failing. Kwiatkowski K. 1990. Recherches sur les Monuments du
Champa. Rapport de la Mission Polono-Vietnamienne
1983–1986. Warsaw.
5 CONCLUDING REMARKS Parmentier M.H. 1904. Les Monuments du Cirque de Mỹ So,n.
Hanoi: F.-H. Schneider.
According to the numerical analyses performed, the Parmentier H. 1918. Inventaire descriptif des monuments
stress in the temple E7 in Mỹ So,n archaeological site čams de l’Annam, Voll. 2, E. Leroux, Paris.
642
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The paper refers to the studies recently carried out on the church of Santa Maria del Mar in
Barcelona and Mallorca cathedral, built during the 14th and 15th centuries. Historical research has provided
significant insight on the construction processes, which were remarkably different in the two buildings. This
information has been considered to investigate the possible link between construction and existing deformation
and damage. As shown by these examples, insight on the construction process (and later man-caused alterations)
may be important for an accurate structural analysis.
643
stabilized soil settlements). Earthquake does not lay
in this group for obvious reasons. Yet another “action”
belonging to this group, which by no means should
be neglected, is the construction process itself and its
related hazards and difficulties.
644
Figure 2. Arrangement of stone blocs in octagonal piers.
3 HISTORICAL INFORMATION
645
Succession War) and the Spanish Civil War (1936– was decided to increase the height of the vaults. The
1939), among other episodes. construction of the main nave developed during the rest
The architectural unity and the consistency of the of 14th and 15th c (with a major interruption from 1460
construction process suggest that a fully developed to 1570). The main façade, of noticeable Renaissance
project should have been well established before 1329, style, was built from 1594 to 1601, when the cathedral
when the construction began. At this moment, not even was consecrated.
the choir of Barcelona Cathedral, frequently regarded Research on the historical books has provided sig-
as the immediate precedent, was finished. nificant hints on the construction process. As shown
Construction of Mallorca Cathedral began by year in Fig. 6, the construction of the nave progressed,
1300 starting with the presbytery (which comprises the bay after bay, from the presbytery towards the façade
so-called Trinity and Royal Chapels). According to the (the last part to be built). Construction of the chapels
most widely accepted interpretation, by 1330 it was was ahead because of the funding provided by noble
decided to build the remaining construction according families or corporations willing them as pantheons or
to a three-nave plan and yet, by the mid of 14th c., it gremial chapels (Domenge, 1997).
646
It has been possible, at least for one of the bays partially collapsed 30 years after its construction.
(the 4th one), to identify the process leading to its com- A significant number of vaults were repaired or recon-
plete construction (Fig. 7). Once again, it started with structed during the 17th, 18th and 19th centuries. Due
the lateral chapels, followed by the piers, then one lat- to the concerning out-of-plumb (about 1.3 m), the orig-
eral vault, then the other and finally the central one. inal façade was taken down replaced by a new one
In the case of this bay, the construction of the vaults during the second half of 19th c. Demolition was
lasted 7 years. It should be noted that during a period decided in March 1851 and hence was not connected
of about 5 years), the lateral vaults were already push- to the earthquake occurring in May the same year.
ing against the pier while the lateral vault was not yet
there to counteract their thrust.
The building has experienced significant prob- 4 EXISTING DAMAGE
lems and repairs. The 4th vault (previously discussed)
Four different types of alterations, visible in both
buildings, are here highlighted and discussed:
1- Cracking in piers. Cracks exist in a few piers
of Mallorca Cathedral and most piers of Santa
Maria del Mar (Fig. 8). Vertical or oblique cracks
have developed across the stone extending, in
some cases, to several rows. They tend to concen-
trate close to the corners of the octagonal section
(the less confined parts) and, in some cases, shape
full wedges partially or totally detached from the
core of the pier. In the case of Santa Maria del Mar,
the cracks are related to additional forms of dam-
age (black patina and crusting, superficial loss of
stone and mortar at joints) which can be clearly
associated with the severe fires experienced. The
reason for similar cracks in a few piers of Mallorca,
randomly distributed, is less clear.
2-Cracking in arches and vaults. In the lateral naves
of Santa Maria del Mar, a longitudinal crack has
developed following the keystones of the transverse
arches (Fig. 8, right). The longitudinal cracks in the
aisles are due to a differential settlement between
the pier and the buttress. The arches do actually
show the deformation and damage which should
be expected from this type of settlement (Fig. 9,
left). In a few cases, initial deformation appeared
Figure 7. Mallorca Cathedral’s 4th bay construction Figure 8. Cracking in piers (Mallorca Cathedral, left, and
sequence. Santa Maria del Mar, repaired, right).
647
Figure 9. Aisle arches of Santa Maria del Mar. Deformation
and cracking (left). Arch deformed due to accidental removal
of centering causing expulsion of unhardened mortar (right).
648
Nevertheless, Santa Maria del Mar was built fol- First, the lateral vaults could have been stabilized
lowing a procedure which would ascertain, at any by means of previously built transverse arches, as in
moment, an easy balance of forces. First, the con- Santa Maria del Mar. If so, the almost 20 m span arches
struction of the entire perimeter structure (including would have needed some stabilizing extra weight to
buttresses, walls, chapels along the choir, sides and resist the thrust of the lateral vaults without experi-
façade) would provide a stiff system able to laterally encing inward deformation or collapse. Extra weight
buttress the rest. Then, the construction of all longitu- actually exists (Fig. 3), although its original purpose
dinal and transverse arches (of both central and lateral might be different. Second, they could have been sta-
vaults) would provide the necessary lateral stability bilized by means of auxiliary devices such as steel or
to the slender piers. The transverse and longitudinal timber ties across the lateral arches or struts across
(or clerestory) arches would also grant full stability the central one. Third, they could have been built
during the formation of the entire vaulted roof. In the without any stabilizing element, the partial structure
longitudinal direction, the stability was ultimately pro- being (precariously) stable by itself thanks mainly to
vided by the choir and façade buttresses, already built. available tensile strength.
This is probably the reason why, yet at present, the Whatever the method, there was some hazard and
piers are showing such a limited drift in spite of their chance for damage and deformation. This is consis-
slenderness. tent with the fact that the lateral deformations at the
The process, however, may have generated an incon- piers, as already mentioned, are very variable (almost
venient side effect. The construction of the vaults “random”) although large in average, and suggests
took place about 30 years after the perimeter struc- that the outcome was very sensitive to the skills and
ture had been built. As should be expected, the lateral methods used by the different builders. Moreover, dif-
structure had by then already experienced most part ferent approaches (as the three ones mentioned) may
of its possible soil settlement. Due to the settlement have been used during the certainly long and irregular
of the piers, the new members built across the aisle construction process.
(arches and vault membranes) experienced a vertical The longitudinal stability at intermediate stages is
deformation (or differential settlement). The lateral even more challenging as the piers had to face the
longitudinal crack along the keystones of the trans- unbalanced longitudinal thrust of both the lateral and
verse arches may have been caused by this effect. In central vaults. According to the historical information
fact, settlements may have been important during the available, a previous construction of all the clerestory
construction because the foundation soil, composed arches (as a way to stabilize the bays at intermediate
of rubble and loose sand, is certainly deformable. stages) should be clearly disregarded. The use of pos-
However, a later significant increment of differen- sible temporary devices (temporary ties or buttressing
tial settlements between lateral structure and piers walls) appears as a likely possibility.
seems unlikely because the rubble existing over the
vaults (a sort of medieval lime concrete supposedly
placed in a later stage but still during the construc- 6 STRUCTURAL ANALYSIS
tion) does not show any crack in correspondence to
the one seen at the intrados of the vaults. More- Structural analysis has been carried out using limit
over, the arches appear more damaged than the vault analysis and non-linear FEM calculations. Previous
membrane, which does not either accompany their results concerning dead loading and seismic per-
significant deformation (figure 9, left); the only expla- formance of Mallorca Cathedral have been already
nation is the later construction of the vaults (after presented (Clemente et al., 2006, Martínez et al, 2006,
the arches had been built) and the possible devel- Martínez, 2008). The present discussion focuses on
opment of most the settlement while the material of the performance of the structures subjected to dead
the vault membranes was gradually placed on the loading and the possible influence of the construc-
arches, but yet with the possibility of correction or tion process. The non-linear FEM analyses have been
almost free deformation while the mortar had not yet carried out using a continuous damage constitutive
hardened model (Clemente et al. 2006). The material proper-
As mentioned, the construction of Mallorca Cathe- ties have been estimated based on laboratory tests
dral followed a different path involving the subsequent on cores taken from the buildings or the original
construction of the bays. The lateral vaults were built quarries. The overall stiffness (value and distribution)
(as in Santa Maria del Mar) before the construction of was assessed by comparing experimental and numer-
the central ones. In this case, historical research has ical natural frequencies, the obtained from ambient
not provided, so far, any hint on the way the structures vibration measurements (Martínez et al. 2006).
were stabilized while the central vault was not yet built In the case of Santa Maria del Mar, the hole drilling
(third stage of Fig. 7). Several possibilities, however, test was used to measure the average compression
may be considered. stress at the base of two piers. The obtained value,
649
7 CONCLUSIONS
650
of Historical Constructions. New Delhi. New Delhi: Martínez, G., Roca., P., Caselles, O., Clapés, J., Canas, J. A.
MacMillan India Ltd. 2006. Characterization of the dynamic response of the
Clemente, R. 2007. Análisis structural de edificios históri- structure of Mallorca cathedral. 5th Int. Conf. of Struc-
cos mediante modelos localizados de fissuración. tural Analysis of Historical Constructions. New Delhi:
Ph. D. dissertation. Barcelona: Universitat Politècnica de MacMillan India Ltd.
Catalunya. Martínez, G., 2008. Vulnerabilidad sísmica de edificios
Domenge, J. 1997. L’obra de la Seu. El procés de construc- históricos de obra de fábrica de mediana y gran luz. Apli-
ció de la catedral de Mallorca en el tres-cents. Palma de cación a la cathedral de Mallorca. Ph. D. dissertation.
Mallorca. Universitat Politècnica de Catalunya: Barcelona.
Fontseré, E. 1971. Recopilació de dades sísmiques de les Vendrell, M., Giráldez, P., González R., Cavallé, F., Roca, P.
terres catalanes entre 1100 i 1906. Barcelona: Generalitat 2008. Santa Maria del Mar. Estudi històric-constructiu,
de Catalunya. materials de construcció i estabilitat structural. Barcelona:
ICOMOS / ISCARSAH (2005). Recommendations for the UB-Patrimoni, Veclus, UPC.
analysis, conservation and structural restoration of archi-
tectural heritage.
651
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
E. Paupério
Construction Institute, Faculty of Engineering, University of Porto, Porto, Portugal
A.G. Costa
Autonomous Section of Civil Engineering, University of Aveiro, Aveiro, Portugal
ABSTRACT: The historical centre of Porto was considered in 1996 UNESCO World Heritage due to the
outstanding universal value of the site, with urban fabric and historic buildings that bear remarkable testimony
to the development over the past thousand years of an important European city”. In order to contribute to the
maintenance of this remarkable set of buildings, with historical, architectural, constructive and heritage value,
in opposition to the continuous built heritage degradation process and the population dislocation from the centre
to the surroundings, NCREP –FEUP have been participating in many rehabilitation projects, analyzing the
conservation state of building structural elements (stone masonry walls and timber floors and roofs). The focus
of the present paper is given to the evaluation of the physical state of timber structural elements, namely to
the various instruments used on the Analysis and Diagnosis phases, in particular to the non destructive (NDT),
medium-destructive (MDT) and, in a certain context, destructive testing techniques.
1 INTRODUCTION
653
careful analysis made by experts from different areas,
involving not only structural and architectural matters,
but also historical and social issues.
With the objective of promoting sustainable
rehabilitation interventions, following ICOMOS
(International Council on Monuments and Sites) rec-
ommendations, NCREP (Nucleus for the Conser-
vation and Rehabilitation of Buildings and Built
Heritage, a team within FEUP – Faculty of Engineer-
ing of the University of Porto) have been participating
in many rehabilitation projects, doing inspections and
structural analysis of several old buildings, in partic-
ular at the Historical Centre of Porto. The work of
NCREP consists on the evaluation of the building
structural elements, which are usually made of stone
Figure 3. Main façades of two XIX century buildings at São
masonry walls and timber floors and roofs, assessing Domingos and Lóios squares, Porto.
their physical state and, consequently, the possibility of
rehabilitation. The paper will focus on the analysis of
the timber structural elements, namely on the Inspec-
tion and Diagnosis phases, which are fundamental for
a correct structural analysis.
The importance of these two phases as support of
the intervention is emphasized in the ICOMOS text
“Principles for the preservation of Historical Timber
Buildings”: “A thorough and accurate diagnosis of
the condition and the causes of decay and structural
failure of the timber structure should precede any Figure 4. Beams and transversal bars of a timber floor. São
intervention. The diagnosis should be based on doc- Domingos building, XIX century (1856), Porto.
umentary evidence, physical inspection and analysis,
and, if necessary, measurements of physical conditions
and non-destructive testing methods”.
defining the building contour: the two façades and the
two lateral walls. The other walls are usually wooden
2 PORTO HISTORICAL CENTRE BUILDINGS type panels, sometimes reinforced by a wooden strut-
and-tie system that improves its in-plane behaviour and
2.1 Architectural and constructive aspects the connection between walls and floors. This internal
The buildings of Porto historical centre were mostly structure, together with the timber floors and the roof
built between the XVII and XIX centuries, evolving trusses, strongly contributes to the buildings efficient
from older constructions. They present a rectangular performance under vertical, but particularly under hor-
plant, with a narrow front, a long depth and a variable izontal loads. As a result, it is important to ensure a
height: an average of 3 to 4 floors and a maximum of 8 good link between these elements in order to guarantee
floors (the top floors being usually added afterwards); a good and stable global behaviour.
the architecture of the inside is quite repetitive, usually The floors that are normally supported on the
with staircases at the centre, toped by a roof skylight, lateral masonry walls consist of a set of circular sec-
one compartment turned to the front and one to the tion beams (diameters varying between 0,15 m and
back of the building, Figure 3. 0,40 m), spaced 0,60 m, and a secondary transver-
These constructions that normally belonged to the sal bar system, Figure 4. Structural elements with
popular and bourgeois branch of the population aren’t rectangular section can also be found, usually as the
exclusive from Porto and exist all over Portugal and in result of interventions performed after the beginning
different European countries. Porto uniqueness resides of the XX century. The most common species used
in the fact that, in the old part of the city the slen- on the construction of floors were chestnut, oak, pine
der houses represent almost the totality of the civil and eucalyptus. However, floors executed with other
buildings. species also exist.
The timber roofs present variable configurations,
depending on the size of the building. As the spans
2.2 Building structural scheme
were normally lower than 6 m, the structure was very
The structural schemes of these buildings are also simple, with wooden trusses constituted by two rafters
quite typical, consisting of four stone masonry walls and a tie beam. In other situations, trusses with two
654
Figure 6. Deformation of timber floor. São Domingos
Figure 5. Timber roof structure. António Carneiro building, building, XIX century (1856), Porto.
XX century (1916), Porto.
655
Figure 7. XX century (1946) refurbishment project of São
Domingos building, XIX century (1856). Porto historical
Figure 8. Use of the videoscope for the visualization of a
municipal archive.
timber floor. Belomonte Palace, XVIII century, Porto.
656
Figure 10. Results from the resistograph on a tie beam of a
wood truss of Belomonte Palace, XVIII century, Porto.
657
Measurements are made on different points of struc-
tural elements (middle spam and supports) and on
different structural elements with the objective of
doing, with a large number of surveyed structures,
comparative studies about the elements integrity.
It is important to refer that, due to the short num-
ber of existing correlations for the different species,
it isn’t possible to reach quantitative values, but only
qualitative, but even so valuable information about the
conservation state of timber elements. Figure 12. Seismograph used by NCREP.
Most of the correlations for the values obtained by
this equipment relates the superficial hardness only application of controlled loading and unloading cycles
with the wood density and not with mechanical proper- with continuous force and displacement monitoring. It
ties, which ends up being a disadvantage. On the other is important to refer that after unloading no important
hand, the majority of the experimental tests showed residual deformations should exist on the structure.
that the moisture content considerably affects the depth Thus, before carrying out such a test, it’s important to
of penetration (Bonamini et al., 2001) and, so, it’s estimate, numerically or not, its behaviour, trying to
important to evaluate correctly the wood moisture avoid surpassing the structure elastic range. In chap-
content. Görlacher (1987) obtained good correlations ter 5 a case of a building in which a load test was
(R2 = 0,74–0,92) between the density and the depth performed is presented.
of penetration of the Pilodyn for an high number of
measurements. Furthermore, Feio (2005) found some
correlation not only for the density, but also for the 5 ANTÓNIO CARNEIRO BUILDING A CASE
modulus of elasticity and the strength of the chestnut STUDY
wood (Castanea Sativa, Mill). However, and although
for the density the correlation is acceptable, for the 5.1 Introduction
modulus of elasticity, Ec,90 , as for the strength, fc,90 ,
the estimated correlations are poor, and the author does In one of the buildings analyzed by NCREP the
not advise its use as a quantitative measure. Turrini and possibility of performing destructive and medium-
Piazza (1983) proposed empirical relations between destructive tests on the floors came up because, due to
the impact force and the modulus of elasticity, adopt- architectonical incompatibilities, the general project
ing a factor of reduction for the modulus of elasticity dictated its demolition. The adoption of destructive
based on the visual classification of the elements: 0,8 tests is naturally to avoid, because it doesn’t allow
for elements without defects and 0,5 for elements with the tested elements to be maintained. However, this
some knots and small degraded areas. opportunity of testing a set of floors that would be
demolished ended up being extremely important, since
it allowed the evaluation of parameters and character-
4.6 Seismographs istics that will surely be helpful in similar structural
The seismographs are instruments used frequently by rehabilitations.
NCREP in the inspection of timber floors. In partic- The surveyed building, built in 1916, has an area of
ular, two 18 bits resolution seismographs have been 12,0 × 30,0 m2 and presents floors with 0,20 m diam-
used, Figure 12. They include tri-axial accelerome- eter beams, spaced 0,60 m and with 6,0 m span. Above
ters, and an autonomous memory. The transference the beams there is a 2,5 cm thick floorboard. Perform-
of the information to a computer allows afterwards ing the transversal link, there are 0,10 m diameter bars,
sign analysis. Readings are done in different loca- spaced 2,0 m.
tions on a floor, supplying its natural frequencies. Besides a campaign of non destructive tests carried
This evaluation permits estimating the floor stiffness out with the Pilodyn and the seismographs, among
and, consequently, the efficiency of its structure or its other instruments, a load test was then performed
physical and (or) material state. together with a very particular test for the evaluation
of the beam-wall friction, using a hydraulic actuator.
4.7 Load tests The campaign of tests, in particular the load test
and the frequencies evaluation, had the main goal
The load tests allow an in-situ realistic evaluation of of assessing the global floor and the single beams
the stiffness of a structure and, in particular, of the behaviour when submitted to a monotonic loading his-
capacity of a structure to resist a certain load. These tory. The setup was also prepared to allow evaluating
are expensive tests that are used by NCREP whenever the effectiveness of the transversal bar system and the
considered convenient. The tests are normally done for floorboard contribution to the stiffness and resistance
loads of about 1,5x the service load and consists on the of the floor.
658
Figure 13. Load test on a timber floor. António Carneiro
building, XX century (1916), Porto.
Figure 14. Load-displacement diagram for two isolated
beams loaded simultaneously (the load refers to the total load
5.2 Load test applied).
659
techniques, it is possible to withdraw helpful informa-
tion about structural timber elements. Nevertheless,
some limitations were verified in their utilization due
to the difficulty to estimate quantitative values for
mechanical and physical parameters.
To be used in a more trustworthy way, the Resisto-
graph and the Pilodyn need experimental studies with
different wood species to create newer correlations.
Nevertheless, these instruments were very helpful in
the accomplishing of qualitative parameters, in partic-
Figure 16. Beam wall friction test using a hydraulic ular in the perception of degraded areas, with voids
actuator. or cracks. In particular, the Resistograph permitted to
find voids inside structural elements that apparently
auxiliary bars, with small inertia, to simulate the floor- were in good state of conservation. On the other hand,
board, linking the beams. Another possible model, that the Pilodyn allowed, through tests on different struc-
tends to approach the floorboard real effect over the tural elements, to conclude about their relative state of
whole structure, consists on considering low thick- conservation through superficial hardness.
ness slab elements to simulate the floorboard. These In addition, the use of seismographs on timber
elements rest on the beams and transversal bars, but floors permits the measurement of natural frequen-
aren’t connected to the resistant walls, as it is usually cies and, consequently, the estimation of the average
observed in buildings (Neves, 2004). modulus of elasticity. In the future, with the infor-
mation collected on different surveyed floors, it will
be possible to organize a database that will allow us
5.4 Beam-wall friction test to establish correlations between the measured values
To evaluate the resistance of the friction connections and the integrity of the floor.
between the timber beams and the masonry walls, a Finally, it is important to refer that medium-
destructive test was carried out. A segment of beam destructive and destructive tests, despite of being
was cut of to install a hydraulic actuator that would applied only under particular conditions, can be an
act along the beam axis. In order to maintain the two important contribution for the knowledge on the mate-
segments of the tested beam in the original position, rials and structures. In particular, the load test that, in
it was conceived a metallic structure with pulleys to this case, had a destructive character, allowed estimat-
support it (not allowing vertical movements but free ing the floorboard contribution to the floor stiffness,
horizontal displacements), Figure 16. To set the actu- as well as verifying the low efficiency of the transver-
ator on the beam, metallic corner cupboards fastened sal bars on the distribution of vertical loads. The
by M12 bolts were used. The test consisted on apply- beam-wall friction test, only possible with the cut of
ing pull out and in movements of the beam in relation the tested beam, allowed concluding about the weak
to the wall. Displacement transducers were installed contact conditions between the timber beam and the
at the extremities of the beam to evaluate the load- masonry wall.
displacement diagram of the beam-wall connection.
With the purpose of verifying any possible horizon-
tal movement of the wall during the test, transducers REFERENCES
were installed on the walls. The results permitted to
conclude about the limited efficacy of the connection Botelho, J. 2006 – Avaliação não destrutiva da capacidade
between the two elements, traduced by a friction force resistente de estruturas de Madeira de edifícios anti-
gos. Tese de Mestrado em Reabilitação do Património
of about 2,0 kN. Edificado. Universidade Porto – Faculdade Engenharia.
Portugal.
Costa, A., Ornelas, C., Paupério, E., Guedes, J., Ilharco, T.
6 CONCLUSIONS 2006. A Inspecção como Instrumento de Decisão. Apre-
sentação de um Caso Prático num Edifício com Valor
To allow the preservation of Porto Historical Centre Patrimonial. Patorreb 2006. Porto.
buildings (UNESCO World Built Heritage), NCREP- ENV 1995-1-4. 1998. Eurocode 5 – Design of Timber Struc-
FEUP has participated in many projects, analysing the tures Part 1-1: General – Common rules and rules for
buildings.
state of conservation of their structural elements. The
Faria, A. 2002, Reabilitação de Coberturas em madeira em
present paper refers to the work developed on the anal- edifícios históricos.A Intervenção no Património; Práticas
ysis of timber structures, floors and roofs, namely on de Conservação e Reabilitação, Porto.
the Inspection and Diagnosis phases. It was concluded Feio, A. 2005 – Inspection and Diagnosis of Historical
that, with the joint use of several non destructive testing Timber Structures: NDT Correlations and Structural. Tese
660
de Doutoramento em Engenharia Civil, Universidade do Multimedia Technologies. Euro-Mediterranean Confer-
Minho – Escola de Engenharia. Portugal. ence Traditional Mediterranean Architecture. Barcelona.
Google Earth, Google. Neves, N., Costa, A., Arêde, A. 2004. Identificação
Gorlacher, R. 1987. Non destructive testing of wood: an in- Dinâmica e Análise do Comportamento Sísmico de um
situ method for determination of density. Holz as Roh- quarteirão localizado na cidade da Horta – Ilha do Faial.
und Werkstoff. Vol. 45. Sísmica 2004. Porto.
Ilharco, T., Romão, X., Paupério, E., Guedes, J., Costa, A. Turrini, G., Piazza, M. 1983. Il recupero dei solai in legno.
2007. Organization of information on Built Heritage using Esperienze e realizzazioni”. Recuperare, Vol. 7.
661
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
D. Makay
Irod M Ltd, – Built Heritage Conservation Research, Planning and Consulting, Cluj, Romania
B. Szabó
Technical University of Cluj, Romania
ABSTRACT: Baroque roof structures encompass important and unique historic values, within: (a) their struc-
tural conformation (trusses and longitudinal bracing systems); (b) the craftsmanship of framing and jointing
(carpenter marks, joints); (c) the authentic, historic timber material that has been preserved for centuries.
Given the high level of safety characterising Baroque roof structures, a relatively significant number of these
survive in Transylvania, though there is no through inventory, yet.
The lecture presents the typology of Transylvanian Baroque roof structures, the PhD research programme
carried out in order to analyse these structures, and through two case studies the level of safety, durability and
the unique character of these structures.
Though professional interest has increased towards historic (roof) structures, Baroque roofs in Transylvania
face a double threat. Scientific research needs to result in a guideline to be used in order to correctly
conserve/reinforce Baroque roof structures throughout Transylvania.
663
Figure 3. Secondary truss of a typical Baroque roof struc-
ture. Additional element: 15 – trimmer 14 × 17.
Figure 1. Early Baroque roof, a transitory structure between
Gothic and Baroque, Franciscan church, Cluj, Romania.
664
– Concerning Romanian (and Hungarian) terminol-
ogy, we need to mention that for almost half a
century (during the Communist Era) both timber
structures and historic buildings were neglected,
therefore their terminology has been forgotten.
– The above mentioned dictionary is trying to reflect
the structural behaviour/role of the various ele-
ments within the structure.
In order to promote international cooperation and
exchange of information, it would be necessary to
elaborate an internationally accepted English termi- Figure 5. Eaves pentagonal purlin within the main truss
joint of: tie beam (1), main/compound rafter (3) and com-
nology for non Anglo-Saxon historic roof structures.
mon rafter (2). Upper (middle) pentagonal purlin within the
The present lecture could present a starting point main truss joint of: straining beam (4), main/compound rafter
in order to discuss the proposed terminology for roof (3), collar beam (6) an upper compound rafter and common
structures with Baroque character. rafter (2).
In order to clearly define roof structures with
Baroque character, contrasted to medieval (Gothic) pentagonal cross section, is the one with Baroque
roof structures, the innovative elements and structural character. The special elements are: the eaves
conformation need to be shown. pentagonal purlin and the upper (or middle) pen-
tagonal purlin (Fig. 5).
There exists a debate concerning the cor-
2.2 Innovations of Baroque roof structures rect designation of these elements as “purlins”.
Both in eclectic and Anglo-Saxon roof structures
Baroque architecture changed the volumes and ratios
purlins transfer gravitational loads from rafters
of buildings, therefore in many cases it was necessary
in secondary axes (trusses) to main trusses. But
to place the eaves height below the keystone of the
in Baroque roof structures generally secondary
vault.
trusses possess either or both tie beams and col-
(a) This has determined one of the innovations gener- lar beams, which actually minimise the transfer
alised within Baroque roof structures, that in most of efforts from secondary trusses to main trusses,
cases secondary trusses do not possess tie beams Therefore the lower/upper pentagonal plate is also
(various solutions within typology, criterion: 4 – a term frequently used for these elements.
see Fig. 8). In less important roof structures or where the
From the point of view of structural conformation, craftsmanship of the carpenters failed, either or
secondary trusses without their own tie-beams both pentagonal purlins can a have simple, rectan-
are not self-supporting systems (contrary to the gular shape.
secondary trusses of Gothic or Romanesque struc- (e) The middle suspension bar/king strut (7) in
tures, where secondary trusses do not transfer Baroque roof structures can be made up of a com-
gravitational loads to main trusses). pound section (double bars), representing the 3rd
(b) The invention of the Baroque straining system (see criterion of the typology.
Fig. 1) – which is actually a “false” or “partial” (f) The typical Baroque longitudinal bracing sys-
straining system. It is working properly as a strain- tem (Fig. 4) is one of the most innovative and
ing system only in the cases of secondary trusses most determinant sub-ensembles of a Baroque
without individual tie-beams and without collar roof structure. It represents an efficient way of
beams. In the other cases the straining systems providing longitudinal stiffness for the structure.
have determinant role in stiffening the structure for The Baroque longitudinal bracing system, like
horizontal loads. (The number of Baroque strain- bracing in Anglo-Saxon roofs, is always placed
ing systems is the 2 criterion of the typology). in the plane of the rafters (Fig. 6). The various
Secondary trusses without individual collar beams patterns of the bracing system are given within
are rare. (The roof structure of the Reformed the 6th criterion of the typology.
church, Cluj, secondary truss type II, Fig. 12, has (g) Unlike medieval roof structures based exclusively
no individual collar beam). on carpenters’ knowledge passed from genera-
(c) Mansard roofs – were developed in France, within tion to generation within the guilds, the Baroque
the Baroque architectural period, and mean roofs roof pattern was designed and drawn. There are
(rafters) with broken, uneven line (plane). This treaties and drawings of architects preserved, such
represents the 1st criterion of the typology. as Franz Ignaz Michael Neumann, The Construc-
(d) The only (continental) historic roof structure in tion Plan for theVaulting and the Roof, republished
Central-Eastern Europe containing elements with by Hansmann (2000).
665
Figure 6. Double St Andrew’s cross Baroque longitudinal
bracing, Jesuit Church, Bamberg, Germany. Figure 7. Roof structure of the Boys’ Dorm, Bethlen Col-
lege, Aiud (Utilitas, Transylvania Trust, 1999–2000). Code
Baroque roofs are mostly surveyed, assessed and stud- according to the typology: B.2.1.-a(3)-II-(1).
ied – as integrated parts of historic buildings – in order
to be repaired, conserved, reinforced or converted for county), to be carried out in 2008. The typology will
a new use. also be open for completion by an eventual through
These researches are objective-oriented and in many Transylvanian inventory.
cases can hardly be used for scientific purposes. The The typology, Figure 8 has been developed using 6
authors promote the elaboration of a guideline for criteria in order to identify the specific characters of a
those working in the professional survey, documen- given Baroque roof structure, as follows:
tation, and conservation or reinforcement planning of (1) continuity / plane of common rafters;
historic roof structures generally, and Baroque ones (2) number and type of Baroque straining systems;
particularly. (3) type / existence of a suspension bar;
(4) horizontal force transmission system;
(5) main / compound rafter;
3 TYPOLOGY (6) pattern of the longitudinal bracing system.
In order to provide a researchable data base, as well Using this classification, all Baroque roof structures
as an organised overview of the roof structures with can be identified by a code made up of 6 letters com-
Baroque character, a typology is a useful tool. The bined into three groups: 1 + 2 + 3 – general + main
typology presented within this lecture is based on the truss characteristics – 4 – identifying the load trans-
measured survey of 50 Baroque roof structures, as well mission from secondary to main trusses – 5 + 6 –
as on visits to Baroque roof structures in Hungary, describing the longitudinal bracing system.
Austria, Germany, and further roofs (not measured, Using the typology, the typical roof presenting ele-
just inspected) in Transylvania. ments of the Baroque roof (chapter 2) has the following
Two main principles were followed when develop- code: A.1.2(c)-b(3)-I. (1).
ing the present typology:
3.1 Classification by the continuity/plane
– It refers only to pure (mature) Baroque roof struc- of rafters
tures; it does not include hybrid or transitory
structures neither between Gothic and Baroque, nor All Baroque roof structures fit into one of the two
between Baroque and Eclectic structures (unlike basic groups: (A) – Continuous/linear common rafters
the detailed topography of roof structures in Mur- (plane); (B) – Interrupted common rafters (mansard
Mürz area, Steiermark, Austria, Caston). roofs).
– It analyses the structural and geometrical confor-
mation of both structural sub-ensembles and of 3.2 Classification by the number and type of
elements. Baroque straining systems
This typology presents a working phase of the Transyl- Up to a 10.00 m span, in many cases Baroque straining
vanian Baroque Roof Structure Typology, which will systems placed in main trusses are not combined with
embody the results of a through inventory (phase I) on other straining systems on upper levels (eventually just
a limited, but representative geographical area (Cluj with an upper collar beam). This represents the basic
666
Figure 8. Transylvanian Baroque roof structure typology, working phase 2007.
667
type of the classification (1), designated by the second
character of the code in the typology.
The most general type though is (2), where the
Baroque straining system on the first level is combined
with other straining systems on one or more upper
levels.
The third version included in the typology (3), is
rare within Transylvania, being used for large spans
and in important buildings. This subcategory includes
roofs with two Baroque straining systems placed one
above the other, combined or not with further straining
systems on upper levels.
In Austria, Germany these types are more usual than
in Transylvania/Hungary, connected to the economic
power of the area/time. Even more there can also be
identified Baroque straining systems on three levels.
This version was not included in the present typology,
as none was identified in Transylvania, yet.
In a number of special cases: large spans, or use Figure 9. Baroque roof structure of the Piarist church in
of the attic space as granary, or deposit functions, the Cluj. Code according to the typology: A.2.2(c).-c(3)-I-(1).
Baroque straining systems are additionally strength-
ened. These special cases are included in the typology
marked with ∗ , at criterion 2∗ .
668
different levels is not considered a defining criterion
in classifying Baroque roof structures.
Within the analysed structures five patterns were
identified in Transylvania, which does not exclude the
existence of further patterns. The shape is marked with
a symbol showing the pattern made by the diagonals.
The number of levels on which these are placed within
the structure is given in brackets. In the great majority
of the studied Transylvanian cases, Baroque longitu-
dinal bracing systems are placed on one level (unlike
examples known from abroad – Figure 6, two levels).
The patterns are the following:
– – the most common solution in Transylvania,
made up of a pair of ascending and a descending
diagonals, fixed in the eaves and the middle pen-
Figure 11. Secondary truss type I (with own tie-beam) of
tagonal purlin, also strengthened by a longitudinal
the Reformed church’s nave – 15.70 m span, Cluj, Romania. element placed at the middle of their height.
Pattern: MT/ST-II/ST-I/ST-II/MT. – Y – is a similar pattern, the only difference being
that the lower end of the diagonals is fixed to the
compound rafters.
– / or \ – a very rare pattern, with only one of the
diagonals present (Fig. 1).
– X – Saint Andrew’s cross pattern, used for more
elaborate constructions, and rare in the analysed
structures, in Transylvania.
– XX – double Saint Andrew’s cross, even more rare
(a single case in more than 50 roofs analysed).
A through survey and the statistical data gathered will
serve as the basis of qualitative assessment of various
structures (rarity, structural conformation, craftsman-
ship of jointing, span, age all will be basic criteria in
rating the value of a structure).
The authors represent the Doctoral School of the
Technical University of Cluj-N. The PhD thesis of
Figure 12. Secondary truss type II (with shoes) of the structural engineer Dorottya Makay, supervised by
Reformed church’s nave – 15.70 m span, Cluj, Romania. Professor Bálint Szabó, The Transylvanian Baroque
Roof Structures, aims at elaborating the rating of these
The main role of the compound rafter in the structures, as well at creating a guidelines for the
Baroque straining system is working for axial forces calculation of Baroque roof structures.
(compression from gravitational loads, compres-
sion/tension from wind loads). Due to the jointing
between the compound rafter and tie-beam, as well as
4 MODELLING BAROQUE ROOF
counterbrace and straining beam, transfer of bending
STRUCTURES – CASE STUDY: REFORMED
moment is also possible (chapter 4).
CHURCH, KOGALNICEANU STREET, CLUJ
Its secondary role is serving as the inclined post
of the longitudinal bracing system, placed parallel to
In the following we shall formulate the main points
the plane of rafters. This solution is marked by the
and questions raised by the modelling and calculus of
character – I – within the third group of characters.
Baroque roof structures, exemplified on the 15.70 m
A less used (cheaper and more rudimental) solu-
span, 56◦ angle roof structure of the Reformed church
tion is having two different bars with smaller sections
in Cluj.
placed one on the other (II).
The safety of roof structures with Baroque char-
acter (being at least 150–200 years old) is proven by
3.6 Classification by the pattern of the
their very existence, having survived extreme condi-
longitudinal bracing system
tions while possessing (in many cases a high number
It refers only to the Baroque longitudinal bracing of) dysfunctional joints and / or biologically decayed
system. The common solution with knee braces on or missing elements.
669
Figure 14. Model of the 3D joint.
670
In the framework of carpentry workshops, in situ
repairs based on the minimum intervention principle
was carried out. (Fig. 16)
A continuity joint assures 50–70% of the capacity
of the element’s full cross section. Therefore the hand-
out of Baroque roof conservation needs to discuss the
estimation of the overall change in stiffness and resis-
tance of the structure through the introduction of a high
number of repair joints.
Baroque roof structures represent exceptional his-
Figure 16. Minimum intervention carpentry repairs,
Bethlen College Aiud, ACTT 2000. toric values which need to be conserved for future
generations. Therefore their inventory and perfect
structural understanding is indispensable in order to
5 THE PRINCIPLE OF MINIMUM assure their protection, conservation and consolida-
INTERVENTIONS – CONCLUSIONS tion, without destroying the historic fabric and values
they carry. So is expertise and craftsmanship, from
Baroque roof structures are safe and durable. The consultant to craftsmen.
roof of the Boys’ Dorm of the Bethlen College, Aiud,
had by 1999 more than 50% of its trusses decayed
REFERENCES
(one or more joints dysfunctional or one or more ele-
ments missing). Nevertheless, the ensemble was still Caston, S.C. Ph. 1998. Dachwerkstopographie für die Steier-
standing. mark. Ein Survey der historischen Dachwerkskonstruktio-
The British practice of minimum intervention was nen in der Mur-Mürz-Furche. Österreichische Zeitschrift
introduced to Romanian practitioners first through the f-r Kunst und Denkmalpflege (Heft 3/4):518–536. Wien.
ACTT 2000 project. This training programme was Hansmann, W. 2000. Barock und Rokoko: 25. Köln: DuMont.
jointly organised by the Transylvania Trust and the Makay, D. 2006. The Safety of Historic Roof Structures with
British Institute of Historic Building Conservation. Baroque Character. Historic Structures – 10th Scientific
(Since 2001 they run the Built Heritage Conservation Conference:. 136–152. Cluj-N: Utilitas.
Szabó, B. 2005. Illustrated dictionary of historic load-bearing
Training Centre within the Bánffy castle in Bontida, structures. Cluj-N: Utilitas.
Romania).
671
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
E.E.Toumbakari
Directorate for the Conservation of Ancient Monuments, Ministry of Culture, Athens, Greece
ABSTRACT: An investigation on the typology of the failures recorded on the first row (orthostate) of the
Parthenon northern wall was undertaken. The analysis of the failure causes was based on the simultaneous
consideration of (a) the kinematic constraints of each block, (b) the applied system of deformations/displacements
on each block and (c) the applied system of deformations/displacements on the wall itself. Moreover, a numerical
analysis was also undertaken, and characteristic results are also reported. The effect of the marble anisotropy
(in terms of the position of the marble soft plane inside the block) was also studied. The main conclusion of the
study is that the mechanical action of the connectors could explain the observed structural pathology, whereas
rust is not a necessary condition for the structural failures to occur.
1 INTRODUCTION
673
3 RECENT OBSERVATIONS and DATA
674
1994), attributable to dowel and anchor mechanical structural elements, instrumental (or not) follow up of
action. the behaviour of the structure and structural analysis”
(ICOMOS 2004).
On the basis of the above, the following hypoth-
The direct observation of the structural pathology
esis was developed: The concentration of failures at
focused on the analysis of the type of action (tension,
the area of the connectors could be attributed to the
shear. . .) that could have produced it. It consisted in
mechanical action of the clamps and dowels, when
the definition of failure modes on the basis of similar
the bearing structure (either entablature or wall) was
and repetitive failure patterns and in the search for a
subjected to strong external (horizontal) actions.
system of deformations/ displacements at the block
Rust could have, and certainly has, occurred, but
level, which could have generated them. This sys-
it is not considered sufficient for the generation of
tem had to be compatible to the kinematic constraints
such large number of failures. Moreover, rust is not
imposed by the connectors as well as to the defor-
the unique condition for the generation of small
mations/ displacements that have been applied on the
cracks, which will later initiate fracture. Those cracks
walls. The latter, subjected to horizontal actions, could
can also be generated if the marble blocks are sub-
be described as plates with partial constraints at three
jected to tension-inducing actions, such as temperature
edges and free at the upper edge, because the roof had
changes, environmental actions and vibrations due to
already collapsed before the explosion. Consequently,
dynamic loading. Recent experimental work on mar-
three criteria for the analysis and interpretation of the
ble properties and behaviour (Vardoulakis et al. 2002)
structural pathology were defined: (a) kinematic con-
established the existence of residual deformations on
straints applied on the building block, (b) system of
marble specimen subjected to tensile load as well as
deformations/displacements applied on the block and
the non-linear behaviour of the material even at an
(c) system of deformations/displacements applied on
early stage of loading. In the study, no explanation
the wall. These criteria had to produce a system of
as to the origin of those deformations is provided.
forces, compatible to the observed pathology. During
It can, however, be assumed that they could be the
the study, it was observed that the failure configura-
result of the opening of small cracks at surfaces con-
tion also depends on the local system of argillosilicic
taining argillosilicic veins, as is empirically known to
veins as well as on the direction of the marble soft plane
practitioners.These cracks can conceptually be assimi-
inside the block. The first do not affect the failure mode
lated to “notches”, which will facilitate (but, of course,
but do affect the precise shape of the failure surface
not generate) crack propagation under the mechanical
and volume of the detached fragment. The latter permit
action of the connectors.
or hinder the development of the one or other failure
Concerning the connectors’mechanical action, it is
mode. The qualitative approach permitted the determi-
important to estimate if the developing pressure is
nation of the type of actions that were applied by the
entirely transferred on the marble walls. It is recalled
clamps and dowels on the anchorage area. It is however
that the void between the connector and the marble was
not sufficient because, if it highlights the mechanisms
originally filled with lead, which, in unconfined condi-
of failure, it still cannot differentiate between the pri-
tions, is an energy-absorbing and deformable material.
mary cause of the failures, namely mechanical action
In the case of ancient connectors, however, lead is in
or rust.
a situation of triaxial confinement. It has to be admit-
A numerical analysis was therefore necessary in
ted that this confinement might not be perfect due to
order to investigate if it is possible for the failures to
constructional imperfections, nevertheless it is certain
occur under the connectors’mechanical action. Exper-
that the deformability of the lead is limited, especially
iments are very important in this respect, and if they
in a horizontal direction, where the filling of the void
are available in the case of dowels (Zambas 1989),
is complete. Thus, doubts are cast about the deforma-
they still need to be completed in the case of clamps
bility of the system under confined conditions. Further
subjected to shear. Two 3D models were created with
experimental work is nevertheless necessary to clarify
the use of solid Finite Elements and the Sofistik
this issue.
code (Toumbakari 2007a, reworked by Filippoupolitis,
2007). They represent marble blocks with length and
4 METHODOLOGICAL APPROACH height equal to 0.6 m and depth equal to 0.5 m (which is
the average depth of the Opisthodomos architraves and
The analysis followed a dual approach, qualitative and orthostate blocks). The choice of these dimensions was
quantitative (ICOMOS 2004, Tassios 2006). The qual- justified by previous numerical analyses (Toumbakari
itative approach is defined as “an approach based 2007a), in which it was shown that the effect of the con-
on the direct observation of the structural pathology nectors does not overpass a distance of 0.25–0.30 m.
and material deterioration as well as on historic and The models are shown in Figures 13 and 14. The
archaeological research”. The quantitative approach load cases and boundary conditions are described in
is based on “the properties of the materials and Tables 2 and 3. In what follows, only the analyses of
675
Table 2. Load cases and boundary conditions of the con-
nectors’ models (s.w. = self weight).
Load
cases Clamp∗
Load
cases Dowel∗
676
Figure 7. Initial position (dotted lines) of one block and
deformed position of two adjacent blocks. The position of
the southern (rear) dowel is shown on the left block, whereas
the northern (front) dowel is shown on the right block.
Figure 5. Typical configuration of Failure Mode I.
677
Figure 8. Detachment of the cover of the southern (rear) Figure 10. Failure of the northern (front) dowel.
dowel.
dowel
678
Table 5. Failure modes of the blocks of the even-numbered
rows of the Parthenon northern wall.
upper surface Failure mode
Row Surveyed
number blocks I II III IV
679
(a survey of the wall architraves was not possible
at the time of the study), it concerns roughly only
the 1/3 of the intermediate blocks.
3. Failure mode IV is rare in all cases. In the orthostate,
it was attributed to a specific direction of the mar-
ble weak plane inside the block (Table 4). It would
be of interest to analyze under which conditions it
appeared in the wall intermediate blocks.
4. Finally, the effect of the weak plane direction does
not only affect the possibility (or not) of generation
of a specific failure mode but its configuration as
well. Indeed, if failure mode I is recorded in prac-
tically all areas of the clamps of the orthostate, Figure 13. Model of the.
architrave and intermediate blocks, the height of
the detached fragment is significantly reduced in
the last case (the intermediate blocks).
680
the x-direction already results in the development of a REFERENCES
tensile stress equal to 4.79 MPa, which is considered
sufficient to initiate cracking. A shear force equal to 1. Archives of the Office for Conservation of the Acropolis
the clamp shear resistance (LC3) generates a tensile Monuments, 2007 (Photographic documentation).
stress equal to 9.22 MPa, which is similar to the mar- 2. CEB (Comité Euro-international du béton) 1994. Fas-
tenings to concrete and masonry structures – State-
ble tensile strength in the second strong direction. The
of-the-Art Report, Eligehausen R. et al. eds, Thomas
more complex situations considered in LC5 through 7 Telford.
result in the development of tensile stresses from 5 to 3. Filippoupolitis, M. 2007. Analysis of the developing
11 MPa in both x- and y-directions, which show that stresses at the architraves of ancient temples under vari-
crack generation is possible. In a similar way, in limit ous stress conditions – application at the Opisthodomos
state situations (LC3, LC6) the developing stresses at of the Parthenon, Diploma Thesis in progress, under the
the dowel area are sufficiently high to produce crack- direction of E.-E.Toumbakari (Prof. F.Karantoni): Univ.
ing (5.14 and 7.03 MPa). These numerical results are of Patras.
compatible to available experimental results on dow- 4. ICOMOS-International Scientific Committee for Anal-
ysis and Restoration of Structures of Architectural
els (Zambas 1994), which have shown failure of the
Heritage 2004. Recommendations for the analysis, con-
dowel anchorage area and not the dowel itself. servation and structural restoration of the architectural
heritage (draft).
5. Korres, M. & Bouras, Ch. 1983. Study for the restoration
8 CONCLUSIONS of the Parthenon, vol.1. Athens: Min. of Culture.
6. Orlandos, A. 1977. The Architecture of the Parthenon.
On the basis of the analysis of the structural failures Athens: Athens Archaeological Society, 2 vol.
in the orthostate of the Parthenon northern wall, the 7. Papadimitriou, G., Tsakiridis, P. & Skarvelis, P. 2007.
following conclusions were drawn: Laboratory examination of one iron clamp from block
14048 of the N.Colonnade, Techn. Report: Lab. of
1. A methodology for the analysis and interpretation Physical Metallurgy, School of Metallurgical Engg,
of the structural damages was proposed, based on NTUA.
the compatibility between (a) kinematic constraints 8. Tassios, T.P. 2006. Structural restoration of Monuments:
applied on the building block, (b) system of defor- Assessment of Values, Athens: Lychnos.
mations/displacements applied on the block and 9. Toumbakari, E.-E. 2006. Study for the Structural
(c) system of deformations/displacements applied Restoration of the Orthostates of the Northern Wall, 2
on the wall. vols. Athens: Office for Conservation of the Acropolis
Monuments.
2. Five failure modes were recognized: (I) clamp area
10. Toumbakari, E.-E. 2007a. Study for the Structural
failure attributable to shear, (II) dowel area failure Restoration of the Opisthodomos (2001–2004), 6 vols
attributable to tension and shear, (III) dowel area (in progress). Athens: Office for Conservation of the
failure attributable to tension, (IV) block separation Acropolis Monuments.
attributable to out-of-plane bending and (V) mass 11. Toumbakari, E.-E. 2007b. Study for the Structural
detachment at the base of the blocks, attributable to Restoration of the upper two rows of the entablature
shear with, occasionnaly, local crushing. of the Parthenon Northern Colonnade, 3 vols. Athens:
3. The configuration of the failure of the area of the Office for Conservation of the Acropolis Monuments.
clamps shows that the latter work mainly in shear 12. Zambas, K. 1989. Study for the restoration of the
Parthenon, vol.2b. Athens: Min. of Culture.
and not in tension.
13. Zambas, K. 1994. Study for the restoration of the
4. The position of the marble soft plane inside the Parthenon, vol.3b. Athens: Min. of Culture.
block was found to significantly affect the genera- 14. Vardoulakis, I., Kourkoulis, S., Exadaktylos, G. &
tion of the failure modes. Rozakis, A. 2002. Mechanical properties and compati-
5. The mechanical action of the connectors could bility of natural building stones in ancient monuments:
explain the observed structural pathology. Rust is Dionysos marble. In M.Varti-Mataranga & G.Katsikis
not a necessary condition for the structural failures (eds), Building Stones in Monuments; Proc. Intern.
to occur. Conf., Athens 2002. Athens: 9 Nov. 2001: 187–210.
681
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The Winter Garden, the largest of the Royal Glasshouses, is situated at the Royal Domain in
Laeken in the northeast of Brussels, Belgium. This glasshouse was built between 1874 and 1876 in order of
king Leopold II, according to the design of architect Alphonse Balat. A two-dimensional analysis demonstrated
the basis structural behaviour under a series of symmetrical load combinations: the structure works as a cupola
with a tension and a pressure ring. The stress levels and deflections in the structure have to be evaluated in a
three-dimensional model which is still in progress. For the time, it seems that the norm values are not exceeded.
The major threat for the metal structure is corrosion. Previous interventions and a very aggressive tropical indoor
climate inflicted heavy corrosion damage. Nowadays, this elaborate plain tour de force must undergo a major
restoration.
1 BUILDING THE WINTER GARDEN Arts in Brussels and the Royal Palace in Laeken, he
also designed most of the glasshouses of the Royal
The typology of glasshouse buildings originates from Domain, including the Winter Garden (Figure 1, no. 3).
the desire for a peaceful and green neighbourhood The Winter Garden (1874–1876) is a significant
in the overpopulated cities during the Industrial cultural legacy and an important artefact of the evo-
Revolution. The first iron glasshouse is the iron hot- lution of structural steelwork and Art Nouveau in
house in Hohenheim near Stuttgart (Germany) dating Belgium. It is the first glasshouse on the Royal Domain
back to 1789 (Kohlmaier & Von Sartory 1991). The with a complete iron and glass covering. The Winter
evolution of glass and iron production techniques in Garden was meant to be a glasshouse for social events
the nineteenth century was essential for the develop- and to this very day, it still performs that task.
ment and spread of the glasshouses. For the design of the Winter Garden,Alphonse Balat
The Winter Garden of Laeken is part of a major referred to the Palm House (Kew Gardens, London,
complex of glasshouses in the Royal Domain in the 1844–1848) designed by Decimus Burton and Richard
northeast of Brussels, Belgium (Figure 1). The com- Turner. Balat used the structural concept of the Palm
plete set of glasshouses takes up an area of 1.5 ha, House for his Winter Garden and applied this to a
covered with 2.5 ha of glass (Goedleven 1988). All circular ground plan. The design of the Palm House
entities are built between 1817 and 1905 in order of in its turn was inspired by the Great Conservatory
king Leopold II, who reigned the country from 1865 (1836–1840) of Joseph Paxton which was unfortu-
until his death in 1909. He is known as the king who nately demolished in 1920. This influence is clearly
fundamentally changed Belgium and Brussels in an shown in Figure 2.
architectural as well as in an urban development man- The structure of the Winter Garden consists of
ner. Despite the fact that Belgium is a small country, 36 arch trusses which are rotated around a central
he wanted it to radiate grand charisma. He used his point and form one large dome (Figure 3). The Winter
personal influence, private funds and the profits from Garden can be divided into two main parts (Figure 4).
the Congo colony to realize many of his ideas. The first part consists of the dome in the middle of the
Alphonse Balat (1818–1895) was the royal architect glasshouse, which is topped off with a small cupola and
from Leopold’s accession until his death. He was one a royal crown. The second part consists of a side aisle
of the architects who lead in the Belgian Art Nouveau around the middle dome. These two parts are separated
movement. Besides the well-known Museum for Old by a circular architrave on sandstone columns.
683
As people would intuitively feel that a cupola geom-
etry (part one as described in the previous paragraph)
produces lateral thrusts,Alphonse Balat added an outer
arch (part two) to each truss of the cupola. By doing so,
he suggests the structure being composed by arches –
structures that can withstand thrusts – rather then a
cupola. The structural surplus value of the outer arches
is still being investigated. However, it is generally felt
that this (visual) addition results in a harmonious and
aesthetically more beautiful construction.
Figure 1. Map of the Royal Glasshouses of Laeken, Brussels Figure 3. Building the Winter Garden, December 30th 1875
(Koppelkamm 1988; translated by the author). (Prentenkabinet of the Royal Library, s III 100752).
684
2 THE WINTER GARDEN TODAY repainted. The glass covering exists of many small flat
rectangular single glass panels. To reduce the heating
2.1 Maintenance of the structure costs, all these panels were replaced by coated glass
jointed with mastic to improve the insulation qualities.
When king Leopold II passed away in 1909, the own-
For the sake of the well-being of the perennial
ership of the Royal Glasshouses passed on to various
plants and trees and for the visitors’ safety, restora-
institutions. The land, purchased by Leopold II, and the
tion works could not be scheduled during winter nor
glasshouses constructed on that land, became prop-
during the two weeks of public opening in spring.
erty of the Royal Donation, a semi-private institute.
Therefore, all work was carried out in different stages,
The other glasshouses, situated on the piece of land
which considerably complicated the work, follow up
that he received to fulfil his duties as a king, are
and coordination.
owned by the Regie der Gebouwen, a governmental
institution. The borderline between these two adjacent
2.2 Pathology of the metal structure
properties can be diagonally traced through the Congo
House (Figure 1, no. 4). The shared ownership of the The metal structure of the Winter Garden is mostly
glasshouses does not facilitate restoration procedures. unchanged, compared to the original construction dat-
Since the death of Leopold II, few of the Royal ing back to 1876. Therefore, great historical value is
Glasshouses were restored. The Winter Garden merely attached to this building and any future restorations
received some refurbishments in the 1980’s, funded by have to be done with the greatest respect to the original
the Regie der Gebouwen. structure.
Some cross-sections were reinforced by welding The major threat for the metal structure is corro-
new parts on to the old ones. However, the weldability sion. By filling the joints between the glass plates
of the initial material was never investigated and there- with mastic, the ventilation of the inside air has
fore the durable efficacy of this intervention remains been dramatically reduced. This well intended inter-
questionable. Heavily corroded profiles were replaced, vention resulted in more condensation on both the
yet without recording these changes in an appropri- glass and the metal structure. Together with the very
ate report. The whole structure was sand-blasted and aggressive tropical indoor climate, this has lead to
increased profile corrosion on some vulnerable spots
(Figure 5-a).
During previous interventions, a series of profile
cross-sections were reinforced by welding new parts
on to the old ones, e.g. on the iron column between
the side aisle and the middle dome (Figure 5-b). The
tension ring at the bottom of this column was cut off
from the column, steel plates were welded on the col-
umn profiles and the tension ring was welded on the
added plates. The tension capacity of these welds is
unknown, as the weldability of the original iron was
never determined.
Some profile connections show missing rivets
(Figure 5-c). These connections have to be repaired,
so loads can be transferred properly.
Figure 4. The structural parts of the Winter Garden (Vieren- During previous restoration works, the whole struc-
deel 1902; notes by the author). ture was sand-blasted and repainted. Nowadays, the
Figure 5. a) corroded metal profiles (2006); b) welded plates on metal column (2007); c) missing rivets (2007); d) removing
the paint and the corrosion (2006).
685
Figure 6. Compound profiles of the arches (2007). Figure 8. Two-dimensional calculation model.
If this is the case, the outer arch of the aisle part will
merely carry minor loads and fulfils purely an aes-
thetic function. This hypothesis will be verified in the
two-dimensional structural model.
This two-dimensional model has its limits. In real-
ity, every load on one of the 36 arches is distributed
to the other arches via horizontal concentric rings as
well. As a consequence, all the arches work together
to bear the loads, especially asymmetric loads. In the
two-dimensional model, this effect is neglected for the
greater part. However, given the specific construction
Figure 7. Three-dimensional structure. and functioning of the metal structure of the Winter
Garden, it is indispensable to determine the stresses
Regie der Gebouwen did again some tests with special and deflections in a three-dimensional finite elements
techniques to remove the paint and corrosion (Figure model.
5-d). The techniques were evaluated on the time it took The two-dimensional calculations were performed
to rub a specific profile, the grade of efficiency of the on two models: one of the inner arch of the side aisle
rubbing technique and the nuisance for the tropical and one of the arch of the middle dome together with
plants inside the glasshouse. The most efficient tech- the outer arch of the side aisle and the column between
nique appeared to be a hammering technique where these two. In the model, the support conditions of the
needles remove paint and corrosion from the iron. structure are modelled as follows:
After cleaning the profiles, they need to be repainted
as soon as possible to avoid new corrosion. – the support at the bottom of the side aisle is fixed
(Figure 8-v);
– the support at the sandstone columns is fixed
(Figure 8-w);
3 STRUCTURAL ANALYSIS OF THE WINTER
– the connection of the outer arch of the side aisle and
GARDEN
the arch of the middle dome is fixed (Figure 8-y1 );
– the connection of the column and the arch of the
The three-dimensional structure of the Winter Garden
middle dome is hinged (Figure 8-y1 );
is quite refined. Every single member of the metal
– the connection of the inner arch of the side aisle
load-bearing structure consists of richly ornamented
and the column is hinged (Figure 8-x).
compound profiles (Figures 6 & 7).
In order to understand the basic structural behaviour The following paragraph describes how the
of the Winter Garden under symmetric load combina- three-dimensional behaviour was simulated in the
tions, a two-dimensional model was investigated. The two-dimensional model. The support of the two-
historical reference to the Palm House in Kew and dimensional arch by the horizontal concentric rings
the Great Conservatory in Chatsworth implies that the was first modelled by sliding supports. This way,
structure works as a dome with a pressure ring at the the concentric rings act as elements without vertical
top and a tension ring at the bottom of the cupola. stiffness and an infinitely large horizontal stiffness.
686
– design value of the tensile strength = 100 MPa as
the common historical value (de Bouw unpub-
lished);
– Young modulus = 210 kN/mm2 from Eurocode 3.
The two-dimensional calculations according to the
Eurocodes reveal that the stress levels, the instabilities
and the deflections of the structure do not exceed the
norm values. The maximum stress level appears at the
bottom of the inner arch of the side aisle and runs up
to 53 N/mm2 (which is 23% of 235 N/mm2 and 53%
of 100 N/mm2 ). The maximal buckling risk is 90%
of the buckling capacity of a profile under compres-
Figure 9. Side view of the Winter Garden (2007). sion and bending and occurs near the bottom of the
arch of the middle dome. The horizontal and vertical
deflections are less than 10% of the laid out deflec-
tion limits. Consequently, the buckling of the arches
The analysis of the model with these support reac-
is the most critical check of the metal structure. How-
tions showed that the above assumption was too much
ever, the global supporting effect against buckling of
of a simplification. To model the three-dimensional
the concentric rings can only be taken into account
behaviour in a more accurate way, the stiffness of the
in a three-dimensional model. At present this three-
concentric rings was determined and introduced into
dimensional model is still under development and no
the two-dimensional model by springs (Figure 8).
final conclusions can be made about the maximum
The two-dimensional model was then analysed
occurring stresses and deflections.
under symmetrical loading combinations made up of
self-weight and symmetrical snow load. The snow load
only applies to the structure that carries the glass cov-
ering, namely the arch of the middle dome and the
4 CONCLUSIONS
inner arch of the side aisle (Figure 9). The asym-
metric snow load, the wind load, the maintenance
The Winter Garden of the Royal Glasshouses of
load and temperature effects will be evaluated through
Laeken is part of the important Art Nouveau move-
three-dimensional calculations.
ment in Belgium. The current metal structure of the
The aim of the two-dimensional analysis consists of
Winter Garden is only minimally modified, compared
assessing whether the pressure ring at the top (Figure
to when it was built from 1874 to 1876. Currently,
8-z) and the tension ring at the bottom of the cupola
this masterpiece must undergo a far-reaching restora-
(Figure 8-x) can carry all the lateral thrusts. These
tion. The principal threat for the metal structure is the
findings would give an indication of how the struc-
corrosion of the compound profiles.
ture works. If these rings cannot withstand the lateral
In the first two-dimensional simplified model, the
thrust, the outer arch of the side aisle is needed for the
structure without the outer arch of the side aisle can
structure’s stability and therefore does not only have
withstand all the applied symmetrical loading combi-
an aesthetic function as assumed before. The pressure
nations (in terms of stress levels and horizontal and
ring is located at the boundary of the small cupola
vertical deflections). The second model, where the
and the middle dome. The tension ring is situated at
outer arch was added to the structure, shows that only
the bottom of the column. The other concentric rings
a minimum of stresses occur in the additional arch.
connecting to the arch of the middle dome (Figure 8-
The following conclusions regarding the structural
yi ) will all help carrying the lateral thrusts. The stress
behaviour of the Winter Garden can be drawn from the
levels in the structure without the outer arch of the
simplified two-dimensional analysis.
side aisle are investigated as well as the stresses in the
First, the outer arch acts as a backup structure for
outer arch itself when the complete model of the metal
the arch of the middle dome and for the inner arch of
structure is considered.
the side aisle. This conclusion reinforces the impres-
These stresses need to be compared with the mate-
sion that architect Alphonse Balat added the outer
rial properties of the used iron, which are currently
arch of the side aisle solely to ease the viewer’s mind.
unknown. These properties will be determined through
The viewer was not yet visually acquainted with the
tensile tests and a metal analysis in the course of the
very thin innovative metal constructions emerging at
next year. The current interpretation of the results of
that time.
the two-dimensional calculations is therefore based on
Secondly the stress levels and the horizontal and
the standard material properties:
vertical deflections do not exceed the limiting val-
– yield limit = 235 MPa from Eurocode 3; ues as prescribed in the Eurocodes. This conclusion
687
is however only valid for the applied symmetrical load Koppelkamm, S. 1988. Glasshouses and Wintergardens of
combinations. the Nineteenth Century. Berlin: Granada.
Further work implies a three-dimensional finite ele- Lauriks, L. 2007. Structurele herberekening van de Winter-
ments analysis under asymmetric load combinations to tuin van het complex van de Koninklijke Serres van Laken
in het kader van de restauratie. Brussel: unpublished
confirm the structural suitability of the Winter Garden master thesis.
with the Eurocodes as well as material testing to obtain Polet, O. & Hamerlijnck, E. 2006. De Koninklijke Serres van
the material properties of the used iron. Laken. Brussel: Polet Photo Production.
Ranieri, L. 1990. Leopold II, Zijn urbanistische visie, zijn
monumentale bouwwerken. In Balthazar, H. & Stengers, J.
REFERENCES (eds.), Dynastie en cultuur in België: 173–212. Antwer-
pen: Mercatorfonds.
de Bouw, M. Assessment of the metallic roof trusses in Vandewoude, E. 1987. De bouw van de Wintertuin te Laken,
Belgian school buildings between 1860 and 1914. Unpub- 1874–1876, een initiatief van Leopold II. In Album Carlos
lished PhD theses. Wyffels: 457–473. Brussel: Algemeen Rijksarchief.
Goedleven, E. 1988. De Koninklijke Serres van Laken. Tielt: Vierendeel, A. 1902. La construction architecturale en fonte,
Lannoo. fer et acier. Leuven: Uystpruyst.
Kohlmaier, G. & Von Sartory, B. 1991. Houses of glass: a
nineteenth-century building type. Cambridge: MIT.
688
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: Traditionally, the diagnosis of damage is largely based on the intuition and experience of the
surveyor. Despite its benefits, this approach to diagnosing can be subjective, and difficult to communicate. This
paper discusses the limitations of the current intuitive approach, illustrated with the results of a questionnaire,
and suggests a method to increase its quality and efficiency. This method proposes a systematic description of
the development of structural damage, based on a uniform vocabulary, and a syntax that distinguishes between
damage patterns, the processes leading to damage, and the factors that influence course and extent of these
processes. By using this approach, it is possible to share the knowledge that is currently embedded in intuition
and experience, and enhance the retrievability of a diagnosis. The benefits are clear: the opportunity to learn
from cases, and make this knowledge available to the profession.
689
damage. We can classify these studies into five groups, studies point out that a basic need is still to be fulfilled:
according to their main goal: the need for better communication between all par-
ties involved in a restoration project. Suggestions have
– Improvement of the data on damage; been made to improve the exchange of data by using a
– Improvement of the knowledge on damaging pro- consistent terminology and standardised survey forms
cesses; (Van Balen 2001, Kelley & Sparks 2006).
– Improvement of the analysis of information; The difficulties in information exchange mentioned
– Improvement of the evaluation of information; above are part of a larger problem: they show the
– Improvement of the exchange of information. limitations that are inherent to our traditional diag-
Many studies belong to the first group. Their aim nostic process. The development of new techniques
is to improve the completeness and quality of data on and methods for data collection, analysis and evalua-
the symptoms or the context of damage. These studies tion has certainly contributed to the improvement of
have resulted in new and better techniques for on-site diagnosis. However, less attention has been paid to the
investigation, for on-site and laboratory testing, and improvement of the diagnostic process itself and to a
for monitoring. Special attention has been paid to the more efficient, practicable distribution of knowledge.
development of a range of non-destructive tests (Binda This paper discusses the limitations of the current intu-
et al. 2000, Binda & Saisi 2002). A good overview of itive approach, and suggests a method to increase its
the techniques currently available is given in the doc- quality and efficiency.
umentation of the Onsiteformasonry project (2005).
Tragedies such as the sudden collapse of the civic 2 DIAGNOSING DAMAGE: LIMITATIONS OF
tower of Pavia in 1989, the Umbria-Marche earthquake EXPERIENCE AND INTUITION
of 1997 and the Molise earthquake of 2002 have led
to a second group of research projects. This group The current approach to diagnosing damage is, for
has aimed at extending our knowledge on specific most experts, an intuitive one. The deduction is a men-
causes of damage. Especially the creep behaviour of tal process that strongly depends on the intuition and
masonry and the effects of seismic actions on build- the experience of the investigator. These abilities are
ings have been investigated (Ignoul et al. 2006, Binda highly valued, but the subjectivity of the approach also
et al. 1992). brings about some inherent limitations.
The third group of studies has focused on the Our brains are fast processors, and they are well-
development of techniques to help the investigator trained in finding references. When diagnosing dam-
to analyse the present behaviour of a structure and age, we use this ability to evaluate and compare a new
interpret the way damage has occurred. The grow- situation with our personal experience. It helps us to
ing possibilities in computation have led to a range judge and weigh different types of information, and
of methods and strategies such as various methods for it makes us capable of readily formulating the most
numerical modelling (Lourenço 2002). For the anal- probable hypothesis.
ysis of collapse mechanisms due to seismic action, However, despite its benefits, this method has some
the macro-element approach has commonly been used disadvantages. First of all, we select our references
(Doglioni et al. 1994). One of the applications of this more or less at random. This is a quick way to retrieve
approach can be found in the digital, didactic hand- our knowledge, but it is also subjective, irreproducible,
book MEDEA (Manuale di Esercitazioni sul Danno Ed and sometimes inconsistent.
Agibilità), which gives an overview of damage patterns Then, experience can also make us biased. It could
in masonry and concrete structures that have suffered make us jump to a conclusion by paying more attention
from an earthquake (Papa & Zuccaro 2004). to facts that support our initial ideas, while we overlook
Furthermore, a fourth group has proposed meth- facts that are contrary. In this way, we would fail to
ods to facilitate the evaluation of information. One notice alternative hypotheses that could explain the
major aspect in this is the evaluation of the vulnera- type of damage as well.
bility of structures, with the aim to predict their future Moreover, our intuitive approach may keep us
behaviour and the related risks (Augusti et al. 2001, unaware of the assumptions we make during the
Speranza et al. 2006). The Multi-Hazard Assessment deduction, wittingly and unwittingly. Because com-
of Vulnerability method (MHAV (historic buildings)) municating this approach is difficult, the facts and
uses the macro-element approach to link building suppositions on which the diagnosis is based may
typologies to their characteristic vulnerabilities to nat- remain veiled to others. This makes it hard to judge
ural hazards such as earthquakes, storms, and floods. the reliability of a diagnosis, since the conditions under
The method evaluates the expected loss of both mate- which the deduction is valid are insufficiently clear.
rial and cultural significance (D’Ayala et al. 2006). To support and improve the diagnostic process, we
Finally, a fifth, smaller group of studies has focused need to understand how the process of diagnosing
on the improvement of information exchange. These works. Therefore, a questionnaire was held among
690
experts. This questionnaire contained pictures of four experts immediately referred to the classic ‘back-of-
buildings with visible damage, and for each case the an-envelope’ yield line pattern, which is the failure
experts were asked to answer the following questions: mode of a rectangular slab, supported around its
perimeter, and uniformly loaded perpendicular to its
– What could be the cause of this damage?
plane. The crack pattern, thus, gives a clear indication
– On what information do you base this hypothesis?
of the collapse mechanism. In accordance to the plate
– What data would you need to be able to verify your
bending mode, the loading must have been horizontal,
hypothesis?
out-of-plane, and directed outwards. In this way, the
– If no initial hypothesis could be established on the
symptoms of the damage led to hypotheses such as
basis of the photograph(s), what would be the next
wind load (wind suction) and horizontal impact load
step(s) in the investigation process?
(explosion), but also hindered volume changes of a
The results of this questionnaire were surprising. concrete floor incorporated into the wall.
On the basis of a first impression, the experts have The experts used the context of the damage to
found different clues, made different assumptions, and exclude some specific causes. The local character of
suggested a range of possible causes. In the next para- the damage and its distance to the ground led them to
graphs, the results of two of these cases are discussed omit the possibility of settlement as initiator. Further-
(Figs 1–5). more, some experts (correctly) assumed the building
The damage in the sidewall of building A, as shown to be located in the Netherlands, which made them
in Figure 1, forms a typical damage pattern. Most conclude that an earthquake would be a less likely
cause.
Figure 2 gives an overview of the different hypothe-
ses and anti-hypotheses that were brought forward by
the participants. It also shows the arguments and the
assumptions that the experts used to frame and support
their hypotheses. It appears that, despite the typicality
of the damage pattern, the cause of damage is not so
obvious. Therefore, in response to the third question,
experts asked for more data on the properties of the
building, on its construction history, on the nature of
the location, and on the characteristics and distribution
of the damage.
The damage in building B is less evident than the
damage in building A. It can be described as an in-
Figure 1. Building A, one of the cases in the questionnaire. plane deformation of the façade (Fig. 3). The skewing
Figure 2. Overview of hypotheses and supporting arguments brought forward for building A.
691
Figure 3. Building B: overview of façade. Figure 4. Building B: Detail of damage.
of the windows at the first floor, visible in Figure 4, is alternative hypotheses resulting from the combination
accompanied by severe crack development in the lin- of the responses shows.
tels and sills of the windows, and in the arch above the Furthermore, the results emphasize that our line of
door opening. The edges of the cracks show displace- reasoning is influenced by unconscious assumptions.
ments and demonstrate that the parts on either side of For example, most experts interpreted the damage
the fracture have undergone a clockwise rotation. in building A as caused by out-of-plane loading of
Concerning the information on the context of the a slab supported around its perimeter. Apparently,
damage, the participants particularly focused on the they assumed that the perimeter of the crack pattern
construction of the building. For example, they pointed matches with the borders of a room behind the wall.
out the difference between the timber-framed con- Yet, in reality the floor lies at the level of the horizontal
struction of the sidewall, and the stonework façade. In crack, which omits the possibility of wind suction and
addition, the relatively large openings in the façade, the explosions as a cause, and increases the likeliness of
apparent age of the building, and speculations on the hindered volume changes of the floor itself as a hypoth-
adjacent building on the left side and the open area on esis. It is clear that assumptions like these should be
the right side led them to the assumption that building mentioned explicitly when formulating a hypothesis.
B may have little or no lateral stability. This assump- Based on the results of the questionnaire, we can
tion helped them to explain the lateral displacements now make some suggestions to improve the current
in the façade. approach to diagnosing:
Figure 5 shows that, in this case, both the symp-
toms and the context of the damage were used to – Pay more attention to alternative hypotheses;
formulate hypotheses. These hypotheses vary from – Place more emphasis on explicitly mentioning the
environment-related processes such as differential set- facts and assumptions on which the diagnosis is
tlement, to overloading due to a lack of horizontal based;
restraint, and hindered volume changes due to cor- – Find ways to join and share our experience; and
rosion of the anchors. The hypothesis of differential – To obtain these goals, our knowledge needs to be
settlement appears to be supported by many arguments accessible, retrievable, and communicable.
and assumptions. Nevertheless, its root cause remains
unclear, although several options were expressed. 3 DATA ACCESS
Summarising, the results of this questionnaire
demonstrate the benefits of our current approach. Even To improve the accessibility of our knowledge on
on the basis of only some photographs, experts have damaging processes, we should examine how our
shown that their skill and experience make them able to experience-based reference system could be supported
formulate a hypothesis that could explain the damage. by a more systematic way of data cataloguing. A pos-
The use of photographs in this questionnaire did sibility is the use of knowledge-based systems. These
hinder the evaluation of damage and context. Never- systems are not a new phenomenon: They are widely
theless, this method has been chosen with precisely used in health care, and also for building pathol-
this goal: It forced the participants to be more aware ogy some systems do exist. Knowledge-based systems
of the facts on which they based their hypotheses. In consist of a framework to contain data, and a tool to
this way, the responses also show the limitations of the make these data accessible.This tool could be an index,
intuitive approach. but also a deduction function: a set of rules composing
One of these limitations is that experience can a ‘wizard’that guides a user through a list of questions,
sometimes make experts biased. This could be over- to determine which specific information is asked for,
come by joining our experience, as the range of or to find a solution to a specific problem. In building
692
Figure 5. Overview of hypotheses and supporting arguments brought forward for building B.
pathology, a knowledge-based system could contain method for analysing and diagnosing damage. MDDS
an overview of damaging processes, including clear helps to collect and order all relevant data on symptoms
selection criteria how to distinguish between them.The and context, and supports the interpretation of these
benefits of such a system are obvious: It supports data by offering background information on damage
the investigator in the framing of a hypothesis, while types and patterns, damaging processes and methods
the systematic approach makes the method unam- of testing (van Hees et al. 2005).
biguous and, therefore, communicable. Prerequisite is Until recently, MDDS mainly contained informa-
a consistent use of terminology, to make the stored tion on damage related to the interaction between
knowledge retrievable and suitable for exchange. materials and environmental factors. To fill this
An example of a knowledge-based system is the gap, a module on structural damage is now being
expert system and decision support tool MDDS (Mon- developed. In the next section, it is explained
ument Damage Diagnostic System, successor of the why a diagnostic instrument for structural damage
Masonry Damage Diagnostic System). This system should be based on a clear distinction between pro-
aims to facilitate a minimal intervention approach, cesses and factors, and on the use of a consistent
by offering a structured, transparent and consistent terminology.
693
Figure 6. Diagram of a structural failure mechanism.
694
Figure 7. Model for an overview of structural damaging processes; settlement-related processes discussed in more detail.
695
6 CONCLUSIONS and moisture. Proc. Sem. Soluble salts in the walls of
old buildings, Damages, processes and solutions, Lisbon:
Like in medicine, understanding the course of dam- 16.1–16.11.
aging processes is indispensable for finding ways to ICOMOS ISCARSAH 2005. Recommendations for the anal-
ysis, conservation and structural restoration of architec-
treat and prevent structural defects in buildings. The
tural heritage. Istanbul: Maya Basin Yayin.
results of the questionnaire have shown that, although Ignoul, S., Schueremans, L., Tack, J., Swinnen, L.,
intuition and experience allow us to readily analyse Freytons, S., Binda, L., Van Gemert, D. & Van Balen, K.
damage and formulate a hypothesis, we should be 2006. Creep behaviour of masonry structures – Failure
aware that our current approach to diagnosing can be prediction based on a rheological model and labora-
subjective and difficult to communicate. tory tests. In P.B. Lourenço et al. (eds), Proc. 5th Int.
To improve this diagnostic process, this paper has Conf. Structural Analysis of Historical Constructions,
presented a method of describing damaging processes New Delhi, India: 913–920. Delhi: Macmillan India Ltd.
in such a way that it allows for a systematic overview Kelley, S.J. & Sparks, S.P. 2006. The challenges of struc-
tural stabilization following the hurricane Katrina disaster.
of all alternative hypotheses that could be formulated
In P.B. Lourenço et al. (eds), Proc. 5th Int. Conf. Struc-
for a certain type of damage. This overview can serve tural Analysis of Historical Constructions; Possibilities
as the basis of a knowledge base. of Numerical and Experimental Techniques, New Delhi,
More work on relating damages to processes is still India: 261–268. Delhi: Macmillan India Ltd.
to be done. Important is also the formulation of infer- Lourenço, P.B. 2002. Computations on historic masonry
ence rules, which will partly be based on objectified structures. Progress in Structural Engineering and Mate-
results of the questionnaire. rials 4(3): 301–319.
The outcome is meant to serve as a basis for a Naldini, S., Vent, I.A.E. de, Hees, R.P.J. van & Binda, L.
knowledge-based system that could be beneficial to 2007. Definitie van constructieve schadepatronen; De
MDDS constructieve schadeatlas. In J.M. van der Veen
overcome and prevent bias, by supporting the formula-
(ed.), Praktijkboek Instandhouding Monumenten; Deel II-
tion of alternative hypotheses. Furthermore, it should 4 Buitenwanden (21): 1–20. Den Haag: Sdu Uitgevers. In
help to make a diagnosis more explicit, so that it is Dutch, English article in preparation.
easier to communicate. Such an expert system would Onsiteformasonry 2005. Results and research methodologies
be applicable in practice, both as a learning tool for of Onsiteformasonry; On-site investigation techniques
novice surveyors, and as a support tool for experts. for the structural evaluation of historic masonry build-
Moreover, it should help in sharing knowledge among ings. Berlin: Federal Institure for Materials Research and
all professionals. Testing (BAM). Cd-rom.
Papa, F. & Zuccaro, G. 2004. MEDEA: A multimedia
and didactic handbook for seismic damage evaluation.
In M. García-Fernández & A.B. Walker (eds), Proc.
REFERENCES
XXIX General Assembly of the European Seismological
Augusti, G., Ciampoli, M. & Giovenale, P. 2001. Seismic Commission, Potsdam, Germany.
vulnerability of monumental buildings. Structural Safety Speranza, E.,Viskovic,A. & Sepe,V. 2006. Integrated method
23(3): 253–274. for the assessment of the structural vulnerability of his-
Binda, L., Gatti, G., Mangano, G., Poggi, C. & Sacchi toric towers. In P.B. Lourenço et al. (eds), Proc. 5th
Landiani, G. 1992. The collapse of the civic tower of Int. Conf. Structural Analysis of Historical Constructions,
Pavia: a survey of the materials and structure. Masonry New Delhi, India: 651–658. Delhi: Macmillan India Ltd.
International 6(1): 11–20. Van Balen, K.E.P. 2001. Learning from damage of masonry
Binda, L., Saisi, A. & Tiraboschi, C. 2000. Investigation structures, expert systems can help! In P.B. Lourenço &
procedures for the diagnosis of historic masonries. Con- P. Roca (eds), Historical Constructions 2001, Possibilities
struction and Building Materials 14(4): 199–233. of Numerical and Experimental Techniques, Proc. 3rd Int.
Binda, L. & Saisi, A. 2002. State of the art of research on Seminar at Guimarães, Portugal, University of Minho:
historic structures in Italy. ARIADNE 11 Workshop ‘His- 15–28. Lda; Multicomp.
toric structures’ – programme of the EC 5th Framework Vent, I.A.E. de & Hobbelman, G.J. 2007a. Failure mech-
activities, Praha, Dejvice, 20-26/05/2002. anisms as a method for the assessment of structural
D’Ayala, D., Copping, A. & Wang, H. 2006. A conceptual damage. In C.A. Brebbia (ed.), Structural Studies, Repairs
model for multi-hazard assessment of the vulnerability of and Maintenance of Heritage Architecture X: 439–448.
historic buildings. In P.B. Lourenço et al. (eds), Proc. 5th Southampton: WIT Press.
Int. Conf. Structural Analysis of Historical Constructions, Vent, I.A.E. de & Hobbelman, G.J. 2007b. Determining the-
New Delhi, India: 121–140. Delhi: Macmillan India Ltd. oretical failure mechanisms for structural damage. In
Doglioni, F., Moretti, A. & Petrini, V. 1994. Le chiese e il A. Zingoni (ed.), Proc. 3rd Int. Conf. Structural Engineer-
terremoto. Trieste: Edizioni LINT. ing, Mechanics and Computation: 219–220, full version
Hees, R.P.J. van, Naldini, S. & Sanders, M. (2005). An expert on cd-rom. Rotterdam: Millpress Science Publishers.
system for analysis of damage to plasters due to salt
696
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
From geometrical and crack survey to static analysis method: The case study
of Santa Maria del Quartiere dome in Parma (Italy)
F. Ottoni
Department of Civil Engineering and Architecture, University of Parma, Parma, Italy
ABSTRACT: Aim of this work is to show the research on the structural problems of the dome of a XVII century
church, Santa Maria del Quartiere, in Parma (Italy), as an exemplar case study on numerical analysis method
connected to geometrical and cracks survey. In order to to plan the future intervention for its consolidation, a
study of the hexagonal structure was carried out by using a finite element model which reproduced its general
geometry and various cases of load and restraint, starting from the equilibrium approach for masonry domes.
1 INTRODUCTION
697
Figure 2. The laser-scan survey. Longitudinal section of the
church.
698
Figure 5. Plan of the church and of the buildings leaned
against.
699
wall, we can find other two modules of 6 braccia (1
pertica) each.
The three consecutive hexagons can be inscribed in
as many circles, which repeat the progression 18-6-6
in the plan distribution. As it is well known, numbers 6
and 12 were fundamental in the constructive system of
the XVII century, which was based on the duodecimal
system instead of the actual decimal one.
This perfect drawing is also maintained vertically:
in fact it is possible to identify the exact classical order
proportion when examining, in the vertical section of
the building, the succession of pedestal, column, and
trabeation. All the architectonic elements are linked
by a precise proportion, based on the original unit of
measurement (Fig. 6).
The dome too is inserted in this perfect design, with
its centres lying exactly 3 braccia under its impost
plane. It is evident from the drawing the presence of
a modular repetition, both in plan and in elevation.
However, it’s very interesting to note, as revealed from
the geometrical survey, that the measurement of the
dome doesn’t fit in the ideal model of perfect mea- Figure 6. Scheme showing the transversal section of the
surement ratio. In fact, the height of the dome was dome and the underlying structures. The metrology and the
found to be 20 cm less than expected, when compared identification of proportions in order to obtain a historical
to the expected measurement. This discrepancy con- monitoring system.
cerning the height of the dome could further support
the collapse mechanism hypothesis.
700
and architects, following the observation of the Vatican
dome. (Fig. 7) In keeping with this, it’s interesting to
read the debate that took place in 1743 among the three
mathematicians Jacquier, Le Seur and Boscovitch, and
also the Poleni analysis untitled “Sulle colonne fesse”,
in 1743 and 1748). Starting from these observations on
the functioning of the domes, it becomes easily under-
standable the typical crack pattern of these structures,
which highlights meridian cracks.
701
Figure 9. Visualization of the FEM results on the intact
model of the dome, Smax principals. The tensile stresses are
evident in the corners, where, in fact, the real cracks have Figure 10. Visualization of the FEM results on the cracked
been actually developed. model of the dome, Smax principals. The tensile stresses are
very high in the terminal part of the cracks, where it has been
installed a recent automatic monitoring system.
702
whole height of the dome, we can observe very high
tensile stresses on the limits of the cracks. Now, these
values are controlled by a recent installed precision
monitoring system, which would give information on
the trend of the cracks, also considering the seasonal
variations.
Obviously, the presence of the cracks has gradu-
ally modified the structural behaviour of the dome.
The structure has changed from a substantial axial-
symmetric functioning (in which each parallel slice
gives a contribution to the meridian ones) to that of 6
different parts (6 panels) which are linked at the top
and which are supported by the common abutment.
The cracked model (Case 3) simulates the real
behaviour of the structure, even if it uses a linear-
elastic constitutive law for the finite elements. In fact,
the insertion of the real cracks of the dome in the Figure 11. Visualization of the FEM results on the cracked
model allows to not consider the masonry extremely model of the dome, with the insertion of the encircling tie.
low resistance to tension and shear strain: in this way Here are highlighted the axial force on the tie.
the non-linearity is concentrated in the cracks.
It’s interesting to note that the load due to the pointed
weight of the roof, applied to the cracked model of
Then, the thrust force value, which has been cal-
the dome (case 4) doesn’t provoke significant varia-
culated both by empiric-experimental method and
tions in the horizontal global thrust. This result further
by finite element analysis, has been used for the
supports the collapse mechanism hypothesis, which
dimensioning of the steel encircling tie, which has to
pointed the global thrust of the dome, due to its self
be put in work on the dome.
weight, as the main responsible of its damage.
The area of the tie section could be easily calculated
by the equation At = H/σ adm.
(where A is the area of the tie, H is the thrust force
3.5 On the linear model choice of the dome and σ adm is the still admissible tension).
The section of the tie has to be the lowest, to reduce
After these analysis, we can say that the elastic linear
the reaction of the steel to temperature variations.
model, with the empiric insertion of the cracks, could
Therefore, in order to determine the better solution
represent a good approximation of the reality, of easy
for the consolidation of the structure, the experimen-
application and control. The presence, in some models,
tation on this type of intervention suggests to compare
of tensile stresses, has evidenced the existing discrep-
the performances of different materials.
ancy, but it has also allowed to identify the fracture
In keeping whit this, we hypothesize to use a
zones.
tie constituted by an high resistance steel, e.g. type
In fact, on valuable historical buildings, it’s useful
DYWIDAG (which has a σ adm = 6500 kg/cm2). At
to know, not only the limit collapse last resistance, but
the end of the calculation we have found that the use
even the rise of the actual damage state level. Indeed, a
of this type of steel allows a significant reduction of the
linear model permits to simply simulate the behaviour
tie diameter: from 2,6 cm to 2,15 cm large. This could
of the structural organism and then to clearly identify
assure a lower dilatation during the seasonal variations.
the risk zones.
Then, in the last case analyzed by finite element
method (Case 5) we have determined the effects of an
encircling tie on the dome. In order to gain the better
4 THE EXPERIMENTAL INTERVENTION results for consolidation intervention, different type of
materials have been simulated, with different diameter
4.1 The steel encircling dimensions.
As stated before, the historical debate about theVatican Precisely, an encircling intervention has been sim-
dome cracks (during the XVII and XVIII centuries) ulated, through the insertion of a tie on the line of the
has for good clarified that a dome exercises a radial windows in the dome. The numerical results which
thrust force which has to be nullified. we have obtained stress the different reactions to
Thanks to the symmetry of the structure, the solu- the temperature variations and the different restraint
tion of this constitutive problem can be solved by conditions.
the insertion of a constraint element, which would It is in course of study an hypothesis of using fiber
neutralize this force: a tie. reinforced material, instead of the projected steel tie,
703
for the encircling of the dome. Some model have been REFERENCES
produced which are waiting the comparison with the
experimental results. Benvenuto, E. 1981. The historical developments of science
of constructions, Sansoni (ed), The Sahara and the Nile:
21–35. Rome: Italy.
Blasi, C. Foraboschi, P. 1989. The masonry arch: a finite ele-
5 CONCLUSIONS ment approach by no-tension-friction-elements and check
of the method. In Proc. AIMETA, IV Italian Conference
This analysis by numerical method has been organized on Computational Mechanics, Padova: Italy.
in a route from linear and quite simple model, build Castigliano, C.A.P. 1879. Theorie de l’equilibre des systemes
with a geometrical simplification of the hexagonal elastiques et ses applications,A. F. Negro Publisher, Turin,
dome, up to definition of a more precise simulation. Italy.
The used linear model describes masonry through a Chiarugi, A., Fanelli, M., Giuseppetti G. 1983. Analysis of
linear elastic constitutive equation, which assumes the a Brunelleschi-Type dome including thermal loads, Proc.
IABSE Symposium, Strenghtening of building structures
material’s lock of resistance to tension and its limited diagnosis and therapy, Venice, Italy, 169–178.
compressive strength. Chiarugi, A., Fanelli, M., Giuseppetti G. 1993. Diagno-
Thanks to a fusion between historical analysis, sis and Strenghtening of the Brunelleschi dome, Proc.
precision geometrical surveys, experimental investi- IABSE Symposium, Strenghtening of building structures
gations and numerical modelling, the static behaviour diagnosis and therapy, Rome, Italy, 441–448.
of the monument has been determined, in order to Couplet, P. 1729. De la pousse des voutes, Historie de
plan future intervention for its consolidation and l’Academie Royale des Sciences, Paris, France.
conservation. Di Pasquale, S. 1984, Masonry solids static: theory and expe-
riences, In Proceedings of the Construction Department,
Florence University, News year VIII, n.5, Florence, Italy.
Di Pasquale, S. 1996, Art of construction between knowledge
ACKNOWLEDGMENTS and science , Essays Polis – Marsilio (ed), Venice, Italy.
Heyman, J. 1966. The stone skeleton Int J. Solid structures,
2, 249–279.
This research has been supported by a grant from Heyman, J. 1980. The safety of masonry arches, Proc. Inst.
the Foundation Cassa di Risparmio di Parma e Pia- Civ. Eng. University of Cambridge, Cambridge, England.
cenza (Pres. Prof. C.Gabbi) to the Department of Mery, E. 1840. Sur l’equilibre des voutes en herceu, Annales
Civil Engineering and Architecture of the University de Ponts et Chaussess, Paris, France.
of Parma (Research Group Resp. Prof. Ivo Iori and Poleni, G. 1748. Memorie istoriche della Gran Cupola del
Prof. C. Blasi). Tempio Vaticano. Padova, Italy.
704
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
P. Morer
Department of Mechanic, University of Navarra, Tecnun, San Sebastián, Spain
R. Goñi
Department of Structural Analysis and Design, University of Navarra, Pamplona, Spain
ABSTRACT: This paper presents a benchmarking study of two methodologies used in the analysis of non-
reinforced structures: numerical and graphical methods applied to the structural behavior of domes. The structural
analysis was performed through the Finite Element Analysis software ABAQUS. The graphical methods are based
on the theory of limit analysis and they were carried out through CAD software tools Pro-Engineer and AutoCAD.
In order to test both methodologies, they were applied to the structural behavior of the Pantheon in Rome. The
results obtained through the numerical methods were compared with Mark & Hutchinson’s results. The graphical
analysis seeks to obtain and visualize the line of thrust and the Stability Factor. Results obtained were compared
with Lancaster’s results. The aim of the paper is to validate and to compare both methods.
705
and the benefit of utilizing each one under different travertine plus-tuff to brick plus-tuff, alternating lay-
circumstances. ers of pieces of tuff and broken tiles or bricks also
in the same mortar. The uppermost level (840 cm) of
the wall consists of concrete, predominantly of bro-
2 SHORT DESCRIPTION OF THE PANTHEON ken bricks in mortar. The stepped rings are composed
of layers of brick fragments set in mortar. The unit
2.1 History weight of the mass of concrete is calculated to be about
1,600 kg/cm2 . The tufo giallo and lightweight volcanic
The first Pantheon was built byAgrippa in 25–27 BD. It
scoria is used for the top of the dome, the unit weight
suffered severe damage in 80 AD during the great fire
being 1,350 kg/cm2 .
of Rome. Although it was rebuilt by Emperor Domi-
tian, it was struck by lighting in 110 AD, during the
reign of Trajan. The present Pantheon was built for the 2.4 Cracking
third time under Emperor Hadrian between 118–125
AD. That date is known by stamps on the bricks. The dome and walls are cracked. Terenzio, during
his inspection of the Pantheon in 1930, documented
cracking in the wall and the dome. The cracks that he
2.2 Geometry and structure of the Rotunda documented continue up the dome to an average of
about 57◦ above the springing, according to Mark and
The Rotunda is a circular space covered by a concrete Hutchinson. Cowan theoretically placed this point at
hemisphere. The internal height of the circular wall is 37◦ degrees from the axis.
equal to the radius of the sphere of the dome. The dome
is held up by a 6-meter thick wall. The wall is made of
concrete and is covered with about 60 cm of brick. The 3 STRUCTURAL ANALYSIS METHODS:
wall is not solid; it contains cavities, chambers and is STRUCTURAL BEHAVIOR OF THE DOME
open towards the inside with large exedras arranged
on three levels (Lancaster 2005). A dome is one of the most difficult architectural forms
Inside the wall there is an archway of bricks, known to construct. It is designed by rotating an arch 360◦
as a relieving arch, to support the upper wall over the around its central vertical axis. It is a vault with double
openings. The relieving arch is a semicircle of thin curvature, capable of covering a large space without
bricks standing radially on end extending into the con- any interior supports. They have great structural influ-
crete wall. This arch distributes upper loads to the piers ence; the double curvature improves the behavior of
during the long time it takes for the concrete to cure, the compression-curvature forces. The study of their
but after curing, it becomes an integral part of the wall. structural behavior is based on knowledge of material
This archway of bricks was only part of the wall and capacities, structural properties and construction tech-
did not extend into the dome. The entire structure is niques. In ancient times, the construction was based
resolved with arches incorporated in the wall. Exter- on geometrical rules. The relation between geometry
nally, the lower part of the dome is hidden by the upper and structural behavior was decisive. The most impor-
storey of the cylinder, while the visible part is raised in tant thing was to ensure the stability and safety of the
seven stepped rings. The first ring is 2.25 m thick. Six structure.
other exterior step-rings present a slope 1:10 inwards. The dome studied in this article is an unreinforced
The last ring starts at 47◦ from the axis. concrete structure. For the equilibrium analysis, it is
assumed that it is a non-tension material, has virtually
infinite compressive strength and sliding failure will
2.3 Materials not occur (Heyman 1995). The only load applied is
The composition of The Rotunda has been documented its own weight. This study was carried out using two
by de Fine Licht and MacDonald. The rotunda rests different methods: graphical and numerical.
on a ring foundation made of opus caementicium,
7.30 m wide and 4.50 m deep. The concrete is made
3.1 Graphical method
of travertine fragments in layers in a mortar of lime
and pozzolana. The unit weight of the mass of con- In this article this is based on Wolfe’s method for
crete is calculated to be about 2,200 kg/cm2 . The lower dome analysis (Wolfe 1921). He developed a graphi-
level (1,310 cm) consists of alternate layers of traver- cal method based on membrane theory. He analyzed
tine fragments and fragments of tuff (the caementae) in the dome as a radial series of segments, subdivided
a mortar of lime and pozzolana. The core is faced with into voussoirs. The weight of each voussoir increases
a thickness of about 60 cm of brick, the unit weight from crown to base and rests on the center of grav-
being 1,750 kg/cm2 . The concrete of the upper level ity cut by the axis plane. The force polygon allows
(890 cm) of the wall is lightened by changing from one to predict meridional and hoop forces. A recent
706
forces allowed ring by ring construction of the dome.
For a dome of zero thickness, the membrane theory
predicts that the hoop forces change at 51◦ 49 from
compressive forces to tensile forces (Heyman 1995).
A typical failure consists of the formation of radial
cracks along its meridians that divide the dome into
segments, due to hoop forces.
707
Table 1. Internal hoop and meridional stress, at the base of
the dome (kg/cm2 ).
Model 1 Model 2
Heavy density Light density
Hoop Meridional Hoop Meridional
forces forces forces forces
∗ ∗
Mark & H 4.9 3.3
Dome (FEA) 4.13 −3.7 2.2 −2.9
Dome (GA) 4.37 −4.6 2.68 −2.98
∗
these values were not provided by Mark & Hutchinson.
2
Figure 2. Model 2: maximum tensile stress (kg/cm ).
708
Figure 4. Models 3–6 analyzed with Abaqus.
4.2.2 Study carried out with graphical analysis Mark and Hutchinson studied the cracked dome
Models 4 and 6 were analyzed with the graphical anal- and modified their model by changing the boundary
ysis method. The profile of the dome has changed. In conditions. The model studied in this article has been
Model 4 the dome begins where the uppermost level created by dividing the base of the dome into twenty-
ends (Figure 5a). The embrace angle is approximately four meridians. In this way, the internal hoop forces
70◦ , taken from axis. are eliminated.
In Model 6 the weight of each stepped ring has been
4.3.1 Study carried out with Abaqus
considered and the profile of the dome is taken, as
Three models have been elaborated. In Model 7 the
shown in Figure 5b.
dome has been divided into twenty-four segments,
The values obtained with the Graphical Analysis
with a length similar to that in Terenzio’s study.
method were reached with the steps indicated in 4.2
This model is proposed with the sole purpose of
and are taken at the center of gravity of each voussoir.
showing the cracked dome. The lower part of the
dome is divided into twenty-four slides (each fifteen
4.3 Models 7–9: cracked model degrees). Figure 6 shows the dome deformed under its
own weight.
The models with cracks completely change the behav- In Model 8 the dome has a pier at the uppermost
ior of the dome. The internal hoop forces disappear. level, with light concrete. In Model 9 the stepped-rings
The dome behaves like an array of arches. are added to Model 8.
709
Figure 6. Study of the cracked dome with Abaqus.
Mark & H * ** * ** * **
FEA (base) * −16.7 * −9.4 * −2.6
FEA (1) * −4.85 * −4.25 * −1.6
FEA (2) * −3 * −3.84 * −1.58 Figure 7. Thrust line for profile Model 9 and Model 8. A
FEA (3) * −1.15 * −1.24 * −1.52 comparison.
*Hoop stress does not exist in this part of the dome.
**M & H do not give values. factors of SF8 = 2.9 and SF9 = 3.9 were obtained for
HF = Hoop Forces, MF = Meridional Forces. model 8 and Model 9, respectively.
In order to know the influence of different densi-
ties used in the construction of the Pantheon, three
The results are shown in table 3. The values are variables were taken for Model 9 in function of the
taken in three points (interior surface of the dome) unit weight: a) the dome at 1,350 and 1,600 kg/m3 , b)
coinciding with the rings and at the base of the dome. the dome at 1,350 kg/m3 (lighter) and c) the dome at
1,600 kg/m3 (heavier). The stability factors obtained
4.3.2 Study carried out with graphical analysis are: SF9 a = 3.9, SF9 b = 3.1 and SF9 c = 2.9. Those
The stability or equilibrium approach is the most values are compared with the values obtained by
important concept to assess the safety of these struc- Lancaster and are reflected in table 4.
tures. Equilibrium can be visualized using a line of
thrust, the theoretical line or inverted catenary, which 4.3.3 Thrust line in a geometrical context
represents the path of the resultants of the compressive We can propose another question. What is the max-
forces throughout the structure. The forces’ line passes imum height allowed for the Rotunda? What is the
through the pier up to the ground-line. At this point the relation between geometry and the thrust line? Figure
Stability Factor (SF), or Rankine factor, is obtained 8 shows different stability values for model 9 in func-
(SF = b/2c, b being the pier weight). This factor was tion of the height of the wall. That factor is taken at
determined by W. J. Rankine in the mid-nineteenth the base of the pier and takes a value 1, indicating the
century. If that factor takes a value 1, it indicates the collapse point of the pier. The actual height is a diam-
collapse point of the pier: a value 3 is considered safe. eter of one sphere (73.75 Roman Feet) inscribed in its
That study was conducted with a Graphical Analysis interior. If the high value increases 82.35 Roman Feet
method. the SF is 1 and the thrust line passes out of the pier.
Various suppositions have been made. Firstly, two If the high increases 27 Roman Feet the SF value is 2,
thrust lines were found for both models. where stability starts to be critical.
Figures 7 shows the two thrust lines found for each The thrust line is tangent to the extrados at point
supposition of the dome profile of the Pantheon. The A. That point coincides approximately with the ver-
weight of the uppermost level is 13.10 N. Stability tex of the so-called Diophantine triangle. Point A is
710
Table 4. Comparative graph between Lancaster and
Morer & Goñi
711
5 DISCUSSION structure is analyzed with the graphical method, the
objective will be to find a thrust line contained inside
5.1 Results obtained with Abaqus the thickness of the arch. At this moment we can say
that this structure will work properly. The different
On the one hand, the results found here with numeri-
thrust lines obtained show how the use of different
cal methods correspond to the same order of the values
densities benefits structural behavior. The whole dome
gathered by Mark and Hutchinson, with minor devi-
at 1,350 kg/m3 gives a stability factor lower than if
ations. The values found are fairly consistent, with
the dome were at 1,350 and the lower part at 1,600,
minor deviations that may be due to a large extent to
improving it by 20%.
differences in the mesh and type of elements used.
The stability factor obtained for different models
Mark and Hutchinson do not point out where the
varies from 2.9 to 3.9. If the dome had been built
maximum values given were measured. Those results
without the third level, the uppermost level, it would
validate our method.
have been outside the limits of stability (Model 3,
On the other hand, those values have served as a
SF = 1.38), with the uppermost level Model 8 improv-
base for comparing results obtained with graphical
ing and doubling this value, resulting in a valid thrust
analyses. Then the results obtained in first Models
line. This value increases to 3.9 with stepped-rings and
1–4 allow validation of the methods utilized and at the
can be considered safe and stable. The rings are appar-
same time, give some information about the influence
ently used for two purposes: for constructive reasons
of different unit weights of the materials. The use of
and for improving the structural behavior of the dome,
lighter-weight concrete towards the crown of the dome
but not as a buttress.
improves the structural behavior, reducing the internal
We assume that the dome is acting in a minimum
hoop and meridional forces.
thrust state, as reflected by the cracks. These cracks are
Models 5–6 show the influence of the stepped-rings
irregular and narrower in shape and recorded as being
and their different densities.
“in situ” by Alberto Terenzio in 1930. The lower part or
According to MacDonald, “the rings add to the load
base of the dome is built similarly to The Rotunda, with
over the critical or launch portion of the great vault and
an archway of bricks covered by concrete. That zone
function as buttresses, helping to bring the structure
does not act as a cupola. The thrust line is tangent to
into stability through compression.” After that study,
the extrados at approximately 51.7◦ and in the intrados
we can observe that the rings have been favorable for
at 17◦ . The tangent points on the intrados coincide
increasing the compression value, but the dome was
with the archway. The crowns of the arches are set at
working under compression before the step-rings came
great tensile stress. The arches unload the stress on
into play. Their function as buttresses is not clear. It is
eight piers. It can be observed that the cracks coincide
likely that the rings had a constructive function at the
noticeably with the construction if the arches.
moment when the concrete is poured and spread.
The last model evaluated, Model 10, shows how
the interior arches work with greater tensile stress in
the central voussoir. The hoop forces reach a value 6 CONCLUSIONS
of 9.4 kg/cm2 and the meridional forces a value of
−0.1 kg/cm2 . Those values indicate where and why This paper presents a comparison between two meth-
the dome is cracked. ods of structural analysis. Those methods have been
applied to ten different models of The Pantheon in
Rome.
5.2 Results obtained with graphical analysis The results obtained have been compared to Mark’s
The utilization of both methods allows a comparison results. The values gathered by Mark referred only
of the information obtained. The graphical analysis to hoop forces. The deviations obtained were small.
utilized is a simple, intuitive, fast and easy method to These studies confirm the numerical method utilized
implement. The three most important structural cri- and offer additional information about the hoop and
teria for a masonry structure are: resistance, rigidity meridional forces, specifically the values of the merid-
and stability. In the case of the Pantheon, the first ional forces and the points of the dome where the
two criteria are irrelevant because the deformations are values were taken. Models have been elaborated with
very small. The third criterion, stability, is more rele- greater precision than before.
vant. The numerical method gives information about The use of different densities in the dome improves
the first two criteria, but the graphical method gives the stability factor and reduces the values of both hoop
information about the third criterion. and meridional forces. The stability factor increases
Those structures work under their own weight by 20%. The studies performed with graphical anal-
rather than under traction. Non-reinforced concrete yses confirm the results obtained and give additional
works only under compression. They are hyper-static information about the stability of The Rotunda and the
structures where numerous solutions exist. When the angle where the hoop forces change from compression
712
to tension. It is confirmed that the uppermost level and Heyman, J. 1977, Equilibrium of Shell Structures, Oxford:
stepped-rings improve the stability factor. Oxford University Press.
The cracks defined by Terenzio coincide with the Heyman, J. 1998, “Hooke’s cubico-parabolical conoid,” Pro-
position of the archway and with the points where the ceedings Royal Society, London. pp. 39–50.
Heyman, J. The Stone Skeleton, Cambridge: Cambridge
thrust line is tangent at the intrados and extrados of University Press. 1995.
the cupola. There is a close relation between the geom- Heyman, J., “1988, Poleni’s Problem,” Thomas Telford Jour-
etry inherent toThe Pantheon, the thrust line calculated nals, Vol. 84, Issue 4, pp. 737–759.
for this geometry and the material used. Huerta S., 2006, “Galileo was Wrong: The Geometrical
Lastly, it is concluded that the numeric model is Design of Masonry Arches,” Nexus Network Journal, Vol.
more complete and informative, but still is a theo- 8, Num. 2, pp. 25–51.
retical model and is based on numerous assumptions Lancaster, Lynne C., 2005, “Concrete vaulted construction
which can never be known. When the mesh is fine in Imperial Rome,” Published Cambridge, UK; NewYork,
and the elements have many nodes, it requires power- ISBN 0521842026.
Livesley, R.K. 1992, “A computational model for the limit
ful computers to post-process the information given. analysis of three-dimensional masonry structures,” Mec-
This method offers values for stress and deforma- canica, Vol. 27, Issue 3, pp. 161–172.
tions, but does not give information about stability. Mark, R., and Hutchinson, P., 1986, “On the Structure of
Nevertheless, the graphical model offers information the Roman Pantheon.” The Art Bulletin, Vol. 68, No. 1,
about the stability of the dome and the position of pp. 24–34.
cracks. Cracks can be simulated with numerical analy- Moore, D. 1995 “The Roman Pantheon: The Triumph of
sis, but require models of materials, information about Concrete,” Mangilao, Guam, University of Guam Station.
material behavior, long procedures and non-linear O’Dwyer, 1999, “Funicular analysis of masonry vaults”,
analysis. Computers and Structures 73, pp. 187–197.
Ochendorf, J., 2006, “Real-time limit analysis of vaulted
masonry buildings,” Computers & Structures, Vol. 84,
Issues 29–30, pp. 1841–1852.
REFERENCES Oppenheim, I.J., 1992, “The masonry arch as a four-link
mechanism under base Motion” Earthquake Engineering
Block, P., Ciblac, T., and Ochsendorf, J. 2006, “Real-time & Structural Dynamics, Vol. 21, Issue 11 , pp. 1005–1017.
limit analysis of vaulted masonry buildings,” Computers Terenzio, A., 1934, “La restauration du Pantheon de Rome,”
and Structures,Volume 84, Issues 29–30, pp. 1,841–1,852. La conservation des monuments d’art & d’historie, Paris.
Boothby,T.E., 2001, “Analysis of masonry arches and vaults.” Wanda, L., 2006. “Equilibrium Analysis of Masonry,” PhD.
Prog. Struct. Engng. Mater. Vol. 3, pp. 246–256. thesis.
Como, M., 1992, “Equilibrium and collapse analysis of William Lloyd, McD. 1976, “The Pantheon: design, meaning
masonry bodies.” Meccanica, Vol. 23, No. 3, pp. 185–194. and progeny.” Published Cambridge. Harvard University
Cowan, H.J., 1977, “A History of Masonry and Concrete Press, ISBN 0-674-65345-9.
Domes in Building Construction.” Building and Environ- Wolfe, W.S., 1921, “Graphical Analysis: a text book
ment, Vol. 12, pp. 1–24. Pergamon, Press Printed in Great on graphic statistics,” New York. McGraw-Hill. Book.
Britain. pp. 250–253.
de Fine Licht, K., 1968, The Rotunda in Rome: A Study of
Hadrian’s Pantheon, Copenhagen.
713
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
E. Cosenza
Department of Structural Engineering, University of Naples Federico II, Naples, Italy
ABSTRACT: Preservation of historical structures often includes evaluation of timber roof trusses. According to
recent research achievements and European building design codes, connections can play a role in the response of
timber structures, but have paramount relevance for assessment of historical constructions under serviceability
and ultimate loading conditions. Whenever semi-rigid behaviour of connections is concerned, investigation
within rotational properties of connections through refined numerical modelling is required. The objective of the
present paper is to study the response under service loads of typical connections of ancient timber structures, by
using finite element-based modelling. The model is calibrated against experimental and numerical results found
in relevant literature and sensitivity analyses are carried out. An application to different connections of the roof
trusses of the Royal Palace in Caserta is presented.
Finite Element analysis seems to be able to simulate semi-rigid behaviour of joints within the elastic range.
Numerical moment-rotation diagrams are calculated for the different connections found on the reference roof
structures as a basic step for the global structural assessment under serviceability loading conditions.
715
Figure 1a. View of the Royal Palace in Caserta, Italy.
716
Figure 4. Connection between king post and main rafters. Figure 6. Connection between the rafters and the main
chord.
717
cross section ratios of the elements averaged the value Conventionally, transverse force, moment and rota-
B/H = 0.50. tions have been assumed positive as the skew angle
At present time, timber trusses are characterised was reduced. The main chord of the joint was con-
by different types and stages of degradation basically strained in the vertical displacements at the base, and
due to environmental factors, lack of maintenance, in the horizontal displacements at the vertical right-
and effects of repeated interventions. Visual inspec- side border. A structured mesh is used for the rafter
tions revealed confined damaged areas due to frequent and the chord, whereas an irregular transition mesh is
exposure to rapid changes in moisture content, or used in the vicinity of the connection between rafter
to fungal attack. However, in any case the state of and chord.
conservation lead to a loss of serviceability of the It is worth noting that wood exhibits anisotropic
whole roof structure (Ceraldi & Russo Ermolli, 2006). elastic and inelastic behaviour, and the characteris-
These problems were already detected in the past, tic inner structure which include several defects (e.g.
and different interventions were performed. In fact, knots, slope grain). The use of refined non linear
besides the partial replacement of the covering sys- orthotropic criteria is essential for detailed numerical
tem (including the purlins), the trusses were subjected simulations of timber joint, since different strengths
to repeated replacement of members and/or part of and softening/hardening characteristics in orthogonal
members. In particular, the original heels, rafters, king directions can be accounted (Lourenço et al. 2007,
post and truss supports were frequently replaced with 1997).
new species of timber elements, which exihibited a For the purposes of the paper, an elastic orthotropic
mechanical properties degradation due to the absence model is accounted for wood elements (Parisi & Piazza
of specific protection from moisture and biological 2000; Bodig & Jayne 1982). Strengths and elastic
attacks. This circumstances lead to a variability of stiff- materials properties have been derived from available
ness and strength properties of wood truss members, tests given in Parisi & Piazza (2000) and represent
which influence the structural response at global level. average values.
In order to properly fit elastic properties for the
wood members, which are characterized by different
2 MODELING OF CONNECTIONS
grain directions, a local coordinate system was intro-
duced. The 1-axis and 2-axis are related to the direction
2.1 Calibration of the finite element model
parallel and perpendicular to the grain, respectively;
In the following, results of finite element (FE) anal- the 3-axis is orthogonal to the 1–2 plane.
yses performed on a typical birdsmouth connection Therefore, the elastic parameters included the elas-
of Chestnut wood specie, are presented. In partic- tic moduli E1 = 9200 MPa and E2 = 310 MPa par-
ular, calibration of FE model is performed on the allel and orthogonal to the grain, respectively; the
base of available data which can be found in the shear modulus G12 = 580 MPa and the Poisson’s ratio
relevant literature (Parisi et al. 1997, and Parisi & v12 = 0.4. Nonlinearities were concentrated at the
Piazza 1995, 1998, 2000). In this work, the connec- contact interfaces.
tion behaviour is simulated with the use of the FE The rafter and the tie were connected by point
software package STRAUS7 (1999), and results are contact elements which allowed relative tangential
investigated in terms of both force-displacement and displacements of facing surfaces and limited normal
moment-rotation diagrams. displacements. In particular, the Normal type contact
The geometric dimensions of the both main rafter implemented in STRAUS7 was used, which provides
and chord were: length of the elements L = 1.90 m; ele- stiffness in compression and not in tension. In addition,
ment cross section 190 × 225 mm and 200 × 200 mm, a friction coefficient was introduced in order to con-
skew angle equal to 30◦ ; rafter cross section notch trol the amount of lateral force that can be transferred
depth tv = 35 mm and notch length lv = 200 mm. thorough the contact surface.
Two loading pattern were considered. The first one Let α the angle between the horizontal direction and
consisted of a vertical pressure uniformly distributed the normal of the contact, the axial elastic stiffness of
at the free end of the rafter; subsequently a trans- the contact elements has been estimated as EαA/L,
verse force was applied perpendicular to the axis of where A is the area of influence of each point con-
the rafter. The second one consisted of an initial axial tact, L is the distance between the rafter and the chord
pressure uniformly distributed over the cross section (1 mm) and Eα is the elastic modulus at the angle α.
of the rafter; from this, a linear pressure distribution The modulus Eα is calculated for each contact by tak-
over the cross section of the element was applied in ing into account the elastic moduli E1 and E2 of the
order to simulate a bending moment at the end of the timber elements as follows (Lekhnitsii 1968):
rafter. With reference to the axial compression stress
in the rafter, two values were accounted: fwr equal to
1.0 MPa and 1.5 MPa, respectively. These stress were
kept constant during the analysis.
718
0 0.6 1.2 1.8 2.4 3
0
Force [kN]
1
Experimental curve
2 Parisi & Piazza (2000)
3 STRAUS7
7
Displacement [mm]
8
719
0 0.5 1 1.5 2 2.5 3 3.5 0 0.8 1.6 2.4 3.2 4 4.8 5.6
0 0
Moment [kNm] Moment [kNm]
0.0005 0.0005
0.001 0.001
µ=0.3
0.0015 0.0015 µ=0.4
Figure 11. Comparisons of moment vs rotation curves Figure 13. Results of sensitivity analysis for µ = 0.3 and
for joint members 200 × 200 mm for fwr = 1.0 MPa and µ = 0.4 for joint members 200 × 200 mm – fwr = 1.5 MPa.
fwr = 1.5 MPa.
an increase of the slope of the first part of the response
of about +10% (Figure 13).
720
0 0.5 1 1.5 2 2.5 3 3.5
0
Moment [kNm]
0.00025
0.0005
721
elastic phase, as demonstrated by deformed mesh in the Joints. Two Cases Study.Proc. ICOMOS IWC – XVI
Figure 15 (a). International Symposium Material to Structure, Mechan-
Contours of compressive stress in the 1-axis direc- ical Behaviour and Failures of the Timber Structures,
tion are also plotted up to the elastic limit value of Florence, Venice and Vicenza, Italy.
Emy, A.R. 1842. Traité de l’Art de la Charpenterie. Liège.
rotation for each connection. It is easy to recognise a Avanze & Comp (eds.)
concentration of stresses at the right end of the carving Lourenço, P.B., Feio A. & Machado, J.S. 2007. Chestnut
in the chord. wood in compression perpendicular to the grain: Non-
destructive correlations for new and old wood. Construc-
3 CONCLUSIONS tion and Building Materials, 21(8), 1617–1627.
Gelfi, P., Giuliani, E. & Marini, A. 1998. Comportamento
della connessione a piolo nelle travi miste in legno e
The present paper dealt with the structural response calcestruzzo: modellazione teorica e confronti sperimen-
of historical wooden trusses present in historical con- tali. Proc. III Workshop Italiano sulle Strutture Composte,
structions. In particular, stiffness and moment-rotation Ancona, Italy. (in Italian)
performances of connections surveyed on the roof of Giordano, G. 1999. Tecnica delle costruzioni in legno. Hoepli
the Royal Palace in Caserta, Italy have been studied by Ed., Milano. (in italian)
FEM numerical analyses. An advanced, but common, Izenour, G.C. 1992. Roofed Theaters of Classical Antiquity.
numerical tool has been used to explore the capabil- Yale University Press, New Haven.
ities of general purposes program for assessment of Lekhnitskii, S.G. 1968. Anisotropic Plates. Gordon and
Breach (eds.), New York (S.W. Tsai, T. Cheron, Transl.).
historical trusses under serviceability loads.
Lourenço, P. B., De Borst, R. & Rots, J. G. 1997.A plane stress
Model has been preliminarily calibrated against softening plasticity model for orthotropic materials. Inter-
numerical and experimental data available in techni- national Journal of Numerical Methods in Engineering,
cal literature. Satisfactory results have been found, so 40, 4033–4057.
that relevant information for real application have been Mésange, M. 1754.Traité de Charpenterie et de bois de toutes
derived for a variety of connections found in Caserta espéces, 2nd vol., Paris.
Royal Palace. Parisi, M.A. & Piazza, M. 1995. Strutture lignee tradizionali
Obviously, results are not comprehensive, but repre- di copertura: modellazione e comportamento sismico. Atti
sent an initial and crucial step for the global structural 7◦ Convegno Nazionale ANIDIS, Siena, Italy.
Parisi, M.A., Piazza, M. & Modena, R. 1997. Model-
assessment of roof trusses using advanced non linear
lazione e sperimentazione di giunzioni lignee di coperture
numerical tools. tradizionali in zona sismica. Atti 8◦ Convegno Nazionale
ANIDIS, Taormina, Italy,
REFERENCES Parisi, M. A. & Piazza, M. 1998. Seismic Behavior and mod-
eling of traditional timber roof structures. Proc. 11th Euro-
Belidor, B.F. 1739. La science des Ingènieurs dans la con- pean Conference on Earthquake Engineering, Rotterdam:
dite des travaux de fortification et d’architecture civile. Balkema.
Charles-Antoine Jombert (ed.), Paris. Parisi, M. A. & Piazza, M. 2000. Mechanics of plain and
Blondel, J.F. 1771. Cours d’Architecture civile (6 volumes), retrofitted traditional timber connections. ASCE Journal
Desaint (ed.), Paris. of Structural Engineering, 126(12), 1395–1403.
Branco, J., Cruz, P., Piazza, M. & Varum, U. 2006. prEN 1995-1-1 Eurocode 5 – Design of timber structures –
Strengthening Techniques of Portuguese Traditional Tim- Part 1-1: General – Common rules and rules for buildings.
ber Connections. Proc. Stuctural Analysis of Historical prEN 338 (1985) – Structural timber – strength classes:
Constructions, New Delhi 2006, P.B Lourenço, P. Roca, CEN/TC 124, Brussels, Belgium.
C. Modena, S. Agrawal (Eds.). Rondelet, J.B. 1802. Trattato teorico e pratico dell’arte di
Candelpergher, L. & Piazza, M. 2001. Mechanics of tra- edificare. Tomo V, Caranenti (ed.), Mantova, Traduzione
ditional connections with metal devices in timber roof italiana con note aggiunte di B. Soresina (in italian).
structures. Bodig, J. & Jayne, B.A. 1982. Mechanics of Seo, J.M., Choi, I.K. & Lee, J.R. 1999. Static and cyclic
wood and wood composites. Van Nostrand (ed.), New behaviour of wooden frames with tenon joints under
York, 110–118. lateral load. ASCE Journal of Structural Engineering,
Caroselli, M.R. 1967. La Reggia di Caserta: lavori, costo, 125(3), 344–349.
effetti della costruzione. Milano, Italy. STRAUS7 theory manual – version 2.2.3. (1999). Finite
Ceraldi, C. & Russo Ermolli, E. 2006. Timber Coverings of Element Analysis System.
palatine Chapel in Caserta Royal Palace. Proc. Stuctural Tampone, G. 2007. Mechanical Failures of the Timber Struc-
Analysis of Historical Constructions, New Delhi 2006, tural Systems. Proc. ICOMOS IWC – XVI International
P.B. Lourenço, P. Roca, C. Modena, S. Agrawal (Eds.). Symposium Material to Structure, Mechanical Behaviour
Chierici, G. 1984. La Reggia di Caserta. Ist.Poligrafico e and Failures of the Timber Structures, Florence, Venice
Zecca dello Stato-Archivi di Stato Coll. Arte Medievale e and Vicenza, Italy.
Moderna, Roma. Proc. STREMAH 2001, 7th Interna- Vanvitelli, L. 1756. Dichiarazione dei disegni del reale
tional Conference, Southampton, 415–424. palazzo di Caserta alle Sacre Reali Maestà di Carlo re
Derinaldis, P.P. & Tampone, G. 2007. The Failure of the Tim- delle Due Sicilie e di Maria Amelia di Sassonia regina.
ber Structures Caused by Incorrect Design-Execution of Milano, Il Polifilo, s.d., 128 pp.
722
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
G.A. Anese
Department of Public Works, Municipality of Portogruaro, Italy
A. Busetto
Busetto Engineering, Pordenone, Italy
ABSTRACT: The Civic Tower in Portogruaro is a 51 m high masonry bell-tower, originally built in the XIII
century, which today leans more than a metre out of plumb. Since 2003, the building inclination has been
continuously monitored with an optical inclinometer in an effort to see whether the tilt is still in progress. When
the monitoring started, it was thought highly unlikely that the Tower would tilt further. After three years of
monitoring and historical investigation, this idea was completely overturned. We show here how the initial view
evolved to a final awareness via a probabilistic model, based on Bayesian logic. We show how the joint use of
instrumental monitoring and historical documentation allowed timely recognition of signs of ongoing tilting and
accurate calculation not only of the mean inclination trend, but also the reliability level of this information.
1 INTRODUCTION small and the monitoring period too short to state with
certainty whether the trend calculated was a sign of an
In September 2002 the Municipality of Portogruaro, ongoing process.
a town in North-Eastern Italy, asked the University of During the third year of monitoring, the Univer-
Trento to install and operate instrumentation to mon- sity recovered some unpublished historical documents.
itor the tilt of its Civic Tower. This building is an Among these, an original design project revealing an
ancient masonry bell-tower, of height 51 m overall, elevation built in 1879, and documents reporting old
which today shows an inclination measured as 1.197 m measurements of the inclination dating 1962 and 1997.
at the north-east corner. Although the inclination was Based on this new information and on the fresh instru-
and is striking, at the time there was no real specific mental data, the University concluded, in the annual
concern about stability because the Tower had always monitoring report issued in September 2006, that an
been known to be leaning, as documented in local increasing inclination was very likely.
chronicles. Also, there was no evidence of inclination This final judgment, in sharp contrast with the
still in progress, and such an occurrence was judged initial view, might appear surprising. In reality, the
at the time to be very unlikely. change of opinion is the result of a rigorous and quan-
The monitoring system started recording the incli- titative logical analysis of the information available at
nation in October 2003. In September 2004, the Uni- the time.
versity reported to the Municipality that the changes in This paper follows this logical route again, showing
inclination recorded during this period were too small how the initial view evolved to final awareness via a
to raise concern over the short-term safety of theTower. probabilistic model, based on Bayesian logic. In the
In September 2005, after almost two years of next Section we introduce the Tower at issue, its his-
recording, the University told the Municipality of tory and the monitoring system installed; Section 3
Portogruaro that analysis of the data acquired to date formulates the algorithm used to update the posterior
allowed calculation of a possible linear trend of 1.7 mm judgment based on the information acquired; the appli-
per year. In particular, the direction of the possible cation of this procedure to the Civic Tower is reported
motion was very close to the maximum lean direc- in Section 4; finally, the outcomes are discussed at the
tion. Nonetheless, the leaning progression was too end of the paper.
723
Figure 1. Overview of the Tower (a); North and East elevation, cross-section and plan views at different levels.
2 THE TOWER AND ITS MONITORING the out of line of the spire is 0.31 m, corresponding to
an angle of 35.47 × 10−3rd .
As seen today, the bell-tower of the Cathedral of Observing the outer masonry of the tower we can
S.Andrea in Portogruaro, also known as CivicTower, is see at least four different masonry textures; this is a
a 59 m tall leaning campanile. The building was prob- clear sign of successive building phases in the past.
ably started in the XIII century. The tower was recently Nevertheless, at the time monitoring started, little
surveyed by Busetto & Romanin (2001) using theodo- documentation was available about this work. Local
lite and diastimeter, and the main geometrical features chronicles report a reconstruction of the spire, in 1879.
are shown in Figure 1. From the architectural point of The Municipality conserved incomplete documenta-
view the Tower has a masonry column, a belfry and tion of work done by the architect Mario De Goetzen
a spire. The column has a roughly square cross sec- between 1962 and 1963; this included reinforcement
tion, of size varying with height, from 7.30 m on a of the tower with concrete ring beams and steel ties,
side at ground level to 6.45 m at the top. The walls are and consolidation of the lower level of masonry at the
masonry infilled, with thickness varying from 1.3 m North and East sides. As visible today, this consolida-
at the base to 0.9 m at the top. There are four wooden tion involved replacing the original low quality yellow
floors at levels 5.58 m, 12.07 m, 18.45 m and 22.74 m, brickwork of the outer layer, with red brick of better
and an old masonry cross vault, now reinforced with a mechanical characteristics.
concrete slab, at level 26.20 m. The main column ends Between 2002 and 2003, the Tower underwent an
with a second cross vault, similar to that below, which extensive experimental campaign including material
supports the floor of the belfry, at level 31.43 m. The sampling, core drillings, endoscopies, flat jacks and
belfry balcony carries an octagonal tambour 5.45 m chemical analysis of the outer masonry foil. The results
in diameter and 4.9 m height, in turn topped by a of the investigation are reported in detail in the report
pyramidal spire 15.84 m in height overall. by Molteni (2003). The core drillings (labeled C1 to
The tower leans to the North-East. However the C3 on Fig. 1) revealed that the masonry is infilled:
angle of slant of the Tower is not constant over the external leafs consists of a single layer of bricks,
the height, because the higher parts were apparently while the quality of the infill is extremely poor and
rebuild with lower inclination in an effort to compen- inhomogeneous. The flat jacks tests showed the better
sate for a pre-existing tilt. Based on the 2001 survey, qualities of the new brickwork with respect to the orig-
the out of plumb respect to the north east edge of the inal: a compression strength of 8.5 MPa was measured
balcony, at level 36.62 m, is 1.197 m, corresponding to at the North side (M1 on Fig. 1), against a strength of
an inclination of 35.47 ×10−3rd (equal to 2.03◦ ); while 3MPa measured at the South side (M2 on Fig. 1).
724
Figure 2. View of the pendulum inclinometer wire, including the network cameras (a); plan-view of the Tower highlighting
the thermocouples and plumb wire position (b).
2.1 Details of the monitoring system of plumb in x and y directions recorded in the same
period.
Since October 2003, the tilt of the tower has been
observed continuously by a monitoring system. The
main instrument installed is an inclinometer, based 2.2 Discovery of historical documentation
on a pendulum hanged from the ceiling of the upper In late 2005, the discovery of unpublished documents
vault at level H = 29.90 m. The pendulum consists of a threw new light on the recent history of the Tower.
1.5 mm diameter steel wire with a brass mass attached: A first step was the retrieval of a geometrical sur-
the mass hangs in a water tank located at ground level, vey of the Tower carried out by De Goetzen in October
to dampen pendulum motion, as shown in Figure 2a. 1962, immediately before the restoration work. One
The position of the pendulum is permanently recorded of the designs reports an out of plumb of 740 mm
by two digital network cameras, carried on a steel to the East and 760 mm to the North, respect to the
frame at level 1.1 m which in turn is fastened to the North-East edge of the balcony, at level 36.74 m.
floor. The corresponding angles are 20.14 × 10−3rd East and
The cameras permanently acquire pictures of the 20.68 × 10−3rd North. The same survey also reports
wire and transmit them though the Internet every 10 an out of plumb of the spire of 150 mm North and
minutes to the monitoring station, physically located 150 mm East, apparently taken over a difference of
at the University of Trento. Using image recogni- level of 22.12 m: i.e. inclinations of 7.90 × 10−3rd East
tion software, these images are real-time processed and North. During an interview with the first author
in order to calculate the position of the wire with of this paper, in September 2005, De Goetzen speci-
respect to the background. The position of the pendu- fied that these measurements were taken using a plumb
lum is returned in the form of two coordinates, x and wire, thus with a precision that can be estimated in the
y, representing the shifts in direction West-East and order of two centimeters. In the same interview, De
North-South, respectively, with respect to the inter- Goetzen said that a many documents were collected
section of the axes of the two cameras. In addition few years ago by a student of the University of Padova,
to the pendulum inclinometer, the system records the as part of the research work in preparation for his
temperature at four thermocouples, two (T1 and T2) Thesis.
installed on the outer surface of the masonry, the other The first author of this paper met the former stu-
two (T3 and T4) on the inside, as depicted in Figure 2b. dent, now a professional engineer in November 2005.
The temperature data is acquired through a National Massimo Zanet, the author of the Thesis, had some
Instrument Field Point device, this also remotely con- important things to say during the interview. First:
trolled by the monitoring station at the University of as reported in his Thesis (Zanet 1997), he surveyed
Trento. the Tower in July 1997, and measured an out of
The server processes all the data, and publishes plumb of 770 mm to the East and 760 mm North
the state of the Tower in real time on the website using the same method and reference system as De
http://smartstructures.org/portogruaro. As an exam- Goetzen. These values correspond to inclinations of
ple, Figure 4a shows the time history of the temper- 20.96 × 10−3rd East and 20.68 × 10−3rd North. Sec-
ature θ 1 recorded at thermocouple T1 from October ond: during his research, Zanet was able to access
2003 to September 2007, while Figure 4b plots the out and examine the original designs of the restoration
725
work carried out between 1877 and 1879 by the civil that accompanies the design, Bon mentions an existing
engineer Antonio Bon. tilt of the original tower, without specifying its extent.
Based on Zanet’s advice, we were able to retrieve Also, there is no mention of the fact that the new spire
these projects in the archives of the Municipality. From would have to be built with a different inclination to
analysis of the designs, it is clear that the XIX century that of the tower below.
work was not just simple for repair, but rather a rad- Table 1 compares the inclinations surveyed in 1962
ical modification of the existing campanile. Figure 3 by De Goetzen, in 1997 by Zanet and in 2001 by
reproduces a sketch from Bon’s preliminary project, Busetto. Even though the precision of the first two
dating 1877, where the original campanile is compared measurements is probably quite low, it is remarkable
with one of the new proposals. We note that the orig- to observe that only one of the measurements is smaller
inal Tower was only 46,85 m tall, considerably lower than the preceding values.
than the height we see today. In the technical report
De Goetzen October 1962 20.14 × 10−3 20.68 × 10−3 7.90 × 10−3 7.90 × 10−3
Zanet July 1997 20.96 × 10−3 20.68 × 10−3 – –
Busetto & Romanin January 2001 21.50 × 10−3 24.49 × 10−3 6.01 × 10−3 12.65 × 10−3
∗
At the North-East corner.
726
In practice, we can divide the domain of the pos- where the notation Normal{x; µ, σ} indicates a normal
sible structural response into a mutually exclusive distribution with mean value µ and standard deviation
and exhaustive set of scenarios (S1 , S2 , . . . ., SNd ), each σ calculated in x. As long as errors are assumed to be
defining the structural behavior under a specific con- uncorrelated for each time and sensor, the likelihood
dition. The structural response n rj (n p, θ(t), t) for time for the whole measure set Mk is obtained combining
t and sensor sj in scenario Sn assuming environmen- the likelihoods of all samples for all sensors and time
tal condition θ(t) is controlled by a certain number of intervals recorded:
parameters (e.g.: stiffness, degradation), represented
by vector n p. The structural response is completely
defined by specifying a scenario and values for
the environmental conditions and for the correlated
parameter set. Here, as in the Bayesian model selection The likelihood of scenario Sn is then calculated
theory (Gregory, 2005, Bretthorst, 1996), the discrete by marginalization of parameters n p, i.e. by integrat-
scenario can be seen as a meta-parameter which qual- ing parameter likelihood on the whole domain Dn p,
itatively identifies the type of response function (e.g.: using their prior distribution PDF(n p|Sn ) as weighting
constant, linear, exponential) which in turn is specified function:
by a parameter set.
Assuming scenario Sn to be correct and after appro-
priate selection of the scenario’s parameter n p, the
observational response can be expressed as:
As the scenario set is complete and mutually exclu-
sive, evidence of measures Mk is simply obtained by
summing the scenarios:
where ej (t) is an error that accounts for both instru-
mental noise and the unavoidable imprecision of the
model assumed. Because from an epistemic point of
view these uncertainties are random, ej is modeled as
an uncorrelated zero-mean Gaussian noise. Its stan- Bayes’ theorem also allows an estimate of the pos-
dard deviation n (σe )j can be assumed independent of terior distribution of the corresponding parameter n p,
time, but generally changes with the sensor. Evidently using:
n
(σe )j changes with sensor type, but we may also
expect a dependency, for example, on sensor position
or precision.
Once measurements Mk and k become available
from the monitoring system, Bayes’ theorem allows
calculation of the updated, or posterior, probability for In summary, given a fresh set of measurements
each scenario Sn , from prior probability prob(Sn ), sce- Mk , this procedure allows real-time update of the
nario likelihood PDF(Mk |Sn ) and evidence PDF(Mk ), probability of each scenario using Eqs. (2) to (7).
using the following expression:
727
equal to 1 ϕ0,x . Conversely, in scenario S2 the compen- prob(S2 ) = 1/1000, and therefore prob(S1 ) = 99.9%
sated inclination is a linear function trend wj and an for the no-trend scenario. We must also define the prior
offset 2 ϕ0,x . In equations, we can write for direction x estimate of distribution of trends wx and wy . Based
(i.e. West-East) on the limited documentation available in October
2003, the only hint is that at the time of construc-
tion, say XIV century, the body Tower was reasonably
straight. Compared to Busetto’s measurements, car-
ried out in 2001, and assuming a linear trend, we can
roughly estimate average shifts of −1.5 mm·year−1 in
and similarly for direction y (i.e. North-South).
both directions, corresponding to inclination trends of
Response measurements are in this case the out of
−40.8 × 10−6rd ·year−1 . In summary, given the scarce
line components, x and y, measured at the base of the
information available at that time, this was assumed
pendulum while vector θ collects the 4 thermocouple
as the most likely value of current inclination trend,
measurements.
if any. Of course, this information is very impre-
According to Eq.(1), we can see the out of line mea-
cise: to account for this uncertainty, we can assume
sured by the pendulum as the sum of a term depending
a prior scatter of, say, 4 mm·year−1 , corresponding to
on tower inclination, a term linearly correlated with
an angular trend of 108.9 × 10−6rd .year−1 . Note that
the temperatures recorded and a noise:
this large scatter simply reflects our almost total prior
ignorance of the trends: the specific value selected is
not critical, in view of the final result of the updating
process.
Based on this prior information (i.e., without con-
where H = 29.90 m is the reference level for the out sidering the historical information discovered in 2005)
of line measurement, ax is the linear transformation we can apply Eq. (7) to update our knowledge using the
that correlates the temperatures to the out of plumb, data acquired real-time by the monitoring system. The
while indices 1 or 2 indicate that this vector gener- light plot on Figure 4e shows how the monitoring data
ally assumes different values in different scenarios. modifies the system perception of having a trend. We
As stated in the equation, the linear relation between can see that during the first two years of monitoring the
temperature and wire measurements is independent of probability of the trend scenario is always close to zero.
the scenario, and vectors aj , are considered as nuisance Only during the third year the monitoring information
parameters, that is, to be marginalized to gain infor- starts overturning the initial perception, to the point
mation on the relevant parameters wx and wy . Random that in April 2006 the data is sufficient to convince the
noise e is assumed to be Gaussian with zero mean system that the Tower is tilting.
and standard deviation n (σe )j equal to 10 mm for every Similarly, the light plots of Figures 4c and 4d show
sensor and scenario: this value, much larger than that the evolution of the distributions of trends wx and wy :
strictly related to the instrumental noise, also takes into we see that the trend estimates, which are very uncer-
account the uncertainties of the correlation model. tain during the first two years, rapidly converge to more
It is worth noting that the two scenarios have dif- reliable values.
fering degrees of complexity: scenario S2 involves The documentation acquired in 2005 radically
parameters wx and wy , so it can follow the measure- changed the initial judgment on the stability of the
ments more closely. Furthermore, scenario S1 can be Tower. Anyone with common sense can understand
regarded as a special sub-case of scenario S2 , when wx that this new information supports the idea that the
and wy are null. The reader might argue that, because Tower is tilting. However, the problem is how to quan-
of this, the probability of scenario S2 will always be tify the impact of this information on the probability
greater than that of S1 . Actually, this is not neces- of there being further tilt.
sarily the case. Indeed, by tuning the parameters of The approach we followed is to cluster the histori-
scenarioS2 , we can always obtain a better fit than that cal information in a number of separate datasets, each
related to scenario S1 . However, according to Bayesian associated with a specific distribution, and use recur-
logic, the ratio between the best and the average fit (the sively the same Eq. (7) to update the likelihood of
so-called Ockham factor) plays a fundamental role too Scenario 2. The first dataset identified (labelled A)
(Gregory 2005). consists in the three out of plumb measurements taken
To implement the Bayesian updating procedure, on years 1962, 1997 and 2001. In these cases the mea-
we have to define quantitatively the prior knowl- surement techniques adopted suggest we consider a
edge. As mentioned, in October 2003, the probability noise of 20 mm. A second set (B) includes the com-
of the Tower tilting further was reputed very low: parison of inclination measurements of the spire taken
we can formalize this initial perception assuming in 1962 and 2001; in this case a scatter of 50 mm
for the tilting scenario a prior probability equal to is assumed. The third data set (C) is basically the
728
Figure 4. Temperature measurements at thermocouple T4 (a); out of plumb measurements (b); posterior distribution of
angular trend wx (c); posterior distribution of angular trend wy (d); posterior probability of scenario S2 (e).
notion that the spire was rebuilt vertical in 1879; this available, while the standard deviations of the angular
conjecture is assumed true with a likelihood of 50%; if trend decrease to 17.3 × 10−6rd year−1 .
true, the standard deviation of the original inclination Mixing this new prior knowledge with the moni-
is assumed equal to 8 × 10−3rd . toring data, we obtain, day by day, the posterior trend
Table 2 summarizes the outcomes of the prior infor- distributions and the probability of scenario S2 , plotted
mation update using recursively the three historical in bold in Figures 4c, 4d and 4e.
data sets. With respect to the initial judgment, prob(S2 ) In detail, Figure 4d illustrates that, considering the
increases up to 3.7% when all the historical datasets are historical data, the system would theoretically become
729
Table 2. Update of prior information based on the historical information.
Prior knowledge 0.1% −40.8 × 10−6 108.9 × 10−6 −40.8 × 10−6 108.9 × 10−6
Considering dataset A 2.0% −30.8 × 10−6 18.0 × 10−6 −58.5 × 10−6 18.0 × 10−6
Considering dataset B 1.3% −27.0 × 10−6 17.8 × 10−6 −57.3 × 10−6 17.8 × 10−6
Considering dataset C 3.7% −29.2 × 10−6 17.3 × 10−6 −60.1 × 10−6 17.3 × 10−6
aware of the trend as early as September 2005; there- Portogruaro initiated an investigation campaign to ana-
fore, we can say that the historical information is lyze the state of the Tower foundations and to plan a
roughly equivalent to 6 months of instrumental mon- reinforcement to prevent any possible future risk of
itoring. Note that, because of the consistency of the collapse. The work is currently under way.
Bayesian logic, the outcome of the updating process
is invariant with respect to the order of information
processing. Thus, the effect of finding the historical
documentation is to shift from the light plot to the bold ACKNOWLEDGEMENTS
plot at the time this information is available, which in
our case is November 2005. The installation and operation of the monitoring sys-
Worthy of note is that after three years of moni- tem was funded by the Municipality of Portogruaro.
toring, the chance of continuous tilt increase of the The Bayesian algorithm reported in this paper has been
Tower is close to certainty, regardless of the his- developed with the financial contribution of the Ital-
torical data assumed. In the same way the angular ian Ministry of Education (MIUR). The authors wish
trends converge to similar values after the first years to thank all those who contributed to the success of
of monitoring. This means that, after a certain time, this work, and specifically Stefano Toffaletti, Marco
the information acquired by the monitoring system Molinari and Nicola Naldo.
becomes dominant over any type of prior knowl-
edge. In detail, the trends identified at the end of
the monitoring are wx = −34.6 × 10−6rd ·year−1 and REFERENCES
wy = −52.1 × 10−6rd year−1 with a standard deviation Bayes, T. 1763. An Essay toward solving a Problem in the
of 6.7 × 10−6rd year−1 : in terms of out of plumb respect doctrine of Chances. Philos. Trans. R. Soc. London 53:
to level H = 29.90 m, this is to say 1.03 mm year−1 pp. 370–418.
West-East and 1.56 mm year− South-North, with a Beck, J.L. & Au, S.-K. 2002. Bayesian Updating of Structural
scatter of only 0.2 mm year−1 . Models and Reliability using Markov Chain Monte Carlo
Simulation. J. Eng. Mech. 128(4): pp. 380–391.
Beck, J.L. & Katafygiotis, L.S. 1998. Updating Models And
5 CONCLUSIONS Their Uncertainties, I: Bayesian Statistical Framework. J.
Eng. Mech. 124(2): pp. 455–461.
In the case presented, the joint use of instrumen- Bretthorst, G.L. 1996. An introduction to model selection
tal monitoring and historical documentation allowed using probability theory as logic. In G.R. Heidbreder (ed),
Maximum Entropy and Bayesian Methods, Springer New
timely recognition of signs of an ongoing tilt and accu- York, NY.
rate calculation not only of the mean value of the Busetto, A. & Romanin, L. 2001. Studio preliminare per
inclination trend, but also the level of reliability of this l’individuazione del percorso atto a definire lo stato
information. The general Bayesian methodology intro- patologico della Torre Civica, Technical report for the
duced deals flexibly with all the uncertainties involved Municipality of Portogruaro.
in the recognition problem: measurement noise, uncer- Collins, M.P. & Mitchell, D. 1991. Prestressed Concrete
tainness on the model and inaccurate prior infor- Structures. Englewood Cliffs – New Jersey: Prentice Hall
mation. Moreover, it lets us quantitatively combine Gregory, P. 2005. Bayesian Logical Data Analysis for the
information of completely different nature, including Physical Sciences. Cambridge: Cambridge University
Press.
incomplete datasets or unproved conjectures. Jaynes, E.T. 2003. Probability Theory: The Logic of Science.
Based on this logical path, the University con- Cambridge, UK: Cambridge University Press.
cluded, in September 2006, that a progressive tilt MacKay, D.J.C. 2003. Information Theory, Inference and
was very likely. Following these conclusions, the Learning Algorithms. Cambridge, UK: Cambridge Uni-
Department of Public Works of the Municipality of versity Press.
730
Measures, R.M. 2001. Structural Monitoring with Fiber Skilling, J. 1998. Probabilistic Data Analysis; An intro-
Optic Technology. NY: Academic Press. ductory Guide, Journal of Microscopy 190(1–2),
Metropolis, N., Rosenbluth, A., Rosenbluth, M., Teller, A. pp. 28–36.
& Teller, E. 1953. Equation of state calculation by Sohn, H. & Law, K-H. 2000. Bayesian Probabilistic Dam-
fast computing machines. J. of Chemical Physics 21: age Detection of a Reinforced Concrete Bridge Column.
pp. 1087–1092. Earthquake Engineering and Structural Dynamics 29(8),
Molteni, M. 2003. Torre Civica Duomo di S. Andrea: Cam- pp. 1131–1152.
pagna di studi, analisi, indagini e prove sperimentali per Zanet, M. 1997. Studio storico strutturale della Torre Civica
la valutazione dello stato patologico della torre campa- della città di Portogruaro. Laurea Thesis, Facoltà di
naria: Relazione tecnica conclusiva. Technical report for Ingegneria, University of Padova.
the Municipality of Portogruaro. Zonta, D., Pozzi, M., Colombo, A. & Bursi, O.S. 2006.
Papadimitriou, C., Beck, J.L. & Katafyogiotis, L.S. 1997. Industrial Design of Instrumented PRC Elements for the
Asymptotic Expansion for Reliability and Moments of Condition Monitoring of Civil Infrastructure. In: Smart
Uncertain Systems. J. Eng. Mech. 123(12): pp. 380–391. Structures and Materials 2006: Smart Sensor Monitor-
Sivia, D.S. 2006. Data Analysis: a Bayesian Tutorial. Oxford: ing Systems and Applications. SPIE proc., Vol. 6167,
Oxford University Press. pp. 131–141.
731
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Author index
733
Giardina, G. 413 Kelley, S.J. 25 Mazzolani, F.M. 383, 817, 1157,
Ginocchio, F. 1095 Kemper, J.H. 1079 1225
Giordano, A. 1391 Kim, N.H. 861 McCabe, B.P. 1357
Giorgi, L. 191 Kirizsán, I. 143, 1375 McCombie, P. 781
Giráldez, P. 643 Klemenc, I. 789 Mele, E. 565, 1261
Giuriani, E. 13, 413 Kolymbas, D. 109 Meli, R. 825
Giustina, I. 957 Korkmaz, A. 1303 Mezzi, M. 1061
González, J.L. 643 Krstevska, L. 383 Miglioli, M. 1277, 1287
González, R. 643 Kruschwitz, S. 109 Milani, E. 1277
Goñi, R. 705 Kutnyi, A. 183 Mileto, C. 1467
Gramatikov, K. 383 Kwiecień, A. 931 Miller, J.D. 1133
Grande, E. 1017 Köck, B. 235 Miltiadou, A. 621
Grazzini, A. 339 Köpp, Ch. 109 Miltiadou-Fezans, A. 71, 1025,
Gu, X.L. 1537 1149
Guadagnuolo, M. 1391 Lacidogna, G. 367, 375 Minch, M.Y. 1455
Gubana, A. 949 Laefer, D.F. 1207 Miranda, L. 875
Guedes, J. 875 Lagomarsino, S. 517 Miri, M. 909
Guedes, J.M. 653 Lakshmipriya, T. 1491 Modena, C. 403, 755, 763, 1345
Landolfo, R. 383, 431 Modi, S.M. 175
Harkness, J. 781 Lanza, S. 109 Morales, K. 1095
Hastemoglu, H. 1303 Lanzoni, L. 1277, 1287 Morer, P. 705
Hayden, N.J. 1369 Laorenza, C. 715 Mormone, V. 915
Heath, A. 781 Lauriks, L. 683 Mosseri, A. 1363
Heinemann, H.A. 151 Lawrence, R.M.H. 885 Mouzakis, H.P. 621
Hendrickx, R. 3, 973 Lee, Y.W. 861 Munari, M. 1277
Hennen, Ch. 109 Li, S. 393 Mundell, C. 781
Henriques, D.F. 1101 Li, X. 1537 Muñoz, A.M. 243
Herle, I. 109 Linscott, K. 121
Hickey, S. 1207 Lopez Izquerdo, N. 1277 Nagai, T. 1399
Hirai, T. 629 Loughlin, B. 1207 Nascosi, M. 1287
Hobbelman, G.J. 689 Lourenço, P.B. 95, 311, 739, 771, Niglio, O. 251
Hoirisch, M. 137 833, 1345 Nijland, T.G. 151
Holzer, S.M. 235 Lucchin, F. 403 Norden, J.F. 1107
Hong, S.G. 861 Lutenegger, A.J. 1079 Nunes, L. 1101
Hughes, T.G. 909 Lutman, M. 1311
Hwang, J.K. 861 Ochsendorf, J. 593
Machado, J.S. 833 Odeyale, T.O. 991
Ignoul, S. 525, 747, 1087, 1345 Macorini, L. 1035 Oliveira, B.T. 283
Ilharco, T. 653 Madhoushi, M. 469 Oliveira, D. 771
Imbimbo, M. 1017 Maeda, T. 629 Oliveira, D.V. 739, 1345
Indirli, M. 1277, 1287 Maierhofer, Ch. 109 Olosz, E. 473
Invernizzi, S. 367, 375, 941 Makay, D. 663 Orduña, A. 1297
Ivorra, S. 349 Mammana, O. 383 Ospina, Y.M. 243
Iwaki, C. 1095 Manuello, A. 375 Ottoni, F. 697
Marcari, G. 715 Ozakin, R. 1141, 1167
Jain, J. 289 Mariani, G. 1481 O’Farrell, S. 1207
Jankowski, R. 931 Mariani, M. 1123 O’Mahony, G. 1207
Jasieńko, J. 357 Marini, A. 13, 413 O’Sullivan, M. 207
Jäger, W. 1181 Marzi, T. 941
Jeong, S.J. 861 Marzo, A. 817 Pallarés, F. 349
Joo, S.J. 861 Marzullo, M. 1061 Pane, A. 1527
Masoero, E. 867 Panizza, M. 763
Kakkinou, S. 1025 Massarelli, T.M. 1461 Panão, A. 817
Kalagri, A. 1025, 1149 Matracchi, P. 191 Papadopoulos, K. 1113
Kaup, J. 1245 Matteo, J.A. 1245 Paredes, J.A. 243
Kayser, C. 1423 Maus, H. 1423 Parisi, M.A. 1253
Keersmaekers, R. 1053 Mazzanti, S. 1287 Parodi, S. 1237
734
Passerini, A. 555 Scaini, S. 867 Valletta, G. 1061
Paupério, E. 653 Schoonjans, Y. 1383 Valluzzi, M.R. 33, 755, 763
Peng, B. 1537 Schueremans, L. 525, 747, 1321, Van Balen, K. 3, 965, 973
Peña, F. 95 1345 Van Gemert, D. 525, 747, 965,
Piazza, M. 459, 507, 843, 1253 Screpanti, A. 1277 973, 1053
Pini, D. 1287 Sextos, A. 611 van Hees, R.P.J. 151, 689
Pitacco, I. 533 Seçkin, N. 267 Van Rickstal, F. 1053
Pittas, M. 1499 Shang, D.F. 1537 Vargas, J. 1095
Plizzari, G.A. 957 Shi, Y. 489 Varum, H. 507
Podestà, S. 441, 1237 Sikka, S. 1069 Vasconcelos, G. 95, 771
Popa, I. 169 Silva, R.A. 739 Vassallo, E. 259
Porter, D.W. 1369 Simonini, G. 1277, 1287 Vegas López-Manzanares, F.
Portioli, F. 383 Skłodowski, M. 423 1467
Posen, D. 1053 Smars, P. 1199 Vendrell, M. 643
Pozzi, M. 723 Sorrentino, L. 217, 1431 Verbeeck, B. 683, 1175
Preciado, A. 1297 Sparks, S.P. 451 Verstrynge, E. 525, 747, 1321,
Proença, J.M. 817 Speranza, E. 1509 1345
Puglisi, C. 1277 Speranzini, E. 981 Vignoli, A. 601
Sperbeck, S.T. 1043 Virsta, A. 109
Qing, C. 1441 Stacey, M.S. 1357 Virág, J. 169
Stammers, C. 489 Viskovic, A. 1445
Ramos, L.F. 95, 311 Stylianidis, K.-A. 611 Vitti, M. 1445, 1509
Resemini, S. 441, 517 Swailes, T. 207
Ribeiro, R.T.M. 137 Swann, J.D. 1339 Walker, P. 781, 885
Rinaldis, D. 799 Szabó, B. 143, 473, 663, 1375 Weiss, P. 789
Riotto, G. 441 Szołomicki, J.P. 1455 Wevers, M. 747
Riva, P. 413
Wouters, I. 683, 1175, 1383
Rizzi, E. 583 Taliercio, A. 635
Roca, P. 643 Tardini, C. 1253
Romanelli, F. 1277 Tashkov, Lj. 383 Yamada, S. 1399
Romano, A. 547 Teston, S. 1287 Yamamoto, T. 629
Ruocci, G. 575 Thelin, C. 121 Yao, L. 393
Russo Ermolli, E. 915 Tiraboschi, C. 1331 Yasmine, J. 1407
Russo, V. 259 Tocci, C. 1431, 1519 Yeomans, D.T. 63
Toker, S. 1303 Yergün, U. 999
Sacco, E. 1017 Tomasi, R. 843 Yuan, J. 393
Saisi, A. 1331 Tomasoni, E. 47
Sala, E. 957 Tomaževič, M. 109, 789 ˛ B. 931
Zajac,
San-José, J.T. 851 Touliatos, P. 621 Zangheri, F. 481
San-Mateos, R. 851 Toumbakari, E.E. 673 Zanon, P. 723
Santaniello, R. 1261 Tsai, P.-H. 331 Zanzi, L. 1331
Sanyal, S. 161 Tulliani, J.M. 941 Zarogianni, E. 1025
Sassu, M. 1481 Ziagou, A. 1025
Savvidou, M. 1149 Ungaro, L. 1445, 1509 Zonta, D. 723
735
STRUCTURAL ANALYSIS OF HISTORIC CONSTRUCTION
PROCEEDINGS OF THE SIXTH INTERNATIONAL CONFERENCE ON STRUCTURAL
ANALYSIS OF HISTORIC CONSTRUCTION, 2–4 JULY, BATH, UNITED KINGDOM
VOLUME 2
Editors
Dina D’Ayala & Enrico Fodde
Department of Architecture & Civil Engineering,
University of Bath, Bath, UK
CRC Press/Balkema is an imprint of the Taylor & Francis Group, an informa business
All rights reserved. No part of this publication or the information contained herein may be reproduced, stored
in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, by photocopying,
recording or otherwise, without written prior permission from the publishers.
Although all care is taken to ensure integrity and the quality of this publication and the information herein, no
responsibility is assumed by the publishers nor the author for any damage to the property or persons as a result
of operation or use of this publication and/or the information contained herein.
Table of Contents
Preface XV
Committees XVII
VOLUME 1
Invited papers
Preservation of workmanship or workmanship for preservation 3
K. Van Balen & R. Hendrickx
Experiences from the Northern Italy 2004 earthquake: Vulnerability assessment and
strengthening of historic churches 13
E. Giuriani & A. Marini
The structural behaviour of masonry vaults: Limit state analysis with finite friction 47
D.F. D’Ayala & E. Tomasoni
Cultural Heritage protection against flood – A European FP6 research project 109
Ch. Maierhofer, Ch. Köpp, S. Kruschwitz, M. Drdacky, Ch. Hennen, S. Lanza, M. Tomaževič,
D. Kolymbas, I. Herle, A. Virsta, L. Binda & P. Askew
V
Analysis of archaic fireproof floor systems 129
D. Friedman
Intervention limits in floor wood structure restoration of a historical building 137
R.T.M. Ribeiro & M. Hoirisch
The safety of historic load-bearing structures 143
B. Szabó & I. Kirizsán
Concrete: Too young for conservation? 151
H.A. Heinemann, R.P.J. van Hees, T.G. Nijland
Significance of a Mughal Mall and its role in achieving safety 161
S. Sanyal
Historical constructions – Authenticity and adaptation to the modern demands 169
I. Bucur-Horváth, I. Popa, M. Bulbuk & J. Virág
Champaner-Pavagadh – Managing conflicts – A conservation challenge 175
S.M. Modi
VI
Structure as a document 289
J. Jain
VII
Diagnosis and analysis of two king-post trusses 459
J.M. Branco, P.J.S. Cruz & M. Piazza
Non destructive and laboratory evaluation of strength of decayed wood members in a historic
construction located in Gorgan (North of Iran) 469
M. Madhoushi, J. Eimanian & M.P. Ansell
The structural behaviour of spires 473
E. Olosz & B. Szabó
In-situ and laboratory investigations on materials and structures for the static improvement:
“Torre Quadrata del Circo Romano” (Milan) 481
S. Bortolotto, E. Ciocchini, F. Zangheri & P.M. De Marchi
Dynamic multi-body behaviour of historic masonry buildings models 489
D. D’Ayala, Y. Shi & C. Stammers
Assessment and strengthening masonry arch bridges 497
C.L. Brookes
Field tests of a timber queen-post truss and numerical analysis 507
J.M. Branco, P.J.S. Cruz, M. Piazza & H. Varum
VIII
Earthquake based condition monitoring of the Katholikon of Dafni Monastery 621
H.P. Mouzakis, A. Miltiadou, N. Delinikolas, P. Touliatos & J. Dourakopoulos
Continuous and discontinuous modeling of the structures in Bayon temple, Angkor 629
T. Maeda, T. Yamamoto & T. Hirai
Safety assessment of temple E7 in Mỹ So,n, Viet Nam 635
P. Condoleo, A. Taliercio & L. Binda
Construction process, damage and structural analysis. Two case studies 643
R. González, F. Caballé, J. Domenge, M. Vendrell, P. Giráldez, P. Roca & J.L. González
Analysis and diagnosis of timber structures of Porto historical centre 653
T. Ilharco, J.M. Guedes, A. Arêde, E. Paupério & A.G. Costa
Baroque roof structures in Transylvania – Research and analyses 663
D. Makay & B. Szabó
The Athens Parthenon: Analysis and interpretation of the structural failures of the orthostate of
the northern wall 673
E.E. Toumbakari
Structural assessment of the Winter Garden of the Royal Glasshouses of Laeken, Belgium 683
L. Lauriks, I. Wouters, S. Adriaenssens, M. de Bouw & B. Verbeeck
Towards a systematic diagnosis of structural damage 689
I.A.E. de Vent, R.P.J. van Hees & G.J. Hobbelman
From geometrical and cracks survey to static analysis method: The case study of Santa Maria
del Quartiere dome, in Parma (Italy) 697
F. Ottoni
A benchmarking study of the analysis of non-reinforced structures applied to the structural
behavior of domes 705
P. Morer & R. Goñi
Structural analysis of the Caserta Royal Palace timber roof connections 715
G. Fabbrocino, G. Marcari, C. Laorenza & E. Cosenza
Real-time Probabilistic health monitoring of the Portogruaro Civic Tower 723
D. Zonta, M. Pozzi, P. Zanon, G.A. Anese & A. Busetto
VOLUME 2
Experimental studies
On the strengthening of three-leaf stone masonry walls 739
R.A. Silva, D.V. Oliveira & P.B. Lourenço
Long-term behaviour of historical masonry – A quantitative acquisition of the damage evolution 747
E. Verstrynge, L. Schueremans, D. Van Gemert, S. Ignoul & M. Wevers
Experimental assessment and modelling of in-plane behaviour of timber floors 755
M.R. Valluzzi, E. Garbin, M. Dalla Benetta & C. Modena
Bond behaviour of CFRP and GFRP laminates on brick masonry 763
M. Panizza, E. Garbin, M.R. Valluzzi & C. Modena
Experimental shear behavior of stone masonry joints 771
G. Vasconcelos, P.B. Lourenço & D. Oliveira
IX
Large scale testing of drystone retaining walls 781
C. Mundell, P. McCombie, A. Heath, J. Harkness & P. Walker
Seismic isolation, strengthening of walls with CFRP strips and heritage masonry buildings 789
M. Tomaževič, I. Klemenc & P. Weiss
Experimental survey on seismic response of masonry models 799
A. Baratta, I. Corbi, O. Corbi & D. Rinaldis
Bond performance of fiber reinforced grout on brickwork specimens 809
I. Carbone & G. de Felice
Experimental analysis of rectangular shaped sleeve connectors for composite
timber-steel-concrete floors: Bending tests 817
L. Calado, J.M. Proença, A. Panão, F.M. Mazzolani, B. Faggiano & A. Marzo
Shaking table testing of a typical Mexican colonial temple 825
M. Chávez & R. Meli
Capacity of a traditional timber mortise and tenon joint 833
A.O. Feio, P.B. Lourenço & J.S. Machado
Behaviour of refurbished timber floors characterized by different in-plane stiffness 843
M. Piazza, C. Baldessari, R. Tomasi & E. Acler
Stone masonry walls: Strengthening with TRM (I) 851
J.T. San-José, D. García, R. San-Mateos & J. Díez
The effect of friction joint and Gongpo (bracket set) as an energy dissipation in Korean
traditional wooden structure 861
J.K. Hwang, S.G. Hong, N.H. Kim, Y.W. Lee, S.J. Jeong & S.J. Joo
Optimization of cutting processes in archaeological sites 867
C. Cennamo, B.M. Chiaia, E. Masoero & S. Scaini
Experimental and numerical study on the application of the flat-jack tests to masonry walls 875
M. Acito, L. Binda, G. Cardani, J. Guedes & L. Miranda
The impact of the water/lime ratio on the structural characteristics of air lime mortars 885
R.M.H. Lawrence & P. Walker
Soil mortar’s evaluation characteristics for their use in conservation practice 891
G.E. Bei
Experimental and numerical analyses for static retrofitting intervention on the “Cappella
dei Principi” in Firenze 899
A. Antonelli, G. Bartoli & M. Betti
Comparison the effect of different repair method applying on masonry arch bridges 909
M. Miri & T.G. Hughes
X
Cross laminated timber panels to strengthen wood floors 949
A. Gubana
Lime mortar with natural pozzolana: Historical issues and mechanical behavior 957
E. Sala, I. Giustina & G.A. Plizzari
Blended lime-cement mortars for conservation purposes: Microstructure and strength
development 965
O. Cizer, K. Van Balen, D. Van Gemert & J. Elsen
Assessing workability of mortar by means of rheological parameters and desorptivity 973
R. Hendrickx, K. Van Balen & D. Van Gemert
Consolidation and reinforcement of stone walls using a reinforced repointing grid 981
A. Borri, M. Corradi, E. Speranzini & A. Giannantoni
Innovative and sustainable local material in traditional African architecture – Socio cultural
dimension 991
T.O. Odeyale & T.O. Adekunle
Imported construction materials and techniques in 19th century Ottoman architecture 999
U. Yergün & A. Çiftçi
Advance composites applications on historical structures in Italy: Case studies and future
developments 1007
P. Casadei & E. Agneloni
XI
Special structural solutions for adaptive use in grand hotel in Norwich, CT, USA 1107
J.F. Norden
The restoration of the foundation in the north side of the temple of Apollo Epikourios 1113
K. Papadopoulos
Structural consolidation methods for the Temple of Santa Maria della Consolazione in
Todi (Perugia, Italy), damaged by landslides and earthquakes 1123
M. Mariani
The diagnosis and arresting of settlement within Westminster Hall in the Houses of Parliament, London 1133
J.D. Miller
Istanbul – Fatih, Millet Library / Feyzullah Efendi Madrasah restoration 1141
R. Ozakin & A. Erdem
Application of non-destructive techniques at the Katholikon of Dafni Monastery for mapping
the mosaics substrata and grouting monitoring 1149
Ph. Côte, X. Dérobert, A. Miltiadou-Fezans, N. Delinikolas, O. Durand, J. Alexandre,
A. Kalagri, M. Savvidou, D. Chryssopoulos, L. Anamaterou & F. Georganis
The PROHITECH research project 1157
F.M. Mazzolani
The column of Marcian/Kıztaşı in Istanbul and an evaluation of its restorations 1167
A. Erdem & R. Ozakin
Restoration and strengthening strategies for 19th century iron pedestrian suspension bridges 1175
S. Adriaenssens, B. Verbeeck, I. Wouters & M. de Bouw
Reconstruction of the Sistani House at Bam Citadel after the collapse due to the earthquake 2003 1181
W. Jäger & C. Fuchs
XII
The effects of temperature on historical stone masonry structures 1271
C. Blasi & E. Coïsson
A GIS platform on main natural hazards for Valparaíso city (Chile) and vulnerability studies
for some historical constructions and urban sectors 1277
M. Indirli, F. Geremei, C. Puglisi, A. Screpanti, D. Blersch, L. Lanzoni, N. Lopez Izquerdo,
E. Milani, M. Miglioli, G. Simonini, M. Munari & F. Romanelli
The construction of a GIS database for the historical center of San Giuliano di Puglia 1287
S. Mazzanti, M. Miglioli, M. Nascosi, G. Simonini, M. Indirli, L. Lanzoni, S. Teston,
C. Alessandri & D. Pini
Vulnerability assessment of churches at Colima by 3D limit analysis models 1297
A. Orduña, A. Preciado, F. Galván & J.C. Araiza
Vulnerability screening by visual inspection of churches of North-East Anatolia 1303
A. Korkmaz, S. Toker & H. Hastemoglu
Seismic resistance assessment of heritage masonry buildings in public use in Ljubljana 1311
M. Lutman
Conservation practice
Relation between sub-soil and masonry structure of Angkor monument 1399
S. Yamada, M. Araya & T. Nagai
Beaufort Castle, Lebanon; conservation versus restoration project 1407
J. Yasmine
Structural faults in earthen archaeological sites in central Asia: Analysis and repair methods 1415
E. Fodde
XIII
Maria Birnbaum – Construction history, conservation history 1423
R. Barthel, H. Maus & C. Kayser
The structural strengthening of early and mid 20th century reinforced concrete diaphragms 1431
L. Sorrentino & C. Tocci
Strengthening design of Ganxi’s Former Residence 1441
C. Qing
The Trajan Markets and their Great Hall – The conservation problems and the structural
intervention for the improvement of the seismic safety 1445
G. Croci, A. Viskovic, A. Bozzetti, L.Ungaro & M. Vitti
Structural strengthening of the Dobrzyca Palace, Poland 1455
M.Y. Minch & J.P. Szołomicki
Structural analysis and restoration of the Guglia della Madonna di Bitonto 1461
T.M. Massarelli
Non conventional solutions for the consolidation of bell towers 1467
C. Mileto, F. Vegas López-Manzanares
Strengthening of a wooden covering built in the 18th century 1475
N. Augenti
Experiences of consolidation on archaeological UNESCO sites in the Sultanate of
Oman: The Fortress of Al Balid and the Citadel of Sumhuram 1481
M. Andreini, C. Cei, G. Mariani & M. Sassu
Conservation and restoration of a Ta Prohm temple 1491
T. Lakshmipriya
Restoration of a historical building for Cyprus Technical Chamber 1499
M. Pittas
The great hall of the Trajan Markets throughout the centuries 1509
L. Ungaro, M. Vitti & E. Speranza
The XIX century opera theatre of Catania: Conservation and seismic strengthening interventions 1519
C.F. Carocci & C. Tocci
House-towers in Campi Flegrei: History, conservation and re-use 1527
G. de Martino, R. de Martino, A. Pane & F. Delizia
Structural inspection and analysis of former British Consulate in Shanghai 1537
X.L. Gu, B. Peng, X. Li & D.F. Shang
XIV
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Preface
The successful preservation of an historic building, complex, or city, unless assigned to museum status, depends
on its continued use and the daily care and maintenance. The possibility of continued use depends on the
adaptation of the building to modern standards and practice of living, and this might require changes of its
constructional or structural features.
Conservation engineering can be defined as the process of understanding, interpreting and managing the
architectural heritage to safely deliver it to posterity. The term ‘safely’ embodies the assumption that the bodies
and individuals responsible for the care of historic buildings work towards ensuring maximum private or public
utility vis a vis minimum loss of fabric and value.
The conflict between safety standards and conservation philosophy usually stems from the fact that not
just the standards themselves, but the practice of achieving the standards, are based on, and refer to, modern
materials, techniques and processes. An historic building is considered of value not only because of its age,
but most importantly because of its uniqueness, its deviation from the norm and hence, to a certain extent,
from what is standard. This constitutes the building’s significance or authenticity, a quality vital to conserve.
With increasing global interest in conservation worldwide it is essential to open the debate on more inclusive
definitions of significance and on more articulated concepts for safety and for acceptable and reliable techniques,
in an attempt to reconcile and integrate further the activity of all the professions involved in conservation.
It is in this framework and with these aims that the Department of Architecture and Civil Engineering of the
University of Bath set out to organize and deliver the sixth international conference on Structural Analysis of
Historic Construction to be held in Bath’s Assembly Rooms between 2 and 4 July 2008.
A grade I listed building designed in 1769 by John Wood the Younger, the Assembly Rooms are located in
the heart of the World Heritage city. Their function was to host events and gatherings and this makes it the
perfect venue for the conference, which is the last of a series that is organized every three or two years. Previous
conferences were organized in New Delhi (2006), Padova (2004), Guimaraes (2001), and Barcelona (1998 and
1995).
The event is sponsored by CINTEC International Ltd, Gifford, English Heritage, ICOMOS, IStructE, and the
University of Bath. This sponsorship has made possible the participation to the conference of authors from other
parts of the world which could have not attended otherwise. Their contribution is critical to our understanding
of practical and theoretical issues in different contexts and to the widening and inclusivity of the debate.
The joint Department of Architecture and Civil Engineering of the University of Bath has a long standing
tradition of teaching conservation and sustainability within the built environment. The MSc in Conservation
of Historic Buildings provides education on the fundamental principles of conservation and is one of the most
successful in the country. Furthermore, the Department offers the only course on Conservation Engineering in
the UK, combining structural aspects with the study and conservation of the built heritage.
The main objective of the conference is to promote debate on an international scale, by the exchange of knowl-
edge and approaches among researchers and practitioners in the different professions involved in conservation and
from diverse cultural backgrounds. The conference is structured into eleven themes: ‘Definition of Significance
and Attribution of Value’, ‘History and Documentation’, ‘Monitoring, non Destructive Evaluation and Testing’,
‘Assessment and Analytical Techniques’, ‘Experimental Studies’, ‘Use of Traditional, Alternative and Innova-
tive Materials’, ‘Novel Conservation Engineering Techniques, Restoration and Strengthening’, ‘Vulnerability to
Natural Hazards, Preparedness and Retrofitting’, ‘Codes, Guidelines and Methods for Safeguarding Safety and
Significance’, and ‘Conservation Practice’.
A total of 180 contributions are collected in the two volumes of proceedings, of which 10 are invited papers
presented by speakers of international renown to the conservation audience. The reading of these proceedings
shows that the discipline of structural analysis of historic buildings is nowadays critical to the success of many
conservation projects and the skills required are increasingly complex and more articulated. The papers collected
all show the need for multidisciplinary approach in conservation research and practice and the continually
evolving need for more sophisticated solutions where both safety and significance play a critical role. An
increasing number of papers tackle issues not covered in previous conferences, such as the structural conservation
of modern materials. It is hoped that the papers presented here can form the basis for future research and the
XV
formulation of relevant issues that need to be advanced. This we hope will constitute the basis for a lively debate
during the 3 days of the conference and will inform and stimulate the directions for future conferences.
The editors are grateful to Marion Harney for managing the organization of the conference and to the depart-
mental office for the great support. The editors are also indebted to the Advisory Committee members for
providing the opportunity and the needed guidance to organize this conference, to the Organizing Commit-
tee members for the efforts and support given to make the conference possible, and finally but crucially to all
members of the Scientific Committee which have agreed to advise in the papers’ selection and reviewing process.
March 2008
Dina D’Ayala, Enrico Fodde
XVI
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Committees
Advisory Committee
Paulo Lourenço, University of Minho, Portugal
Claudio Modena, University of Padua, Italy
Pere Roca, Politecnico de Cataluña, Spain
Organising Committee
Dina D’Ayala, University of Bath, UK
Marion Harney, University of Bath, UK
Enrico Fodde, University of Bath, UK
Alex Copping, University of Bath, UK
Michael Forsyth, University of Bath, UK
Peter Walker, University of Bath, UK
Susan Denyer, Secretary ICOMOS-UK
Geoff Clifton, Chairman, Gifford, UK
Terry Girdler, Chief Engineer, English Heritage, UK
Scientific Committee
Takayoshi Aoki, Nagoya City University, Japan
Görün Arun, Yildiz Technical University, Istanbul, Turkey
Koenraad van Balen, Katholieke Universiteit Leuven, Belgium
Luigia Binda, Politecnico di Milano, Italy
Marcial Blondet, Catholic University of Peru, Peru
Miloz Drdácký, Institute of Theoretical and Applied Mechanics, Czech Republic
Peter Elliott, Elliott & Company, UK
Xianglin Ling Gu, Tongji University, China
Mehrdad Hejazi, University of Isfahan, Iran
Peter James, Cintec International Ltd, UK
Debra Leafer, University College Dublin, Ireland
Giorgio Macchi, Università di Pavia, Italy
Christiane Maierhofer, Fed Inst Material Research and Testing (BAM), Germany
Roberto Meli, Universidad Nacional Autónoma de Mexico, Mexico
John Ochsendorf, MIT, USA
Daniel Oliveira, University of Minho, Portugal
Maurizio Piazza, Universitá Degli Studi di Trento, Italy
Jan G. Rots, Technical University of Delft, The Netherlands
Marek Sklodowski, Institute of Fundamental Technological Research, Poland
Tom Swailes, University of Manchester, UK
Miha Tomazevic, Slovenian National Building and Civil Engineering Institute, Slovenia
Maria Rosa Valluzzi, Universita’ di Padova, Italy
Elizabeth Vintzileou, National Technical University of Athens, Greece
David Yeomans, ICOMOS, UK
XVII
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Conference sponsors
XIX
Experimental studies
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: This paper is devoted to the experimental characterization of the structural behaviour of three-
leaf stone masonry walls. The first part of the experimental results described here was presented during the last
SAHC Conference (Oliveira et al. 2006). In total ten walls, plain and strengthened resorting to transversal tying,
injection and both techniques applied simultaneously, were tested aiming at capturing the detailed structural
behaviour. Globally, all strengthening techniques described here showed to be effective in different ways.
739
The transversal tying technique is aimed at improv- The experimental results described in a previous
ing the connection among leaves, in particular between paper (Oliveira et al. 2006) are here summarized and
the external ones, in order to reduce the transverse all new developments made are described in detail.
deformation. For this purpose, stainless steel bars or
FRP bars can be used. The bars are inserted into 2 CHARACTERIZATION OF WALL
drilled holes through the thickness of walls and then COMPONENTS
anchored. In case of FRP bars, the anchoring can be
achieved by using special anchoring elements (like The mechanical characterization of some of the com-
angle bars or connector developed on purpose) or ponents has been already provided in a previous paper
relying on the bond behaviour between the FRP bar (Oliveira et al. 2006), but for the sake of simplicity, it
and the mortar, developed along the thickness of the will be briefly repeated here.
external leaf. In order to improve this last anchoring
mechanism, a local grout injection around the tie can 2.1 Stone
be applied instead (Oliveira et al, 2006). This tech-
nique can also be applied in combination with other A locally available granite stone was used to build
techniques. all walls. Its mechanical characterization was per-
Note that the aforementioned techniques try to solve formed in cylindrical specimens of dimensions
a problem of localized behavior, i.e. at the element Ø100 × 200 mm2 , where the following average values
level, and not a global behavior problem, i.e. at level were obtained: compressive strength of 52.2 N/mm2 ,
of the structural global behavior (connections between Young’s modulus of 20.6 × 103 N/mm2 and Poisson’s
walls, floors, etc.) that require the application of other ratio of 0.24.
strengthening techniques, which are out of the scope
of this paper. 2.2 Mortar
The work presented here deals with the strength- A 1:3 binder/sand ratio and a 0.8 water/binder ratio
ening of three-leaf stone masonry walls, using dif- were selected (all ratios in weight) in order to obtain
ferent strengthening techniques: transversal tying by a representative mortar composition. The binder was
means of GFRP bars, injection and combination of the composed by 25% of hydrated lime and 75% of
two previous techniques. The main objectives of this metakaolin (pozzolanic material). In addition, a poz-
research are: characterization of the behaviour of zolanic drier (10% on binder weight) was used to
three-leaf walls under different strengthening configu- obtain a faster mortar drying and, therefore, to improve
rations and development of a suitable numerical model the construction procedure of the walls.
able to interpret and explain the structural behaviour The mechanical behaviour of mortar was assessed
of such type of walls. This paper presents the exper- using cubic specimens of 50 × 50 × 50 mm3 sampled
imental results concerning the testing of three series during the construction of the walls and tested under
of walls (ten walls), where the three aforementioned compressive loading at the ages of 7, 28 and 90 days.
strengthening techniques were applied. Strengthening Average compressive strengths of 0.5 N/mm2 ,
was applied without any previous induced damage. 2.9 N/mm2 and 2.2 N/mm2 were measured at the
The test program is summarized in Table 1. aforementioned ages, respectively.
Information about the materials used and mechan-
ical properties of masonry components (external and 2.3 GFRP bar and grout
inner leaves) is also provided.
The transversal tying technique was applied by means
of GFRP bars placed transversally to the wall and
Table 1. Testing of three-leaf walls. anchored along the thickness of the external leaves, by
means of injected grout.The bond strength between the
Wall Wall series Strengthening technique bar, the grout and the masonry was considered enough
to transmit the load from the external leaf to the GFRP
1W1 1 U bar. Furthermore, the low stress state together with
1W2 1 U the high tensile strength of the GFRP bar (a value
2W1 2 U of 760 N/mm2 was provided by the manufacturer)
2W2 2 T excluded its brittle tensile failure.
2W3 2 T
A commercial lime-based grout was used, for both
2W4 2 T
3W1 3 U wall injection and bonding of the GFRP bars to
3W2 3 I masonry. Its mechanical behaviour was assessed by
3W3 3 I means of cubic specimens of 50 × 50 × 50 mm3 sam-
3W4 3 T+I pled during injection. For the walls strengthened with
GFRP bars within series 2, the grout samples pro-
U – Unstrengthened wall/T – Transversal tying/I – Injection. vided an average compressive strength of 17.6 N/mm2
740
and an average tensile strength of 0.3 N/mm2 . For the An average compressive strength of 4.1 N/mm2 was
walls injected within series 3, grout specimens reached attained with a variation coefficient of 12%. The
an average compressive strength of 13.7 N/mm2 (no improvement observed was of about 14 times the
tensile tests were performed). strength of the plain inner leaf. The low variation
coefficient shows that injection allowed for a homog-
2.4 External leaf enization of the inner leaf strength.
In Figure 1b it is shown the crack pattern of a tested
The mechanical behaviour of the external leaf was
prism, where it is possible to observe cracks contouring
characterized through a set of representative stone
the stone elements.
masonry prisms (see Oliveira at al. 2006). An average
compressive strength of 9.2 N/mm2 and a coefficient
of variation of 19% were computed for specimens 3 WALL TEST PROCEDURES
coming from the three series.
3.1 Wall specimens
2.5 Inner leaf
As exposed above, the entire ten wall specimens were
Representative core specimens were also built using built with three leaves, two granite masonry external
granite scabblings poured into alternate layers with leaves and an inner leaf built with granite scrabblings
mortar and avoiding any compaction, aiming at rep- and mortar, without any compaction, aiming at obtain-
resenting the construction procedure followed for ing an amount of voids representative of this kind of
the walls. During the construction of the walls, a walls, able to allow injection. The average thickness
total of ten cylindrical specimens of dimensions of each leaf was about 100 mm, which wasn’t always
Ø150 × 300 mm2 were built. The specimens were possible, due to the variable size and shape of the
tested under uniaxial compressive loading at a dis- stones used. This feature also influenced the number of
placement control rate of 5 µm/s. An average com- courses, which ranged from six to eight courses. The
pressive strength of 0.3 N/mm2 and a coefficient of global dimensions adopted for all wall specimens were
variation of 45% were obtained. As expected, a very 600 mm long, 300 mm thick and 1100 mm high, simi-
low strength was achieved. lar to dimensions found in previous works (Vintzileou
et al. 1995, Toumbakari 2002, Valluzzi et al. 2004)
2.6 Injected inner leaf and aiming at representing 1/2 scaled models of real
walls. It is worth to mention that no stones connecting
The injected inner leaf was also characterized. For that,
the external leaves were used, in order to both assure
seven prisms were sampled from the inner core of the
the worst conditions found in ancient buildings and
injected walls (3W2 and 3W3) during the dismantling
provide leaf interfaces for all specimens as similar as
procedure, see Figure 1a. The samples were extracted
possible.
with average dimensions of 80 × 80 × 160 mm3 (h/d
ratio of 2) in order to obtain representative core 3.2 Strengthening
specimens.
The specimens were tested under uniaxial compres- The transversal tying technique was applied and tested
sive loading at a displacement control rate of 2.5 µm/s. in the second and third series of walls. After drilling
two holes with a diameter of 20 mm, through the
entire wall thickness, at one third and two thirds of
specimen’s height, GFRP bars with 10 mm diame-
ter were placed and the holes were injected with the
aforementioned grout.
The injection technique was applied only in the third
series of walls. Due to the low wall thickness, injec-
tion was applied just in one side of the walls. For that,
the subsequent procedure was followed (see Figure 2):
(a) drilling of slightly inclined holes with 14 mm
diameter. Their distance varied between 100 mm and
200 mm, depending on the location of masonry joints;
(b) introduction of small plastic tubes with a diameter
of 10 mm in the holes; (c) sealing of holes and major
mortar joint absences, caused by the drilling process,
with silicone to prevent grout leakage; (d) injection
Figure 1. Injected inner leaf: (a) general view of a wall’s of water to verify which tubes were active and to
inner leaf; (b) crack pattern of a sampled specimen after wet the inner leaf; (e) sealing of the transversal sides
testing. with polyurethane foam to prevent grout leakage and
741
Table 2. Summary of results of the unstrengthened walls.
fc εa,p E0 E[30–60]
Wall (N/mm2 ) (mm/m) (N/mm2 ) (N/mm2 )
742
Figure 4. Unstrengthened walls: evolution of the λ param- Figure 6. Axial stress – axial strain curves relative to the
eter with regard to the axial stress. transversal tied walls.
2W2 3.3 9.35 1954 722 series, however, the hypothesis of damage induced by
2W3 2.6 5.93 1707 603 the drilling procedure cannot be totally excluded. This
2W4 3.5 9.37 1160 675
feature will be later discussed.
Average 3.1 8.21 1607 667 Table 3 also shows that the existence of the transver-
CV (%) 27 24 25 9 sal ties does not prevent the important stiffness
reduction observed previously.
to plates and the middle height of the wall. Vertical The axial stress – axial strain curves of the tied
cracks contouring the masonry stones are also visible, walls are displayed in Figure 6. For these walls it is
see Figure 5. possible to observe smooth and continuous stiffness
degradation with increasing applied load.
The evolution of the λ parameter with respect to
4.2 Walls strengthened with transversal tying
the applied stress is displayed in Figure 7 for the tied
Results concerning the three walls strengthened with walls. A sudden increase of the variation rate of λ
transversal GFRP bars are summarized in Table 3. occurs for a stress level close to the peak, while for
The presence of the ties allowed a 55% increase the unstrengthened walls, this sudden change occurred
of the average compressive strength in relation to the much sooner.
unstrengthened walls. It is believed that the lower value From above, it can be concluded that the transversal
of the Young modulus of the tied walls (using both tying technique does not prevent stiffness degradation
criteria) is due to differences associated to distinct but makes it happen in a more smooth way.
743
Figure 8. Typical crack pattern of a tied wall (wall 2W4).
fc εa,p E0 E[30–60]
Wall (N/mm2 ) (mm/m) (N/mm2 ) (N/mm2 ) Figure 9. Axial stress – axial strain curves concerning
injected walls.
3W2 3.9 7.00 2172 1604
3W3 3.3 6.87 1844 1114
Average 3.6 6.93 2008 1359
CV (%) 11.0 1 12 26
744
Figure 11. Crack pattern of injected wall 3W3.
fc εap E0 E[30–60]
Figure 13. Combined strengthened wall: evolution of the
Wall (N/mm2 ) (mm/m) (N/mm2 ) (N/mm2 )
λ parameter with regard to the axial stress.
3W4 3.8 4.86 3539 1849
745
biggest improvement was the combination of injection the POCI/ECM/58987/2004 project is gratefully
and transversal tying, with an improvement of 90%, acknowledged.
followed by the injection technique with 80% increase
and finally the transversal tying with 55% increase of
compressive strength. REFERENCES
Due to scattering and differences among series,
it was observed that the Young’s modulus decreased Anzani A., Binda L., Fontana A., Pina Henriques J., 2004,
An experimental investigation on multiple-leaf stone
when shifting from plain to tied walls, which might be
masonry. 13th International Brick and Block Masonry
considered an anomalous behaviour. However, when Conference, Amsterdam, July 4–7, 10 pp., CD-ROM.
comparing walls built in a same series, it is possible to Binda L., Modena C., Baronio G., Gelmi A., 1994, Exper-
observe that, on average terms, strengthening usually imental qualification of injection admixtures used for
causes an increase of the initial stiffness, around 13% repair and strengthening of stone masonry walls. 10th
for tied walls, 19% for injected walls and 127% for International Brick/Block Masonry Conference, Calgary,
the combined strengthened wall. It is worth to men- Canada, Vol. 2, pp. 539–548.
tion that, due to scattering, these figures need further Binda L., Baronio G., Penazzi D., Palma M., Tiraboschi C.,
support based on more experimental results. 1999, Characterization of stone masonry walls in seismic
areas: data-base on the masonry sections and materials
The transversal tying technique promoted the leaves
investigations (in Italian), L’ingegneria Sismica in Italia,
detachment control, which caused changes in the stiff- 9th National Conference, Turin, Italy, 14 pp., CD-ROM.
ness degradation evolution as well as in the crack Binda L., 2006, The difficult choice of materials used for
pattern, with the absence of major horizontal cracks. the repair of brick and stone masonry walls, 1st Inter-
In the same way, injection changed the behaviour of national Conference on Restoration of Heritage Masonry
the walls, allowing all leaves to work together till near Structures, Cairo, Egypt, CD-ROM.
peak load, by increasing both the connection between ICOMOS, 2001, Recommendations for the analysis, con-
leaves and the inner leaf strength, which was noticeable servation and structural restoration of architectural her-
by the diffuse crack pattern and by the presence of big itage, International Scientific Committee for Analysis and
Restoration of Structures of Architectural Heritage.
cracks in the inner leaf.
Modena C.,1997, Criteria for cautious repair of historic build-
The results obtained for combined strengthened ing. Evaluation and strengthening of existing masonry
wall seem to show that this technique gathered the ben- structures, Binda L. and Modena C., Ed. RILEM.
efits of both techniques applied separately, however Oliveira D.V., Lourenço P.B., Garbin E., Valluzzi M.R.,
more tests are need to confirm these aspects. Modena C., 2006, Experimental investigation on the struc-
Finally, it is important to refer that scattering is a tural behaviour and strengthening of three-leaf stone
key issue when dealing with historical constructions masonry walls, Structural Analysis of Historical Con-
as well as with natural and handmade materials. structions, New Delhi.
Toumbakari, E.E., 2002, Lime-pozzolan-cement grouts and
their structural effects on composite masonry walls.
Ph.D. Thesis, Katholieke Universiteit Leuven, Heverlee,
ACKNOWLEDGEMENTS Belgium, pp. 364.
Valluzzi M.R., da Porto F., Modena C., 2004, Behaviour and
The authors would like to thank the technical staff at the modeling of strengthened three-leaf stone masonry walls,
Laboratory of Civil Engineering, University of Minho, Materials and Structures, Vol. 37, pp. 184–192.
Valluzzi M.R., Binda L., Modena C., 2005, Mechanical
for the help provided.
behaviour of historic masonry structures strengthened
Acknowledgements are also due to the companies by bed joints structural repointing, Construction and
Fradical, Mapei and Augusto de Oliveira Ferreira for Building Materials, Vol. 19, pp. 63–73.
providing raw materials and workmanship. Vintzileou E., Tassios T.P., 1995, Three-Leaf Stone Masonry
The financial support provided by the Por- Strengthened by Injecting Cement Grouts, Journal of
tuguese Science and Technology Foundation through Structural Engineering, pp. 848–856.
746
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
S. Ignoul
Triconsult NV, Lummen, Belgium
M. Wevers
Department of Metallurgy and Materials Engineering, KULeuven, Heverlee, Belgium
ABSTRACT: Within the framework of some dramatic collapses of historical masonry structures, a test program
has been set up including short-term and long-term creep testing of masonry, composed with different types of
mortar. These experimental creep tests are very adequate to gather more insight into the described phenomenon,
but other tools are necessary in order to evaluate the damage evolution of historical masonry. Therefore, acoustic
emission detection has been used as a non destructive technique in order to evaluate the damage evolution during
experimental creep testing. This paper discusses the data obtained during the creep tests and first results of
the use of the acoustic emission technique for detecting the damage evolution within masonry under persistent
compressive load, based on laboratory tests on small specimens.
1 INTRODUCTION that the masonry was heavily loaded and in bad condi-
tion. This crack pattern was found on several of the
During the past decades, experimental research has supporting elements of the tower and a monitoring
been carried out to get better insight into the long-term campaign was set up, including the monitoring of a
behaviour of historical masonry structures. The initial number of cracks by means of a mechanical strain
motivation for the rising interest within this field of gauge.
research were a few dramatic collapses of historical Within a period of two weeks, some crack openings
masonry buildings (Civic Tower of Pavia, 1989; Noto showed an increase of 1 mm and more and the advice
Cathedral, 1996) (Binda 1992). Recent collapses of was given to close the church and to start immedi-
masonry buildings in Belgium confirm the importance ate strengthening and repair actions. Just two weeks
of this research. later, while the technical staff (who was removing valu-
able sculptures before the restoration) was having a
lunch break, the tower collapsed. Fortunately, making
1.1 Bell tower of the Sint-Willibrordus church, no casualties.
Meldert, Belgium
The bell tower of the Sint-Willibrordus church col-
1.2 Medieval tower “Maagdentoren”,
lapsed on 7th July 2006 (Fig. 1). The three-leaf
Zichem, Belgium
masonry of the tower was composed of two outer leafs
in sandstone and an inner core with rubble masonry of The Maagdentoren partially collapsed on 1st of June,
smaller sandstone chunks and large amounts of lime 2006 under comparable circumstances. Again, no sud-
mortar. The red-brown, rather porous sandstone con- den disturbance of the acting forces, such as an
tains iron oxide, which is responsible for its colour. earthquake, could be pointed out as having caused the
The sandstone is known for having a large scatter on its collapse. The tower was also composed of three-leaf
strength characteristics and easy absorption of water, masonry in sandstone, which appeared to be in a bad
making it rather vulnerable for weathering influences. condition. A high dead load acted on the base of the
A visual inspection of the supporting corner pil- tower, especially in the area where the cross-section
lars of the church tower, carried out before collapse, of the wall was reduced by a staircase opening. It is at
showed a large amount of vertical cracks, indicating this point that the wall collapse was triggered.
747
These experimental creep tests are very adequate to
gather more insight in the described phenomenon, but
other tools are necessary in order to evaluate the dam-
age evolution of historical masonry on site. Therefore,
the acoustic emission technique has been used as a non
destructive technique for monitoring the damage evo-
lution during experimental creep testing. These tests
are discussed in the second part of the paper.
748
Figure 3. Strain-time graph of accelerated creep tests on Figure 4. Volumetric strain-time graph of accelerated creep
three hydraulic lime mortar wallets. tests on three hydraulic lime mortar wallets. Positive volu-
metric strain indicates volume increase.
2.1 Monotonic compression tests
Only during the last load step of each test, also the
Compression tests were performed in strain-controlled third creep phase (increasing strain rate) was reached.
conditions to gather information on the compressive At two successive steps, the strain rate of the sec-
strength of the different masonry types. Based on these ondary phase increased, indicating a relation between
ultimate strengths, the stress path to be followed dur- the sustained load level and the strain rate.
ing accelerated and long-term creep tests could be The evolution of the volumetric strain initially
calculated. shows a volume decrease. During the last stages of the
tests, the volumetric strain becomes positive, indicat-
2.2 Accelerated creep tests ing the dilatancy behaviour of the masonry just before
Accelerated, short-term creep tests were performed to failure occurs.
assess the damage accumulation in masonry under
sustained stresses. The aim is to obtain data during 2.3 Long-term creep testing
a short-term test, which can be used to deduce the In order to assess the long-term behaviour of masonry
parameters, necessary to simulate and assess the strain under persistent loading, 9 long-term creep tests were
and long-term damage evolution, for example during set up. For each mortar type, respectively hydraulic
the life-time of a masonry structure. lime mortar, cement mortar and lime-cement mortar,
Three accelerated creep tests were performed for three masonry wallets were tested. These three wallets
each type of mortar, starting at a stress level of were initially loaded at 50, 65 and 80% of the compres-
50–60% of the compressive strength and increasing sive strength, obtained by averaging the results of the
with stress increments of 5% until failure of the spec- monotonic compression tests. These long-term tests
imen. In between two stress increments, a constant are still ongoing.
stress level was maintained during a period of at least
2 hours, but even then, failure often occurred during
the stress increment and no tertiary creep phase was 3 ACOUSTIC EMISSION TECHNIQUE
reached.
The strain-time graphs of the accelerated creep tests The acoustic emission technique is a non-destructive
on the columns with hydraulic lime mortar are pre- technique which detects and locates damage at the
sented in Figure 3. During one of the tests, the stress moment of occurrence. Acoustic emissions (AE) are
level could not be kept constant at the end of the test- high frequency transient sound waves, which are emit-
ing period due to a technical problem (indicated with ted during local stress redistributions caused by struc-
arrow in Figure 3). tural changes, such as crack growth. The technique is
The time increment of two hours is relatively short proposed here as a monitoring technique for the detec-
in order to reach long-term creep effects, but the tests tion of damage initiation and assessment of the rate of
still indicated some interesting results. During each damage evolution during creep deformation.
load step, a primary (decreasing strain rate) and a sec- When a set of several sensors is applied for the
ondary (constant creep rate) creep phase was reached. acoustic emission measurement, localisation of the
749
surrounding test equipment to name a few. Therefore,
the necessary precautions have to be taken in order to
keep these boundary conditions as much constant as
possible. Concerning this remark, it also follows that
not the absolute amount of detected events, but rather
the change in detection level or event detection rate is
a determining factor for the assessment of the damage
accumulation. The experimental research, discussed
below, indicates that the event rate is related to the rate
of damage increase within the masonry.
750
(a) (b)
751
Figure 8a. Stress-strain curve for a cyclic accelerated creep
test on masonry cores.
752
Figure 11. Evolution of elastic modulus and AE counts
during successive load steps for test core 4.
753
The first results of the use of the acoustic emission
technique for detecting the damage evolution within
masonry under increasing and persistent compressive
load were discussed. These results have shown that
the rate of detected acoustic emission can be related
to the rate of damage evolution. As the time incre-
ments were too short to reach secondary creep during
all the load steps, further testing is necessary includ-
ing also long-term testing in order to evaluate the
use of AE technique to monitor damage accumula-
tion at constant stress level. Even so, these first results
prove that the acoustic emission technique is a valu-
able tool in assessing damage evolution of masonry
under persistent compressive load.
ACKNOWLEDGEMENTS
Figure 13. AE counts per second in function of stress level
during constant stress for two cyclic accelerated creep tests.
The stress is indicated relatively in relation to the maximum The authors express their thanks to the Flemish Fund
stress. for Scientific Research (FWO) for the doctoral grant,
offered to Els Verstrynge.
this behaviour. During the last steps, the AE count
increases for both cyclic accelerated creep tests pre- REFERENCES
sented. This means that the damage accumulation rate
during constant stress depends on the load level. The Anzani, A., Binda, L., Melchiorri, G. 1995. Time dependent
stress in Figure 13 is indicated relatively in relation to damage of rubble masonry walls, 4th Int. Masonry Conf.,
the maximum stress obtained during the creep tests, London, Vol 2 (7): 341–351.
which is indicated as 1. Binda, L., Gatti, G., Mangano, G., Poggi, C., Landriani, G.S.
As not all constant stress intervals had an equal 1992. The Collapse of the Civic tower of Pavia: A survey
of the materials and structure. Masonry Int. Vol 6 (1):
duration, the AE count rate (counts per second) is indi- 633–642.
cated in stead of the total amount of counts during a Binda, L., Anzani, A. 1993. The time-dependent behaviour of
period of constant stress. masonry prisms: an interpretation. The masonry society
During the last load step, a very high, unstable dam- journal. Vol 11 (2): 17–34.
age growth is detected which leads to failure of the Carpinteri, A., Lacidogna, G. 2007. Damage evolution of
specimens. three masonry towers by acoustic emission. Engineering
structures 29: 1569–1579.
Colombo, S., Main, I.G., Forde, M.C. 2003. Assessing dam-
4 CONCLUSIONS age of reinforced concrete beams using b-value analysis
of acoustic emission signals. Journal of materials in civil
engineering. 15 (3): 280–286.
An extensive test program was set up to evaluate the
Eberhardt, E., Stead, D., Stimpson, B., Read, R.S. 1997.
long-term behaviour of masonry under monotonic and Changes in acoustic event properties with progressive
sustained loading. This test program involved short- fracture damage. Int. J. Rock Mech. & Min. Sci. Vol 34
term monotonic compression and accelerated creep (3–4), paper nr. 071B.
tests and long-term creep tests on masonry wallets with Grossi, C.M., Esbert, R.M., Suarez del Rio, L.M., et al. 1997.
three types of mortar compositions. The results of this Acoustic emission monitoring to study sodium sulphate
test program are presented briefly. crystallization in monumental porous carbonate stones.
These experimental creep tests are very adequate to Studies in conservation. Vol 42: 115–125.
gather more insight into the described phenomenon, Tomor, A.K., Melbourne, C. 2007. Monitoring masonry
arch bridge response to traffic loading using acoustic
but other tools are necessary in order to evaluate the
emission techniques. Proc. of the 5th int. conference on
damage evolution of historical masonry. Therefore, arch bridges, Madeira, Portugal, 12–14 september 2007:
acoustic emission has been used as a non destruc- 281–288.
tive technique in order to evaluate the damage evo- Verstrynge, E., Ignoul, S., Schueremans, L., Van Gemert, D.
lution during experimental creep testing on small 2008. Modelling of damage accumulation in masonry sub-
masonry cores. The tests were performed consider- jected to a long-term compressive load. Proc. of the 6th int.
ing the knowledge gained during previous creep tests seminar on structural analysis of historical constructions.
on masonry and therefore, both accelerated creep tests Bath 2–4 July 2008.
and cyclic accelerated creep tests were performed.
754
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: In the paper, the assessment of the in-plane behavior of timber floor samples is proposed by
experimental laboratory tests, aimed at characterizing the basic properties of simple construction floor typologies,
common in existing masonry buildings, for the proper selection of structural improvement solutions. On the basis
of the analysis of the testing procedures and systems proposed in literature, a specific set-up has been designed
and realized, in order to minimize undesirable and/or uncertain effects, able to simulate both monotonic and
cyclic shear behaviour. The first experimental results on unreinforced floors are discussed, together with the
calibration of a preliminary FE model.
755
et al. (2004), a monotonic and cyclic bending test
set-up on horizontal floor samples, treated as simple
supported or double fixed ends beam, was presented.
This procedure enables the investigation of the shear
behaviour of the floor as a part of the flexural defor-
mation, as in the beam model of Timoshenko (1934).
By this system it is possible to perform tests without
the disadvantage above, but it requires specimens in
real scale with large dimensions (about 4 × 8 m, as in
Peralta et al. 2004).
Within this context, an experimental program has
been planned at the University of Padua, in order to
evaluate the shear behaviour of original and strength-
ened timber floors, by using traditional materials. A
new test set-up has been designed and realized, in
order to perform both monotonic and cyclic tests, and
by minimizing many undesired effects and the dimen-
sional limitations. The system has been preliminary
tested with two pilot floor samples, made by five
beams and two types of boarding (raw-finished and
tongue-and-groove shaped in the thickness). Follow-
ing this preliminary phase, one of the two boarding Figure 1. Design of test set-up and structural scheme
systems will be used as basic sample for various provided by steel frame.
strengthened conditions (double-boarding, diagonal
metal stripe or timber board, etc.). The steel articulated quadrilateral was designed
In the paper, the description of the test set-up and such that out-of-plane motion at the specimen bound-
the results of the preliminary pilot samples are pre- aries (bottom, lateral and top) is prevented, as it
sented. The local behaviour measured during the tests happens in buildings. On the contrary, it allows uplift
has been used to calibrate a numerical FE model, able and in-plane deformations of the sample, by minimiz-
to reproduce the global behaviour of the floor, to be ing frictional effects. Floor specimens are inserted in
used for predictive analysis on the further planned the steel frame with the beams placed horizontally,
tests. The validation of the system is also discussed on thus allowing the free rotation of the boarding.
the basis of reliable drifts expected in masonry build- The bottom joist is fully fixed to the bottom steel
ings subjected to horizontal cyclic loads, as available reaction beam by means of bolts. The steel reaction
in literature. beam is connected to the steel basement trough three
load cells by means of mechanical sliding connections,
thus providing the three simple supports. In the hori-
2 EXPERIMENTAL PROCEDURE zontal direction, a hydraulic actuator was used, to load
the specimens at the level of the top beam.Tests are per-
2.1 Test set-up formed in displacement control, and the force applied
For real scale floor specimens, a specific testing at the top of the samples is measured through a load
machine was on purpose designed and realized, to cell. Finally, horizontal and vertical load cells apply the
adequately simulate the monotonic and cyclic shear shear and the bending moment to the floor specimen
behaviour of floors. in its plane.
Starting from the typical advantages of the vertical In Figures 1 and 2, the main characteristics of the
diagonal test, the proposed system consists in a ver- test system and some views of the experiment are
tical shear-bending test similar to the configurations given.
used for testing woodframe shearwalls (Gatto & Uang A suitable instrumentation was placed on both sides
2003). The structural scheme provides three simple of each specimen, to capture the global and local
supports applied at the floor base (Figure 1); it is equiv- shear-bending behaviour. Several transducers were
alent to a cantilever beam, which can apply bending used to record horizontal, vertical and diagonal relative
and shear deformations, as set in the Timoshenko beam displacements (Figure 3).
theory (1934). Thanks to this structural scheme, the
geometrical dimensions of the floor specimens could
2.2 Floor specimens
be almost half of those used in Peralta et al. (2004),
and about a quarter of those usually present in existing The pilot tests were performed on two timber floor
masonry houses. elements (2,2 × 2,2 m), namely F1.M and F2.M, built
756
Figure 2. View of testing machine with sample, and detail
of connection of base beam to frame.
757
Table 1. Results obtained from tested connections.
758
Figure 10. Comparison among load displacement curves of Figure 11. Comparison between in-plane floor and in-plane
specimens F1.M and F2.M, and schematization proposed by masonry wall deformations.
UNI EN 12512.
and an ultimate stiffness respectively 256% and 42%
Table 2. Results of shear test on floor specimens. higher than those provided by F1.M.
A first estimation of the effect of the in-plane defor-
F1.M F2.M mation of timber floors in the behaviour of masonry
subjected to lateral actions can be also suggested,
Fmax (30 mm) (N) 1047 1435
Fy,est (N) 774 901
by considering a simplified case, represented by a
Vy,est (mm) 8,60 2,67 masonry box having common dimensions of walls
Ki = tgα (N/mm) 80,93 288,29 (4 m long and 3 m high, as in Figure 11) and covered
tgβ (N/mm) 13,49 19,22 by a plain floor. The estimation of drifts (δ) normal-
tgα/tgβ – 6 15 ized to the height of the wall (H) for masonry piers
available from literature, led to the following values:
0,2 ÷ 0,3% is the common range for the achievement
samples is graphically compared in Figure 10. Both of the first cracking, 0,4% for the shear cracking,
specimens showed a failure characterized by a relative 0,5 ÷ 0,6% for the attainment of the maximum load,
rotation between the boards and the beams (Figure 9), whereas 0,8% refers to the rocking behaviour (da Porto
which involved the resistant mechanism of the moment 2005, Tomaževič et al. 1996, Tomaževič & Lutman
of the couple provided by the pair of nails fastened at 1996, Tomaževič 1999).
each beam. By dividing the shear load by the span length
The contribution of additional friction due to the (L = 4 m), and referring it to the ratio between the
tongue-and-grove connection during the relative slip maximum relative floor shear displacement (δf ) and
at the board sides, influenced the better performance of the height of the wall (H = 3 m), see box in Figure 11,
the F2.M sample, in comparison with the F1.M one. By the previous limits become: 0,23% for first cracking;
considering the conventional displacement of 30 mm 0,30% for shear; 0,45% for maximum load; and 0,60%
as reference (UNI EN 12512), an increment of strength for rocking. It can be seen how the already poor capa-
of 37% was found (Figure 10 and Table 2). bility in redistributing the horizontal seismic forces is
To analyze the results, by assuming that monotonic worsened by the loss of shear stiffness in correspon-
curves usually constitute the envelope of cyclic tests, dence of the damage zone of masonry walls, when the
for general timber structure built with metal fasteners, redistribution needs are very desirable to assure a good
the UNI EN 12512 standard can be also used (Ceccotti seismic behaviour (Figure11).
et al. 2005). Since the F2.M sample showed better performances
The results are summarized in Table 2 and repre- than the F1.M one, its results will be considered for
sented in Figure 10, where symbols are as follows: the calibration of a FE numerical model, as described
Fmax is the maximum load at 300 mm of displacement, in the following. Moreover, the boarding with tongue-
Fy,est is the estimated load at yielding, Vy,est is the dis- and-groove connection will be used as basic type for
placement correspondent to Fy,est , and α and β are the the planned strengthened cases.
angles of the slope of first and the second branch of
the rectified curve, respectively, built as described in
the box of Figure 10. 4 NUMERICAL ANALYSIS
Also in terms of yielding load the F2.M sample per-
formed better (Table 2), with an increment of 16% A simplified numerical model was developed in order
of the reference load. Moreover, it showed an initial to identify the behaviour of the plain sample F2.M.
759
Figure 12. Numerical model: (a) detail of connection
between board and beam, (b) general view of frame.
Figure 14. Comparison between experimental and numeri-
cal data for the two big diagonal transducer 7 and 8 located
on the boarding side.
760
A simplified numerical model, calibrated with the
experimental results, has shown a very good agree-
ment with the experiments, both at local and global
level, and will be considered for the design and the
predictive analysis on the further samples.
The research will prosecute with monotonic and
cyclic tests on timber floor panels strengthened with
different methods (double boardings, metal diagonal
strips or diagonal board, etc.).
ACKNOWLEDGEMENTS
REFERENCES
Abrams D.P., Tena-Colunga A. (1994). Structural evaluation
of low-rise masonry buildings with flexible diaphragms
(a) subjected to earthquakes, X IB2MaC, Calgary, Canada.
Barbisan U., Laner F. (1995). I solai in legno. Franco Angeli
Ed.: Milano, Italy.
Binda L., Cardani G., Saisi A., Valluzzi M.R. (2006). Vul-
nerability analysis of the historical buildings in seismic
area by a multilevel approach. Asian Journal of Civil
Engineering, Vol. 7, n. 4, pp. 343–357.
Ceccotti, A., Follesa, M., Lauriola M.P. 2005. Le strutture di
legno in zona sismica. CLUT: Torino, Italy.
Cohen G.L., Klingner R.E., Hayes J.R., Sweeney S.C. (2001).
Seismic Response of Low-Rise Masonry Buildings With
Flexible Roof Diaphragms, US Army Corps of Engineers.
Corradi, M., Speranzini, E., Borri, A., Vignoli, A. 2006. In-
plane shear reinforcement of wood beam floors with FRP.
Composites: Part B, Iss. 37, pp. 310–319.
da Porto, F. 2005. In-plane cyclic behaviour of thin layer joint
(b) masonry walls. Ph.D. Thesis, University of Trento: Trento,
Italy.
Figure 17. Principal stresses at 30 mm displacement: Foraboschi P. (2005). Consolidamento statico e adeguamento
(a) tension, (b) compression. sismico. Annotazioni per le costruzioni in muratura,
L’Edilizia, n. 138, Italy.
Galasco A., Lagomarsino S., Penna A. (2001). Analisi sis-
Monotonic tests on two plain floor samples have mica non lineare a macroelementi di edifici in muratura,
ANIDIS 2001, Italy.
evidenced the better performances of tongue-and- Gattesco, N., Marcorini, L. 2006. Strengthening and Stiff-
grove connections among boards. The analysis of ening Ancient Wooden Floors with Flat Steel Profiles,
results pointed out the need of strengthening and 5th Structural Analysis of Historical Constructions, Proc.
stiffness of the horizontal structures, when floors are intern. symp., Delhi, India, 6–8 November 2006. India:
considered restrained as in common buildings. MacMillan, pp. 405–412.
761
Gattesco, N., Macorini L., Benussi F. 2007. Intervento sui OPCM 3431 (2005). Ordinanza 3431/2005 del 3 Maggio
solai lignei per l’adeguamento sismico di edifici storici 2005: Norme tecniche per il progetto, la valutazione e
con tecniche caratterizzate da elevata reversibilità, ANI- l’adeguamento sismico degli edifici. P.C.M., Italy.
DIS 2007, Italy (CD-Rom). Piazza, M., Tomasi, R., Modena, R. 2005. Strutture in Legno.
Gatto, K. & Uang, C.M. 2003. Effects of Loading Protocol on Ulrico Hoepli: Milano, Italy.
the Cyclic Response of Woodframe Shearwalls. Journal of Peralta, D.F., Bracci, J.M., Hueste, M.B.D. 2004. Seismic
Structural Engineering, Vol. 129, n. 10, October 1, 2003. behaviour of wood diaphragms in pre-1950s unrein-
ASCE, pp. 1384–1393. forced masonry buildings. Journal of Structural Engi-
Giordano, G. (1999). Tecnica delle costruzioni in legno. neering, Vol. 130, No. 12, December 1, 2004, ASCE,
Hoeply: Milan, Italy. pp. 2040–2050.
Giuffrè A. (1993). Sicurezza e conservazione dei centri prEN 1995-1-1 (Final Draft, Dec. 2003): Eurocode 5. Design
storici: il caso di Ortigia, Laterza: Bari, Italy. of timber structures - Part 1-1: General–Common rules
Giuriani, E, Plizzari G.A. 2003. Studio sperimentale sul com- and rules for buildings.
portamento di solai in legno rinforzati con lastre di acciaio Sorrentino, L., Monti, G., Kunnath, S., Scalora, G. 2007.
per resistere alle azioni sismiche, IV Workshop Italiano Un modello meccanico semplificato accoppiato nel
sulle Strutture Composte, Italy. piano-fuori del piano per valutare il ruolo di solai, immor-
Giuriani, E. 2004. L’organizzazione degli impalcati per gli sature, qualità muraria e muri di controvento, ANIDIS
edifici storici. L’Edilizia, Speciale Legno Strutturale, n. 2007, Italy (CD-Rom).
134, pp. 30–43. Tampone, G. 1996. Il restauro delle strutture di legno. Ulrico
Kim S., White D.W. (2004). Nonlinear analysis of a one- Hoepli: Milano, Italy.
story low-rise masonry buildings with flexible diaphragms Timoshenko S. 1934. Theory of Elasticity. McGraw-Hill:
subjected to seismic excitation, Engineering Structures, n. New York, US.
26, Elsevier. Tomaževič M. (1991). The influence of rigidity of floors on
Lee D., Kim H., Chun M. (2002). Efficient seismic analysis the seismic behaviour of old stone-masonry buildings,
of high-rise building structures with the effect of floor European earthquake engineering, Vol. 5, n. 3.
slabs, Engineering Structures, n. 24, Elsevier. Tomaževič, M. 1999. Earth quake resistano design of
Liberatore D., Perillo G., Spera G. (2004). L’intervento a solai masonry buildings. Imperial College Press: London, UK.
deformabili per il miglioramento sismico di strutture in Tomaževič, M., Lutman, M., Petkovic, L. 1996. Seismic
muratura, ANIDIS 2004, Italy. behaviour of masonry walls: Experimental simulation.
Mariani, M. 2004. Consolidamento delle strutture lignee con Journal of structural Engineering, Vol. 122, n. 9, 1996,
l’acciaio, DEI srl Tipografia del Genio Civile, Roma, ASCE, pp. 1040–1047.
Italy. Tomaževič, M. & Lutman, M. 1996. Seismic behaviour
Mezzina M., Dentamaro C. (1997). Adeguamento e miglio- of masonry walls: Modelling of hysteretic rules. ASCE
ramento degli edifici in muratura ordinaria, ANIDIS 1997, Journal of structural Engineering, Vol. 122, n. 9, 1996,
Italy. pp. 1048–1054.
Modena, C. 1997. Criteria for cautious repair of his- UNI EN 26891. 1991. Timber structures. Joints made with
toric building. A valuation and strengthening of existing mechanical fasteners. General principles for the determi-
masonry structures, Binda L. and Modena C., Ed: RILEM. nation of strength and deformation characteristics. UNI:
Modena, C., Tempesta, F., Tempesta, P. 1998. Il recupero a Milano, Italy.
secco di impalcati in legno. L’edilizia 3/4, pp. 38–45. UNI 9091-1. 1987. Wood. Determination of moisture content.
Modena, C., Valluzzi, M.R., Garbin, E., da Porto, F. 2004. Electrical method. UNI: Milano, Italy.
A strengthening technique for timber floors using tra- UNI EN 12512. 2006. Timber structures. Test methods.
ditional materials, 4th Structural Analysis of Historical Cyclic testing of joints made with mechanical fasteners.
Constructions, Proc. intern. symp., Padova, Italy, 10–13 UNI: Milano, Italy.
November 2004. Rotterdam: Balkema. pp. 911–921. Valluzzi M.R., Michielon E., Binda L., Modena C. (2001).
Moon S.K., Lee, D.G. (1994). Effect of inplane floor flex- Modellazione del comportamento di edifici in muratura
ibility on the seismic behaviour of building structures, sotto azioni sismiche: l’esperienza Umbria – Marche,
Engineering Structures, vol. 16, n.2. ANIDIS 2001, Italy (CD-Rom).
762
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: In the last decade, innovative technologies have been developed using Fiber Reinforced Polymer
(FRP) as strengthening and repair of masonry structures. Bond of FRP to substrate is crucial for the effectiveness
of the technique especially to masonry substrate, which can have a wide variability. Few contributions are available
concerning debonding problems on masonry. In this paper the main results of an experimental campaign on the
local behaviour of externally bonded FRP, applied on clay bricks, are presented. Double-lap Push-pull Shear Tests
have been performed by using carbon and glass fiber reinforcement. Experimental results, in terms of failure load,
have been compared with predictive bond-strength models proposed in literature, mainly available for concrete.
Based on the measured strength, interface fracture energy has been calibrated. A simplified analytical model,
fitted on the experimental data, has been proposed as bond-slip law. Finally, a bilinear function, as commonly
adopted by some guidelines, has been calibrated.
763
Table 1. Bricks mechanical properties.
764
Figure 5. Experimental failure loads per unit width.
Figure 4. Brick and composite surfaces after the test. elastic behaviour of the composite, it was also possible
to evaluate Young’s modulus of the reinforcement for
each sample through a best fitting (Eq. 1):
Table 3. Experimental results for carbon reinforcement.
Ef Pu Pu /2bf σu
Specimen MPa N N/mm MPa
ShC1 164,419 31,884 318.8 1932 where σ = nominal tensile stress, P = axial load
ShC2 336,439 34,233 342.3 2075 on the composite strips, bf = single strip width,
ShC3 284,991 35,325 353.3 2141 t = equivalent textile thickness, ε = axial strain mea-
ShC4 277,511 39,210 392.1 2376 sured by the strain-gauge and Ef = composite elastic
ShC5 338,456 40,301 403.0 2442
modulus.
Mean value 280,363 36,191 361.9 2193 Tables 3–4 also report the composite elastic mod-
Stand. dev. 70,696 3505 ulus values, Ef , and the nominal tensile stresses at
COV 25.2% 9.7% failure, σu . The experimental mean elastic moduli
result higher than the producer’s values (22% for car-
bon reinforcement and 25% for glass one). The mean
Table 4. Experimental results for glass reinforcement. maximum stress reached by reinforcement is around
64% of tensile strength for carbon, and 68% for glass.
Ef Pu Pu /2bf σu As the reinforcement axial stiffness per unit width,
Specimen MPa N N/mm MPa Ef tf , was known for each specimen, failure loads per
unit width, P/2bf , were tabulated versus the axial stiff-
ShG1 50,934 23,380 233.8 1017 ness. Trend lines were fitted, referring to all data or to
ShG2 87,014 27,940 279.4 1215
each set (carbon and glass). The expression adopted
ShG3 80,545 27,300 273.0 1187
ShG4 102,598 26,400 264.0 1148 for the trend lines is given in Equation 2:
ShG5 84,842 28,360 283.6 1233
Mean value 81,035 26,676 266.9 1160
Stand. dev. 18,817 1985
COV 23.2% 7.4%
where c1 e c2 are regression constants, which values
are reported in Table 5. It can be observed that the
It was assumed that strain is uniform on the com- fitting of all data shows a better correlation than the
posite cross section, and that is possible to refer fitting of each single set.
mechanical properties to the dry woven: this approach By assuming the relationship between the failure
is accepted by many authors (Chen & Teng 2001) load and the square root of the axial stiffness per unit
and by some guidelines, such as Italian CNR-DT width (Eq. 7), it was possible to reduce the number
200/2004. of free parameters in Equation 2, by imposing the
Therefore, it was possible to calculate the nominal exponent value c2 equal to 0.5. Results (carbon data
tensile stress on the textile, defined as the load and set, glass data set and all data) are given in Table 5,
cross section area ratio. By coupling the strain mea- whereas Figures 6–7 compare the trend lines with the
sures (obtained from the strain-gauge on the unbonded experimental data. Regression coefficient for GFRP
region) with the load values, and by assuming the linear are slightly higher than CFRP (around 16%) and this
765
Table 5. Regression constants of load vs axial stiffness Table 6. Predictions of failure load.
(both per unit width) trend lines.
CFRP GFRP
Data set c1 c2 R2
Model Pu /2bf Error Pu /2bf Error
All data 12.876 0.310 0.876 N/mm N/mm
Glass data only 27.197 0.234 0.622
Carbon data only 35.063 0.218 0.439 Tanaka 166 −54.0% 166 −37.6%
All data (Square Root) 1.759 0.5 n.a. Hiroyuki and Wu 158 −56.2% 158 −40.6%
Glass data (Square Root) 1.953 0.5 n.a. Maeda 254 −29.7% 174 −34.9%
Carbon data (Square Root) 1.681 0.5 n.a. Khalifa 248 −31.4% 170 −36.2%
Yang 192 −47.0% 143 −46.3%
Sato 415 +14.6% 116 −56.5%
Iso 271 −25.1% 161 −39.5%
Izumo 557 +53.9% 224 −15.9%
Neubauer and R. 283 −21.8% 179 −32.7%
Chen and Teng 245 −32.2% 156 −41.6%
Monti et al. 321 −11.3% 204 −23.6%
Lu et al. Bilinear 220 −39.3% 139 −47.7%
Brosens and V. G. 359 −0.9% 228 −14.7%
CNR−DT 200 263 −27.4% 167 −37.5%
Nakaba et al. 350 −3.3% 222 −16.7%
Savoia et al. 328 −9.4% 208 −22.0%
Neubauer and R. 266 −26.4% 169 −36.6%
Dai and Ueda (1) 326 −9.9% 207 −22.5%
Figure 6. Failure loads per unit width versus reinforcement Dai and Ueda (2) 322 −11.0% 202 −24.2%
axial stiffness per unit width: experimental data and trend Lu et al. Precise 220 −39.3% 139 −47.7%
lines. Lu et al. Simplif. 220 −39.3% 139 −47.7%
Mean experim. 362 – 267 –
766
where bf = reinforcement width, Gf = interface frac-
ture energy, αT = constant value, Ec tc = axial stiffness
per unit width of the concrete substrate.
Yuan (Chen & Teng 2001) proposed a modified con-
stant value (Eq. 6) that takes into account the width (bf
and bc ) ratio of the bonded materials:
Gf from Eq. 7 Gf from S.R. fit By applying these formulas to the experimental data of
Reinforcement type N/mm N/mm this work, it emerges that taking or not into account the
parameters αT or αW leads to a difference lower than
Carbon fibers 1.42 1.41 2%. It has to be noticed that Equation 7, demonstrated
Glass fibers 1.91 1.91 in some cases (Wu et al. 2002; Dai et al 2005), is
assumed in every case of regular interface law (Savoia
et al. 2003b).
It is considered significant also for the clay substrate
reinforcement, underestimate the mean experimental adopted in the present work.
failure load. Moreover, all formulations provide an Accordingly with Equation 7, it was possible to cal-
estimation closer to test results in case of CFRP than ibrate fracture energy Gf through the mean values of
GFRP, except for the Tanaka and Hiroyuki models. failure load and elastic modulus; results are reported in
However, results show large differences from model Table 6. The estimated value, for glass reinforcement,
to model: they vary between 44% and 154% of exper- is around 35% higher than carbon one.
imental mean failure load, for carbon reinforcement, Moreover, the fitting parameter c2 given in Table 5,
and between 43% and 85%, for glass one. where c1 was imposed equal to 0.5 (Square Root
based fitting), allowed to evaluate Gf , as shown in
Equation 8. The results (Table 6) show no signifi-
cant difference from values obtained by means of
5 FRACTURE ENERGY CALIBRATION
Equation 7.
The interface fracture energy mode II, Gf , is defined
by Equation 4 as the definite integral of the tangential
stress, τ, expressed as function of the mutual slip of
composite and substrate, s:
6 CALIBRATION OF A BOND-SLIP LAW
767
Figure 9. Some bond-slip law shapes available in literature.
768
It is useful to rewrite the law, herein labeled UniPd Table 8. Significant values for local bond of CFRP.
curve, in a normalized form (Eq. 16).
Gf τmax s0 sf
Curve N/mm MPa mm mm
where s0 = 1 / B and τmax = τ (s0 ) are the coordinates Monti et al. 1.11 5.37 0.046 0.415
of the point of maximum tangential stress. Lu et al. Bilinear 0.52 3.73 0.048 0.280
Brosens and V. G. 1.39 2.71 0.012 1.025
After the optimization of the Unipd curves, in case
CNR 0.75 7.46 0.056 0.200
of carbon reinforcement and glass one, it was possible Nakaba et al. 1.32 7.08 0.065 –
to calibrate a bilinear law, whose analytical form is Savoia et al. 1.16 7.08 0.051 –
reported in Equation 17. Neubauer and R. 0.77 5.69 0.270 –
Dai and Ueda (1) 1.15 8.58 0.103 –
Dai and Ueda (2) 1.12 6.41 0.061 –
Lu et al. Precise 0.52 3.73 0.054 –
Lu et al. Simplif. 0.52 3.73 0.048 –
where sf is the ultimate strain, related to null stress. Table 9. Significant values for local bond of GFRP.
This form is commonly proposed by some guidelines
(fib Bulletin 14 2001; CNR-DT 200 2004). Gf τmax s0 sf
Since the bilinear function depends on more param- Curve N/mm MPa mm mm
eters, to reduce them the peak tangential stress value,
obtained from the fitted Unipd curve, and the cali- UniPd fitting 1.91 6.33 0.111 –
Bilinear fitting 1.91 6.33 0.048 0.603
brated fracture energy were imposed.
Figures 10–11 show the optimized curves and the Monti et al. 1.11 5.37 0.046 0.415
experimental stress-slip data. It can be noticed that car- Lu et al. Bilinear 0.52 3.73 0.048 0.280
bon reinforcement interface seems to be slight stiffer Brosens and V. G. 1.39 2.71 0.012 1.025
than glass one. CNR 0.75 7.46 0.056 0.200
Tables 8–9 report the significant values (fracture Nakaba et al. 1.32 7.08 0.065 –
Savoia et al. 1.16 7.08 0.051 –
energy, peak tangential stress with related slip, and Neubauer and R. 0.77 5.69 0.270 –
ultimate slip) calculated by the fitting process; they Dai and Ueda (1) 1.15 7.10 0.107 –
have been compared with the values estimated through Dai and Ueda (2) 1.10 5.69 0.067 –
the eleven models based on the fracture energy predic- Lu et al. Precise 0.52 3.73 0.054 –
tion, mentioned in chapter 4. Estimated values vary Lu et al. Simplif. 0.52 3.73 0.048 –
into a quite wide range.
It has to be noticed that not all models provide
significant differences from carbon to glass reinforce-
ment; in particular, the difference between the fracture substrate. All predictions, except two in case of CFRP,
energy values experimentally calibrated, in case of underestimate the results of the tests; the models,
CFRP and GFRP, is not in agreement with most except two in case of GFRP, seem to work better for
predictions. the carbon reinforcement. However, the strength pre-
dictions vary into an wide range (between 44% and
154% of experimental mean failure load for CFRP,
7 CONCLUSIONS 43% and 85% for GFRP).
From the measured failure loads, different fracture
The bond behaviour of the composite-clay brick inter- energy values have been derived, around 35% higher
face has been investigated by means of Double-lap in case of glass reinforcement than carbon one.
Push-pull Shear Tests, for both high-strength carbon To analyze stress and slip from strain-gauges mea-
(CFRP) and alkali-resistant glass (GFRP) reinforce- surement, when instruments are not uniformly spaced,
ment. discrete equations have been used, consistent with
Far from being exhaustive, the results show a better central finite difference methods.
performance of carbon reinforcement than glass one, Moreover, a mathematical function easy to integrate
around 36% higher in the first case. and derive is proposed as bond-slip law. This function
The experimental strength has been compared with has been fitted in case of both carbon and glass rein-
twenty-one predictive models developed for concrete forcement; beside these fittings, two bilinear functions
769
have been also calibrated. The optimized functions CNR-DT 200/2004. Guide for the Design and Construc-
seem to show an interface local behaviour of CFRP tion of Externally Bonded FRP Systems for Strengthening
slightly stiffer than GFRP. Existing Structures. National Research Council, Italy.
However, further investigations are needed, to better Dai, J.G. & Ueda, T. 2003. Local Bond stress Slip relations for
FRP Sheets-Concrete Interfaces. Fibre-Reinforced Poly-
inquiry the bond behaviour of composite-clay bricks mer Reinforcement for Concrete Structures FRPRCS-6;
interface; this is a first step in order to take into proc. int. symp., Singapore.
account, in the future, the role of the mortar joints, Dai, J.G., Ueda, T. & Sato, Y. 2005. Development of the
characteristic of masonry structures. Nonlinear Bond Stress-Slip Model of Fiber Reinforced
Moreover, the reliability of the experimental set-up Plastics Sheet-Concrete Interfaces with a Simple Method.
needs to be verified; despite of its simplicity, the actual ASCE J. of Composites for Construction, 9 (1): 52–62.
distribution of the load should be more clarified, as it De Lorenzis, L., Miller, B. & Nanni, A. 2001. Bond of
could cause undesired bending moments inside the test FRP laminates to concrete. ACI Materials Journal, 98 (3):
specimen. 256–264.
fib Bulletin 14 2001. Externally bonded FRP reinforcement
for RC structures – Technical Report on the design and
use of externally bonded fibre reinforced polymer rein-
ACKNOWLEDGEMENTS forcement for reinforced concrete structures. Fédération
Internationale du Béton.
The authors acknowledge Andrea Cartolaro and the Karbhari, V.M., Niu, H. & Sikorsky, C. 2006. Review and
technical staff of the Laboratory of Material Testing Comparison of Fracture Mechanics-based Bond Strength
of the Department of Construction and Transportation Models for FRP-strengthened Structures. J. of Reinforced
Plastics and Composites, 25 (17): 1757–1794.
of the University of Padova, where tests have been Lee, Y.J., Boothby. T.E., Bakis. C.E. & Nanni. A. 1999. Slip
performed. Modulus of FRP Sheets Bonded to Concrete. ASCE J. of
This activity has been partially supported by the Composites for Construction, 3(4): 161–167.
National Italian Project ReLUIS. Lu, X.Z., Teng, J.G., Ye, L.P. & Jiang, J.J. 2004. Bond-
All materials, including fibers and adhesion system, slip models for FRP sheet/plate-to-concrete interfaces.
have been supplied by Modern Advanced Concrete Advanced Polymer Composites for StructuralApplications
(MAC S.p.A.) of Treviso, Italy. in Construction – ACIC 2004; proc. 2nd int. conf., Surrey,
UK.
Lu, X.Z., Teng, J.G., Ye, L.P. & Jiang, J.J. 2005. Bond-
slip models for FRP sheets-plates bonded to concrete.
REFERENCES Engineering Structures, 27: 920–937.
Nakaba, K., Kanakubo, T., Furuta, T., Yoshizawa, H. 2001.
Aiello, M.A. & Sciolti, M.S. 2005. Bond analysis of masonry Bond Behavior between Fiber-Reinforced Polymer Lam-
structures strengthened with CFRP sheets. Construction inates and Concrete. ACI Structural Journal, 98 (3):
and Building Materials, 20: 90–100. 359–367.
Bizindavyi, L. & Neale, K.W. 1999. Transfer Lengths and Savoia, M., Ferracuti, B. & Mazzotti, C. 2003a. Non
Bond Strengths for Composites Bonded to Concrete. linear bond-slip law for FRP-concrete interface. Fibre-
ASCE Journal of Composites for Construction, 3 (4): Reinforced Polymer Reinforcement for Concrete
153–160. Structures FRPRCS-6; proc. int. symp., Singapore.
Briccoli Bati, S., Rovero, L., Tonietti, U. 2007. Adesione fra Savoia, M., Ferracuti, B. & Mazzotti, C. 2003b. Una legge di
blocchi in laterizio e rinforzo in CFRP (in Italian). Materi- interfaccia non lineare per placcaggi con lamine in FRP
ali ed Approcci Innovativi per il Progetto in Zona Sismica (in Italian). 16th AIMETA Congress of Theoretical and
e la Mitigazione della Vulnerabilità delle Strutture; proc. Applied Mechanics; proc., Ferrara, Italy.
nat. workshop ReLUIS, Salerno, Italy. Täljsten, B. 1996. Strengthening of concrete prisms using the
Camli, U.S. & Binici, B. 2007. Strength of carbon fiber plate bonding technique. Int. J. of Fracture, 82: 253–266.
reinforced polymers bonded to concrete and masonry. Täljsten, B. 1997. Defining anchor lengths of steel and
Construction and Building Materials, 21: 1431–1446. CFRP plates bonded to concrete. Int. J. of Adhesion and
Cartolaro, A. 2004. Volte in muratura di mattoni rinforzate Adhesives, 19: 319–327.
con FRP. Studio sperimentale dei meccanismi resistenti Valluzzi, M.R., Tinazzi, D., Garbin, E. & Modena, C. 2003.
alle interfacce (in Italian). MS Dissertation, University of FEM modelling of CFRP strips bond behaviour for bed
Padova, Italy. joints reinforcement techniques. Computer Methods in
Chajes, M.J., Finch, W.W.Jr., Januszka, T.F. & Thomson, T.A. Structural Masonry; proc. 6th int. conf., Rome, Italy.
1996. Bond and Force Transfer of Composite Material Wu, Z.,Yuan, H. & Niu, H. 2002. Stress Transfer and Fracture
Plates Bonded to Concrete. ACI Structural Journal, 93 Propagation in Different Kinds of Adhesive Joints. ASCE
(2): 295–303. Journal of Engineering Mechanics, 128 (5): 562–573.
Chen, J.F. & Teng, J.G. 2001. Anchorage Strength Models for Yao, J., Teng, J.G. & Chen, J.F. 2004. Experimental study on
FRP and Steel Plates Bonded to Concrete. ASCE Journal FRP-to-concrete bonded joints. Composites: Part B, 36:
of Structural Engineering, 127 (7): 784–791. 99–113.
770
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The mechanical characterization of the shear strength properties takes a major role on the evalu-
ation of the lateral strength of masonry shear walls by means of simplified methods or when numerical analysis
based on micromodelling approach is to be followed. Thus, the present paper presents an overview of the results
obtained from direct shear tests of different types of masonry joints: dry and mortar masonry joints. Besides the
shear strength parameters, a good insight was achieved in the evaluation of the complete shear stress-shear load
displacement diagrams.
771
(a) (b) Plan view Elevation view
Figure 1. Masonry specimens; (a) dry joints; (b) mortar Figure 2. Arrangement of the LVDTs for measuring of the
joints. relative horizontal and vertical displacements.
ensuring mechanical properties for numerical simula- In addition, two thin sheets of Teflon were interlay-
tions of the in-plane behavior of stone masonry wall ered between the steel platens and the specimens to
structures, the adopted testing program provides also minimize bending effects. In order to guarantee right
the fundamental information about the shear behavior angle surfaces, the specimens were suitably ground
of two different masonry joints. using a rectifying machine. The same procedure was
used in the contact surfaces between both units of the
specimens to ensure the maximum contact area in case
2 EXPERIMENTAL DETAILS of dry joints. In fact, according to Hansen (1999),
the uneven stress distribution can also be attributed
2.1 Test specimens and procedure to the non-uniform distribution of the material along
Although triplet tests have been adopted as the Euro- bed joints. In both types of specimens, when necessary,
pean standard method (EN1052-3 2002) to perform a thin layer of glue was placed at the surface in con-
shear tests in mortar joints, the shear strength proper- tact with the steel platens in order to provide perfectly
ties of dry and mortar joints were obtained by means leveled surfaces. The confinement of the specimens
of direct shear tests conducted on couplet specimens, was improved for load reversal by means of a couple
see Figure 1 (Vasconcelos 2005). In fact, in the triplet of steel rods fixed to the steel plates through metal-
test, the two joints do not fail at the same time and the lic bolts. This arrangement is particularly useful in the
analysis of the experimental results is rather complex, cyclic tests but was also used in case of monotonic
Lourenço et al. (2004). tests. The numerical assessment of the effectiveness of
The shear tests were carried out in a servocontrolled the test setup was performed by Lourenço and Ramos
universal testing machine CS7400S. This equipment (2004) based on a finite element model. It was con-
is composed by two independent hydraulic actua- cluded that although a deviation on the stresses occurs
tors used to transmit normal and shear loads, able in the zone adjacent to the steel plates, an almost uni-
to operate under force or displacement control. The form normal and shear stress distribution is achieved
features of the testing equipment and the existing in 63% of the extent of the bed joint.
loading platens imply that the most suitable testing In order to simulate the usual range of normal
sample is composed by two units with geometry and stresses existing in ancient masonry structures three
dimensions indicated in Figure 1 and a single dry or distinct pre-compression stress levels were applied
mortar joint, similarly to Pluijm (1999) and Hansen under force control, σ = 0.5 N/mm2 , σ = 0.75 N/mm2
et al. (1998). The surface of the dry masonry units and σ = 1.0 N/mm2 in dry joints under monotonic and
adopted here is smooth resulting from sawing the cyclic loading. An additional pre-compression stress
specimens, whereas the joint surface of the units of level corresponding to σ = 1.25 N/mm2 was consid-
the mortar assemblages presents enough roughness ered for the monotonic tests carried out on unit-mortar
to achieve appropriate adherence conditions and thus assemblages. Three specimens were tested for each
more realistic masonry can be simulated. The detailed level of pre-compression for both types of masonry
experimental characterization of this type of granite joints. The possible influence of the moisture content
can be seen inVasconcelos et al. (2007).The specimens on the shear response of dry masonry joins was inves-
were placed between two thick steel plates and attached tigated by considering dry and saturated conditions.
to the steel platens by steel bolts, so that shear force Although the horizontal actuator is servo-controlled,
could be transmitted. Thin steel sheets were attached the control was made using the horizontal LVDT adja-
to the steel plates to concentrate the shear load as close cent to bed joint because a more stable response was
as possible in the bed joint, aiming at preventing bend- found. The disposition of the LVDTs for measuring
ing moments and provide a more uniform shear stress the horizontal and vertical displacements of the joint
distribution. is depicted in Figure 2.
772
The relative horizontal displacement of the joint was
measured by the horizontal LVDTs placed at each side
of the specimen. Although the LVDTs were fixed to
the unit through the supports that were glued to it,
the influence of the shear deformation of the units
should be marginal in the measured final deformation
(Hansen 1999). The vertical displacements of the joint
were measured by the LVDTs placed at the opposite
corners of the specimen. The relative vertical dis-
placements were monitored in case of dry joints for
assessing the possible dilatant behavior of the joints.
Nevertheless, technical problems did not allow to mea-
sure relative vertical displacements in mortar joints, as
tests were conducted in a subsequent phase. Both shear
and normal stresses were measured and recorded by
the horizontal and vertical load cells of 22 kN capacity.
(a)
3 ANALYSIS OF RESULTS
773
(a)
Figure 5. Relation between total vertical displacement and
shear displacement.
2%). Note that the shear tests are conducted under hor-
izontal displacement control. This further assesses the
validation of the test setup.
In spite of the fact that the relative local vertical
displacements of the joint could not be measured due
to technical problems with the LVDTs, the total verti-
cal displacement was recorded by the internal LVDT
located inside the vertical actuator. The evolution of
this displacement with the shear displacement is dis-
played in Figure 5, where in the greater number of the
tests two distinct phases can usually be distinguished.
Firstly, the uplift of the joint is expressed by increasing
positive vertical displacements, which is particularly
(b) remarkable for low normal stresses. The nonlinear evo-
lution of the displacements provides variable dilatancy
Figure 4. Shear behavior of mortar joints; (a) shear assuming decreasing values as the shear displacement
stress-shear displacement diagram; (b) pre-peak nonlinearity. increases. This behavior is connected to the changes
on the interfaces due to surface wearing. It is observed
that the shear displacement associated to the maximum
recorded by the two horizontal LVDTs located at each value of the vertical displacement is close to the hori-
side of the specimen and the shear stress is calcu- zontal displacement corresponding to the stabilization
lated according to eq. 2. The general shape of the of the shear stress.
shear stress-shear displacement is characterized by a The dilatant behavior reflects, to great extent, the
sharp initial linear stretch. The peak load is rapidly distinct shear failure modes obtained in the specimens
attained for very small shear displacements. Similarly submitted to different normal stresses. In fact, for low
to what was reported for dry masonry joints, non- levels of pre-compression, shear failure occurs at the
linear deformations develop in the pre-peak regime, unit-mortar interface along one unit face or, more fre-
see Figure 4b. quently, divided between two unit faces, see Figue 6.
After peak load is attained there is a softening For the larger normal stress level (σ = 1.25 N/mm2 ),
branch corresponding to progressive reduction of the the failure is only localized in the mortar and a larger
cohesion, until reaching a constant dry-friction value. amount of small mortar particles were found to be
This stabilization is followed by the development of detached. No damage was visible in the stone units
large plastic deformations. in all cases.
As required, the shear tests were carried out with- During the regime of pure friction the verti-
out significant fluctuations of vertical load (less than cal displacement remains constant or progressively
774
Figure 6. Typical Failure mode for the lower compression
level (σ = 0.5 N/mm2 ).
775
(a)
Figure 9. Correction of the measured shear displacement-
dry specimens.
776
(a)
Figure 11. Relation between peak and residual shear
stresses with normal stress.
777
respectively. This means that for the range of pre- Table 1. Shear strength properties for different unit-mortar
compression levels tested, the peak and residual shear assemblages.
strength of the bed joint can be reasonably described
by means of the Coulomb’s friction law given by Source units mortar c (MPa) µ
eq. 1. Therefore, the linear approach provides shear
Atkinson Old clay 1:2:9 (13) 0.127 0.695
strength characteristics of the mortar joint, cohesion,
(1989) Old clay 1:2:9 (7) 0.213 0.640
c, and friction coefficient, µ.A value of cohesion about New clay 1:1.5:4.5 0.811 0.745
0.36 N/mm2 and the tangent of the friction angle, tanφ,
Amadio and Solid Cement 0.65 0.723
equal to 0.63, corresponding to a friction angle of
Rajgeli bricks Lime-
32.2◦ , were attained for the peak strength. The residual (1990) cement
shear strength can be calculated with reasonable accu-
Magenes Solid Hydraulic 0.206 0.813
racy from a friction coefficient of 0.78. This value can
(1992) bricks lime 0.081 0.652
be used for evaluation of the shear sliding resistance Lime mortar
of walls or piers submitted to seismic action failing
Binda et al. Sandstone Hydraulic 0.33 0.74
along horizontal sliding joints. The strength values,
(1994) Calcareous lime 0.58 0.58
particularly the bond strength, are greatly dependent Hydraulic
on the moisture content and porosity of the units and lime
on the strength and composition of mortar as well as on
Roberti et al. Bricks Hydraulic 0.23 0.57
the nature of the interface (Amadio & Rajgelj 1990). (1997) lime mortar
Binda et al. (1994) pointed out that when strong mortar
Lourenço Hollow Micro- 1.39 1.03
is considered, the strength of the units can also regu-
et al. (2004) bricks concrete
late the shear behavior of the joints. This yields that a
wide range of shear strength values have been pointed This study Granite Lime mortar 0.359 0.630
out for various combinations of units and mortar.
Mann & Muller (1982) indicated a mean friction
coefficient of approximately 0.65 on brick-mortar
assemblages and a cohesion ranging from 0.15 up to
0.25, depending on the mortar grade. From the results nearly 5% was recorded on the friction coefficient
of direct shear tests carried out by Pluijm (1999), the changing the moisture condition from dry to saturated.
coefficient of internal friction ranges between 0.61 Low differences were found between the peak and
and 1.17, whereas cohesion varies from 0.28 up to residual friction angle for dry masonry joints under
4.76, depending on different types of units and mortar. distinct moisture conditions. No dilatancy was found
Table 1 summarizes other results published in literature to characterize the shear behavior of dry masonry
referring to the shear strength properties for different joints.
combinations of materials. Similarly, an extended plastic branch character-
izes the residual post-peak shear behavior of mortar
masonry joints. After a reduction of approximately
25% of the peak strength, the shear stress stabilizes
4 CONCLUSIONS with considerable level of plastic deformations. The
measured total vertical displacement revealed a ten-
The experimental characterization of masonry assem- dency for the mortar joint dilation to decrease as the
blages and masonry components used in the shear normal stress increases. Only for a pre-compression of
walls discussed in Chapter 6 is addressed, focusing 0.5 N/mm2 a systematic expansive trend was verified
on the strength properties of dry masonry joints and up to peak stress. For this case, it was found that dila-
the unit-mortar interface, as well as on the compressive tancy decreases significantly as the shear displacement
properties of the masonry. increases.
A set of direct shear tests was conducted on cou-
plet specimens, either considering dry or mortar joints.
From these tests it was possible to derive the shear REFERENCES
strength properties, namely cohesion and friction coef-
ficient. Besides, the complete shear stress-shear dis- Amadio, C., Rajgelj, S. 1991. Shear behavior of brick-mortar
placement diagrams enabled a better insight into the joints, Masonry International, 5 (1), 19–22.
Atkinson, R.H., Amadei, B.P., Saeb, S., Sture, S. 1989.
shear behavior of these assemblages. An elastic per- Response of masonry bed joints in direct shear, Journal of
fectly plastic diagram was found to characterize the Structural Engineering, 115 (9), 2277–2296.
monotonic and the cyclic envelope of shear tests Binda, L., Fontana, A., Mirabella, G. 1994. Mechanical
conducted in dry masonry joints. No significant dif- behavior and stress distribution in multiple-leaf stone
ferences in the frictional behavior of dry joints under walls, 10th International Brick Block Masonry Confer-
distinct moisture contents were found. A reduction of ence, Calgary, Canada, 51–59.
778
Calvi, G.M., Kingsley, G.R., Magenes, G. – Testing masonry Lourenço, P.B., Barros, J.O., Oliveira, J.T. 2004. Shear test-
structures for seismic assessment, Earthquake Spectra, ing of stack bonded masonry, Construction and Building
Journal of Earthquake Engineering Research Institute, Materials, 18, 125–132.
12(1), 145–162, 1996. Lourenço, P.B., Ramos, L.F. 2004. Characterization of cyclic
EN 1052-3, Methods of test for masonry: Part 3 – Determi- behavior of dry masonry joints, Journal of Structural
nation of initial shear strength, 2002. Engineering, 130 (5), 779–786.
Geerstsema, A.J. 2002. The shear strength of planar joints in Mann, W., Müller, H. 1982. Failure shear-stressed masonry
mudstone, International Journal of Rock Mechanics and – an enlarged theory, tests and application to shear walls,
Mining Sciences, 39, 1045–1049. Proc. British Ceramic Society, 30, 223–235.
Hamid, A.A., Drysdale, R.G. 1980. Behavior of brick Misra, A. 2002. Effect of the asperity damage on shear behav-
masonry under combined shear and compression loading, ior of single fracture, Engineering Fracture Mechanics, 69,
Proc. 2nd Canadian Masonry Conference, 314–320. 1997–2014.
Hansen, K.F. 1999. Bending and shear tests with masonry, Pluijm, R.V.D.1999. Out-of-Plane bending of masonry,
SBI Bulletin 123, Danish Building Research Insti- behavior and strength, PhD thesis, Eindhoven University
tute, p. 36. of Technology. ISBN 90-6814-099-X.
Hansen, K.F., Nykänen, E., Gottfredsen, F.R. 1998. Shear Riddington, J.R., Ghazali, M.Z. 1990. Hypothesis for shear
behavior of bed joints at different levels of precompres- failure in masonry joints, Proc. Instn. Civ. Engrs, 89,
sion, Masonry International, 12 (2), 70–78. 89–102.
Homand, F., Belem, T., Souley, M. 2001. Friction and degra- Samarasinghe, W., Hendry, A.W. 1980. The tensile of brick-
dation of rock joint surfaces under shear loads, Interna- work under biaxial tensile and compressive stress, Proc.
tional Journal for Numerical and Analytical Methods in 7th International Symposium on Load Bearing Brickwork,
Geomechanics, 25, 973–999. London, 129–139.
Huang, T.H., Chang, C.S., Chao, C.Y. 2002. Experimental Vasconcelos, G. 2005. Experimental investigations on the
and mathematical modeling for fracture of rock joint with mechanics of stone masonry: characterization of granites
regular asperities, Engineering Fracture Mechanics, 69, and behavior of stone masonry shear walls, PhD thesis,
1977–1996. University of Minho, Portugal.
Lee, H.S., Park, Y.J., Cho, T.F., You, K.H. 2001. Influence Vasconcelos, G., Lourenço, P.B., Alves, C.A.S, Pamplona, J.
of asperity degradation on the mechanical behavior of 2007. Prediction of the mechanical properties of gran-
rough rock joints under cyclic shear loading, International ites by ultrasonic pulse velocity and Schmidt hammer
Journal of Rock Mechanics and Mining Sciences, 38, hardness, 10th North American Masonry Conference, (in
967–980. press).
779
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
John Harkness
School of Civil Engineering & the Environment, University of Southampton, UK
Pete Walker
BRE Centre for Innovative Construction Materials Dept. of Architecture & Civil Engineering,
University of Bath, UK
ABSTRACT: There are numerous historic drystone retaining walls in the UK, but the analysis of these eco-
nomically important walls is restricted by the lack of experimental data. As part of an ongoing investigation into
the stability of drystone retaining structures, two full-scale walls have been constructed and tested to collapse.
Details of each test are described, including set-up, wall construction, instrumentation and data collection. Initial
findings are presented and analysed, along with the continuing aims and goals of the project.
781
led by Jean-Claude Morel and Boris Villemus built
and tested five large scale test walls at ENTPE,
Lyon.[5]. The walls were of various sizes ranging from
2 m–4.25 m high, up to 1.8 m thick and between 2 m
and 3 m long. Each wall was subjected to hydrostatic
forces via a PVC-lined water filled bag, in order to load
the wall using purely horizontal pressures that could
be precisely known at all times.
The main aim of Morel and Villemus’ study was to
identify the internal failure angle within the walls at
yield. As each test wall was only a short section (2–3 m
long), the wall ends were left exposed and were consid-
ered to reveal the internal actions throughout the entire
length. The tests could not proceed to collapse because
Figure 1. Burgoyne’s test wall geometry[4]. of the way pressure was applied to them, but monitor-
ing of the end faces allowed relative movements within
instrumentation is used to monitor the walls’ response the wall to be measured as yielding took place. The
through to their collapse. angles of the resulting shearing surfaces to the hori-
The main aim is to have sufficient data to verify zontal were related to the pattern of construction of the
existing models and theories, and assist with the gener- walls.
ation of new analysis techniques, more accurate mod-
elling tools and assessment guidelines for drystone
structures. The study is funded by the Engineering & 4 TEST SETUP
Physical Sciences Research Council (EPSRC) and
is conducted in collaboration with the University 4.1 Testing rig
of Southampton. The creation of these guidelines,
together with appropriate engineering tools to back For the purposes of this project, a bespoke test rig
them up, could save a substantial part of the estimated was required. The rig required the ability to impose
£1 billion estimated cost of replacing the walls in the various conditions upon the test walls, necessary to
UK alone [2]. undermine their stability. These conditions, including
relative wall settlement, backfill compaction and sur-
charge loading, mimic the real-life situations to which
3 RELATED RESEARCH many drystone walls are subjected.
To simulate settlement of the foundations or the
As previously mentioned, physical test data regard- backfill, the ability to raise, lower or tilt the wall
ing drystone retaining wall structures is sparse. The is required. To achieve these movements, the wall
largest reported tests to date were conducted in itself is constructed on a 1.4 m × 4 m steel platform,
1834 by Lieutenant-General Sir John Burgoyne [3], which is in turn mounted on four 20 tonne mechani-
who constructed four full size test walls in Dun cal screwjacks. This arrangement can produce vertical
Laoghaire, Ireland. Each wall was built using the movements of up to 400 mm, tilt up to 17◦ forwards
same overall volume of square cut granite blocks, or backwards, as well as intermediate combinations of
but arranged in different sections (Fig. 1). Testing these movements.
consisted of backfilling each wall until the full reten- A steel frame spans the central portion of the wall
tion height (6 metres) was achieved, unless premature and backfill, under which a hydraulic jack is mounted.
collapse occurred. This jack allows a patch surcharge to be applied to
From this work Burgoyne proved that wall geometry almost any location behind the wall, simulating addi-
has a substantial impact on overall stability, although tional localised loadings of up to 200 kN (Fig. 2). The
the use of highly worked granite blocks perhaps walls were designed to be sufficiently long to minimise
caused the walls to act more monolithically than would the edge effects from both induced settlements and sur-
otherwise be observed in more traditional drystone charge loading on the behaviour of the wall which was
structures. Regardless, his findings and observations a serious limitation of the Burgoyne experiments.
remain the basis for the validation of almost all of
the numerical studies carried out to date on drystone
4.2 Instrumentation
retaining walls, despite consisting solely of dimen-
sional measurements and visual observations reported To ensure as much data as possible are gathered from
19 years after the tests. each test, a combination of techniques were imple-
Work of such a physical nature was not conducted mented. Load cells monitor the forces acting on the
again until 2005, when a French engineering team platform, and the applied patch surcharge. Simple
782
the nature of the tests allows sufficient time for the
Total Station to be used at any point up to the final
moments of collapse. Although the final failure of
the wall cannot be captured using this technique, it
is still very useful for examining bulge formation and
the movements prior to failure.
In addition to the Total Station readings, draw wire
transducers were attached to the centre line of the wall.
These allow continuous high resolution monitoring of
wall face movements during testing, facilitating con-
trol of test procedures. The transducers are anchored
to a stanchion attached to the loading platform, so
that measurements represent deformation of the wall
directly, without the need to correct for movement of
Figure 2. Test rig setup. the platform.
Given the scale of variations within the construc-
tion of drystone walls, it is not possible to guarantee
button-type compression cells are used for the screw-
that the transducers and load cells will be located on
jacks, with S-type tension/compression cells located
parts of the structure where key events unfold. To
on the radial arms. Data-logging records readings at
ensure detailed recording of mechanisms, this project
appropriate intervals throughout the different stages
employs a range of visual tools to supplement the
of construction and testing.
instruments. In front of the test wall face, on a fixed
It is necessary to monitor both local movements
pedestal, are a pair of Nikon D40x digital SLRs, pro-
of the gravel and the distribution of earth pressures
viding stereo photographs of the wall face throughout
within the backfill. Monitoring of movement is dif-
testing. An additional camera is mounted behind the
ficult to accomplish during testing, as there are few
backfill to record back face and backfill conditions in
non-destructive/non-invasive means to achieve accu-
response to the surcharging.
rate results. For the purposes of these tests, several
Southampton University used Particle Image
layers of steel ball bearings have been laid in grids
Velocimetry (PIV) to record movements of the cen-
within the backfill, with each ball carefully placed and
treline of the face to sub-millimetre accuracy. High
surveyed into position using a reflectorless Total Sta-
resolution digital photographs were taken sighting
tion. The polished surfaces of the balls ensure that the
along the line of the wall to targets mounted on
measuring beam is reflected back from a point on the
brackets attached to a number of individual stones.
surface in line between the instrument and the centre
A computer program was then used to determine
of the ball, so that a correction for the radius is all that
changes from one image to the next.
is needed to determine the accurate positions of the
The whole test process has also been captured on a
centres. After failure of the wall, the ball bearings are
Canon HV20 HD camcorder. This final piece of equip-
carefully recovered using metal detection equipment,
ment ensures that the final moments of the test are
and again their positions recorded, thus revealing any
recorded in a high-definition video format. Addition-
movements and aiding identification of failure planes
ally, it allows any sudden movements or noises to be
within the backfill.
recorded, information that might otherwise be missed.
Backfill pressures are measured using pressure
cells buried within the fill. These consist of parallel
100 mm × 100 mm steel plates separated by 2500 N
or 5000 N load cells. These are placed within the
5 EXPERIMENTAL TESTING
fill either vertically or horizontally, allowing continu-
ous monitoring of lateral or vertical pressures. These
5.1 1st test wall
sensors are particularly useful to indicate how the sur-
charging force is distributed through the backfill. The The first test wall was constructed in June 2007, requir-
pressure cells are spread out through the backfill, with ing approximately 35 tonnes of limestone. The wall
the majority concentrated along the centre-line of the was 2.5 m high, tapering from 600 mm at the base
wall where the loads are expected to be highest. to 400 mm at the coping, with a vertical rear face
To monitor the wall face itself, a combination of and a battered front face. This initial wall was con-
techniques have been used. Around 350 small targets structed with two fair faces, so that the wall could be
have been scored into the wall face in five vertical constructed without the need to be building and back-
lines. The points are relocated throughout testing using filling simultaneously. The wall contained three layers
a reflectorless Total Station. Although a slow process, of through-stones, at 0.5 m, 1 m, and 1.5 m high. As
with each round of readings taking 15–20 minutes, is standard practice amongst most UK masons, these
783
Figure 4. Test wall 1 (prior to failure).
784
5.3 Materials testing
Following tests an intact area of wall was measured,
weighed and removed to ascertain the void percentages
and densities. A void percentage of 28% was mea-
sured, giving an average wall density of 17.9 kN/m3 .
Two further test sections, each of dimensions
0.5 m × 0.5 m × 0.7 m, were constructed by the
masons to investigate this further. One was built to the
highest standard that could be achieved with this stone,
but using fill pieces of a normal size. The second was
built to a functional standard, but to represent a struc-
ture built quickly without taking care to achieve an
optimal fit between stones. Upon disassembly, the void
percentages for these two volumes were 21% and 37%,
with overall densities of 19.7 kNm−3 and 15.6 kNm−3
respectively.
Other material tests are scheduled to be carried out
within the time frame of this project, including labora-
tory investigations into the interface between the wall
and the backfill, and sliding between wall beds.
6 ANALYSIS
785
2.25 4
3.5
2
3
1.75
Depth (m)
2.5
1.5
1.25
Height (m)
1.5
1
1
2 2.5 3 3.5 4 4.5 5 5.5 6
Distance (m)
1st Level Initial 1st Level Final 2nd Level Initial 2nd Level Final
0.75 3rd Level Initial 3rd Level Final 4th Level Initial 4th Level Final
5th Level Initial 5th Level Final Possible Failure Plane
0.25
of the platform rotation and early loading near the top
of the wall succeeded in pushing the centre of gravity
0
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
very far forwards.
Displacement (m) Although the second wall also toppled, the manner
1st Wall Initial 1st Wall Final 2nd Wall Inital in which it initiated this failure was different. Block
2nd Wall Final Initial Position sliding through the lower courses occurred to the point
where a significant portion of some blocks were over-
Figure 6. Test walls 1 & 2 total station data (horizontal scale hanging the course below. This in turn allowed an
exaggerated). overhanging block to begin to rotate. In the final phases
of test two, specific blocks near the base of the wall
than in the second test, leading to much higher loads could be identified in both the photography and the
before slip planes were formed. This also had the effect video which had clearly slid forwards over the blocks
of transmitting lateral stress from the surcharge much beneath them.
higher up the wall than originally desired, encouraging As the wall is pushed further forwards, the resultant
toppling rather than bulging. thrust line will similarly move forwards until it is pass-
After dismantling the first wall, the ball bearings ing through the ends of these overhanging blocks. The
within the fill were unearthed, located with the Total overturning of the wall was then triggered by the rota-
Station, and then analysed for changes in position tion of one of these overhanging blocks. At this point
(fig. 7). The stiffness of the fill allowed very lit- the thrust line will fall outside of the wall at this level
tle movement away from the wall face; indeed the and toppling would occur. The movement of this rota-
only movements which were recorded coincided with tion along the course of stones could be seen clearly in
a failure wedge originating at the rear edge of the the video and in the sequence of stereo photographs.
surcharging plate.
The other significant impact that the angle of back-
fill friction has upon the failure is the magnitude 7 CONCLUSIONS
of active pressure exerted on the wall. The first test
with the higher friction angle has the added benefit Although only two of the four programmed tests have
of a much greater vertical component upon the wall, as yet been carried out, substantial results have already
whereas the lower friction angle has a relatively greater been generated. Each wall has been constructed to
horizontal component. emulate real walls as closely as possible, with care-
It is difficult to be certain exactly how much of the fully limited changes between successive wall tests to
backfill friction is mobilised against the wall. How- facilitate meaningful comparison of the results.
ever by initially raising the platform until the platform An initial aim was to create the distinctive ‘belly
loads peak, and given the roughness of the back of the bulges’ found in many walls today, and both test walls
walls, it is reasonable to assume that it is almost fully showed some bulging deformation. The reasons why
activated. these particular failure mechanisms formed are now
Both walls eventually failed via toppling. This was better understood, highlighting issues that were not
to be expected for the first wall, as the combined effects previously considered.
786
Though these first two wall tests do not provide data support this project would not have been possible:
applicable to a wide range of wall failures, the mech- Engineering & Physical Sciences Research Council;
anisms observed most probably occur, and every dry Southampton University; Network Rail; Wiltshire
stone retaining wall that is still standing must necessar- County Council; Gloucestershire County Council;
ily be resisting failure by these mechanisms. The tests County Surveyors Society; Bradford Metropolitan
also provide detailed observations of the behaviour of District Council; Cornwall County Council; Richard
real walls which can be used to verify the results of Tufnell Dry-Stone Wallers; and, colleagues at the
numerical modelling, including that currently being University of Bath.
carried out at Southampton University. This numeri-
cal modelling, once verified, will be used to investigate
other aspects that cannot be safely or effectively inves- REFERENCES
tigated by full scale testing, such as the effect of major
changes in pore water pressures. The tests therefore 1. Powrie, W., Harkness, R.M., Zhang, X. and Bush D.I.
represent a clear step towards a greater understand- (2002). “Deformation and failure modes of drystone
ing of these structures, and also towards the eventual retaining walls”, Geotechnique, Vol. LII, No. 6, August,
goal of formulating guidelines and accurate analysis 435–446.
2. O’Reilly, M.P., Brady, K.C. and D.I. Bush, Research on
techniques. At the very least these tests prove beyond masonry-faced retaining walls, in 2nd European road
all doubt that even walls such as these, designed to research conference. 1999: Brussels.
deform and fail, are incredibly resilient and able to 3. Burgoyne, J., Revetments or retaining walls. Corps of
adapt to significant loads with little affect on overall royal engineers, 1853. 3: pp. 154–159.
long term stability. 4. Harkness, R.M., Powrie, W., Zhang, X., Brady, K.C. and
O’Reilly, M.P. (2000). “Numerical modeling of full-scale
test on drystone masonry retaining walls”, Geotechnique,
ACKNOWLEDGEMENTS Vol. 50, 165–179.
5. Villemus, B., Etude des murs de soutenement en macon-
nerie de pierres seches, in L’institut national des sciences
The authors acknowledge support of the follow- appliquees de Lyon. 2004, ENTPE: Lyon.
ing organisations and individuals, without whose
787
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The efficiency of improving the seismic resistance of heritage masonry buildings by means of
seismic isolation and strengthening of structure with CFRP laminate strips has been investigated. Five models of
a simple two-storey brick masonry building with wooden floors without wall ties have been tested on the shaking
table. Besides control model, two models, isolated by either damp proof course or seismic isolators, have been
tested. Models four and five have been strengthened with CFRP laminate strips, simulating the wall ties placed
horizontally and vertically at floor levels and corners of the building, respectively. One of the CFRP strengthened
models has been placed on seismic isolators. Tests have shown that the damp proof course, unless adequately
designed, cannot be considered as seismic isolation. The isolators have also not improved the behavior in the
case of the building without wall ties. However, both models confined with CFRP strips exhibited significantly
improved seismic behavior. They did not collapse even when the accelerations of the shaking table exceeded the
accelerations, measured when tested the control models without wall ties, by three times.
789
Figure 2. Typical model during construction.
Figure 1. Scheme of laying the bricks and position of
wooden joists and lintels.
790
Figure 5. Calibration test of seismic isolator.
791
Table 1. Maximum accelerations and displacements of the
shaking table, measured during individual test runs (average
values for the groups of models).
dmax dmax
Duration Test run amax (g) (mm) amax (g) (mm)
792
Figure 8. Mechanism of collapse of non-strengthened con- Figure 9. Mechanism of collapse of non-strengthened
trol model M1. model M2 with damp proof course.
3 TEST RESULTS
793
Figure 11. Damage to CFRP laminate strengthened, non-
isolated model M4 at the end of shaking table tests.
794
Figure 14. Detachment of isolator at the end of tests of
model M5.
Before
Model test R50 R75 R150 R300
Figure 15. Response of control model M1 during test run
M1 f (s−1 ) 15,6 12,3 12,3 – –
R075.
ζ(%) 13,5 16,7 15,5 – –
M2 f (s−1 ) 19,0 15,3 13,9 – –
ζ(%) 5,5 13,4 13,8 – –
which reduced the response and prevented structural
M4 f (s−1 ) 21,2 20,6 19,5 18,9 12,6
ζ(%) 3,8 5,0 9,5 10,2 8,9 damage.
M5 f (s−1 ) 2,2 2,1 2,1 2,1 2,1 As an indication of efficiency of confining the
ζ(%) 11,6 10,2 10,1 10,0 10,3 structures wth CFRP laminate strips, measured top
acceleration and relative first storey displacement
responses of two of the tested models are shown in
Figures 15 and 16. As a measure of intensity of exci-
measured response. Fourier analysis of acceleration tation, shaking table acceleration time history is also
and displacement records has been used to obtaine plotted in each figure. In the figures, the responses of
these data. Unfortunately, dynamic characteristics of control model M1 and CFRP laminate strengthened
isolated, non-strengthened model M3 have not been model M4 to seismic excitation of the same intensity
measured. are compared (test runs R75 and R300, respectively
As expected, a trend of degradation of the first (see Table 1 for comparison of maximum shaking table
natural frequency of vibration and increase in val- accelerations).
ues of coefficient of equivalent viscous damping can Maximum values of the base shear evaluated on the
be observed with increased intensity of excitation in basis of the measured acceleration responses of the
all cases, except in the case of isolated model M5, models during each test run are compared in Table 3.
which has not suffered damage during testing. The Base shear is given in terms of the base shear coef-
differences in initial values, especially in the case of ficient BSC, i.e. the ratio between the base shear
models M1 and M2 can be mainly attributed to rela- BS developed in the model during shaking and the
tively large scattering of deformability characteristic weight of the model above the base W : BSC = BS/W .
of used model masonry materials. The increase in stiff- Base shear has been calculated as the sum of prod-
ness in the case of model M4, confined with CFRP ucts of masses, concentrated at the levels of floors
laminate strips, however, can be attributed to the effect mi and measured average maximum values of accel-
of confining the model structure horizontally and ver- erations at the same level ai : BS = mi ai . Since
tically with rigid elements. No changes in the first all models were equal, the mass concentrated at the
natural frequency of vibration and damping in the case second floor was taken as m2 = 287 kg, mass at the
of isolated model M5 are the result of seismic isolators first floor m1 = 448,4 kg, whereas the total mass of
795
the case of model M4 without isolators developed in
the model at full intensity of shaking table motion.
4 CONCLUSIONS
796
Gayevoy, A.V., Lissel, S.L. (2004). Monitoring of CFRP pre- Schwegler, G. 1995. Masonry construction strengthened
stressed masonry walls. Proceedings, 13th International with fiber composites in seismically endangered zones.
Brick and Block Masonry Conference, Vol. 2, Amsterdam. Proceedings, 10th European Conference on Earthquake
Hamilton, H.R., Dolan, C.W. 1998. Strengthening unrein- Engineering, Balkema.
forced masonry with FRP. Proceedings, Structural Engi- Triantafillou, T.C., Fardis, M.N. (1997). Strengthening of
neering World Wide, Elsevier Science Ltd.: Paper No. historic masonry structures with composite materials.
T119-5 (CD-ROM). Materials and Structures, 30 (10).
Page, A.W. (1995). The shear capacity of membrane type Triantafillou, T.C. 2001. Seismic retrofitting of structures
damp-proof courses in masonry.Australian Civi Engineer- with fibre-reinforced polymers. Progress in Structural
ing Transactions, Vol. CE37, No.1. Engineering and Materials, 3 (1).
Sarrazin, M., Moroni, M., Boroschek, R., Soto, P. 1996. Zhou, X., Miao, H. 1996. Optimum design of resilience-
Results from a base-isolated experimental building in friction-slide base isolation system for low cost build-
Santiago-Chile. O. Proceedings, 11th World Conference ings. Proceedings, 11th World Conference on Earthquake
on Earthquake Engineering, Acapulco, CD ROM. Engineering, Acapulco, CD ROM.
797
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
D. Rinaldis
ENEA Casaccia Research Center, Italy
ABSTRACT: This paper is focused on the elaboration of the results recorded during some dynamic experi-
mental tests executed on a masonry arch placed on a shaking table facility at Laboratory of the ENEA Casaccia
Research Center. The experimental results are correlated to the theoretical results obtained by means of some
calculus codes, which have been elaborated for applications on masonry structures by the researchers of the
Department of Structural Engineering of the University of Naples “Federico II” and suitably modified for the
study model. The results have shown that the curves solved by the static calculus may be given an exponential
form as the dynamic curves by the tests; moreover, it is evident that the static degradation is much faster than in
the seismic response, probably thanks to kinetic energy absorption that helps the structure to resist inertia forces.
799
Figure 3. Admissible funicular line.
800
The structure is made by yellow tufa bricks tied
by a poor mortar, which is a most common masonry
encountered in South Italy. The intrados profile of the
arcade is semicircular with a radius of 100 cm. The
arcade is composed by two rows of bricks determining
a masonry thickness of 20 cm; the two pillars which
support the arcade have a rectangular base 50 cm and
an height 70 cm; the depth of the whole is 100 cm. The
wing walls, continuing in height the pillars, which have
to contain the overload, are characterized by thickness
35 cm and an height 110 cm.
Moreover, some steel tie-beams fixed by means of
flexible elastic ties are placed between the structure
and the wing walls used for containing the overload,
in order to guarantee the stability during the tests. The
total weight of the masonry structure is 5,1 tons. On the
top of the portal arch an overload of material composed
by crushed tufa and lime and having a weight of 1,4 ton
is applied, in order to simulate the structural context
where the real arcade is included. Sliding is prevented Figure 5. Sensors locations on the masonry arch by
through steel profiles attached to the shaking table. Buffarini et al. (1997) modified.
The total weight of the structure plus the overload
and the steel bars is 7,0 tons. are fixed to some steel trestles integral with the shak-
ing table. A scheme of the arch with the locations of
the sensors composing the monitoring equipment is
3.2 Characteristics of the shaking table and shown in Figure 5 (Buffarini et al., 1997; Clemente
of the recording instruments et al., 1999).
As described in the above, during the tests, the seis-
mic input is transmitted to the structure by means of 3.3 Description of the laboratory tests
a shaking table; the table has the following techni-
cal characteristics: dimension of 4 × 4 m, maximum At first the tuning of the shaking table is realized in
supported weight of 10 ton, six degrees of freedom, order to check the response of the table. To this pur-
frequency range of 0–50 Hz, maximum peak accel- pose, the table is loaded with a fictitious structure
eration of 3 g, maximum velocity of 5 m/sec and simulating the testing structure both in its total weight
maximum span of 25 cm. and in its barycentre position, and the fictitious struc-
In order to evaluate the time histories of the accel- ture is subject to the selected profile, consisting in
eration and displacement some recording instruments the selected time history of the acceleration. The final
are located at some “critical” positions on the arch, objective of the tuning phase is the realization of the
which are fundamental for the determination of the test profile (Buffarini et al., 1997).
structures’ response.
The recording instruments consist of two different 3.3.1 Tests developed on the 1st portal arch
typologies of accelerometers: The first arch is tested in two phases up to the collapse.
In the first phase (Phase A) only an horizontal input is
• n. 20 piezoelectric accelerometers with feed- used; thereafter, in order to approach the collapse con-
through band of 2–15000 Hz (± 10%), and nominal dition of the structure, in the second phase (Phase B),
sensitivity of 10 pC/g, a vertical input is also added to the horizontal one. In
• n. 8 transducers of displacement (LVDT) subdi-
both the two phases the applied excitation acts in the
vided in: plane of the structure, orthogonally to its side at the
– transducers with nominal sensitivity of 0,1 Vmm-
bottom of the portal arch.
1 (± 3%), maximum displacement of ± 2 inch,
In the first phase (Phase A) the characterization of
and feed-through band of 50 Hz,
the arch having an overload on the top of the arcade
– transducers with nominal sensitivity of 0,2 V/mm
fixed to 14,1 kN is developed by means of a white
(± 3%), maximum displacement of ± 1 inch,
noise with amplitude of 0,1 g (as well known, the
and feed-through band of 50 Hz.
theoretical white noise is a disturbance with infinite
The accelerometers are directly applied on the energy uniformly distributed over the whole frequency
masonry arch, while the transducers are located in cor- range. After the initial characterization of the struc-
respondence of the external sides of the two piers, and ture, an excitation corresponding to the time history
801
Figure 6. Curve of the structural frequencies f vs the seismic
coefficient c during the laboratory tests on the first arch.
802
Three cycles of tests with different overload and Table 1. Scheme of the laboratory tests executed on the
excitation are developed in order to study the behavior second arch pre-consolidation.
of the second arch:
Test Solicitation Note
Cycle 1) in the first cycle an overload of 8,5 kN is
applied on the top of the portal arch, and the seis- Cl white noise Overload 8.5 kN. Tests
mic input consists of a white noise for the seismic start (cycle 1)
Pl pink noise_01
characterization of the structure, a pink noise and TI test 01
two tests of 0,1 and 0,2 g. C2 white noise
Cycle 2) In the second cycle an overload of 14,1 kN P2 pink noise_02
(equivalent to the one acting on the first arch) is T2 test 02
C3 white noise
applied and the structure is subjected to two seismic
ClA white noise Overload 14.1 kN. New
tests with peak 0,1 and 0,2 g; the cycle is then inter- start of tests (cycle 2)
rupted because of the detachment of some stones PIA pink noise_0l
at the intrados. T1A test 01
Cycle 3) In the third cycle, started after the restora- C2A white noise
P2A pink noise_02
tion of the broken bricks, an overload of 14,1 kN is T2A test 02
kept, the pink noise which is probably the cause of C3A white noise
the masonry slack is removed, and the seismic tests P3A pink noise_03 Test stop for breaking
are continued up to the peak acceleration of 0,7 g. off of some stones
C1R white noise Restoration of the
By analyzing the results, on the basis of the test broken bricks and
program shown in Table 1 (from the laboratory tests’ new start of the test
schemes by Buffarini et al., 1997 modified) and (cycle 3)
T1R test 01
Figure 8, some considerations can be outlined. C2R white noise
After the dynamic characterization (C1) developed T2R test 02
on the undamaged structure, the first test (T1) has been C3R white noise
developed with the acceleration time history scaled to T3R test 03
C4R white noise
a peak of 0,1 g. At any step the peak value is increased T4R test 04
by 0,1 g, and so on up to a maximum peak value of C5R white noise
0,7 g (T7R), about twice the real peak acceleration T5R test 05
recorded to Sturno, which is equal to 0,34 g. The test C6R white noise
T6R test 06
with a peak of 0,7 g (T7) was interrupted because of C7R white noise
the detachment of some stones at the intrados and then T7R test 07 Test stop for breaking
repeated (T7R) (Clemente et al., 1999). After any test off of some stones.
a characterization of the structure has been developed C8R white noise Characterization
following to the
by operating a white noise input (C1 to C9) plus a pink previous test
noise having an increasing peak value of 0,1 g at any C8RB white noise Characterization
step (P1 to P8). before the testing
In the first three characterizations (C1 to C3) an start
C8RATT white noise Characterization with
overload of 8,5 kN has been imposed on the top of the a scale spectrum 1/2
arch, then the tests have been interrupted and started T7RB test 07 Final test
again by a peak value of 0,1 g and an overload of C9R white noise
14,1 kN has been imposed (C1R to C9R) in order to
make tests homogeneous to the 1st arch.
3.4 Comparison between the dynamic results subjected, approximately according to an exponential
in the laboratory tests curve. Nevertheless the dynamic curve of the first arch
decreases more quickly than the curve of the second
By the comparison between the curves in Figures 6 arch. Probably this effect is due to the different pro-
and 8, which plot the results by the dynamic tests in gram of the shaking sequences to which the arches
laboratory on the masonry arches, some general con- have been subjected during the tests: in the tests of the
siderations can be made about the behavior of the two second arch an increasing pink noise has been cou-
arches (Fig. 9). pled to the Sturno input, with the result to anticipate
The laboratory tests put to evidence some differ- the damage of the arch.
ence between the two arches. Both the arches show This different behavior is pointed out in Figure 9
a typical behavior where the trend of the recorded where the decay of the first own frequencies of the
frequencies decreases with respect to the seismic two arches is shown. The fact is however that the trend
intensity (the coefficient “c = ap /g”) the arch has been of both curves can be approximated by an exponential
803
Figure 8. Dynamic curve of the structural frequencies f(Hz)
vs the seismic coefficient c = ap /g during the laboratory tests Figure 10. Trend of the seismic coefficient c vs the maxi-
for the second arch. mum displacements umax by the static calculus.
curve, the curve relevant to the first arch decreases c = ap /g) can be inferred up to the collapse condition
more quickly than the second one. for c = cf (Fig. 10).
By deriving the seismic coefficient c one gets
the trend of the tangential stiffness K = K(c) = c (u)
as a function of the displacement umax and of the
4 AN INSIGHT ON VULNERABILITY
coefficient c (Fig. 11).
FORECASTING OF A MASONRY
It can be conceived that the own frequency of the
STRUCTURE
structure is a function of the type f2o = α2 Ko , where Ko
is the stiffness of the structure. Since the structure has a
4.1 Procedure for vulnerability evaluation
NT non-linear behavior (Fig. 10), one expects that the
In order to have an estimate of the seismic vulnerability frequency depends on the stress intensity, that is mea-
of a existing masonry structure under a seismic input, sured by “c”. So, referring to the tangential stiffness
an attempt is made to draw conclusions on the basis of K(c), one grossly puts f2 (c) = α2 K(c).
dynamic identification coupled with static analysis, in After dynamic characterization on the structure in
the light of the above experimental survey. site (e.g. by soliciting the fabric with a white noise
At first, the static analysis yielding the fundamental or other) one identifies the own frequency fo before
elements of the examined masonry structure can be any earthquake strikes on the √ building. This allows to
solved by considering only the geometric dimensions estimate the coefficient α = fo Ko , after calculating
and some other easily collectable data. So the trend of the initial stiffness
√ Ko . It is found that, heuristically, the
some characteristic displacement parameter uc (e.g. curve f(c) = α K(c) follows a exponential proceeding
the maximum displacements of points of the structure (Fig. 12). This curve is referred to in the following as
umax ) with respect to the seismic coefficient “c” (with the “static damage progression”.
804
where “c” is the intensity of the worst earthquake the
structure has suffered in the past and to which it has
survived. At the same time the own frequency “fo ”, i.e.
the easiest parameter that can be evaluated by dynamic
identification, evolves with increasing the damage of
the structure.
So, if one considers the derivative of the expression
in Eq. (4)
805
Figure 14. The experimental evidence shows that dynamic Figure 15. Experimental absorption of inertia forces in
interaction with inertial forces strongly improves the seismic accordance with Eq. (12).
performance of the structure.
806
5 CONCLUSIONS REFERENCES
By comparing the dynamic curves of the tested arches Baratta, A. 1995. The no-tension approach for structural
(Fig. 9), which can be well approximated to an expo- analysis of masonry buildings. Proc. of the Fourth Inter-
national Masonry Conference, British Masonry Society,
nential form, one observes that the curve relevant to
London.
the first arch decreases more quickly than the second Baratta, A. 1998. Scale influence in the static anc dynamic
one. Moreover, by the comparison of the exponen- behaviour of no-tension solids. Proc. SMART 98 Confer-
tial approximations of the static and dynamic curves ence, Varsavia.
(Fig. 12), it is evident that the static degradation is Baratta, A. 2007. Active control technology for dynamic
much faster than the dynamic one. This effect is prob- shaking-table tests on scaled models of masonry struc-
ably caused by the opening of the fractures in the arch tures. Proc. of Thematic Conference on Computational
that produces an increasing absorption of the oscil- Methods in Structural Dynamics and Earthquake Engi-
lation energy as kinetic energy, mainly due to the neering, ECCOMAS, M. Papadrakakis, D.C. Charmpis,
N.D. Lagaros, Y. Tsompanakis (eds.), Rethymno, Crete,
progressive activation of a collapse mechanism, rather
Greece.
than as elastic energy. The considerations illustrated in Baratta,A. & Corbi O. 2003. The No Tension Model for
Sec. 4.3 enable a transformation of the static analysis the Analysis of Masonry-Like Structures Strengthened by
results to produce a "seismic response expectation" as Fiber Reinforced Polymers. Intern. Journal of Masonry
in Fig. 16. Looking at the results of the static analysis International, British Masonry Society, vol. 16 No. 3:
elaborated to yield seismic expectation (Fig. 17) the 89–98.
second arch has to produce larger displacements than Baratta, A. & Corbi O. 2005. On Variational Approaches in
the first one, to store a given amount of elastic energy. NRT Continua. Intern. Journal of Solids and Structures,
This means that a larger part of the external energy Elsevier Science, Vol. 42, pp. 5307–5321.
Bintrim, J.W., Laman, J.A. & Boothby, T.E. 1998. Dynamic
displayed by the earthquake has to be transformed by
testing of masonry arch bridges. In Sinopoli A. (ed.)
the second arch in kinetic energy through the activa- Arch bridges, Proc. Second International arch Bridge
tion of a mechanism. The larger the amplitude of such Conference held in Venice: 295–303, Rotterdam, Balkem.
mechanism the faster the arch approaches collapse. Brown G., Pretlove A.J., Ellick, J.C.A., Hogg, V. & and
Choo, B.S. 1995. Changes in the dynamic characteristics
of a masonry arch bridge subjected to monotonic load-
ACKNOWLEDGEMENTS ing to failure. In Melbourne C. (ed.) Arch bridges, proc.
First Internationa Arch Bridge Conference held in Bolton,
The laboratory tests were funded by the ENEA London:375–383.
Casaccia Research Center (New technology Depart- Buffarini, G., Clemente P. & Rinaldis D. 1997. Analisi speri-
ment, Rome, Italy) and by the Authority of the Monu- mentale su tavola vibrante di un arco in muratura. Internal
Report of ENEA Casaccia Research Centre, Rome, Italy
mental Heritage Supervision in Benevento and Caserta
(in italian).
Provinces. The recent advancements have been per- Clemente, P., Buffarini, G., Rinaldis, D. & Baratta, A. 1999.
formed thanks to the financial support by the Depart- Changes in the dynamic characateristics of a masonry arch
ment of Civil Protection of the Italian Government, subjected to seismic actions. Proc. of the Fourth Euro-
through the RELUIS Pool (Convention n. 540 signed pean Conference on Structural Dynamics, EURODYN’
07/11/2005, Research Line no. 8). 99, Prague, Czech Republic.
807
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: This study aims at investigating the bond performance of fiber reinforcement with cement-based
matrix applied on masonry structures. The preliminary results of an experimental campaign addressed to explore
the influence of anchorage length, the effect of the discontinuity induced by mortar joint and the consequences
of fiber stiffness and roughness are presented; different specimens, made either with brick or with brickwork,
and reinforced either with carbon fibers or with steel fibers, have been tested when varying the anchorage length.
The failure modes that take place have been detected together with the corresponding debonding force and the
slip at collapse. Besides, τ-slip relationships have been deduced from experimental data.
809
Figure 1. Steel cords on a polyester net. Figure 2. Carbon fibers on a polyester net.
810
Figure 4. Test setup.
811
Figure 5. Failure modes.
812
Table 2. Experimental plane.
A-SRG1 3995 a –
A-SRG2 3883 a –
A-SRG3 3455 a –
A-CFRCM1 2425 b –
A-CFRCM2 2175 b –
A-CFRCM3 2237 b –
B-SRG1 6368 b –
B-SRG2 6190 b 0.36
B-SRG3 6358 b 0.27
B-SRG4 7516 b –
B-SRG5 6754 b –
B-SRG6 6462 b 0.23
B-CFRCM1 3875 c 0.58
B-CFRCM2 4074 c 1.18
C-SRG1 6166 b 0.72
C-SRG2 5977 b –
C-SRG3 5589 b 0.82
C-SRG4 5438 b 0.42
C-SRG5 5309 b 0.69
C-SRG6 5599 b 0.57
813
has occurred; so it can be deduced that the effective
anchorage length of SRG on masonry specimens lays
between 167 and 297 mm. In the case of brick speci-
mens, as shown by the other strain profiles in figure
11, a similar effective length is found for SRG rein-
forcement, while an effective anchorage length not
higher than 150 mm can be estimated for C-FRCM
reinforcement. As expected, a higher stiffness of the
reinforcement is the consequence of a shorter effective
anchorage length.
814
Borri, A. Castori, G. Menestò, F. 2007. Interventi con Matana, M. Nanni, A. Dharan, L. Silva P. Tunis, G. 2005.
SRG sulle volte di palazzo Jacobilli in Foligno. XII Bond performance of Steel Reinforced Polymer and
Convegno L’Ingegneria Sismica in Italia ANIDIS. Pisa: Steel Reinforced Grout. Proceedings of International
10–14 June. Symposium on Bond Behaviour of FRP in Structures.
Briccoli Bati, S. Rovero, L. Tonietti, U. 2007. Adesione tra International Institute for FRP in Construction.
blocchi in laterizio e rinforzo in CFRP. Workshop: Materi- Mazzotti, C. Savoia, M. Ferracuti, B. 2007. The effect of
ali ed Approcci Innovativi per il Progetto in Zona Sismica surface preparation on FRP-concrete bond strength, Work-
e la Mitigazione della Vulnerabilità delle Strutture. Uni- shop: Materiali ed Approcci Innovativi per il Progetto in
versità degli Studi di Salerno, Consorzio Reluis. Fisciano: Zona Sismica e la Mitigazione della Vulnerabilità delle
12–13 February. Strutture. Università degli Studi di Salerno, Consorzio
Di Tommaso, A. Focacci, F. Mantegazza, G. Gatti, A. 2007. Reluis. Fisciano: 12–13 February.
FRCM vs FRP composites to strengthen RC beams: a Mosallam, A. 2007. Out-of-plane behaviour of unreinforced
comparative analysis. FRPRCS-8, University of Patras. brick walls strengthened with FRP composites, Compos-
Patras: 16–18 July. ites Part B: engineering 38: 557–574.
Ferracuti, B. Savoia, M. Mazzotto, C. 2007. Interface law Neale, K.W. 2000. FRPs for structural rehabilitation: a survey
for FRP-concrete delamination. Composite Structures 80: of recent progress, New materials in construction, Prog.
523–531. Struct. Engng Mater. 2: 133–138.
Huang, X. Birman, V. Nanni, A. Tunis, G. 2005. Properties Triantafillou, T.C. 1998. Strengthening of masonry struc-
and potential for application of Steel Reinforced Poly- tures using epoxy-bonded FRP laminates. ASCE Journal
mer and Steel Reinforced Grout composites. Composites of Composites for Construction 2(2): 96–104.
Part B: engineering 36: 73–82. Valluzzi, M.R. Valdemarca, M. Modena, C. 2001. Behaviour
Lu, X.Z. Teng, J.G. Ye, L.P. Jiang, J.J. 2005. Bond-slip mod- of brick masonry vaults strengthened by FRP lami-
els for FRP sheets/plates bonded to concrete. Engineering nates, ASCE Journal of Composites for Construction 5(3):
Structures 27: 920–937. 163–169.
815
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The paper deals with the laboratory tests on timber-steel-concrete composite beams, equipped
with an innovative connection system. Tests are part of a more comprehensive experimental campaign, aiming at
the evaluation and optimization of the connection devices. In this paper both monotonic and cyclic bending tests
on six composite beams are presented. In particular, each specimen consists of a timber beam with rectangular
cross section, connected to a concrete reinforced slab by means of steel collars. Specimens differ for the type
of collar-timber contact and the collars spacing, in order to determine the more effective solution. The system
performance has been evaluated in terms of both maximum force-displacement relationship and accumulated
dissipated energy at failure.
817
Figure 2. Bending tests on timber beam: static scheme [cm].
Specimen 1, 2, 3 4, 5, 6
818
Table 4. Characteristics of tested specimens.
819
Figure 6. Force-time histories for specimens n. 1–5.
The experimental results are presented in terms of load 5 COMPARISON AMONG RESULTS
vs absolute or relative displacement curves, according
to the aforesaid displacement parameters, measured by 5.1 General
a number of 3 to 16 transducers. Comparison among experimental results are presented
The sign convention used for the collar-beam slip- in terms of load vs both absolute and relative displace-
ping is given in Figure 10: such as, if the top of a device ment curves (Fig. 15).
moves to the right, the transducer at its right will have Aim is the evaluation of the effects of the type of
a positive displacement. beam-collar contact (rubber or rough steel), the spac-
For the sake of synthesis, all diagrams related to ing between devices (constant or variable) and the
specimen n. 1 are plotted in Figures 11, 12, 13; in additional bolted connection between steel collar and
particular in Figure 13 each diagram shows the results timber beam.
820
Figure 11. Specimen 1: a) load vs beam deflection curves;
b) load vs beam-slab split curves.
821
Figure 13. Specimen 1: load vs relative displacement
curves.
822
Table 5. Values of the strength and stiffness parameters.
Fu δu Fy δy E1 E2 N. Er E40
Spec. kN mm kN mm GPa GPa cycles kNm kNm
823
demonstrates the efficiency of the collar as a con-
nection system;
– variable spacing of the collars (according to shear
distribution) provides a better behaviour as respect
to the constant one;
– direct contact between the steel rough surface and
the timber beam improves the behaviour of the com-
posite beam as respect to the contact between the
surfaces through the rubber layer;
– the connection of the collar with internal rough sur-
face to the beam by means of a mechanical device
provides a small increase of both strength and stiff-
ness; besides the collar induces a significant damage
Figure 18. Effect of the type of collar-beam contact. of the beam contact surface, which does not justify
the choice of this solution.
contact surface. The accumulated dissipated energy – the composite system by collars provides an incre-
up to failure for SP6 is about 35% larger than for SP3. ment of strength and stiffness of more than two times
In order to evidence the increase of both strength as respect to the timber beam.
and stiffness of the composite system as respect to the
simple timber beam, reference has been made to the Further investigations, both experimental and
second moment of area. For the composite section this numerical ones, are ongoing, in order to achieve
has been related to a measured elastic deflection equal the system optimization and its structural character-
to 5 mm, which corresponds to a 10 kN applied force ization, which would allow to set up a analytical
(Fig. 15a), according to the static scheme plotted in formulation for design purposes.
Figure 2, for all the specimens. As a result, for the com-
posite and the timber beams the second moments of
area are equal to 7415 cm4 and 3413 cm4 respectively, ACKNOWLEDGEMENT
the increment factor being about 2.2.
The research activity is developed within the PRO-
HITECH International project “Earthquake Protection
6 CONCLUSIVE REMARKS
of Historical Buildings by Reversible Mixed Technolo-
gies”, coordinated by F.M. Mazzolani.
The present paper illustrates both monotonic and
cyclic bending tests carried out on timber-steel-
concrete composite beam realized by using a com-
pletely reversible connection device. The base system REFERENCES
is a steel collar, located astride the timber beam with or
without the interposition of a rubber layer. Collars are Faggiano B., Marzo A., Formisano A., Mazzolani F.M. 2007.
made in two parts bolted together at ad hoc wings. The Innovative steel connections for a composite wood –
steel – concrete floor: a numerical investigation. Com-
function of connector is held by a steel stud welded at puter & Structures (Elsevier – K. J. Bathe and B.H.V.
the superior side of the collar. The final configuration Topping editors, ISSN 0045-7949), submitted.
of the adopted collars was obtained from a previous Faggiano B., Marzo A., Mazzolani F.M. and Calado L. 2007.
experimental campaign on a single connection system. Experimental analysis of rectangular shaped sleeve con-
Tests illustrated concern the composite timber- nectors for composite wood-steel-concrete floors: push-
steel-concrete system with a rectangular shaped beam out tests. Proc. of the 9th International Conference
cross section. The timber beam is made of new pine “Modern Building Materials, Structures andTechniques”,
wood, therefore no irregularities at the surfaces were Vilnius, Lithuania, 16–18 May.
present, except for knots. Several configurations of Giurani, E. 2002. Timber beam floors strenghtened by means
of a collaboratine concrete slab: design criteria. Edilizia
the composite system have been analysed by varying (4): 32–40 (in Italian).
both collar spacing and type of collar-beam contact. Gelfi, P.; Giurani, E. and Marini, A. 1998. Behaviour of
Finally, the set up composite systems have undergone stud shear connectors in the composed timber-concrete
both monotonic and cyclic bending tests. beams: theoretical modelling and experimental compar-
The comparison among experimental results has isons. Proc. of the 3rd Italian Workshop on the Composed
been performed in terms of force vs absolute and Structures, Ancona, Italy, 29–30 October (in Italian).
relative displacements. It has been evidenced that: Mungwa, M. S., Jullien, J. F., Foudjet, A. and Hentges, G.
1999. Experimental study of a composite wood-concrete
– relative displacements between the beam-slab beam with the INSA-Hilti new flexible shear connector.
and the beam-collar are almost negligible, what Construction and Building Materials (13): 371–382.
824
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: To study the dynamic behavior of typical temples of the 17th and 18th Centuries stone masonry
construction in Mexico, a shaking table test program is being performed on a 1:8 scale model built with the
same materials and techniques than its prototype. Results of a first stage of the research in which the model was
submitted to horizontal and vertical motions of increasing intensities are presented in this paper. The test was
stopped at an intensity corresponding to a base shear force of about 40% of the total weight, causing a damage
that was considered as possible to be repaired. Vertical component of the motion showed to significantly affect
the response and to increase the damage of the model. Damage patterns were similar to those observed in actual
temples. Main features of the measured response are compared to those computed from a linear elastic, finite
element model. The degree of damage and the variation in the amplification of the response are correlated to
the variation of the fundamental vibration period of the model.
825
a) Plan view
b) Front view
826
Figure 3. Construction of the model.
827
8 16
6 14
12
Acceleration (m/s2)
4
Amplitude (m/s2)
2 10
0 8
9.281775
0 5 10 15 20 25 30 35 40 45 50 55 6
-2
4
-4
2
-6
0
-8
0 0.2 0.4 0.6 0.8 1
Time (s)
Period (s)
6 16
5 14
4
Amplitude (m/s2)
12
Acceleration (m/s2)
3
2 10
1 8
9.281775
0
-1 0 6
5 10 15 20 25 30 35 40 45 50 55
-2 4
-3 2
-4
-5 0
Time (s) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Period (s)
by fixing displacement transducers to two stiff steel state (0.088 s), but also for the longer periods that
frames placed outside the shaking table. are expected when the structure be damaged by the
The total weight of the model was 11.96 t; an addi- earthquake.
tional weight of 2 t was evenly distributed on the vault Response to ambient vibration and to a white noise
in order to increase the inertia forces on the main shaking with small amplitude, was measured before
body of the model, as well as the vertical compressive starting tests with the selected ground motions. Initial
stresses on the walls, thus producing a slight correc- natural periods of vibration were derived from these
tion of the difference with respect to that required by tests as 0.088 s.
the dimensional analysis. When initiating the seismic tests, a malfunction-
Acceleration time-histories to be applied to the table ing of the software controlling the actuators gave rise
in the horizontal and vertical directions were derived to a strong, abrupt shaking of the table that produced
from a strong motion record obtained in 1985 near the significant damage to parts of the building: diagonal
epicenter of a large magnitude earthquake (Ms 8.1). cracking of the facade, flexural cracking on the right
The seismic records were modified in their time scale longitudinal wall and buttresses, and some disloca-
in the order to enhance their effects in the model. tion of the vault. The model was carefully repaired,
The horizontal motion was applied in the transverse first restituting its initial geometry by tensioning some
direction of the model. Their acceleration scale was external steel tie rods, and then injecting cracks with a
increased to almost reach the capacity of the table. lime grout. Natural period measured at the end of the
The time scale was reduced to produce highest spec- process were close to those of the undamaged model
tral amplitudes in the neighborhood of the fundamental (0.08 vs. 0.088 s).
period of vibration of the structure. Some modifica-
tions were made to the initial part of the vertical record 5 BEHAVIOR OF THE MODEL DURING THE
in order its maximum acceleration occur at the same SEISMIC TESTS
time than that of the horizontal record. The resulting
time histories and their acceleration response spec- Intensity of the selected seismic input was scaled down
tra for 5% damping are shown in Fig. 4. As it can to different fractions of the reference records, start-
be appreciated, the combination of acceleration his- ing with 5% and rising up to 60%. For each level
tories is very severe for the structure under study. of intensity, first only the horizontal component was
The maximum acceleration is very high in both direc- applied, and then horizontal and vertical motions were
tions; the duration of the strong motion is very large simultaneously induced.
(about 50 s); their frequency content gives rise to The description of the set of tests performed and the
large spectral amplitudes not only for the fundamental maximum acceleration applied at the shaking table for
period of the structure corresponding to its undamaged each intensity level are shown in Table 1.
828
Table 1. Test program and main results.
Seismic Horizontal
Test intensity Applied
No. % components m/s2 Vertical Damage description
6 MEASURED RESPONSE
829
Table 2. Maximum measured accelerations for different Table 3. Maximum measured displacements for different
intensities of applied horizontal motion. intensities (mm).
A E A E A E A E
Intensity
% (mm)
830
9
W. noise Cal. H 20% 0.13
f = 9.03 Hz f = 11.15 Hz
8
Cal. H 10% 0.12
Cal. H 40% f = 12.06 Hz
f = 9.56 Hz W. noise
0.11
Period [s]
f = 12.51 Hz
7 Cal. H 5%
Cal. H 60% f = 12.46 Hz 0.1
f = 7.76 Hz
6 0.09
5 0.08
0.07
4
0 5 10 15 20 25 30 35 40 45 50 55 60 65
3
Seismic intensity [%]
2
a) Horizontal
1 0.09
0 0.08
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Frequency [Hz] 0.07
Period [s]
0.06
Figure 6. Ratios of Fourier’s spectral amplitudes measured 0.05
at the top of the vault to those at the base. 0.04
0.03
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Table 5. Vibration periods derived from response to differ-
Seismic intensity [%]
ent intensities of the motion.
b) Vertical
Lateral vibration Vertical
vibration Figure 7. Variation of vibration period with the intensity of
H HV HV the motion.
Condition Period
Test of the Experimental Analytical
no. Description model (s)
7
∗ 6
Analytical 0.097 0.06
∗∗ 5
Amplitude
831
Table 6. Damping derived from response to favored their overturning, and most importantly on the
different intensities of the motion. vault, where the vertical inertia forces produced an
important vertical vibration causing variation in thrust
Intensity
[%] Damping
on the supports, and an opening and closing of the vault
that increased lateral displacements in one side of the
10 0.07 supporting walls thus affecting their stability. It can be
20 0.08 assumed that buildings of this kind would be particu-
40 0.11 larly vulnerable to earthquakes with nearby epicenters,
60 0.14 which show high vertical accelerations of the ground.
A good correlation was found among the amount
those commonly found in modern structures, indicat- of damage and the corresponding period of vibration
ing a large energy dissipation, which for low intensity of the structure, which constitute an effective index of
motion could be attributed to plastic behavior of the residual capacity of the structure. The gradual decrease
mortar, and for higher intensity to sliding along the of frequency reflected the loss of model stiffness that
profuse cracking of masonry members. reached a factor of about two, for the maximum applied
intensity.
Lateral displacements of the model did not exceed
9 CONCLUSIONS
few millimeters; it can be assumed that due to their
high stiffness and to their low period of vibration,
The model and its testing program are providing valu-
displacements of structures of this kind remain quite
able information about the seismic response of stone
small until they are near to failure.
masonry construction in general, and of this kind of
Response computed by finite element linear anal-
colonial temple in particular. Despite of the damage
yses is similar to that measured in the tests, only for
suffered by the model before its testing, the rehabili-
rather low levels of shaking intensity. For predicting
tation returned the model near to its initial conditions,
the response in near to failure conditions, use of non
though with some stiffness increase in the zones
linear models is necessary.
injected with grout.
Damage produced in this phase of the is consid-
The motion applied to the shaking table was par-
ered as repairable; then, the next stages of the research
ticularly severe for this type of building, because of
program will aim at evaluating the effectiveness of
its high spectral ordinates in a wide range of periods,
some commonly used rehabilitation techniques, and
corresponding to the initial and damaged states of the
the ability of non linear methods of analysis to predict
structure. Maximum induced accelerations reached
the seismic capacity and general vulnerability of this
4.53 m/s2 at the roof and 7.75 m/s2 at the top of the
kind of structure.
towers, and the maximum base shear coefficient was
almost 40%.
The amplification of the acceleration at the roof REFERENCES
was rather small, not exceeding 40%, nevertheless, the
acceleration at the top of bell towers was almost three Chávez M. 2005, Estudio experimental de las propiedades
times the acceleration of the base for low intensity base mecánicas de mamposterías de piedra natural (Experi-
motions. mental study on mechanical properties of stone masonry),
Damage patterns induced in the model corre- Master Dg Thesis, UNAM, México.
sponded to those observed in actual temples hit by Lourenco, P. B., 2002. Computations on historic masonry
intense earthquakes. Most vulnerable parts of the structures. Progress in Structural Engineering and Mate-
rials 2002, John Wiley & Sons. Ltd.
structure showed to be: the bell fry suffering great
Meli R. & Peña F. 2005, On elastic models for evaluation of
amplifications due to its flexibility; the main body the seismic vulnerability of masonry churches, Structural
of the towers which is prone to separating from the Analysis of Historical Constructions-Modena, Lourenço
facade and failing by shear; the facade itself is prone & Roca (eds), November, Padova, Italy.
to in plane shear failure; the central part of the nave Morales H., 2004, Análisis de la respuesta sísmica de un tem-
whose lateral drift produces bending damage in walls plo colonial típico en cruz latina (Analysis of the seismic
and buttresses, as well as dislocation of the vault. It response of a typical Latin cross colonial temple), Master
must be taken into account that other modes of failure Dg thesis, UNAM, México.
due to the ground motion in the longitudinal direction Rinawi,A. & Clough, R. 1992, Improved amplitude fitting for
frequency and damping estimation, Proc. 10th Int. Modal
of the temple were not included in the study.
Analysis Conference. Soc. for Exp. Mechanics, Bethel,
The effect of the vertical component of the motion Conn, 1992, pp. 893–898.
showed to be rather severe. The main differences with Tomazevic M. & Velechovosky T. 1992, Some aspects of
the case where only the horizontal component was testing small-scale masonry building models on sim-
applied, were noticed in the bell towers where the ple earthquake simulators, Earthquake Engineering and
reduction of the net compressive forces on the columns Structural Dynamics, Vol. 21.
832
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Artur O. Feio
Madeicávado – Madeiras, S. A. University Lusíada, Famalicão, Portugal
Paulo B. Lourenço
University of Minho, Guimarães, Portugal
José S. Machado
LNEC, Lisboa, Portugal
ABSTRACT: The safety and quality of traditional timber structures mostly depends on the conditions of
their connections. A testing campaign has investigated the behaviour of mortise and tenon timber joint. The
present paper addresses the quantification of the strength capacity of wood-wood mortise and tenon joint by
physical testing of full-scale specimens. In addition, the performance of different non-destructive techniques
(NDT) for assessing global strength is also evaluated. For this three non-destructive methods (ultrasonic testing,
Resistograph and Pilodyn) are considered and the possibility of their application is discussed based on the
application of simple linear regression models. New (NCW) and old (OCW) chestnut wood, obtained from
structural elements belonging to old buildings, is used. Finally, a nonlinear model has been formulated and the
predicted behaviour compared with the behaviour observed in the full-scale experiments, in terms of failure
mode and the ultimate load. The study has shown very good agreement with the experimental values.
833
Table 1. Average values of density (one specimen for each
timber element).
Density (kg/m3 )
834
Table 2. Average values of density (one specimen for each
timber element).
Ultimate
Force (kN) Average Std. Dev. Group
835
240 load-displacement diagrams. It is evident as addressed
220
before, that the scatter of the OCW group is much
200
larger than the scatter of NCW group, due to J_7 spec-
180
160
imen. From the load-displacement diagrams obtained
140
the following relevant remarks can be drawn. In a
Force (kN)
120
a constant stiffness, which is higher than the loading
100
stiffness.The justification of this behaviour is probably
80
Max. Displacement = 11.87 mm attributed to the nonlinear behaviour of the interface
60 between rafter and brace, which exhibits a closure phe-
40 nomenon. Finally, after the conventional maximum
20 OCW load the displacement increases rapidly with a much
0
0 1 2 3 4 5 6 7 8 9 10 11 12
lower stiffness, due essentially to the compressive
Vertical Displacement (mm) failure of the wood in the rafter around the joint.
(b) For the purpose of a more refined numerical anal-
ysis, the true load-displacement diagrams were cor-
Figure 4. Envelope of load-displacement diagrams for: rected with an offset that eliminates the upward curve
(a) the NCW group, and (b) the OCW group. related to the nonlinear behaviour of the joint previous
to full contact (joint closure).
The specimens were executed avoiding the presence Due to the nature of load-displacement diagrams for
of large defects but some small defects were present. these joints, choosing the linear portion of the curve
During the tests it was observed that the longitudinal includes some subjectivity. To reduce the subjectivity
and radial cracks of moderate width in the rafter did not the elastic stiffness was calculated between 1/3 and 2/3
have a considerable influence in the ultimate strength of the ultimate load. The line plotted between these two
and in the global behaviour of the joints. The longitu- points to visually analyze the quality of the fit to the
dinal cracks show the tendency to close their thickness linear portion of the plot, indicated that the proposed
and the radial cracks show the tendency to open. This procedure is adequate.
effect is more salient when the cracks are close to
the joint. On the other hand, the cracks present in the
brace, namely the longitudinal ones, show a tendency
to propagate and to open during the tests. Neverthe- 4 DESCRIPTION OF NON-DESTRUCTIVE
less, it seems difficult to quantify the influence of these TEST PROCEDURES
cracks in the ultimate strength of the joint.
In order to investigate possible correlations and the
3.2.2 Load-displacement diagrams validity of using NDE as a tool to assess the joint
The difference in the results between old and new wood strength, different non-destructive techniques (NDT’s)
is very low, which seems in agreement with the val- have been carried out. The adopted NDT’s are the Pilo-
ues of density found for the sample, where the NCW dyn, the Resistograph and the ultrasonic tests, which
group present slightly higher values of density (≈5%) were carried out in both timber elements. Average
in comparison with the OCW group. values were considered in all measurements and two
The results of all tests in terms of load-displacement readings per specimen, per side, were generally made
diagrams, given by the vertical force vs. vertical but a third one was added if the two first readings
displacement, are given in Figure 4. differed significantly. Here, it is noted that Pilodyn
Figure 4a and Figure 4b show typical individ- and Resistograph have been carried out in samples
ual load-displacements diagrams and envelopes of removed from the elements ends, in order not to affect
836
Table 3. Average results of the Resistograph and Pilodyn 180
Fult, joint = -94.13 + 0.051 x V
Tests (values in bits/mm and mm, respectively). 170
Fult, joint = -306.57 + 0.115 x V
r2 = 0.61 r2 = 0.68
160
Resistograph Pilodyn
150 Fult, joint = -116.67 + 0.055 x V
130
J_1 449.5 449.5 8.0 8.0 NCW
120
J_2 367.7 367.7 7.8 8.8
J_3 365.0 365.0 8.0 7.3 110 Rafter and Brace
Rafter
J_4 463.6 463.6 8.0 7.3 100 Joint
Linear Fit (Rafter and Brace)
J_5 391.7 391.7 8.0 8.2 OCW Linear Fit (Rafter)
90
J_6 332.0 332.0 8.0 7.3 Linear Fit (Joint)
J_7 396.6 396.6 9.0 8.8 80
3200 3400 3500 3800 4000 4200 4400 4600 4800 5000 5200
J_8 323.1 323.1 8.7 8.2
Velocity (m/s)
837
Table 4. Adopted elastic and inelastic material properties. 220
200
Ex Ey Gxy νxy 180 Kinfinite KSpring, fit
160
2 2 2
800 N/mm 8500 N/mm 1500 N/mm 0.3 140
fc,x fc,y β γ
Force (kN)
120 Kfit
7 N/mm2 45 N/mm2 −1.0 3.0
100
80
60
40 Numerical
Nonlinear analysis is used to trace the equilib- Experimental
20
rium path up to and beyond the first critical point,
0
at which the structure becomes unstable. There is one 0 1 2 3 4 5 6 7 8
838
220
200
180
160
140
Force (kN)
120
100
80
Experimental
60 Numerical (kn=0.5)
40 Numerical (kn=1.0)
Numerical (kn=2.0)
20
0
0 1 2 3 4 5 6 7 8
200
180
marginal. The usage of infinite stiffness for the inter-
160
face (rigid joint) results in an increase of the slope
140
of the first part of the response, from 30 kN/mm to
Force (kN)
120
80 kN/mm (+266.7%). The ultimate strength of the
100
joint, given by an offset of the linear stretch by 2% in
80
terms of strain values, also changes from 130 kN to Experimental
60
152 kN (+17%), once the joint becomes fully rigid. 40
Numerical (ks=0.5)
Numerical (ks=1.0)
Figure 7 shows the contour of minimum principal 20 Numerical (ks=2.0)
stresses at the end of the analysis. It is possible to 0
observe a concentration of stresses in a narrower band 0 1 2 3 4 5 6 7 8
with peak stresses at the joint (zone where the interface Vertical Displacement (mm)
elements were placed). (b)
With this concentration of stresses one may say that
failure is clearly governed by wood crushing where, for Figure 8. Effect of the variation of parameter: (a) kn , and
a late stage of the analysis, the compressive strength (b) ks on the model response.
of the wood in the joint is completely exhausted. This
situation is also confirmed in the experiments. an offset of the linear stretch by 2%, increases from
127.2 kN to 135.0 kN (+7%).
The reduction/increase of the normal stiffness of the
7 EFFECTS OF THE MATERIAL PARAMETERS interface also affects the global stiffness of the joint:
the global stiffness of the joint decreases as the normal
A strong benefit of using numerical simulations is stiffness of the interface decreases, being more sensi-
that parametric studies can be easily carried out and tive to this variation when compared with the ultimate
the sensitivity of the response to the material data strength. The reduction of 50% of the kn parameter,
can be assessed. There are a total of six key param- results in a decrease of the slope of the first part of the
eters in the present model and the effect of each response, from 32 kN/mm to 26 kN/mm (−23%).
parameter on the global response will be analyzed sep- On the other hand, the multiplication by a factor
arately. It is noted that moderate variations (±25%) of 2 of this parameter results in an increase of the
are considered for the strengths and large variations slope of the first part of the response, from 32 kN/mm
(division/multiplication by two) are considered for the to 41 kN/mm (+28%). Because this parameter sets
stiffness values. These assumptions are rooted in the the relation between the normal traction and the nor-
fact that strength is usually better known than stiffness. mal relative displacement, the obtained results were
expected a priori.
839
220 220
200 200
180 180
160 160
140 140
Force (kN)
Force (kN)
120 120
100 100
80 80
Experimental Experimental
60 60
Numerical (Ex=0.5) Numerical (fc,y=0.75)
40 40
Numerical (Ex=1.0) Numerical (fc,y=1.0)
20 Numerical (Ex=2.0) 20 Numerical (fc,y=1.25)
0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Vertical Displacement (mm) Vertical Displacement (mm)
Figure 9. Effect of the variation of the elastic modulus of Figure 10. Effect of the variation of the compressive
elasticity (Ex ) on the model response. strength (fc,y ) on the model response.
decreases as the ks parameter decreases. The reduction (+23%). However, the global stiffness of the joint is
of 50% of the ks parameter, results in a decrease of the insensitive to the variation of the compressive strength
slope of the first part of the response, from 32 kN/mm perpendicular to the grain.
to 28 kN/mm (−14%).
On the other hand, the multiplication by a factor of 8 CONCLUSIONS
2 of this parameter results in an increase of the slope
of the first part of the response, from 32 kN/mm to Despite the wide use of mortise and tenon joints
37 kN/mm (+16%). in existing timber structures scarce information is
available for design and in situ assessment. The objec-
7.3 Elastic modulus tive of the present study was to quantify its strength
The effect of the variation of the elastic modulus of capacity by physical testing of full-scale specimens.
elasticity parallel and perpendicular to the grain was Also, the performance of different NDT for assessing
considered individually. Figure 9 indicates that the ulti- global joint strength is evaluated. Finally, the adequacy
mate strength is almost insensitive to the variation of of an anisotropic failure criterion to represent the
the elastic modulus of elasticity for wood (± 4%). behaviour of a traditional mortise and tenon joint was
The inclusion of the effects of the elastic modulus assessed from the comparison between experimental
of elasticity does change significantly the elastic stiff- and numerical results.
ness of the joint. Therefore, decreasing the parameter The difference in the results for the ultimate load
E decreases the global stiffness of the joint. The reduc- between the two groups is very low, which is in agree-
tion of 50% of the Ex parameter, results in a decrease ment with the values of density found for the sample.
of the slope of the first part of the response, from Thus, safety assessment of new and existing timber
32 kN/mm to 28 kN/mm (−14%). On the other hand, structures can be made with similar mechanical data.
the multiplication by a factor of 2 of this parameter With respect to the usage of NDT for the predic-
results in an increase of the slope of the first part of tion of the ultimate strength, the dispersion found
the response, from 32 kN/mm to 36 kN/mm (+13%). for the density, Resistograph and Pilodyn do not rec-
ommended the usage of the related parameters for
quantitative mechanical assessment. On the contrary,
7.4 Compressive strength
ultrasonic testing provides good correlations. Novel
The ultimate strength and the global stiffness of the linear regressions have been proposed in this study.
joint are insensitive to the variation of the compressive The different failure mechanisms observed in the
strength of wood in the parallel direction. experiments are well captured by the model, which is
Figure 10 indicates higher sensitivity of the ultimate the most important validation of any simulation. It is
strength of the joint to the variation of the compres- striking that such excellent agreement is obtained also
sive strength of wood in direction perpendicular to in the load-displacement diagrams.
the grain, as expected: with a reduction of 50%, the A preliminary analysis considering an infinite stiff-
ultimate strength of the joint, given by an offset of ness of the interface, assuming a fully rigid connection,
the linear stretch by 2‰, decreases from 130 kN to indicates that such an assumption provides too stiff
100 kN (−30%); multiplying by a factor of 2 the ulti- results. Another conclusion is that the normal stiffness
mate strength of the joint, given by an offset of the of the interface elements has considerable influence
linear stretch by 2‰, increases from 130 kN to 160 kN in the yield strength of timber joints. The numerical
840
results, in terms of force-displacement diagrams, with REFERENCES
the adjusted stiffness for the interface elements, pro-
vide very good agreement with the experimental Ross, R., DeGroot, R., Nelson, W., Lebow, P., 1997 – “The
results both in the linear and nonlinear parts. The relationship between stress wave transmission characteris-
tics and the compressive strength of biologically degraded
influence of the experimental horizontal restraint, wood”. Forest Products Journal, Vol. 47(5), pp. 89–93.
simulated by a linear spring, is only marginal. CEN; 1991 – “EN 26891 – Timber structures. Joints Made
It has been shown that the parameters that affect With Mechanical Fasteners General principles for the
most the ultimate load are the compressive strength of determination of strength and deformation characteris-
wood perpendicular to the joint and the normal stiff- tics”. Office for Official Publications of the European
ness of the interface elements representing the contact Communities. Brussels, Belgium.
between rafter and brace. The tangential stiffness of Feio, A., Machado, J., Lourenço, P., 2005 – Compres-
interfaces and the Young’s moduli of wood have only sive behaviour and NDT correlations for chestnut wood
very limited influence in the response. The compres- (Castanea sativa Mill).
Lourenço, P., 1996 – Computational strategies for masonry
sive strength of wood parallel to the grain has almost structures. PhD thesis, Delft University of Technology.
no influence in the response.
841
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The structural behaviour of an existing masonry building subjected to seismic action, is strongly
affected by the in-plane stiffness of the floors, and by the connections between the horizontal diaphragms and the
masonry walls. The aim of the research is to experimentally evaluate the behaviour of timber floor refurbished
using different techniques, with special regard to the in-plane stiffness. The size adopted for the specimens (5 m
span, 4 m width), is similar to the ordinary dimensions of timber floors in historical buildings in Italy. Taking
into account the size of the specimens, and the need to determine the in-plane strength and stiffness of the floor,
a special test set-up has been designed and adopted in order to allow the free in-plane deformation of the floor
itself subjected to lateral load: the load configuration applied to the floor simulates the effect of seismic action
on the floor. The experimental phase of the research aims to calibrate engineered models that can be used for
studying existing structures.
1 INTRODUCTION
843
masonry walls, and suggests new alternative strength- studied, which are described in the cited Italian stan-
ening techniques for the horizontal diaphragm. Some dard as possible approaches in order to increase the in
of them are presented in the next paragraphs. plane stiffness and therefore the building robustness.
Moreover, in some cases, the existing floor can benefit
of a higher level of out-of plane strength and stiffness,
2 TIMBER FLOOR TYPOLOGY AND depending on the different techniques considered, as
REFURBISHMENT TECHNIQUES in the case of the timber–timber or timber–concrete
composite structure.
In this paper a simple supported timber floor is consid- In the first strengthening technique considered, a
ered, where timber beams have a section of 18 × 18 cm, second layer of wood planks (100 cm width, 3 cm
spaced 50 cm, which is a recurrent configuration for thick) is used, crossly arranged to the existing ones
the floor structure in Italian historical buildings. The and fixed by means of 10 mm steel rods, epoxy glued
deck system is composed with a simple layer of wood into timber beams (Figure 3 b).
planks (3 cm thick), crossly arranged and nailed to the The application of diagonal bracing (45◦ ) on the
timber beams (Figure 3a). Starting with this config- existing wood planks (Figure 3.c), can be done utiliz-
uration, five different reinforcement techniques were ing wide sheets of CFRP (50 mm wide, 1.4 mm thick),
glued to the wood by means of epoxy-based resin, or
adopting light steel plates (80 mm wide, 2 mm thick),
nailed to the planks. The mesh of diagonal bracing
applied to the floor is 705 mm for both techniques
analysed: obviously it is a function of seismic intensity.
Another strengthening technique analysed makes
use of three layers of plywood panels (21 mm thick),
arranged on the existing wood layer, and connected
to timber beams by means of 10 mm steel rods, glued
with epoxy−based resin (Figure 3.d).
A reinforced concrete slab connected to the timber
beams (Figure 3.e) is a strengthening technique which
Figure 2. Role of the diaphragm preventing the overturning
was widely used in the past decades when restoring
modes of masonry walls: (a) the inadequate in-plane stiffness timber floors. Regular concrete slab (50 mm thick)
of the floor causes overturning of the walls perpendicular to is built on the wood planks; the slab reinforcement
the seismic action; (b) a stiff diaphragm allows forces to be is composed by a welded steel mesh (6 mm diame-
transmitted to the walls parallel to the seismic action. ter, mesh 200 × 200 mm). The connections between
Figure 3. Different timber floor in plane shear strengthening techniques: (a) existing simple layer of wood planks on the timber
beams; (b) second layer of wood planks crossly arranged to the existing one and fixed by means of steel studs; (c) diagonal
bracing of the existing wood planks by means of light steel plates or FRP laminae; (d) three layers of plywood panels glued
on the existing wood planks; (e) a stud-connected reinforced concrete slab (all measures in mm).
844
the timber beams and the concrete slab is obtained In the solution proposed by Doglioni (2000), an L-
by means of L shaped connectors (re-bars FeB 44 k, shaped profile is connected to the floor by means of
16 mm diameter, 150 mm and 50 mm edges, epoxy screws; both the end sides of the profile are linked to
glued to timber beams in holes 90 mm deep). the lateral masonry through threaded steel bars (diame-
The in-plane shear behaviour of the horizontal ter varying from 20 to 30 mm), which are chemically or
diaphragm subjected to seismic action can be schemat- mechanically connected to the masonry walls (see Fig-
ically illustrated as in Figure 5. The lateral forces must ure 4). Along the floor border, additional connectors
be transmitted to the shear masonry walls parallel can be placed, in order to guarantee the shear trans-
to the seismic action, and therefore an adequate link mission to lateral walls, and to prevent the possibility
between the floor and the walls must be assured. of out–of–plane mechanisms.
Moreover the in-plane deflection of the slab induces For the concrete slab floor depicted in Figure 3e, the
compression and tension zones in the deck. Compres- concrete ring curb is guaranteed by additional steel
sion stresses can be counteracted by the wood planks, bars inserted along the border of concrete deck (see
while for the tension stresses a reinforcement element Figure 4.a), in order to avoid the insertion of a concrete
must be on-purpose designed. curb “inside” the thickness of the masonry walls, that
Two technological solutions were considered in the can weaken significantly the existing walls.
analysis. For the floor typology depicted in Figures 3.a,
b, c, d, a steel ring curb encloses the perimeter of the
floors. The steel curb has the double role to take up the 3 EXPERIMENTAL CAMPAIGN
tension stresses of the deck, and to transmit the shear
forces on the lateral walls. 3.1 Test set-up
The experimental test apparatus was thoroughly
designed taking into account the specimen dimen-
sions, the boundary conditions and the load config-
uration.
The floors were built in the laboratory adopting
different specimen sizes: the first monotonic pilot
tests were performed on small size floors (1 × 2 m);
then the cyclic tests were performed on real size
floor specimens (4 × 5 m). In both test configura-
tions, real size timber elements were used: beams
(0,18 × 0,18 × 4,2 m), planks (0,2 × 0,03 × l m, with
l varying from 0,6 to 1,6 m), and the previously
described reinforcement elements. The real size
dimensions showed to be necessary in order to sim-
ulate the real contribution of the secondary elements
(planks and reinforcement elements).
Another important aspect concerning the design
of the test apparatus is the boundary conditions of
specimens.
845
Instead of reproducing, in the test apparatus, the The floor specimens and the steel beams form-
actual boundary conditions of the floor in situ, the ing the test apparatus rested on some timber supports
adopted design approach was intended to allow free fixed to the reaction floor. In order to reduce friction
in-plane deformation of the diaphragm. In fact the effects during the tests, Teflon plate were added at the
experimental replication of the interaction between interfaces between materials.
floor and masonry wall was considered unpractical, In the real size specimens, cyclic tests were per-
with many problems involved in understanding and formed, therefore the loading system should work in
analysing the experimental results. On the contrary, both directions: beside the pushing system depicted
the adopted configuration allows an accurate measur- in Figure 7, a pulling system composed by steel plates
ing of the deck in-plane stiffness, which is one of the applied to both ends of the floor beams, and connected
key parameters for the structural restoration design. with pre-tensioned tie-rods, was adopted (Figure 10).
Therefore the floor specimen was linked to the lab- Movements, deflections, and loads were measured
oratory reaction floor by means of two external hinges, at multiple locations on the diaphragm specimens
according to the scheme depicted in Figure 6. The using electronic sensors of various types, connected
hinges are positioned centrally, at the neutral axis level. to a computer controlled data acquisition system.
Particular attention was paid on the load system, In Figure 8, the instrumentation layout is shown. In
in order to reproduce the transmission of seismic order to find the in-plane shear deflection of the floor,
forces through the floor. During an earthquake, the two wire sensors were used, diagonally arranged. At
lateral forces are proportional to the vertical load the ends of the 5 beams, not directly loaded, 5 LVDTs
applied on the floor, which can be considered uni- transducers were used in order to record the floor in-
formly distributed. Therefore a uniformly distributed plane deformation.
horizontal action should be applied to the floor under Three other LVDTs instruments were placed for
experimentation. finding lateral movements of the tested floor (orthog-
The set-up adopted for the force transmission is onally to the jack direction); finally, in the zones of the
depicted in Figure 7, where two levels of steel cross- slab characterised by maximum tension/compression
beams were adopted in order to distribute the load stresses, 4 strain gauges were used, 2 of which applied
of the single hydraulic jack onto four timber beams. on the metallic profile and 2 on the planks, so to find
A preliminary numerical analysis proved this config- the efficacy of the steel.
uration to be able to accurately reproduce a uniformly
distributed load.
Four different load configurations were consid- 3.2 Test protocol for cyclic tests
ered, from the theoretical condition of uniform load Cyclic quasi-static tests were performed under dis-
to the real condition adopted during the test, which placement control, with a loading rate varying
can be estimated sufficiently accurate according to the
numerical results reported in Table 1. It is worth noth-
ing that the real size floor tested is made up of 11
timber beams.
Deviation
Displacement
Load mm mm %
Uniform 25.65 – –
11 beams 23.86 1.79 7.0
6 beams 23.38 2.27 8.8
Figure 6. Set up configuration of the floor adopted for the 4 beams 23.54 2.11 8.2
experimental campaign.
846
from 0,05 e 2 mm/s, according to the test proce-
dure described in the European standard EN 12512
(Figure 9).
In this procedure, cyclic tests are defined as a func-
tion of the yielding value, which can only experimen-
tally be determined through a preliminary monotonic
test. Such preliminary tests were conducted on the
small size specimens, with the hydraulic jack acting
only on the central timber beam.
4 EXPERIMENTAL RESULTS
Figure 10. Global view of the set-up configuration for cyclic test.
847
Figure 12. Experimental results for monotonic and cyclic
test on the timber – concrete composite floor (Figure 3e).
Figure 11. Experimental results for monotonic and cyclic
tests on the “original” simple plank – timber floor (Figure 3a). Table 2. Monotonic test results.
dy Fy k
Floor mm kN kN/mm
Taking into consideration a simple 3 storey masonry
building, where 60 cm masonry walls surround a floor Wood planks 30 29.7 0.99
with dimension 4 × 5 m, the lateral force at the third Concrete slab 13 112.0 8.62
storey level (height 9 m) can be estimated for a peak
ground acceleration (PGA) of 0,35 g (which corre-
sponds to the highest level, in Italian seismic standard).
This force, hereafter indicated as Feq, is the maximum Table 3. Cyclic test results.
lateral force transmitted by the horizontal diaphragm,
in this simple case. From experimental results reported d max F max d eq* F eq*
in Figures 11 and 12, the collapse of the diaphragms Floor mm kN mm kN
happened for higher values (ranging from 1,5 to
2 × Feq). Wood planks 180 203 100 133.8
These simple considerations confirm what observed Concrete slab 52 451 10 142.4
during post earthquake survey, where failure is asso-
* Seismic action on floor PGA = 0.35 g.
ciated with out-of-plane fall down of the support-
ing walls rather than with inadequate strength of
diaphragms. diaphragm can be considered as rigid). For the “origi-
In Table 3, the maximum deformation correspond- nal” timber floor illustrated in Figure 3.a, the deforma-
ing to the Feq force is reported. For the timber – con- tion value corresponding to the Feq force is 100 mm,
crete composite floor, the corresponding deformation which can be awkwardly reputed consistent with the
is 10 mm (according to the Italian standard such safety of the structure.
848
4.2 Role of the steel ring plate
Cyclic test results highlighted the role of the steel ring
curb for the specimen reported in Figure 3.a, both in
terms of stiffness and strength.
The steel ring curb increases the maximum shear
force transmitted by the floor to the walls; at the same
time, resistance is assured against tension and com-
pression forces induced in the slab by the movements
generated by the quake.
This fact can be appreciated considering the values
of the strains locally recorded in the steel L-shaped
plate and in the nearby planks, were some strain gauges
were applied. The stress values in the steel plate and
in the wood planks reported in Figure 13.b, c, were
computed considering the mechanical and geometrical
properties of the different components, and the axial
strains recorded by means of the strain gauges.
Figure 13.d reports the maximum lateral deforma-
tion of the simple plank – timber floor: the values
confirm that the floor can carry out free in-plane defor-
mation, without any lateral constraint, which was the
base assumption for the adopted set-up. Figure 14
shows the effectiveness of the steel curb in the existing
timber floor with a simple layer of wood planks.
849
the research program carried out for the Italian Agency
for Emergency Management.
REFERENCES
Borri, A., Corradi, M., Grazini, A. 2005. A method for
flexural reinforcement of old wood beams with CFRP
materials. J Compos Part B, 36/2:143–53.
CEN, EN 1995-1-1:2004. Eurocode 5: Design of timber
structures – Part 1-1: General – Common rules and rules
for buildings.
CEN, EN 12512:2006 Timber structures – Test methods –
Cyclic testing of joints made with mechanical fasteners.
Corradi, M. Speranzini, E., Borri, A. & Vignoli, A. 2006. In-
plane shear reinforcement of wood beam floors with FRP.
J Compos Part B; 37: 310–319.
D’Ayala, D., Speranza, E. 2002. An integrated procedure
for the assessment of seismic vulnerability of historic
buildings. Proc. Of the 12th European Conference on
Earthquake Engineering, London, Elsevier Science, paper
n. 561 (CD-ROM).
Doglioni, F. (Editor) 2000, Codice di pratica (linee guida) per
la progettazione degli interventi di riparazione, miglio-
ramento sismico e restauro dei beni architettonici dan-
neggiati dal terremoto umbro-marchigiano del 1997,
Bollettino Ufficiale della Regione Marche, Ancona.
Gattesco, N., Macorini, L. 2006. Strengthening and Stiffening
Ancient Wooden Floors with Flat Steel Profiles. Structural
Analysis of Historical Constructions, New Delhi 2006 P.B.
Figure 14. Effect of the steel curb applied to the existing Lourenço, P. Roca, C. Modena, S. Agrawal (Eds.).
deck system with a simple layer of wood planks (figure 3.a). Griffith, M.C., Magenes, G., Melis G., Picchi L. 2003. Eval-
uation of out-of-plane stability of unreinforced masonry
The final aim of the research is to develop a suitable walls subjected to seismic excitation. Journal of Earth-
equivalent shell modelling of timber floors, to be easily quake Engineering, Vol. 7, Special Issue 1, pp. 141–169.
implemented in global numerical models of traditional Marini, A., Giuriani, E. 2006. Transformation of Wooden
Roof Pitches intoAntiseismic Shear Resistance Diaphragms.
buildings. It will be possible, consequently, to choose Structural Analysis of Historical Constructions, New
adequate strengthening strategies for existing floors Delhi 2006 P.B. Lourenço, P. Roca, C. Modena, S.Agrawal
and to optimize the interventions for each particular (Eds.).
building. Ordinanza, P.C.M. 3431 2005. Primi elementi in materia di
criteri generali per la classificazione sismica del territorio
nazionale e di normative tecniche per le costruzioni in
ACKNOWLEDGEMENTS zona sismica.
Piazza, M., Turrini, G. 1983. Il recupero dei solai in legno.
The authors gratefully acknowledge the Italian Recuperare, n◦ 7. (In Italian).
ReLUIS Consortium for financing the study, within
850
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The purpose of this paper is to present the state of an investigation aimed at strengthening
decayed stone masonry walls with textile reinforced lime-cement mortars (TRM). The proposed solution could
be applied by itself or jointly with other techniques of reinforcement, like sewing or injections, to guarantee the
integrity of the wall. The validation of this strengthening system is carried out by a laboratory test campaign. The
experimental study is based on the typology of the Spanish Romanesque walls, a plentiful constructive element
in this country. The validation of the solution is being done by means of a characterization of the individual
materials (stone, mortar, TRM), designing of the anchor system and construction, strengthening and testing of
1/3 scale stone walls. The wall specimens have been erected with the same configuration of the original ones.
They are three-leaf masonry walls with an inner core made of low quality mortar and rough-cut sandstone pieces
(remains from rough-shaping of the stones), poured between the two external layers, and they are deflected a
common structural damage in these elements.
1 INTRODUCTION
851
This new structural material, namely textile rein-
forced mortar (TRM), was developed as an evolution
of short-fibres reinforced concrete, in order to obtain
very thin-structured concrete elements with a high
strength in compression as well as tension.
The results obtained from preliminary studies on
the use of TRM as a strengthening solution of unre-
inforced masonry walls (Triantafillou, 2001) can be
extended to heritage masonries where the use of epoxy
resins is prohibited. For these masonry elements, a
lower strength TRM system than for reinforced con-
crete structures will be suitable. Hence the use of lower
Figure 2. Scheme of the strengthening solution.
properties fibres than carbon, for example AR-glass or
basalt fibres can be adopted.
The need of a low cost TRM system is obvious,
since there are many thousands of heritage buildings
around the Mediterranean and many of them in coun-
tries where the cost is of major importance for the
governments to fund a heritage monuments strength-
ening project. In a TRM system, mortar’s cost is low
compared to textile cost. Hence the cost of textile is
predominant. For this reason the carbon and aramid
high cost fibres are not suitable.
The expected results of this strengthening system
Figure 3. Wall specimens.
are: an increase in ductility and tension/shear resis-
tance, a decrease of cracks by located tensile loads and design of suitable anchorage system and the construc-
an improvement in the general behaviour, especially at tion, strengthening and testing (under static loads) of
the failure moment. nine real stone walls made at 1/3 scale.
These structural tests were defined from the knowl-
2 CONCEPT OF THE PROPOSED SOLUTION edge of this type of masonry (typology, materials, etc.)
and the structural behaviour connected to the spe-
Related to the TRM application as strengthening sys- cific deficiency to eliminate. Therefore, a particular
tem on historical masonry structures, nowadays, there attention was taken in choosing the geometrical and
are in course multi-year development efforts regard- the morphological characteristics and the constituent
ing its ability to retrofit un-reinforced masonry walls materials of the walls to test, to make them as much
(URM). The objective is to establish the adaptabil- as possible representative of the Spanish Romanesque
ity of a full-compatible and highly durable structural real typology, available in situ.
strengthening system based in TRM, including:
– The strengthening core: a technical fibre textile, 3 TEST CAMPAIGN
adaptable to strengthen different structural elements
(vaults, aches, walls) and substrates (masonry, A set of nine walls was constructed with low strength
adobe bricks and timber). lime-cement mortar and sandstone units, representing
– Fixing system: based on mortars or/and in anchor- Spanish Romanesque walls and similar to walls often
age devices, in case of need. found in historical urban centres., with two external
– Conditioning and finishing mortars: based on mod- leaves and an internal core of rubble material. Both
ified or lime-cement mortars, such as a compatible of them have the external leaf made of ashlar masonry
interface between the substrate and the TRM and and the internal leaf made of rough masonry. The walls
improving its aesthetic integration. The effective- are deflected, a common structural problem in these
ness of any externally applied reinforcement is elements, having the following nominal dimensions:
highly dependant on the bond between the com- 2 m wide, 2 m high, 0,3 m thick and a 5 cm deflection,
posite and the substrate, therefore the interface from the half height to the top.
behaviour is one of the key issues in the structural
analysis. 3.1 Masonry
The validation of the strengthening system is carried Two types of stone were used for the walls. For the
out by a lab test campaign, regarding characterization ashlar masonry leaves, Sandstone1, a uniform fine
of the materials (stone, mortar, masonry andTRM), the grain sandstone rock, was used. It was received in
852
Table 1. Mechanical properties of stone.
0,30
0,25
Strength (MPa)
853
– Ashlar masonry with bed mortar joints possesses
the most uniform behaviour, with a slope of the
strain-stress line practically constant.
854
Figure 7. Basalt fibre anchor device.
855
Table 3. Strengthening solutions.
Wall
specimen TRM Transversal tying
M1 No No
M2 1 layer No
M3 2 layers No
M4 1 layer Yes
M5 2 layers Yes
M6 No No
M7 1 layer No
M8 2 layers No
M9 2 layers Yes
856
Figure 14. Buckling and disconnection of the leaves.
857
200
Strain gauges on the render mortar (West face) capacity, improvement of the failure mode, increase
of the ductility, etc).
150 – A next generation FEM model will be calibrated.
This model will consider the typology of the stone
100
walls, the connection or adherence between leaves
Load (KN)
Compressed
50
Tensioned
and the presence of reinforcements like the applied
ones.
0
-200 -150 -100 -50 0 50 100 150
-50
Microstrain 4 CONCLUSIONS
Gauge1 Gauge2 Gauge3 Gauge4 Gauge5
The effectiveness of TRM as a means of strengthening
Figure 15. Micro strains of mortar face in Wall 1. stone masonry walls is investigated in this study. The
following conclusions are stated from the concluded
works:
Ultimate axial load
250 – The FRP, in its different formats (strips, rods, sheets
or sheets), has probe to be a reasonable solution
200
for the reinforcement of masonry structures. How-
ever, actual trends are focused in a more compatible
150
technique (TRM).
Load (kN)
100
– Materials cost of the intervention is significantly
reduced. Lime-cement matrix is between 6 and 10
50
times cheaper than epoxy resins. Similar researches
(Triantafillou & Papanicolaou 2005) on regular
0 masonry have demonstrated a certain improvement
M1 M5 M6 M9 of the mechanical properties, even with low cost
Wall specimens fibre textiles.
– In order to design the anchorage solution, it is nec-
Figure 16. Ultimate compressive load. essary to consider the substrate material nature, its
conservation status, the load history, the aesthetic
requirements and the reversibility.
other hand, there are a lot of cases where architec- – By the application of TRM solution, reinforced
tural requirements do not allow to cross or to even elements can be obtained without increasing sig-
partially bore the stone units. In these situations a nificantly the original weight, and even without
different approach must be carried on, placing the modifying their external appearance.
rods into the mortar joints and using anchor devices. – The strengthening intervention could be done while
– Strain gauges on the face covered with mortar have the structure is in service.
revealed positive deformations in certain zones of
the wall. The presence of tensile tensions is not
easy to locate by conventional methods when the ACKNOWLEDGEMENTS
structure is damaged or there are structural defor-
mations that affect geometry. These tensile loads This research is being financed by EU Commission
can be partially assumed by the textile of the TRM. through the OPERHA 517765 contract and the Basque
– The combination of surface treatment withTRM and Government through the SISMU contract.
sewing with low modulus CFRP rebar has allowed
to increase the ultimate load a 13% for a wall of the REFERENCES
first typology (M5) respect the reference wall (M1).
ACI Committee 440. 2002. Guide for the design and construc-
Complete results and final conclusions of this tion of externally bonded FRP systems for strengthening
experimental investigation are expected at the begin- concrete structures. ACI 440.2R-02.
ning of 2008. The analysis of the data will provide high Bakis, C.E., Bank, L.C. & Brown, et al. 2002. Fiber-
value information. reinforced polymer composites for construction – State-
of-the-art review. Journal of Composites for Construction:
– A better understanding of the behaviour of the 6(2), 2–73.
masonry, specially the natural stone walls, and Casareto, M., Oliveri, A. & Romelli, A. 2002. Strengthening
the capacities of the promising strengthening tech- of masonry: opportunities and challenges in the use of
niques previously described: surface treatment with composites Center for Infrastructure Engineering Studies
TRM and sewing with rebar (increase of the bearing (CIES) – University of Missouri – Rolla, USA.
858
CEB-FIP. 2001. Externally Bonded FRP Reinforcement for applied to different substrates. Structural Faults & Repair.
RC Structures. London.
Monti, G. & Santini, S. 2002. Reliability-based calibration Triantafillou, T.C. 2001. Seismic Retrofitting of Structures
of partial safety coefficients for fiber-reinforced plas- Using FRPs. Progress in structural Engineering and
tic. Journal of Composites for Construction Vol 6, N◦ 3: Materials, 3 (1): 57–65.
162–167. Triantafillou, T.C. & Papanicolaou, C.G. 2005. Textile Rein-
San-José, J.T., García, D., Garay, A. & Castillo, J. 2007. forced Mortars (TRM) versus Fiber Reinforced Polymers
Novelty FRP strengthening systems applied to different (FRP) as strengthening materials of concrete structures.
substrates: analysis of the anchorage behaviour. FRPRCS- FRPRCS-7, ACI SP-230: 99–118. Kansas City, USA.
8. Ed. Prof. T.C. Triantafillou, University of Patras, Valluzzi, M.R., da Porto, F. & Modena, C. 2004. Behaviour
Greece. and modeling of strengthened three-leaf stone masonry
San-José, J.T., Roca, P., Prendes, P., Mieres, J.M. & Garay, A. walls. Materials and Structures Vol 37: 184–192.
2006, Anchorage study of FRP strengthening systems
859
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
J.K. Hwang
Department of Traditional Architecture, National University of Cultural Heritage, Korea
S.G. Hong
Department of Architecture, Seoul National University, Korea
N.H. Kim
Korea Bridge Design and Engineering Research Center
Y.W. Lee
Department of Architectural Engineering, Kunsan National University, Korea
S.J. Jeong
Department of Architecture, Hannam University, Korea
S.J. Joo
Senior Engineer, Korea
ABSTRACT: The main objective of this paper is to investigate structural behavior of Korean traditional wooden
structures. The main characteristics of traditional wooden structures in East Asian countries are considered as
unique techniques of joint without fastener and relying on friction between components to transfer interaction
forces.
As one of temple structures Bongjeong-sa built in the early stage of the tenth century first is considered as a
typical example of Korean traditional wooden structures. It is classified as a national treasure in Korea. It was
rebuilt several times for restoration. Its basic structural system consists of a column and lintel beam connected
by friction joint and Gongpo (bracket set). The Gongpo system is a unique system of brackets placed on top of
columns to provide additional support to beams or overhanging eaves.
A one-third scale model with the roof-dead load of 46,710 N is designed and prepared on a steel sliding table
of 5 m by 4 m to perform a series of structural tests in a laboratory. Using this model the characteristics of lateral
and twisting displacement on the top of columns of the frame and the dynamic effects of the Gongpo on the
global behavior of whole systems are investigated. It is concluded that the friction joint of column-connecting
beam serves as an energy dissipation device reducing the impact of earthquake,but the Gongpo is not efficient
to dissipate the energy as we have guessed.
861
Figure 1. Exterior front view. Figure 4. Soro(block) and cheomcha(arm).
2 EXPERIMENTAL PROGRAM
862
Figure 9. Sliding table and actuator.
3 VIBRATION TEST
and it kept the percentage of moisture under 17%.
3.1 Random vibration
The steel plate of 46,710 N was used to simulate the
weight of roof (Figs 6–8). It is actuated the sliding table as random signal which
To perform the vibration test, the steel sliding table contains the frequency of from 1 Hz to 3 Hz to figure
of 5.5 m by 4.5 m was made. It was slid in the one-way out the natural frequency of the test model. In process
direction on the LM guide and designed to be able to of the random vibration experiment, we observe that
rotate to perform the test of orthogonal direction. It the natural frequency of test building changes accord-
was slid in the hydraulic dynamic actuator. ing to the magnitude of vibration. To confirm the
863
Table 1. The variation of natural frequency for the RMS
amplitude of random signal.
0.2 2.606
0.3 2.576
0.5 2.258
1 1.962
2
Force (kN)
0
- 40 - 30 - 20 - 10 0 10 20 30 40
-1
-2
-3
Figure 13. TF from ground acc. to connecting beam acc.
-4
Displacement (mm)
of whole system changes according to the variation of
Figure 12. Force-lateral displacement relationship of joint. joint displacement.
The accelerations and displacements are measured
simultaneously during 180 seconds with a sampling
frequency of 100 Hz. Then, stationary portions of
variation of natural frequency, the random vibration records are selected to calculate the transfer function
test was performed for the RMS amplitude of dif- that could evaluate about the predominant frequency
ferent kinds. The natural frequency of X-direction on of the structure. To evaluate the variation of the pre-
the connecting beam, acceleration 3 in Figure 10, and dominant frequency according to the magnitude of
the results of natural frequency are shown in Table 1. vibration and the effect of vibration transmission
According to natural frequency results, the natural between the layers, the transfer functions for accel-
frequency of test building could confirm to change eration between sliding table and architrave were
according to the magnitude of vibration amplitude. calculated. The results of the calculation of transfer
The variation of natural frequency could possible functions are shown in Figure 13.
when weight or stiffness varied. In case of this test The one of the main subjects of this research investi-
model, the weight did not change at all. So, we could gates the energy dissipation characteristics of Gongpo.
suppose that the lateral stiffness of whole frame var- To this end the transfer functions for acceleration
ied. The lateral stiffness of Korean traditional wooden between sliding table and Gongpo are calculated. The
structure is composed for the most part by the stiffness transfer functions of architrave and Gongpo are shown
of column-connecting beam joints. The research result in Figure 14. We can confirm that the transfer functions
for joint stiffness shows that the stiffness of column- of architrave are identical with the one of Gongpo.
connecting beam joints is varied by restraint force and Therefore, we can confirm that the energy dissipation
the magnitude of lateral displacement (Fig. 12). The effect is in the joint of column and connecting beam
stiffness of joints is the linear in the case of small dis- and Gongpo do not perform the role of energy dis-
placement and changes if the displacement of joint is sipation. Also, we can confirm that the predominant
gradually enlarged. By this reason, the lateral stiffness frequencies shift to the left according to the increase of
864
Figure 16. The free vibration record of impact displace-
ment.
Figure 14. TF of on connecting beam and Gongpo. Table 2. Damping ratio by impact.
X-direction impact
displacement (mm) Damping ratio (%)
5 8.94
10 10.02
15 10.19
20 10.30
25 10.19
30 10.24
865
to express their gratitude to the many researchers of [3] S.J. Joo, S.G. Hong, N.H. Kim, Y.W. Lee, S.J. Jeong
the Architectural Research Division. and J.K. Hwang, Seismic Response Characteristics of
the Main Building of Bongjeong Temple, Proceed-
ings of COSEIK Annual Conference 2007, Seoul,
REFERENCES Korea, 2007.
[4] Y.W. Lee, S.G. Hong and J.K. Hwang, Capacity of
[1] Andong-city. The report of total repair for the Main Lateral Load Resistance of Dori-direction Frame with
Building of Bongjeong-sa(temple), 2004. Jang bu-connection in Traditional Wood Structure
[2] Anil K. Chopra. Dynamics of Structures. Prentice-Hall. System, Journal of the architectural institute of Korea,
2001. structure & construction, vol.23 n.2, 2007.02.
866
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
C. Cennamo
Seconda Università di Napoli, Napoli, Italy
S. Scaini
DEXPLO Srl, Parma, Italy
ABSTRACT: Cutting processes in archaeological sites are critical operations which can irreparably damage
precious artefacts. The damage can consist either of irregular and wide cutting grooves and of chemical deterio-
ration triggered by flushing liquids. Therefore, an holistic approach to the optimisation of cutting operations is
an important issue which requires an accurate knowledge of the ancient base materials and theoretical models
which permit to manage the interaction between artefacts and cutting tools. According to this framework, the
results of the chemical analyses and of the mechanical tests performed on the concrete constituting an ancient
Greek Gymnasium are firstly reported in this paper. Then, an essential mechanical model of the cutting process
is proposed. It relies on an energy parameter, the “cutting strength”, which must be the target of optimisations
and can be easily post-calculated starting from data collected on site. Finally, an empirical model for the tool
wear is also proposed.
1 INTRODUCTION: HISTORICAL
BACKGROUND AND MOTIVATIONS
867
Figure 2. Cement core wall in opus reticolatum (left side)
and street paved with tuff (right side).
868
Table 1. Average mineralogical composition of tuff samples
TA and TB.
869
Figure 6. Stress vs. strain curves obtained from axial
compression tests.
870
Figure 10. Material removed during the vertical advance-
ment equal to s (left side) and particular of a tooth (right
side).
so that:
871
Figure 12. Out of regime cutting conditions obtained reduc-
ing (left side) or increasing (right side) the externally applied
normal force.
872
Therefore, applying Equation 6 the corresponding
values of the penetration stiffness k turn out to be
comprised between 1.77 and 3.8 kN/mm.
The vertical displacement wC induced at a point C
on the surface of a semi infinite elastic continuum by
a uniformly loaded square area b × b centered in C is
given by:
873
where Amaxw is the asymptotic value of the worn tooth varying the external driving forces, namely the pene-
base area (characteristic of the specified tooth) and tration force provided by the workman and the tensile
C2 is another constant which incorporates the effects force into the chain provided by the cutting tool engine.
of all the relevant parameters quoted above and which The back analysis performed based on the on site
can be easily evaluated by directly measuring the tooth operations, indicates a value of the cutting strength
base area at different cut advancements. Minimizing for the ancient concrete S = 250 ÷ 600 MPa, which
this constant, increases the teeth’s lifetime. is comparable with that of a modern concrete cut in
By inserting Equation 20 into Equation 19 the wet conditions. Even if the poor mechanical proper-
following wear dependent penetration stiffness is ties of the ancient concrete would suggest a cutting
obtained: strength much lower than that of a modern concrete,
the dry conditions of the cutting processes performed
in Naples, seem to justify the obtained results, due to
the good quality of the ancient mortar.
Moreover, scale effects related to the size of the
teeth (i.e. a smaller scale involves a higher ultimate
Differently from C2 , C1 is hardly directly deter-
quasi-static strength) have been put into evidence.
minable, since the penetration law (Equation 19)
The small amount of experimental data coming
refers to dynamic conditions. Nevertheless, if a cut-
from the site and from laboratory tests (only a small
ting process at constant power We and normal force
number of samples was sacrificed) still leave some
N is considered, Equations 21 and 5 suggest that the
open issues, like the influence of dynamic effects and
optimal penetration u∗ decreases with cutting advance-
the roles of topological heterogeneities into the base
ment, while Equation 16 shows that the speed δ
material.
remains unvaried. This implies (Equation 10) a that
In conclusion, the optimisation of the cutting pro-
the rotational speed of the chain increases with wear.
cess allows to reduce vibrations to a great extent and
to extract straight blocks of material from the masonry
structures, minimizing the groove thickness so that the
6 CONCLUSIONS
structures could be reassembled on situ in excellent
conditions. So, this performance could be used in case
The question of coexistence among the criteria for
the disassembly of a historical situ is really necessary
preservation of old items and the technological
to the aim of its conservation, specifying every time
advancement in cities, is a very serious problem for
the mechanical model according to the characteristics
the historical centres of ancient European towns, and
of the material under consideration.
it is emphasized when situations like the one in Piazza
Nicola Amore, bring the issue to the forefront. It is
therefore necessary to pay more attention, through reg- REFERENCES
ulations including the adaptation of machinery and
the development of new specific technologies. In the Giuffrè A. 1991. Letture sulla meccanica delle murature
specific case, the dry cutting processes performed on storiche. Edizioni Kappa, Roma.
the archaeological site found in Naples, dealt with an Wojtanowicz A.K. & Kuru E. 1993. Mathematical modelling
extremely heterogeneous material. A model aimed to of PDC bit drilling process on single cutter mechan-
the optimisation of the cutting process has been pro- ics. ASME Journal of Energy Resources Technology 115:
posed in Section 3. This model defines the optimal 247–256.
Fischer-CrippsA.C. 2000. Introduction to contact mechanics.
conditions that can be reached during cutting processes
New York USA: Springer Verlag.
and underlines how these can be determined starting Chiaia B.M. 2001. Fracture mechanisms induced in a brittle
from the external forces applied by the workman and material by a hard cutting indenter. International Journal
by the tool, given an opportune description of the base of Solids and Structures 38: 363–371.
material and of the cutting machines. Carpinteri Alberto et al. 2003. Numerical analysis of inden-
As previously shown, the key expedient of the tation fracture in quasi-brittle materials. Engineering
model is the introduction of the cutting strength param- Fracture Mechanics 71: 567–577.
eter (energy required to remove a unitary volume of Glowka D.A. 1989. Use of single-cutter data in the analy-
base material) which has dimensions of compressive sis of PDC bit design: Part 1 – development of a PDC
cutting force model. Journal of Petroleum Technology
strength, but is generally higher since it takes into
33(August): 797–849.
account all the dissipations related to dynamics (vibra- Nabhani F. 2001. Wear mechanisms of ultra-hard cut-
tions, heat, fragmentation, milling, etc.). In this way, ting tools materials. Journal of Materials Processing
all the sources of complexity in the cutting process are Technology 115: 402–412.
collected inside this parameter and the model shows
that also when the cutting strength and the tool power
are constants, different conditions can be obtained by
874
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: Double flat-jack tests were performed on a brick masonry wall using a series of LVDT’s to
monitor the distribution of displacements in the wall above the top flat-jack. The results allowed estimating:
(i) the way the stress induced by the flat-jacks is distributed in the wall; (ii) the main direction of the joints
openings/cracks that in fact extend to the sides of the flat-jack. The analysis of these results shows that actually a
volume of wall larger than the one above the top flat-jack is directly involved in the test. A numerical simulation
of the flat-jack test using a finite element code was also performed.
1 INTRODUCTION
875
Figure 3. Scheme of the masonry texture of the
Figure 2. (a) Slot in the mortar joint before inserting the Cavallerizze.
flat-jack (350 × 250 × 4 mm); (b) Displacement measures
devices.
876
Table 1. Elastic parameters achieved from the test (MST-
J7D).
877
Figure 8. Stress-strain of the single transducers placed
along the vertical line above the top flat-jack (MST-J7D).
878
Table 3. Elastic parameters achieved from the test (MST-
J6D).
Figure 12. Crack below the right edge of the lower flat-jack.
879
Figure 16. Peak strain-horizontal position of the single Figure 18. Pick Strain-height plot of the single transducers
transducers placed along a diagonal line at each loading cycle placed along an horizontal line above the top flat-jack to the
(MST-J6D). right, at each loading cycle (MST-J6D).
Figure 17. Peak strain-horizontal position of the single Figure 19. Pick Stress-strain plot of the LVDTs 01, 02, 03
transducers placed along a diagonal line, at each loading cycle and 04 (tests MST-J7D and MST-J6D).
(MST-J6D).
The elongation on the horizontal LVDTs 19 and 20 that contributes to the in-situ stress. This, of course,
(Fig. 16) shows that the compressed masonry above the corresponds to a very ruff simplification of the reality.
top flat-jack pushes the lateral masonry that becomes Along the horizontal line above the top flat-jack
horizontally compressed.That effect is more important (y = 21,5 cm), the compression strain is quite constant
closer to the flat-jack, inducing the higher compression inside the vertical limits of the flat-jack (LVDTs 06,
strain measured at LVDT 19. 10, 11 and 12, in figure 18).
Comparing the stress vs. strain curves below (aver- On the right side of these limits, LVDT 11 detects
age between LVDTs 02 and 03) and above (LVDTs an important tensile strain which decreases to almost
06 plus 07) the top flat-jack, the results are quite simi- zero when it reaches LVDT 12. These measurements
lar, although the masonry above presents an apparently show once again that the volume of the masonry above
stiffer behaviour. the top flat-jack, pressed during the test, is much
Along the diagonal line (Fig. 17) on the right-up larger than the column inside the vertical limits of the
side of the top flat-jack, LVDTs 17, 11 and 13 detect an flat-jack.
elongation that decreases from the 17 to the 13. It can In conclusion, a comparison between the two in-
be noticed that although at the beginning the elonga- situ tests is presented in figure 19, in a stress-strain
tion of LVDT 12 is slightly higher than the elongation plot of the peak stress mean values measured in the
of LVDT 13, at the end the elongation of the first is area between the flat-jacks (WS).
lower than half of the second. Nevertheless, the strain The comparison between the two masonry walls
of the two LVDTs is rather low. shows that the best mechanical characteristics are more
The fact that there is a crack that goes up, fol- in evidence for the plastered and so better protected
lowing a 45◦ line, until an horizontal distance of masonry wall of the “Siloteca”. Furthermore, in that
50 cm from the extremity of the flat-jack, will give test, the “inversing point” was not observed due to
a column of masonry participating on the test that is interruption of the test at a pressure of 26 bar when a
(50 + 30 + 50)/(30) = 4.3 times wider than the column damage occurred to one of the two fat jacks.
880
3 NUMERICAL SIMULATION
881
flat-jack, the effects extinguished at a height which
is approximately the double of the wall specimen
one (WS). Therefore these results can be eventually
transferred into guidelines for the use of the double
flat-jack test.
ACKNOWLEDGEMENTS
882
the compressive strength of masonry”; Computer and on Computational Methods in Structural Dynamics and
Structures 84, 2006 – 1977–1989. Earthquake Engineering, Crete, Greece, 13–16/6, 2007.
Attard M.M., and Chaimoon K. 2007. Modelling of un- Costa A., Silva B., Costa A., Guedes J., and Arêde A. 2006.
reinforced masonry walls under shear and compression”; “Structural Behaviour of a Mansonry Wall under Hori-
Engineering Structures 21, 2007 – 2056–2068. zontal Cyclic Load; Experimental vs. Numerical Study”,
Silva B., Guedes J., Arêde A., and Costa A. 2007. “Structural 11th International Conference Structural Faults + Repair,
assessment of a stone masonry wall using a contin- Edinburgh, Scotland, 13–15 June 2006.
uum damage model”, ECCOMAS Thematic Conference
883
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: Lime based mortars are now widely acknowledged to be generally superior to cement based
mortars in the repair of appropriate historic infrastructure. Increasingly the benefits of hydraulic lime mortars
are also being realised in new masonry construction. In order to standardise the expected performance of mortars,
designers will specify the type of lime, the type of filler (aggregate), the proportions of each and quantity of
water or the required workability. Limes can be non-hydraulic (calcium or dolomitic) or hydraulic (natural or
artificial).
It is well known that the water/binder ratio has a marked effect on the structural performance of cement-based
mortars. This relationship is known as Abrams’ rule, which states that when a cement mortar is fully compacted,
its strength is inversely proportional to the water/cement ratio. Abrams’ rule has also been demonstrated to apply
to hydraulic lime mortars. The reason for this is that both cement and hydraulic limes require a minimum quantity
of water to produce the chemical set resulting from the hydration of calcium silicates and calcium aluminates.
Surplus water eventually dries out, leaving micropores in the matrix which weaken the resulting set mortar.
It has generally been assumed that the same relationship applies to non-hydraulic lime (air lime) mortars.
This paper reports on results of tests conducted on air lime mortars at early stages of curing. It is known that
the form of air lime and the physical and chemical characteristics of the aggregate have a strong impact on the
structural performance of air lime mortars. Results to date show that the water/lime ratio has a minimal impact
on the structural performance of air lime mortars compared with the impact of lime and aggregate type. 91 day
compressive strengths for air lime mortars with a water/lime ratio of 0.56 (a stiff mix) are identical to those with
a water/lime ratio of 0.875 (a loose slurry). Whereas Abrams’ rule is a key consideration for designers of cement
and hydraulic lime mortars, it has been demonstrated that it requires modification in the case of air lime mortars.
A relationship between form of lime, type of aggregate, water/lime ratio, and age of mortar is proposed. The
resultant equation allows the compressive strength of air lime mortars to be predicted taking into account these
factors. The insights gained from this study will allow practitioners to more confidently design and specify air
lime mortars.
885
Table 1. Water/lime ratios used [by volume]
Specimen Water/Lime
Designation ratio
Air Lime
A1 0.5
A2 0.5625
A3 0.625
A4 0.6875
A5 0.75
A6 0.875
Hydraulic Lime
Figure 1. Relationship between water/lime ratio and H1 0.375
strength (Allen et al. 2003). H2 0.4375
H3 0.5
H4 0.5625
strength through hydration products.Allen et al. [2003] H5 0.625
have shown this relationship in Figure 1.
The data presented in Figure 1 cannot, however,
be taken to be truly representative of Abram’s rule.
This is because the data are based on the compres-
sive strengths of different binder:aggregate ratios – It was found that air lime required more water in
each mortar requiring a different quantity of water to order to make a workable mix than hydraulic lime,
produce a specified flow. This means that two vari- and could accommodate more water before becom-
ables are present in the graph, and it is not clear what ing a loose slurry. This was likely to be a function of
proportion of the compressive strength is affected by the greater capacity of air lime to absorb water than
which variable. There has been conflicting evidence hydraulic lime as a result of having finer particles and
about the applicability of Abram’s rule to air lime therefore a greater surface area.
mortars. Schäfer & Hilsdorf [1993] and Winnefeld & Both mortars were de-moulded after 5 days and
Böttger [2006] present data which show that increased cured in a controlled environment of 60% RH at 20◦ C
water content in air lime mortars does not reduce until testing.
compressive strength. It has been shown that higher Compressive tests on six 50mm cubes were con-
porosity in air lime mortars allows greater access ducted after 28, 56 and 91 days from the date of
to atmospheric carbon dioxide (CO2 ), which pro- manufacture.
motes carbonation and therefore can produce greater
compressive strengths [Lanas & Alvarez, 2003].
This paper describes a systematic evaluation of the 3 RESULTS
impact of the water/lime ratio on the unconfined com-
pressive strength of air lime mortars up to 91 days after The results of compressive tests on the mortars are
manufacture. shown in Figures 2–4. Error bars are included showing
the range of results of the six tests used to produce each
data point.
2 EXPERIMENTAL These data compare well with the data produced
both by Schäfer & Hilsdorf [1993] and Winnefeld &
50 mm × 50 mm × 250 mm prisms of mortar were pre- Böttger [2006].
pared with 1 part of dry hydrated high calcium lime
(CL90) and 3 parts of silicate sand by volume using
a range of different water/lime ratios. 9 specimens 4 ANALYSIS AND DISCUSSION
of each mortar type were prepared and 2 cubes were
taken from each specimen for testing at the appropri- The data for hydraulic lime mortars appear to follow
ate time intervals. For comparison purposes a further Abrams’ rule with the relationship between strength
set of 9 prisms were prepared using 1 part NHL3.5 and water/lime ratio following an approximate hyper-
lime and 3 parts silicate sand with a range of different bolic curve. The data for the air lime mortars, apart
water/lime ratios. from the lowest water/lime ratio, show very little varia-
The amount of water added to each lime type ranged tion in compressive strength when the water/lime ratio
from the minimum quantity needed to make a workable is varied. It is conceivable that the data points for the
mortar to the amount required to make a loose slurry. air lime mortar are all to be found at the lower end of
The water/lime ratios used were as shown in Table 1. the hyperbolic curve, where there would be very little
886
an hydraulic set, which takes up to 28 days, depending
on the hydraulicity of the lime. Subsequent to this the
strength gains are due to carbonation.
Van Balen [1994] has proposed a model for car-
bonation represented by a differential equation with
a ‘sink’ term (R(w, c)) . The factors involved in the
equation include time, the porosity and diffusivity of
the material, the construction method and the presence
of cracks, and the geometrical shape of the surface
exposed to air. Carbonation depth (x) is proportional
√ root of time√(t) ± a constant (e) in the
to the square
form x = k t or x = e + k t, where k is a factor which
does not necessarily correspond to a property of the
Figure 2. Compressive test results on specimens 28 days material.
from manufacture and different water/lime ratios. It has been shown that the compressive strength of
air lime mortars varies in proportion to the extent of
carbonation (Lawrence, 2006a), and it is therefore rea-
sonable to expect that development of compressive
strength will also be proportional to the square root
of time.
The following formula has been developed to model
the variation in compressive strength of air limes as the
water/lime ratio varies.
887
Figure 5. Day 28 air lime data compared with the proposed Figure 8. Proposed equation applied to mortars made with
equation. different aggregates. 4SS3 = silicate sand; 4BN3 = crushed
bioclastic limestone (Ham Hill stone); 4ON3 = crushed
oolitic limestone (Stoke Ground Bathstone).
888
mortars made with oolitic stone shows greater early be done to understand the mechanisms involved which
strength than predicted by the equation. Indeed early produce such significant differences in compressive
strength exceeds that of moderately hydraulic lime strength. Such an understanding will allow the formu-
mortars. It has been shown by Lawrence (2006a) that lation of a more developed equation than that which
no hydraulic effects are involved, and this phenomenon has been proposed.
is the subject of ongoing research.
REFERENCES
Allen, G., Allen, J., Elton, N., Farey, M., Holmes, S.,
5 CONCLUSIONS
Livesey, P., Radonjic, M. 2003. Hydraulic lime mor-
tar for stone, brick and masonry. Shaftesbury, Donhead
The experimental data demonstrate that, unlike cemen- Publishing Ltd.
titious or hydraulic lime mortars, the increase in Lanas, J., Alvarez, J.I. 2003. Masonry repair lime-based mor-
porosity produced by higher water content does not tars: factors affecting the mechanical behaviour. Cement
result in the strength reductions predicted by Abrams’ and Concrete Research 33(11): 1867–1876.
rule. Indeed it can be seen that the choice of aggre- Lawrence, R.M.H., Walker, P., D’Ayala, D. 2006a. Non-
gate has a significantly greater impact on the strength hydraulic lime mortars. The influence of binder and
characteristics of air lime mortars. filler type on early strength development. Journal of
Architectural Conservation 12(2): 7–33.
One key implication of this are that for air-lime
Lawrence, R.M.H. 2006b A study of Carbonation in non-
mortars, water/binder ratios can safely be selected to hydraulic lime mortars, unpublished PhD Thesis, Univer-
produce a mortar with suitable workability charac- sity of Bath.
teristics as demanded by the particular application, Nagaraj, T.S., Banu, Z. 1996. Generalisation of Abrams’ law.
rather than based on the erroneous assumption that Cement and Concrete Research 26: 933–942.
Abrams’ rule applies. Unlike hydraulic limes, with Neville, A.M. 1995. Properties of Concrete. Harlow,
air-lime mortars apart from very stiff mortars made Longman.
with water/lime ratios of 0.5 compressive strengths Schafer, H.R., Hilsdorf, H.K. 1993. Ancient and new lime
are shown to be remarkably unaffected by greater mortars – the correlation between their composition struc-
ture and properties. In M. J. Thiel Conservation of Stone
quantities of water in the initial mix.
and other Materials. London, E. & F.N. Spon: 605–612.
The proposed equation will allow a first approxima- Van Balen, K., van Gemert, D. 1994. Modelling lime mortar
tion of compressive strength of different air lime mor- carbonation, Materials and Structures, 27: 393–398.
tars to be predicted. Continuation of the present study Winnefeld, F., Böttger, K.G. 2006. How clayey fines in aggre-
will allow the creation of empirical constants for types gates influence the properties of lime mortars. Materials
of lime and aggregate (Km and Kl ). More work needs to and Structures 39: 401–411.
889
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
G.E. Bei
Aristotle University of Thessaloniki, Greece
ABSTRACT: The aim of this study was the investigation of the mechanical properties, the drying shrinkage
and the hydroscopicity of soil mortars in order to firstly understand and then classify their behaviour; for their
use as structural mortars or renderings at historical building interventions or modern building structures.
Soil mortar compositions based on traditional data and improved –stabilised mixtures were designed using
additives, admixtures and fibres. Their properties were estimated with the help of mechanical, capillary suction
in pure and salted water and shrinkage tests. Useful conclusions have been revealed, based on the tests results
and the comparisons, related to the studied parameters for the mortars behaviour. Finally, some of the studied
compositions, according the above criteria were selected for structural mortars or renderings and they were
proposed for further durability and compatibility with the substratum measurements.
Soil was one of the main building materials in Greece 2.1 Data – Preliminary tests
from the prehistoric times until the beginning of the
It is obvious that the design methodology of appropri-
20th century, when the concrete appeared in con-
ate intervention materials on the historical buildings
struction. Less than a century time the use of soil
requires the study and the analysis of the existing mate-
in structures has been underestimated and neglected.
rials. This experimental work included the study of the
Nowadays, the environmental consequences of the
soil mortars as they were used in the past. Before the
modern materials use, such as this of concrete (CO2
begging of the production of the compositions, var-
and other toxic emulsions during its production) has
ious preliminary tests were carried out. Mortars of
led the developing world to re-establish the tradi-
different compositions have been selected from tradi-
tional structural materials. Therefore, soil is studied
tional buildings in several regions of Northern Greece
on a scientific base. Its behaviour and improving
(Kozani, Kastoria, Thessaloniki) and analysed.
ways of its properties are systematically investigated.
The results of this study showed that external soil
However, there are still limited experimental data
mortars on existing buildings were demonstrated by
(Venkatarama & Gupta 2005, 2006).
high percentages of CaO and therefore they could be
This work aims at contributing to the study of
characterised as soil-lime mortars. On the contrary,
soil as a building material evaluating soil mortar
internal mortars proved to be pure without admixtures.
compositions. Mixtures based on traditional compo-
Soil mortars used as renderings contained a 0.7 to 1.5%
sitions were examined. Improved-stabilised mixtures
of straw with a length of 2 to 5 mm or small wood sticks
were designed using additives, admixtures and fibres.
with maximum diameter 3mm and 20 mm of length.
Their properties were examined performing mechan-
ical, shrinkage during drying process, absorption and
capillary suction capacity in pure and salted water
tests. 2.2 Evaluation of soil
The tests described above made it possible for direct The above results (§2.1) were taken into account
results to be achieved for an indicative amount of spec- for the design of the new soil mortars compositions
imens, because both the production procedure and the of this project.Two different soil types have been
experimental process are time consuming. used as the basic materials. Both have been obtained
891
Table 1. Chemical analyses of soils type A and B. The magnitude of shrinkage varies with the clay
content and the type of clay (Balderrama & Teuton-
Percentage of Percentage of components ico1983). In fact, it was respected that that soil A
components soluble to 0.1 nHCl composition will shrink more than those of soil B. On
Soils A B A B the basis of mineralogical analyses, soil type A and B
were tested on the clay fraction (2 ÷ 0.2 µm). Soil A
Na2 O 1.4 2.5 0.11 0.28 had a more stable structure than B due to the higher
K2 O 2.8 1.3 0.04 0.03 percentage of illite in its composition. On the contrary,
CaO 1.99 6.2 0.70 1.80 soil B, which had less illite and a lot more of smectite,
MgO 2.63 3.63 0.33 0.44 bulked immediately after the beginning of evapora-
Fe2 O3 8.68 7.1 0.13 0.22 tion. This was observed during the drying process in
Al2 O3 18.68 19.03 0.26 0.30
almost all the specimens with soil B in the mixture.
SiO2 52.8 51.94 0.22 0.17
Weight loss 11.5 8.3 These soils can be described as Clay Loams (CL)
PH 8.5 8.9 according to Atterberg limits. Particularly, soil type A
was described as a medium plasticity clay soil with liq-
uidity limit (WL ) 38.5 and plasticity index (IP) 16.7,
while soil type B as a low plasticity clay soil, with
Table 2. Percentage of salts on soils type A and B. WL 30.1 and IP 11.74. Both types belong to A6 cat-
egory according to the American Society for Testing
Percentage of salts (%) and Materials (ASTM D423-66 & D424-59). A6 indi-
A B cates plastic soils, which usually present an important
volume change, when passing from dry state to humid
Chlorides 0.014 0.001 state. However, it should be pointed out that these clas-
Nitrates 0.0007 0.0008 sifications are not always good indicators of the soil
Sulfates 0.034 0.041 mortars behaviour. For instance, the plasticity index is
computed by means of the separate soil fraction which
is smaller than No 40 mesh and not all of the fractions.
Therefore, the plasticity index value does not neces-
Table 3. Grain size composition (%) of soils type A and B. sarily reflect the actual plasticity of the soil used for
mortars (as in our case).
Grain size composition A (%) B(%)
sand fraction SF 52.485 38.670 2.3 Requirements for the use of mortars in
silt fraction MF 21.515 42.940
conservation practice
clay fraction CF 26.000 18.390
In 1981, for the first time, a tentative list of the char-
acteristics of an ideal mortar for restoration was pre-
sented. These were easy workability, rapid and reliable
from special sites of brick manufacturing; soil A (red setting in both dry and wet environments, slow dry-
colour), from Western Thessaloniki, soil B (brown ing shrinkage during setting (crucial for soil mortars),
colour), from Eastern Thessaloniki. These two sites good mechanical and thermal characteristics, porosity
are areas from where earth has been traditionally similar to the components of the masonry and solu-
obtained and processed since at least the12th century ble salts content as low as possible (ICCROM 1981).
AD (Theocharidou 1988). Renderings of soil that are directly exposed to the
Chemical analyses, grain size composition tests and environment should have high hardening speed, water
Atterberg limits were defined and additionally miner- resistance, low level of microcracks during drying pro-
alogical analyses were realized in order for soils to be cess and good bonding with the substratum. Generally,
classified (Table 1, 2). the production of the soil mortars of this work took into
Both samples contained low percentage of sulfates, account the above requirements and specimens with
allowing us to foresee a fairly low susceptibility to porosity and mechanical characteristics similar to the
efflorescence. Soils A and B showed a high content in traditionals were produced.
SiO2 and Al2 O3 , in accordance with a high content in
clay minerals. Soil B was more calcareous than A, and
2.4 Tested mixtures compositions
that both have abundant Fe+++ .
The grain size distribution of the two tested soils For each soil A and B, seventeen (17) compositions
showed that the clay fraction (philomorphe material) were produced firstly on the basis of traditional mor-
of soil A is higher (26%) compared to B (18.39%) tars (taking into account analyses of soil mortars
(Table 3). from existing structures §2.1) and secondly with the
892
Table 4. Soil mortar compositions for soil A. Table 5. Soil mortar compositions for soil B.
893
The compressive and flexural strength of the speci- Generally, compositions improved with hydrated
mens at different ages (28, 45, 90, 180 days) were lime or with a mixture of hydrated lime with poz-
measured. Ultrasonic test, where the dynamic modu- zolan without fibres, showed a tendency of decrease
lus of elasticity in specimens, was calculated, capillary the strength at the age of 28 and 45 days. This may
suction capacity tests in pure and salted water, drying be due to the interference of the clay material of the
shrinkage (curing conditions predetermined), absorp- soil in the absorption of the available water by lime
tion tests were performed in order to define the open and pozzolan in the mixture. Nevertheless, hydrated
porosity of the soil mortars. As far as shrinkage test is lime and pozzolan, have contributed to the increase
concerned the results of the consequent measurements of the strength especially concerning the mixtures of
are presented as a proportion of the initial volume of soil A which were more argillaceous while the addi-
the specimen. The first measurement at the day after tion of cement (composition B5) contributed better to
drying in the mould (0 day) was considered as the strength on compositions with sandy soils as soil B
plastic deformation (Figure 6). (Table 4 & Table 6 compositions A.4.1, A.4.2, A.4.3,
A11 and compositions B11, B.4.1, B.4.2 and B.4.3).
The open porosity of compositions at the age of
3 RESULTS – DISCUSSION 28 days is presented in Table 6. The porosity was cal-
culated into accordance with the bibliographical data
The experimental investigation focused on the follow- (absorption test) and it was relative to the strength
ing major criteria: The strength of the specimens, the of the tested mortars. In fact, minimum porosity cor-
capillary suction capacity and shrinkage during drying responded to maximum strength (Papayianni 2001),
process. (Mojumbar et al. 2002).The addition of water reducing
or air entraining agents did not influence the poros-
ity of the specimens (compared to the compositions
3.1 Results of mechanical tests without additives). The values of the open porosity in
the soil mortars with lime (from 34.57% to 49,39%
The tests’ results on the Table 6 showed that strength
for compositions A.4.1, B.4.1, A.4.2, B.4.2, A.4.3 and
increased with time (columns (1), (2), (3) and (4))
B.4.3) were similar to the porosity of lime mortars
although in some cases, it seemed that there was a
(with porosity values between 30 and 40%) (Shafer et
decrease of strength especially for specimens with
al. 1992). The similar porosity could be an indicator of
additives. Specimens of soil B presented higher
compatibility between the tested soil and lime mortars.
strength comparing with those of soil A. This may be
due to the lower percentage of clay in the soil B which
makes the soil less “demanding” in the mixing water
3.2 Results on the capillary suction capacity
for the appropriate workability (Table 4).
Natural fibres increased strongly the flexural The compositions with soil B absorbs less water (3%–
strength of the soil mortars as it is shown on the Table 6. 14%) than those of soil A (3%–25%); this may be
It seemed that thin fibres with high dispersion (spe- due to the high content of clay materials (Figure 1
cific surface) were more efficient such as the egyptian & Table 4). The use of air entrainer reduced the per-
cannabis. In compositions with fibres, especially those centage of the rising water in the capillaries, even
of egyptian cannabis, it was not possible to measure though the porosity of the specimens with admixtures
the ultimate compressive stress of each specimen. It was not higher than the free admixtures composi-
was difficult to distinguish whe the specimen broke tions. It seemed a different distribution of voids, to
because the load indicator on the testing apparatus exist which influence negatively the capillary suction
continued measuring after the first crack. The spec- capacity (Figure 2).
imen with the fibres continued to absorb load energy Sand addition reduced the water absorption during
and be deformed. Therefore, it was decided to stop the suction (Figure 3). The addition of lime or pozzolan
measurement of compression test at the 50% of the in the compositions increased significantly the water
height reduction of the specimen. absorption through the capillaries (25% for A.4.1).
In fact, the specimen was plasticly deformed until After the beginning of the evaporation, the weight loss
its height, between the load plates-surfaces, became was immediate (Figure 4).The rate of water weight loss
the half (50% of slum). during the evaporation was slower when cement was
Sand percentage more that 20% in the mixture of added in the mixture. This seemed important because
the soil mortars seemed to decrease the strength. The it showed that the water content of the soil mortar sta-
cohesion between the coarse sand and the clay min- bilised with lime could easily evaporate compared to
erals is lower when the sand percentage is remarkable soil mortar stabilized with cement. It is well known the
(>20%) in the mortar composition. The dynamic mod- facility to water evaporation of lime mortars is the main
ulus of elasticity test detected voids and cracks in the advantage towards cement mortars for their durability
specimens within 60% of sand. and the prevention of frost attack (Papayianni , 1994).
894
Table 6. Compressive/flexural strength and open porosity of specimens.
The content of soluble salts in soil mortars should compared to those particles which sediment in fresh
be extremely low, if not non-existent, due to the well water. Since a highly flocculated structure gener-
known destructive effects of efflorescence. The layer- ates large voids, the mix will have low density that,
ing structure of the clay particles of the soil in relation ultimately, will weaken the strength of the mortar
to the presence of soluble salts underlies another (Balderrama & Teutonico 1983). The capillary suction
important consideration. capacity test of specimens in salted solution presented
Relevant works concerning the structure of soil the unfavorable influence of different components of
stated that clay particle sediments in salt water solu- the compositions in function with the salt elevation.
tions tend to develop a highly flocculated structure The test results showed that the use of air entraining
895
Figure 3. Influence of sand in the capillary suction capacity.
Figure 1. Minimum and maximum percentage of water
absorption among the tested specimens.
896
Figure 8. Influence of pozzolan at the shrinkage.
4 PROPOSALS
897
per u.w. and of 34.7% water per u.w. of the solid Therefore, is very important to identity the soil
materials. These specimens showed the second used through its mineralogical, chemical and physical
highest strength between the tested compositions. properties. If the strength improvement of the mortar is
Cement at proportion lower to 15% proved to be the objective, then the addition of cement in sandy soils
a good binder in soil compositions. Soil type B, or of lime – pozzolan in clayed soils may increase the
cement and water in combination with cannabis values of the mechanical characteristics. The addition
increased strongly the flexural strength (at 2.08 of thin fibres with high dispersion acts beneficially to
MPa in 28 days). The B9 composition provides the the increase of the flexural strength. At this work, rel-
lower shrinkage (4.93%) without visible cracks in atively high water resistance was obtained with sand
the texture and high hardening speed. addition in the mixture or with the combination of lime
B.4.1. specimens were composed of 75% soil, 8.3% with pozzolan because of the high evaporation speed
of lime and of 16.7% pozzolan per u.w. and of that it was achieved. Finally, the combination of addi-
39.1% water per u. w. of the solids. They showed the tive (cement, lime pozzolan) and sand decreased the
highest compressive strength among compositions volume shrinkage.
stabilised by lime and pozzolan
Compositions B7 and B9 could be proposed for
ACKNOWLEDGMENTS
additionally investigation as renderings. More specif-
ically, the main characteristics of the compositions
This research has been conducted at the Laboratory of
(given in order of their shrinkage) are:
Building Materials of the Civil Engineering Depart-
B9 specimens showed the lowest shrinkage (4.3%) ment of Aristotle University of Thessaloniki with the
and absorption of water (8.7%). Open porosity was help of prof. I. Papayianni and her support is gratefully
up to 38.9%, indicating indirectly that the evapo- acknowledged.
ration speed of B9 specimens is high, allowing the
mortar to “breath”.
B7 specimens were composed of lime and poz- REFERENCES
zolan in proportion 1 to 2. The use of cannabis
and sand restrained the microcraks during the dry- Balderrama, A. & Teutonico, J.M. 1983. Notes on manu-
facture of adobe blocks for the restoration of earthen
ing process. The shrinkage test showed 6.3% of
architecture. Rome: (ed.) ICCROM.
contraction which was one of the lowest rates of ICCROM (ed). 1981. International Symposium on Mortars,
shrinkage among the tested specimens. The absorp- Cement and Grouts used in the Conservation of Historic
tion of water during the capillary suction capacity Buildings. Rome. November.
test was 7% and was one of the lowest percentages Mojumbar S.C., et al. 2002. Thermophysical properties of
among the tested compositions. Open porosity was blends from Portland and sulfoaluminate – belite cements.
up to 35.58%. Acta Physica Slovaka, vol. 52.No. 5. October: 435–446.
Papayianni I. et al. 2001. The evolution of Porosity in
lime – based Mortars. 8th Euroseminar on Microscopy
to Building Materials, Athens:. 4–7 September.
5 CONCLUSIONS
Papayianni I. 1994. Durability lessons from the study of old
mortars and concrete., Proc. P.K. Mehta Symposium of
Nevertheless, the variety of soils is very large; rela- Durability of Concrete Nice:May.
tively all the soils, except these with organic admix- Shafer J. & Hillsdorf H.K. 1992. Ancient and lime mortars.
tures, could be used in construction when they are The correlation between their compositions, structures
modified and improved -stabilised in their compo- and properties. M.J. Thael, Proceedings Conservation of
sitions. The improvement of soils in situ may be Stone and other Materials. RILEM (ed): 605–613.
achieved using additives, sand, fibres and admixtures. Theocharidou K. 1988. A Contribution to the study of brick
The criteria of the evaluation of different factors in and Tile in the Byzantine and Postbyzantine Periods.
Bulletin of The Christian Archeological Society, vol. XIII,
the behaviour of soil mortars are determined from
Period IV. Athens. 97–112. (in greek).
the position of the mortar in the structure (internal, Venkatarama R. & Gupta A. 2005. Characteristics of Cement-
external wall or rendering). The strength, the change Soil Mortars, Materials and Structures, 38(280):639–650.
in volume during drying process and the hydroscop- Venkatarama R. & Gupta A. 2006. Strength and Elastic Prop-
icity were used in this research as a first approach erties of Stabilized Mud Block Masonry Using Cement-
on the study of soil mortars. Their properties were Soil Mortars, Journal of Materials in Civil Engineering.,
analysed by means of their mechanical characteristics, vol. 18. issue 3. May/June: 472–476.
their shrinkage and their capillary suction capacity.
898
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The paper reports the results of a research concerning the analysis and the interpretation of the
cracking pattern on the “Cappella dei Principi” (the Medici’s mausoleum) in the church of San Lorenzo (Firenze).
The research was motivated by the sudden collapse of a keystone of an internal arch. Firstly, the principal results
obtained from in-situ surveys and laboratory tests on cored samples will be described. Then the numerical
analyses developed in order both to obtain the static identification of the monument and to identify the possible
causes leading to the observed static problems will be described. The numerical modeling has been performed
by different steps, from linear and quite simple models built with plane elements up to a non linear model with
three-dimensional elements. The non linear FEM, performed with respect to a series of in-situ measures, allowed
to reproduce both qualitatively and quantitatively the behavior of the structure and its static problems in the area
of the main arches covering the side apses.
1 INTRODUCTION
899
Figure 2. Crypt layout.
Figure 4. Tambour layout (+30.0 m).
900
Figure 8. Chapel sector’s identification.
901
On the sectors where on Chapel level a side apse is A complete cracks recognition in the Chapel was
present, there is a crack that develops from the upper quite difficult due to the internal marble covering that
third of the dome, crosses the bell-window and reaches made impossible a global survey. On the other hand,
the keystone of the apse arch. On the sectors where on the external side only partial failures are visible on
the apse is not present (i.e. on the strong chapel walls) the keystone in the outer masonry arches of the apse
the cracks, still developing from the Dome, reach the sides. As mentioned before, the last damage occurred
bell-windows and do not go on over (Figs 9, 10). These on November 4th 1999 when a marble stone of the
cracks are oriented along the typical meridian direc- internal covering in an lateral apse (sector #3 in Fig. 8)
tions with the maximum opening at the windows level. dropped down revealing that one of the keystones of
As to the cracks on the Dome it is possible to the arch supporting the tambour and cover the apse
observe that these fissures, present on the extrados, had moved (Figs 11 and 12).
don’t continue on the vaults intrados. In is interesting
to observe that the crack pattern found on the monu-
ment is quite similar to the cracks layout observed on
similar structures (see e.g. the Dome of Santa Maria
del Fiore described in Bartoli et al. 1996).
Figure 10. Internal crack (sectors 5÷8). Figure 12. Keystone failure (particular).
902
In order to investigate the monument, and to plan the by heterogeneous material (disordered stone tied by a
necessary restoration, all the covering marbles in this good mortar). The second cored sample substantially
sector were completely removed (the marble elements confirmed the quality of the internal fill.
are approximately one square meter with a thickness Tests with single and double flat-jacks were per-
of about 5 cm, fixed on a base of stone about 15 cm formed on the arches covering the apse in sector #3,
thick anchored to the masonry by iron elements). This in order to estimate the stress levels in the masonry
removal permitted to plan an extensive experimental and to assess the elastic in-situ modulus. A thin single
survey. (or double) flat-jack has been inserted in a slot sawn
The same damage found on the stone arch in sector into a mortar joint. After the creation of a slot in the
#3 was also present in the stone arch in sector #5 (the masonry, the compressive stresses present at that point
high altar). cause the closure of the masonry above and below
the slot. Therefore it has been possible to evaluate the
compressive stress state in the masonry by introduc-
3 THE IN-SITU SURVEY ing a flat-jack into the slot and increasing the applied
pressure until the original distance between the points
Both in-situ and laboratory tests have been made in above and below the slot was restored. The test is only
the “Cappella dei Principi” compound. The in-situ slightly destructive, as after the test is completed the
tests were performed to assess the global structural flat-jack can easily be removed and the mortar layer
behavior and the local masonry characteristics; the lab- restored to its original condition. The value of the
oratory tests (crushing tests on cored samples) where pressure p applied by the flat-jack approximately cor-
performed to evaluate the mechanical characteristics responds, taking into account some corrective factor,
of the masonry constituting material, mainly the ulti- to the local pressure in masonry. In particular the aver-
mate strength of the stone as well as its mechanical age compressive stress in the masonry, σ m , could be
properties (Binda et al. 2000). Results of this exper- calculated as follows:
imental survey have been used to tune a numerical
model able to correctly reproduce the actual behav-
ior of the building (Bartoli et al. 1996). The model
has been built starting from results obtained from the In (1) Ka is a factor that accounts for the ratio between
geometrical and architectural survey; in a following the bearing area of the jack in contact with the masonry
phase the mechanical characteristics obtained from and the bearing area of the slot; Km is the factor that
experimental tests have been assigned to the materials. accounts for the physical characteristic of the jack
In particular, two cored samples have been extracted and Pf is the pressure required to restore the original
at the level of the tambour (Fig. 13). The first was taken distance between the gauge points.
from the apse in sector #3 starting from the external The double flat-jack technique has been used to
surface of the building, and it crosses the wall section. evaluate the masonry elastic modulus (Bartoli et al.
The second one was taken inside the masonry wall, tak- 2000). The test is made by using two flat-jacks into
ing advantage of the presence of the windows. Even if parallel (horizontal) slots one above the other; the pres-
is quite difficult to extend results of local inspections sure into the flat-jacks is gradually increased in order
to the whole building, this inspection provided useful to induce a compressive stress state in the masonry
information on the constructive techniques. The first
cored sample permitted to identify the presence of a
multi layered wall. The external face is made by stone
masonry with a thickness of about 33 cm; the other
external layer is made of stone masonry with a thick-
ness of about 75 cm; the internal filling is composed
Figure 13. Log position at tambour level. Figure 14. Flat-jack tests position (arches sector #3).
903
between the two slots. By this procedure the stress–
strain relation has been obtained by measuring the
deformation of the masonry. Tests have been repeated
in different positions.
The first investigated location was the keystone of
the stone arch covering the apse supporting the tam-
bour. In this point, after the creation of the slot where
the flat-jack had to be inserted, an opening of the
masonry above and below the slot was observed, this
meaning that in this area a tensile stress was present.
This arch was also tested at the spring (second loca-
tion) with single and double flat-jack tests. In this
area, the single flat-jack test reported a pressure of
23–25 bar corresponding to a stress of about 2.25–
2.45 N/mm2 .The third investigated location was on the
area close to the spring of the internal masonry arch
(Fig. 7). The aim of this third investigation was to iden- Figure 15. Crack opening under the bell-shaped window.
tify the functioning of this arch (i.e. to assess if this was
a structural arch or if it was just an architectural ele-
ment without structural relevance). The test resulting
in a pressure of 2.3 bar corresponding to a compressive
strength of about 0.22 N/mm2 . This low tension level
means that probably this internal arch does not play a
structural role and that the transfer of the vertical loads
from the Tambour to the ground is mainly left to the
upper stone arch.
904
Figure 17. Finite element model of the “Cappella dei Figure 18. Link10 insertion points.
Principi”.
905
and weak walls (odd sectors in Figure 8) used as lat-
eral apses. Above the chapel these lateral apses are
closed by a complex system of arches constituting
the basis for the Tambour. This geometrical difference
probably has caused (as observed with the numerical
model) changes on the stiffness of the Tambour that
have modified the normal stress state of this archi-
tectural element. The Tambour area above the strong
walls is stronger than the corresponding one over the
weak walls. This could be considered the origin of
the switching of the stress state. Due to this deviation,
unloading on the arches has probably originated the
sliding on the keystone, also due to the specific shape
of this keystone (a parallelepiped instead of a wedge).
The retrofitting needs to take into account these phe-
nomena and it must cover these two points. Firstly, it
Figure 21. Stress state at the arch keystone (kg/cm2 ).
is necessary to create cohesion between the stone ele-
ments of the arch; secondly is important to increase the
global tensile resistance of the arch taking into account
that, probably, in the rest of its life the described phe-
nomena will be amplified. Another aspect that needs to
be considered is that the intervention must be designed
as a long term intervention. The proposals described
in this paper foresee to re-establish the arch cohesion
by means of internal steel bars that connect the three
levels of stone on the arch. At the intrados an external
covering with FRP is suggested to ensure a general
improvement of the arch functioning.
6 CONCLUSIONS
906
cracks on the Tambour-Dome systems: as a matter of Bartoli, G., Chiostrini, S. & Innocenti S. 2000. Problems
fact, the described cracks would have been developed related to the analysis of experimental data from flat-jack
also assuming a fixed base model (i.e. not considering tests. Atti del Convegno CICOP2000. 5th International
structure-soil interaction). On the contrary, when the Congress on Restoration of Architectural Heritage.
Binda, L., Saisi, A. & Tiraboschi, C. 2000. Investigation
latter is assumed, a development of crack on the crypt procedures for the diagnosis of historic masonries. Con-
vaults is also observable (as visible on the monument). struction and Building Materials 14(4): 199–233.
Results herein reported offer a first identification of Chiarugi, A., Bartoli, G. & Morano S.G. 1998. The surveil-
the buildings behavior. A more exhaustive interpreta- lance of Brunelleschi’s Dome in Florence. Proceeding of II
tion of the overall building functioning will require an International Seminar on Structural Analysis of Historical
interaction between several modeling strategies (i.e. Constructions, Barcelona.
limit analysis, non-linear analysis) each of which can Chiarugi, A., Fanelli, M. & Giuseppetti, G. 1983. Anal-
offer significant elements. ysis of a Brunelleschi-Type Dome Including Thermal
Eventually a retrofitting proposal was presented Loads. IABSE Symposium on Strengthening of Building
Structure, Diagnosis and Therapy, Zurich: 169–178.
considering that engineers involved in monumental ICOMOS 2001. Recommendations for the analysis, conser-
buildings assessment are called to pay attention both vation and structural restoration of architectural heritage.
to the economic and cultural relevance of this kind of
building, next to the safety requirements.
REFERENCES
ANSYS INC. 1992. Users’s manual. Swans. analysis
systems.
Bartoli, G., Chiarugi, A. & Gusella, V. 1996. Monitoring
systems on historical buildings: the Brunelleschi Dome.
Journal of Structural EngineeringASCE 122(6): 663–673.
907
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M. Miri
Department of Civil Engineering, University of Sistan and Baluchestan
T.G. Hughes
Cardiff School of Engineering, Cardiff University
ABSTRACT: Small scale centrifuge models were used to study the behaviour of arches repaired using different
techniques. The models under test were 1/12th scale replicas of a 6 meter single span three ring arch. Two types
of arch geometry, with span/rise of 4 and 2, were studied as a shallow and deep arch geometry. The models were
tested in a centrifuge under a steady equivalent gravity of 12 g.
Two types of 2-D and 3-D arch models were studied which had the same geometry but different in the addition
of spandrel walls. The 3-D models were built with spandrel walls but without any wing walls or parapet. The
models were usually tested with fourteen passes of a rolling load and then up to the observation of first signs
of failure to enable them to be suitable for applying a repair method. The repaired models were tested using the
same procedure but up to the full failure load.
Plastic mesh reinforcement, stitching, and concrete slab on top of the soil backfill were applied as repair
methods to the 2-D arch models. Stitching of arch barrel and the barrel to the spandrels, applying partial saddle
concrete and strengthening of spandrel wall using reinforced concrete were tested in the 3-D arch models. The
results are presented and compared with each other in this paper.
909
Table 1. Detail of models under test.
Parameter Dimension
910
significant effect of the repair method in arch load other one in the longitudinal direction. The saddle con-
capacity, the repair was repeated to the deep arch crete is stitched to the spandrels walls to avoid any
geometry. disconnection between the spandrel and barrel.
Two rows of holes with a depth equivalent to one
ring were initially drilled to the arch barrel and 2 mm
4.1 Plastic mesh reinforcement steel rods were installed in them using epoxy resin.
The second 2-D model (S2D-2R) was repaired using These rods had an extra length of about 12 mm from the
plastic mesh reinforcement. Two mesh layers of hole depth through the saddle and connected the barrel
TENAX Promat (TENAX ) which was produced by the extrados to the saddle concrete. A width of 60 mm in
TENAX UK Limited Company for ground reinforce- model scale (that means 720 mm in prototype scale),
ment and erosion control, were placed in the backfill. was used which should be available as a pavement on
Top layer of the backfill was removed from the arch most of arch bridge or may be provided by restricting
barrel, mesh layer was installed and the backfill was a single lane of traffic to a narrow lane.
replaced again.
4.5 Spandrel wall strengthening
4.2 Concrete slab repair The third and fourth 3-D arches were repaired with
Following the initial tests, S2D-3R and D2D-1R the strengthening of their spandrel walls using a rein-
arches were repaired by laying a reinforced con- forced concrete slab. After completing the benchmark
crete slab on top of the backfill. In this case, the tests, models were repaired by applying reinforced
un-strengthened tests had a crown backfill depth of concrete to the inner sides of the spandrel walls. The
13 mm to readily facilitate the placement of the con- location of the concrete was restricted to those parts
crete without overly distributing the damaged arches. of the structure that would in normal circumstances
The overall depth of construction over the crown of be readily accessible for such work. In the present
the arch, including the 17 mm of concrete layer in the study the vertical extent of the reinforced concrete was
strengthened arches, was 30 mm. restricted to 100 mm in depth (about 1.2 m in the pro-
The concrete itself was manufactured with 2.0 mm totype) this would require only limited support during
aggregate as the coarse material, Chelford 95 silica construction.
sand as the fine aggregate and OPC. (BS 12 1991),
with mix proportion of 1:1.8:2.8:0.6 (cement: fine:
coarse: water) by mass. Compressive strength tests 5 TESTS RESULTS
on 25 mm concrete cube samples yielded 56 N/mm2
according to British Standard (BS 1881-116). The Laboratory experiments have successfully been car-
model concrete was nominally reinforced with a mesh ried out on two different geometries of 1/12th scale
of type 304 manufactured of 0.8 mm mild steel at single span centrifuge arch models. The experiments
20 mm centres. provided useful information on the effectiveness of the
tested repair methods on the failure mechanism and
particularly on the service and ultimate load capacity
4.3 Stitching the arch barrel of the arches. Final comparisons of all the repaired
The first 3-D model was repaired by stitching the arches are presented in Figures 2 and 3 for the 2-D and
arch barrel with steel bars between the barrel and 3-D model tests.
the spandrel walls joints. The patterns of holes with The general conclusions are:
an angle of 45 degree were drilled in the arch barrel – The use of plastic mesh reinforcement, test S2D-2R,
and 2.4 mm diameter stainless steel bars type 304 was had no significant effect on arch load capacity and
placed in them. pressure distribution under applied loads.
– The relieving concrete slab on top of the backfill sig-
nificantly increased the ultimate arch load capacity
4.4 Partial saddle concrete
for both shallow and deep arch geometries. The load
Partial saddle concrete to the extrados of the arch was at failure of the shallow strengthened model was
employed to repair the second 3-D arch. Strengthening 3.4 times that of the benchmark model and 2.7 of
the extrados of arch barrel is an appropriate method the average benchmark. The results for the repaired
of repairing arches. The advantage of this method is deep arch were, respectively, 3.7 and 3.2 times as
that it not only strengthens the arch but also improves strong as the benchmark arch. The application of
load distribution and ties together any cracked sections the slab to the surface appears to be at least equiva-
(Department of Transport). lent to application directly to the arch intrados and
A 60 mm width of concrete slab was laid on both extrados.A concrete slab on top of the fill distributes
edge of the arch barrel from one abutment to the the pressure and decreases the recorded pressure on
911
to connect the extrados ring and spandrel walls to
each other. Test results showed no prevention of ring
separation by the stitching bars but an increase of
50% in the ultimate arch load capacity following
application of this method.
– Applying a partial saddle concrete on part of the arch
barrel in addition to steel bar connection of the con-
crete to the spandrel increased the arch load capacity
and the stiffness of the arch. The arch load capacity
was improved by about 215% for the repaired arch,
which is comparable with applying the same con-
crete on top and beneath the barrel in the 2-D arch
model tests.
REFERENCES
Sumon, S.K. Repair and strengthening of five full scale
masonry arch bridges. in Second International Conference
Figure 2. Comparison between benchmark and repaired on Arch Bridges. 1998. Venice, Italy.
2-D Failure load. Melbourne, C., M. Begimgil and M. Gilbert. The load test
to collapse of a 5 M span brickwork arch bridge with
tied spandrel walls. in Arch Bridges. 1995. Bolton: Tomas
Telford.
Ashurst, D. An assessment of repair and strengthening tech-
niques for brick and stone masonry arch bridges. 1992,
Transport Research Laboratory: Crowthorne.
Baralos, P. (2002). “The small-scale modelling of repair tech-
niques for masonry arch bridges using a geotechnical
centrifuge.,” PhD, University of Wales, Cardiff.
Burroughs, P. O. (2002). “A study of parameters that influence
the strength of masonry arch bridges using a geotechnical
centrifuge,” PhD, University of Wales, Cardiff.
TENAX. (2002). “Ground Reinforcement and Erosion Con-
trol, Product Guide and Price list." Tenax UK Limited.
BS 1881-116. (1983). Testing concrete – Method for determi-
nation of compressive strength of concrete cubes, British
Standard Institution, London.
Department of Transport. (1997a). “BA 16/97 – The Assess-
ment of Highway Bridges and Structures.” HMSO,
London.
912
Use of traditional, alternative and
innovative materials
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: A quite speared methodology of restoration envisages the introduction of new timber elements,
connected to the standing structure. An efficient alternative to the most used jointing technologies, which employ
metallic bolted plates, steel pins, or epoxy resin adhesives, is the use of timber pins. The design rules for pinned
joints given by Eurocode 5 are based on Johansen’s theory. The present research has the target of verifying the
applicability of this theoretical model, and of the derived practical formulas, to the case of timber pegs. At this
aim an experimental research has been done, and obtained results have been compared with theoretical ones,
showing as the different nature of the material employed for the pins needs more specific evaluations of the
“yielding moment” and of the failure mode in case of low slenderness. Furthermore, an appreciable ductility of
timber connections made with timber pegs has been detected.
1 INTRODUCTION
gluing and timber pegs made of Azobè wood, finally
giving to this gluing the function of granting the con-
1.1 Employing timber pegs
tinuity to the connection, while timber pegs strength
In the field of ancient timber structures restoration, has been verified in case of exceeding slipping of the
the need of substituting rotten portions of structural gluing or its yielding. The evaluation of the design
elements, preserving the original spatial configura- strength of the double shear plane dowel joint has been
tion without modifying the structural behaviour, is made following Eurocode 5, introducing in the for-
quite frequent. In particular, in ancient trusses and mula calculating the bolt yield moment, the timber
floors, taking off the end portions of rafters or beams, peg characteristic bending stress value. At the IATF
damaged by biotic agents often due to high mois- Institute in Florence an experimental research has been
ture content inside the masonry walls, is sometimes done with specimens having the same dimensions of
inevitable. the real used timber elements and the same wood
A quite speared methodology of restoration envis- essence (silver fir), making glued and bolted joints,
ages the introduction of new timber elements, made with the really employed configuration as well as var-
of massive wood or glulam, connected to the standing ied ones, fastened with timber pegs of Oak and Azobé,
structure. This connection is generally realized with and steel bolts. In the case of timber pegs the rupture
metallic bolted plates, steel pins, or epoxy resin adhe- mode has most frequently been that of peg bending
sives (Mormone & Russo Ermolli 1999). Those joint- failure.
ing technologies show some disadvantages as moisture Another experimental investigation has been done
condensing problems in the interface between timber by Menegotto (2001) at the Architectural Construction
and steel elements; the questionable aesthetic output Department of IUAV in Venice on double shear plane
in uncovered structures; the difficult reversibility of bolted joints made of fir glulam, connected by tim-
the intervention with resin adhesives. ber pegs of beech wood with variable diameters (30
An efficient alternative is the use of timber pegs in and 50 mm). The author has noticed that the most fre-
timber to timber connections as all those handicaps quent rupture mode has been due to tension failure of
can be overcome. timber pegs.
An example of joints made with timber pegs has Actually the design rules for pinned joints given by
been realized in the restoration of the timber trusses Eurocode 5, the normative for timber structures which
in the Pieve of San Marino (San Marino Republic), ought to be acknowledged by each European coun-
(Ceccotti et al. 1998). The extremities of rafters and try, are based on Johansen’s theory, (Johansen 1949),
tie-beams, which showed a high level of decay, have named EYM, (EuropeanYield Model), which assumes
been substituted by timber prosthesis connected with a perfectly plastic behaviour for the connected timber
a tenon – mortise joint, reinforced by an epoxy resin elements as well as for the metallic pins.
915
Table 1. Design formulas of bolts. In Formula 2 the assumed steel behaviour, which is
the same in traction and in compression, induces the
Rd = Peg Design Strength yielding limit value when the limit tension strength is
reached with linear stress distribution on the circular
Mode N
cross section.
I fh,1,d t1 d
II 0.5fh,1,d t2 dβ
fh,1,d t1 d 4β(2+β)My,d 2 EXPERIMENTAL PROGRAM
III 1.1 2+β
2β(1 + β) + fh,1,d dt12
−β
2β At the aim of studying the applicability of this the-
IV 1.1 1+β
2My,d fh,1,d d
ory, and of the derived design formulas, in the case
of timber pegs used as fasteners, the basic hypothesis
and also the matching of the collapse modes envis-
The aim of the present research is verifying the aged by Johansen must be verified. The definition of
applicability of this theory, and of the derived design the “yielding moment” of the timber connecting device
formulas, in the case of timber pegs used as fasteners, is also needed, which is employed when the joint col-
and it’s based on an experimental investigation made lapse passes through the creation of one or two plastic
at the Testing Laboratory on Materials and Structures hinges in the pin.
of the COMMA Department of Naples University So an experimental research in different phases has
“Federico II”. been carried on:
1.2 Design formulas based on Johansen’s Theory – characterization of the two different wood species
initially chosen for making the pegs, as present-
Eurocode 5 (UNI ENV 1-1 – 6.2.1) gives the design ing large specific gravity, durmast and ash, through
formulas to evaluate design bearing strength for fas- measurements of their density, compressive, bend-
tener of cylindrical shape in double shear plane bolted ing and cutting shear strengths;
joints, quoted in Table 1. – characterization of timber to be used for mak-
where: t1 , t2 are timber thickness; fh,1,d (fh,2,d ) is dowel ing the connected elements, fir glulam, through
bearing strength in t1 , (t2 ); β is the ratio fh,2,d /fh,1,d ; measurements of its density and compressive
d: bolt diameter; My,d : yield bending moment of the strength;
bolt. – rupture tests on double shear plane joints, on three
As already said, these formulas have been derived sample, each one constituted of three specimens,
from Johansen theory, dated 1949, relying upon the made of three connected timber boards with vary-
hypothesis of elastic – perfectly plastic behaviour of ing thickness ratios, to induce the different collapse
timber and of the connecting devise. modes envisaged by the theory. At this aim, the
Each evaluation of the bolt design load, for each connections have been designed to avoid the cut-
shear plane, corresponds to an envisaged failure mode: ting shear failure of the timber pegs, obviously
while in the first two formulas the dowel bearing absent in the Johansen’s model with metallic bolts.
strength of the jointed timber is overcome before any Through the readings of mutual slipping, made
plastic deformation of the bolt occurs, in the other two employing LVDT sensors, the behaviour of the con-
the creation of one or two plastic hinges in the bended nections under increasing loads has been studied,
pin takes place. Obviously the bearing strength of the particularly in reference to evaluation of the joint
bolt in the joint is the less of the four values calculated ductility.
with the previous formulas.
The characteristic dowel bearing strength fh,0,k of
the jointed timber elements can be evaluated with the 2.1 Characterization of timber chosen
Formula 1, (UNI ENV 1995 1-1, 6.5.1.): for the pegs
For both wood species the following mechanic char-
acteristics have been evaluated: parallel to grains
compressive strength; bending strength on clear spec-
where ρk is the specific gravity of timber (kg/m3 ); d imens; cutting shear strength; the corresponding mean
is the diameter of the bolt (mm). and characteristic values are quoted in Table 2.
The bolt yielding moment value is given by the Tests have been made with an electronic board uni-
formula in paragraph 6.5.1.2 of the same Code: versal load machine, in displacement control, Sun 5,
produced by Galdabini, with a maximum load of
50 kN, and, for larger loads, with a oil-pressure uni-
versal load machine of 2000 kN max capacity. Tested
where fu,k is the characteristic tension strength of steel. specimens have been conditioned in an ambient with
916
Table 2. Mean and characteristic values for ash and durmast Table 3. Mean and characteristic values of fir glulam.
timber.
Fir glulam
Ash Durmast
Properties Mean Charact.
Properties Mean Charact. Mean Charact.
Density (kg/m3 ) 447 392
Density (kg/m3 ) 652 572 680 597 Compression (MPa) 38.17 31.95
Compression (MPa) 56.29 49.42 51.01 44.31
Bending (MPa) 112.48 84.61 75.81 54.85
Shear (MPa) 31.90 23.86 26.10 20.65
has been loaded in compression to apply shear on the
two bearing cross sections.
The obtained results suggest choosing ash timber to
make the pegs, as even if its density is less than that
of durmast, its mechanical characteristics show higher
values.
917
Table 4. Mean values of the ultimate load and of the
corresponding slipping for double shear plane joints.
Sample R1 R2 R3 R4
918
3.2 Comparison between theoretical and
experimental results
From Tables 4–5 comparisons, the following observa-
tion can be deduced:
– for samples of type B and C, the experimental ulti-
mate load diverges of only 4% (−4.73 and +4.36
respectively) from the theoretical value and the
failure modes observed in the tests are the same
of those obtained by the application of Johansen’s
formulas;
– for sample of typeA, the experimental ultimate load
overcomes the theoretical one of 44.21%; on the
other hand the failure modes shown by tests are not
uniform and only for one specimen the splitting
of the central timber board occurs, as theoretically
predicted. In fact, experimental results point out as
more determining for specimen failure the bending
behaviour of the peg than the dowel bearing strength
of the fir timber boards. Figure 3. Bending of the timber peg.
Even if the reduced number of specimens, only
three for each sample, doesn’t allow meaning gener-
alization, some interesting observations can be done.
Failure loads given by Johansen’s formulas can be
assumed to predict experimental data, in fact even
when they show greater values, as in the case of sample
type B, the difference is negligible. The strong diver-
gence between theoretical and registered values, which
occurs in the case of sample type A, together with the
observed absence of matching of the failure modes,
reveals the inadequateness of corresponding formulas
in the particular case. In fact to R1 and R2 theoretical
values it’s associated a failure mode only due to dowel
bearing strength of the jointed boards timber, while in
two of the three tested specimens the timber peg shows
a relevant bending (Fig. 3), strong enough to give rise
to joint failure.
3.3 Ductility
The analysis of the load – slipping diagrams of
each specimen, obtained by transducers registrations,
allows some evaluations relative to the ductility of
these joints constructed employing timber pegs. On
the diagrams quoted in Figure 4, which are the most
representative of each sample, the loads corresponding
to the change of slope in the load-deformation curve
have been chosen as yield limits.
These loads values and the corresponding deforma-
tions are quoted in Table 6.
The index of static ductility D is defined as the
ratio between the ultimate displacement and the one
corresponding to the yield limit.
In the last column of Table 6 are quoted the values
of D obtained as ratios between the ultimate dis- Figure 4. Load-displacement diagrams, with yield limit
placements of Table 4 and those quoted in Table 6. punts.
919
Table 6. Values of yielding loads, displacements and the First of all, obtained results show how the differ-
ductility parameter D. ent nature of the material used for the connecting
device needs a more specific definition of the “yield-
Load Displacement ing moment” of the pins, which must be inserted in
Sample N mm D
the formulas, if those ones have to be employed as a
design mean confidently.
Type A (A02) 14,700 1.880 5.09
Type B (B02) 21,303 4.156 2.56 Moreover, it’s been noticed that failure loads given
Type C (C02) 26,641 2.935 4.88 by Johansen’s formulas for the three tested samples of
double shear plane joints don’t diverge significantly
from the experimental ones, when the collapse mode
involves bending of the pin. Instead the theoretical and
The more or less dissipative behaviour of a joint is experimental data show absence of matching in those
generally given relating to its slenderness, defined as joints characterized by a large ratio between the diam-
the ratio between the minimum thickness of connected eter of the timber pin and the least thickness of the
elements and the diameter of the pin. connected timber boards. So the behaviour of joints
The Type B sample has the minimum thickness of characterized by low slenderness needs more analysis
the jointed boards given by t1 which is half of the not only from an experimental point of view, even if in
corresponding minimum thickness of Type C sample; the field of restoration low ratios between the jointed
consequently, as the pegs have a fixed diameter, the board thickness and the peg diameter difficultly occur.
two samples show a ratio of 1/2 in their slenderness. Another appreciable characteristic of connections
Tests results for these samples confirm a behaviour with timber pegs, experimentally observed and quite
which is quite expected in case of steel pins: a double important in seismic zones, is the good ductility of
value of slenderness gives an almost double value of these joints.
ductility.
The Type A sample doesn’t follows this scheme
in fact even if it has the less slenderness, an half REFERENCES
of that of type B joint, it shows the greatest ductil- Ceccotti, A., Bonamini, G., Ruffini, M. & Uzielli, L. 1998.
ity, which can be ascribed to the plastic deformation Restauro conservativo di capriate lignee. La Pieve di San
of the timber peg, scarcely opposed by the reduced Marino. Torino: C.L.U.T.
thickness of jointed boards. This behaviour obviously Giordano, G. 2005. Tecnica delle costruzioni in legno. V Ed.
occurs because the jointed elements and the connect- Milano: HOEPLI.
ing device are made of materials with stiffness of the Johansen, K.W. 1949. Theory of timber connections. Interna-
same magnitude. tional association for bridge and structural engineering
9: 249–262.
Menegotto, A. 2001. Al disopra di ogni aspettativa. Prime
indicazioni sulla possibilità di impiego di cavicchi.
4 CONCLUSIONS Adrastea 18: 28–35.
Mormone,V. & Russo Ermolli E. 1999. Palazzo reale di
This research allows some interesting observations Napoli. Interventi di recupero delle coperture. Recupero
about the applicability of Johansen’s theory and the e conservazione 27: 82–87.
derived design formulas in the case of connections UNI ENV 1995-1-1 Eurocodice 5 Progettazione delle strut-
made using timber pegs. ture di legno – Regole generali e regole per gli edifici.
920
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: A Seismic retrofitting of structures belonging to the architectural heritage requires meeting of
constraints which are related to preservation of artistic features. Such a philosophy was applied to the design
process of the retrofitting intervention of a masonry arch, belonging to a 17th century portico built inside a
Roman amphitheatre in the city of Spoleto (Italy), and of two masonry vaults of an ancient building in the city of
Foligno (Italy). This led to the opportunity of investigating the efficiency of an innovative composite material,
based on fine steel cords embedded in a cementitious matrix (Steel Reinforced Grout). To assure an adequate
strength against earthquake, a couple of prestressed SRG strips, in the first case, and a reticular system, made of
transversal and longitudinal prestressed SRG laminates, in the second case, have been applied on the extrados
of the arch and of the vaults, respectively.
921
collapse mechanism does not depends by the mate-
rials’ limit strength, but it is due to the incapability
of the structure to fit the horizontal and vertical dis-
placements of the abutments. As a consequence it is
clear that similar displacements should be considered
when strengthening masonry arches introducing only
systems, which are able to realize an effective “rein-
forcement” without changing the constructive features
of the structure.
Conversely, existing methods of repair often have
been based on the idea of working over to make the
structure resistant to the seismic actions, changing, in
this way, the main features of global behavior. Figure 1. Prestressing device (courtesy Eng. Giannantoni).
According to this, it has to avoid, for instance, that
methods (i.e., concrete slab), which, in order to limit the method is the use of pre-stressed strips. Loading
the effects produced by the thrusts, try to eliminate the vault in radial direction, the SRG reinforcement
them, changing, as a result, the static behavior of the increase its compression and improve its resistance to
structure, that is so reduced to a simple “ceilings”. For pressure-flexure induced by incidental loads.
the same reasons, the use of removing the spandrel The consolidation effect is realised by simply plac-
fill should be assess with care, taking into account the ing one or more strips alongside the extrados surface of
geometric configuration and the cracking pattern of the vault. The strips are fixed to the masonry of the sup-
the vault. porting walls and then pre-tensioned (Figure 1). This
Such problems have led researchers to suggest fact implies the transmission of radial self-equilibrated
strengthening masonry vaults with fiber – reinforced forces between the curved strips and the arch.
polymer (FRP) composites in the form of bonded As the compression resistance of the masonry is
surface reinforcements. There are several advantages usually high, it is possible, and not risky, to strongly
related to this strengthening technique: very low rise the axial internal load in the masonry, avoiding
weight, corrosion immunity, high tensile strength the formation of the four-hinges collapse mechanism.
and low thermal expansion coefficient. Moreover the All the structural section of the masonry will be more
somewhat easiness of execution of the intervention, compressed as in the original state, thus postponing
even in difficult operative conditions, allows a wide the formation of the cracks. Using this technique, the
range of possible applications in several situations reinforcement does not interfere with the in situ mate-
of damage, without considering that the possibility rial and respect the structural behaviour of the existing
of binding or wrapping structural elements made of building.
brittle materials (like masonry) allows, in most cases, The method permits a recognisable sign of con-
to avoid the collapse of the structure and so assure temporary interventions. The reinforcement so applied
the pursued safety conditions. Nevertheless, their lack works as an “active system”, which allows calibrating
of fire resistance and their relatively high cost may the actions as it needs and, if loss of pre-stress takes
represent an obstacle for a widespread use. places, allows re-tensioning.
According to these difficulties, beside the “tradi- Even in case of variation of curvature along the arch
tional” FRP, the use of a new family of composite ring, the reinforcement strips act in a beneficial way,
materials, based on high strength twisted steel wires as the mutual forces applied by the cords perpendic-
embedded within a cementitious grout (Steel Rein- ularly to the surface are maximal exactly in the zones
forced Grout), is more and more considered. The core where the radius of curvature is minimal. It has to
of the project is based on the idea to combine, along be noticed that the proposed consolidation technique
with the traditional advantages of the composite mate- works well only if the piers are able to sustain the lat-
rials, the performances of a material that, using twisted eral thrust induced by the arch (or vault). If they were
steel wires, allows the same applications of FRP mate- too weak, the structure would break in a section some-
rials with lesser costs. Without considering that the where between the springing and the keystone. This
use of SRG, because of the presence of a cementi- means that, in this case, the reinforcement have to be
tious grout, which replaced the polymeric resin of FRP prolonged till the base of the piers. They have to help
materials, allows to increase the fire resistance as well. balancing load across the spans and thus they have to
The flexural limited resistance of the masonry be placed up to the spans. In some case shear failure
vaults or arches can be overcome by introducing may take place in the structure and a sliding failure
“passive” reinforcing steel strip (Borri et al. 2007a), mechanism can occur. The reinforcement, although
making it able to sustain substantial bending moment less efficient than in the flexural induced collapse case,
in addition to axial loads, but a better application of still bring to an increase of ultimate load.
922
Figure 3. Layout of the intervention.
3SX-12
Figure 2. Cloister Portico: plan view.
923
Figure 4. Deflectometer used to register arch crown
deflection.
3.3 Installation
In order to ensure the safety of the structure, the arch
rehabilitation started with the installation of scaffold-
ing on the intrados both of the above mentioned arch
and of the two adjacent arches. This action is neces-
sary to supply, after the removal of the filling material,
the horizontal thrust of such arches and to bear ten- Figure 5. Anchorage region.
sile stresses generated by the prestressing of the SRG
sheets.
After surface cleaning by sanding and water based
solvents and then levelling the surface of the outer
arch area, bedding bands were created using suitable
cementitious grout.
The first step in the assemblage of the prestressing
device consist of fastening the anchoring steel plates to
the arch abutment through the use of two 16 anchor-
ing rods, inserted vertically, long enough (1000 mm) to
reach the height of the pier. Each anchoring plate con-
sist of three parts: a central plate (440 × 300 × 6 mm)
with two round holes for the anchoring rods and two
lateral plates (440 × 150 × 6 mm) joined to the central
plate by a butt hinge. Anchorage of the SRG sheets is
created by bonding the ends of the sheets on the afore-
said lateral plates with a polymeric resin and by fas-
tening two removable steel plates (440 × 150 × 6 mm) Figure 6. Bonding SRG sheets to the anchoring steel plate
of the loading region.
through the use of steel bolts. The removable plates are
fixed to the lateral plates by the tightening of the nuts
in the anchorage region. High pressure must be applied Because of the dimensions of such element, which
to the SRG sheets through steel plates and ten bolts at should not permit prestressing of the two sheets simul-
each end to prevent slipping of the SRG sheets, which taneously, it becomes necessary to use an intermediate
would result in a loss of prestressing force. Test results steel plate (440 × 150 × 6 mm), placed at a position
(Borri et al. 2007b) show that the friction resulting 250 mm distant from the loading device. Using such
from the pressure was sufficient to anchor the sheets a steel plate (Figure 7), it is possible to fasten the two
during prestressing. SRG sheets, fixed to the arch abutment, to the SRG
The next step consist of creating the anchorage of sheet, fixed to the loading device. As in the previous
the SRG sheets in the loading region. cases, the anchorage is created by fixing a removable
At one end, anchorage of the SRG sheets is cre- steel plate (440 × 150 × 6 mm) to the intermediate
ated by using the same anchoring steel plates as the steel plate through the use of steel bolts.
anchorage region (Figure 6). After the anchorage was created, the desired pre-
Conversely, at the other end, the reinforcement is stressing stress can be smoothly achieved in the SRG
fixed directly to the winding axis of the loading device. sheets by tightening the winding axis manually with
924
Figure 7. Fixing SRG sheets to the loading device.
925
Figure 11. Enlargement of the Music room (1712–1713).
926
Figure 14. Anchoring of the SRG sheets.
927
Figure 16. Prestressing of the SRG sheets.
5 CONCLUSIONS
Figure 18. Application of steel flat plates.
The operations carried out, firstly to save and then to
Also, it can be noticed that prestressing SRG strips consolidate and restore the masonry arch of the cloister
permits the reinforcement to follow a regular curve portico built inside a Roman amphitheatre in the city
parallel to the original ideal surface of the undeformed of Spoleto (Italy), and of two masonry vaults of the
vault. According to this, in the Music room, connected Jacobilli building in the city of Foligno (Italy) have
928
all followed the same philosophy. To place the most Foraboschi, P. 2004. Strengthening of masonry arches with
up-to-date techniques and technologies at the service fiber-reinforced polymer strips. Journal of Composites for
of culture, in order to respect the historic value of Constructions, ASCE 8(3), 2004, pp 7–16.
the ancient buildings and to obtain adequate safety Hardwire llc., 2002. What is Hardwire. Product Guide
Specification. Web site: http://www.hardwirellc.com.
levels, whilst changing as little as possible the orig- Huang, X., Birman, V., Nanni, A. & Tunis, G. 2005. Properties
inal structural conception. These technologies, never and potential for application of steel reinforced polymer
applied before in the field of restoration, have been and steel reinforced grout composites. Composites, Part
studied specifically for this occasion, offering new and B, Vol. 36, 2005, pp 73–82.
interesting possibilities for the safeguard of the World Jurina, L. 1997. The reinforced arch: a new technique for
architectural heritage. strengthening masonry arches and vaults using metal tie
bars. Proceedings of 16th National Congress of C.T.A.,
Ancona, Italy ,1997.
Matana, M., Galecki, G., Maerz, N. & Nanni, A. 2005. Con-
REFERENCES crete substrate preparation and characterization prior to
adhesion of externally bonded reinforcement. Proceed-
Borri, A., Casadei, P., Castori, G. & Ebaugh, S. 2007a. ings of International Symposium on Bond Behaviour of
Experimental analysis of masonry arches strengthened FRP in Structures (BBFS 2005), Hong Kong, China, 2005.
by innovative composite laminates. Proceeding of the Triantafillou, T.C. 1998. Strengthening of masonry structures
10th North American Masonry Conference (10NAMC), using epoxy-bonded FRP laminates. Journal of Compos-
St. Louis, Missouri, USA, June 3–6, 2007. ites for Constructions, ASCE 2(2), 1998, pp 96–104.
Borri, A., Castori, G., Giannantoni, A. & Grazini, A. 2007b. Valluzzi, M.R., Valdemarca, M. & Modena, C. 2001.
Performance of reinforced masonry bond beams. Pro- Behaviour of brick masonry vaults strengthened by FRP
ceeding of the 10th North American Masonry Conference laminates. Journal of Composites for Constructions,
(10NAMC), St. Louis, Missouri, USA, June 3–6, 2007. ASCE 5(3), 2001, pp 163–169.
929
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
˛
A. Kwiecień & B. Zajac
Cracow University of Technology, Cracow, Poland
R. Jankowski
Gdansk University of Technology, Gdansk, Poland
ABSTRACT: In the paper, the static and dynamic properties of proposed flexible polymer joints are discussed.
The discussion is based on investigations on special polymer, describing its properties obtained during laboratory
and field tests. The research has allowed us to point potential advantages of the polymer application in cracked
historical masonries. In addition, the rules of the polymer application and the possible conservative mask method
of the constructed joint are presented.
931
Figure 2. The new crack at the vault surface passing around
the mortar injection, too stiff in comparison to masonry
material.
Figure 1. Evolution of uniaxial tension stress-strain curve
(after Oñate et al. 1996). (Ortega et al. 1996), that represents very massive
intervention technique and the installation of shape
under movement caused by settlements, temperature memory alloy devices in the Basilica of San Francisco
changes and structural vibrations. in Assisi (Castellano & Infanti 2004), representing the
It is important that after rupture the surrounding of group of innovative techniques.
the crack remains a weak zone with the micro-cracks
occurring. In this zone the strength of the masonry 2.2 Rules of intervention in heritage structures
material is lower than of the original one what influ-
ences the repair strength the in case of the crack Nowadays, the use of innovative techniques or par-
bonding. ticular solutions, which efficiency must however be
demonstrated, should not be in any case forbidden. On
the contrary, the possibilities and limitations of each
innovative technique should be briefly explained and
2 CONCEPTION OF STRUCTURAL
also estimated and justified by calculating their effect
INTERVENTION IN CRACKED MASONRY
in terms of variations in the global behavior of the
structure. In the first place it is necessary to allow the
2.1 Discussion on applying of only original
structure to manifest a satisfactory global behavior,
materials
by improving the connections between the masonry
The requirements of safety and use are almost per- walls and between the walls and the floors. Interven-
manently in conflict with the respect of the iconic, tions aimed at increasing the masonry strength may
historical and material integrity of the monuments and be used to re-establish the original mechanical prop-
treatment where only using of traditional methodolo- erties lost because of material decay or, alternatively,
gies is acceptable. In most countries the conservation to upgrade the masonry performance. Techniques used
is controlled by officials having an influence on any must employ materials with mechanical and chemical-
action to be undertaken imposing constraints and lim- physical properties similar to the original materials
itations that sometimes appear unreasonable to the (Modena et al. 2006) and neutral from the interaction
engineer. On the contrary, the engineer tends to achieve point of view.
safety by means of solutions or procedures which It is highly preferred to select properly innova-
appear unacceptable to the officials in charge of con- tive materials to make them work together with the
servation. It is evident that some consensus has to be existing materials with better “compatibility” from the
found. On one side, conservation requires the safe- mechanical point of view. This is to reduce as much as
guard of the formal, material and historical integrity of possible high stress concentrations, which occur using
the monument, but also its survival and safe exploita- high stiffness modern bonding materials like high
tion (Viggiani 2006). It is important to answer the strength resin or high strength mortars. It should be
question: do we want to have the heritage object with noted that applying too strong strengthening material
only original intervention materials with the threat of in cracked structures (in comparison to the masonry
being destroyed by an earthquake or do we want to have properties) could cause additional damages. As an
the safe working historical structure for the next gen- example, the new crack passing around the mortar
erations, protected with small amount of non-original injection, too stiff in comparison to masonry material,
intervention materials that are in compatibility with a is presented in Figure 2. The crack is localized at the
monument and invisible for visitors. It is believed that repaired cloister vault of the Archeological Museum in
the second option is suitable. Cracow (Ciesielski et al. 2004). The cracks appeared
As an example of non traditional methodologies is after the mistaken intervention in foundation of the
the steel structure inside the spires of Burgos Cathedral structure. The criterion in selection should concern
932
be of the same nature as the in situ ones and because
of the positive effect of grouting on the mechanical
properties of masonry (Vintzileou 2006).
The polymers used in the FJM are flexible two-
component grouts based on polyurethane resin, hand
and machine applicable. They introduce also the pos-
itive effect on the mechanical properties of masonry,
thus should be accepted as innovative inject technique
Figure 3. The flexible joint made of polymer (by injection), for application in historical structures. They fulfill
bonding disrupted structural elements. requirements of the proper applicability into cracks
and the adequate rheological, physical, chemical and
mechanical properties.
much more deformability than strength, especially in Rheologically, they are injectable, have sufficient
seismic areas (Modena 2004). fluidity and stability. Physically, they do not pro-
duce high temperatures and shrinkage and also have
adequate hardening time and hygroscopic properties.
3 THE FLEXIBLE JOINT METHOD (FJM) They are resistant to temperature from −40◦ C to
+80◦ C (temporary up to +150◦ C) and UV radi-
3.1 Description of the method ation. The polymers are insensitive to moisture and
The Flexible Joint Method (FJM) is the method of also have long life expectance. Chemically, the poly-
repair using deformable elastic-plastic polymers and mers are neutral to the masonry materials (no chemical
reinforced polymers for bonding of damaged struc- reactions between them, related to both durability and
tures. It is dedicated to cracked masonries (of poor mechanical properties). Mechanically, the polymers
quality too), especially to historical structures where are reducing vibration shear-resistant and permanently
minimum intervention is required. The cracks are fill- elastic adhesives and also have adequate strength
ing in with the special binary mixed polyurethane and deformability characteristics. Giving an example,
mass injection, forming the flexible joints bonding the for the soft flexible polymer (of the Young modulus
disrupted structure elements (Fig. 3). This method is E = 4 MPa) described by (Kwiecień et al. 2006a, b) the
registered in the Polish Patent Department with No. tear strength is 1.7 MPa and adhesion to the concrete
P-368173 and was described in papers (Kwiecień et al. surface is 1.2 MPa with elongation of 60%.
2006a, b). The FJM permits safe work of the retrofitted The flexible polyurethane mass shows no signifi-
structure in the new stabilized state of balance. This is cant changes of mechanical parameters after 3 million
the method particularly conductive to objects, in which cycles in the fatigue test (producer data) with elonga-
a redistribution of stress occurred in consequence of tion of 10%, frequency 5 Hz and maximum stress of
damage (cracks). It permits further safe exploitation of 1.0 MPa. Cyclic tension-compression tests under har-
the object under additional static and dynamic loads. monic excitation were also conducted (by the authors
Cracks’ filling in the previously damaged bearing of the present paper) for the soft flexible polymer,
structure with specially designed polymer strengthens with frequencies of 0.1, 0.5, 1, 2, and 5 Hz. The set
the disrupted structural elements. It assures that the deformations of specimens (of 28 mm diameter and
damaged building regains tensile, compression and 28 mm height) were of +/− 10, 20, 30 and 40 %
shear resistance in place where bearing capacity was with the corresponding stresses of +/− 0.16, 0.35,
lost. Especially, the tensile resistance, deformability 0.62 and 0.97 MPa. The example results for 2 Hz fre-
and ductility are of big importance. The deformation quency are presented in Figure 4. The average damping
of the flexible joint under load assures uniform distri- ratio (Jankowski 2003) has been calculated as equal to
bution of stress along the lap joint over the total contact D = 0.06 for all obtained data.
surface, equalises deformation and damps vibrations. In addition, the field test (producer data) confirmed
The polymer mass bonds structural faults in places that a direct fixation of rail base-plates using flexible
of cracks and limits development of the new stress polyurethane mass presented only marginal changes in
concentrators. flexibility (6%) after exposure to weathering and fre-
quent dynamic loading over 28 years. The differences
obtained during long-term dynamic behaviour in situ
3.2 Properties of the polymers
were so minimal that the flexible joint can be expected
Grouting constitutes one of the most common tech- to last for many more years.
niques applied to historic masonries. Although grout- The polymers reach the full strength in 24 hours
ing is a non-reversible technique, it is well accepted after application. Depending on the actual state of
even for monuments of high historical and architec- masonry the proper polymer with required properties
tural value, since the materials added to masonry may can be chosen. The application rules of the polymers
933
Figure 5. Peaks of the stress concentration visible in the
photo-elastic research on a concrete specimen.
934
Figure 6. The pull-off device and the specimens after
damage.
935
Figure 12. Diagnostic dynamic exciters: the Vibrosejs of
20 T mass and the modal hammer of 5 kg mass.
4.4 Loading of the structure with static forces Figure 14. Maximum horizontal accelerations measured
on the top corner of the masonry (stationary harmonic
The building was loaded statically with the horizon- excitation).
tal force imposed in place where structure was excited
dynamically by caterpillar. The maximum generated
force was of 32 kN and caused permanent opening (0.4 range of 6–30 Hz were conducted. The analysis of the
mm) of the non bonded small crack (9/H), after over- building response (presented for horizontal vibrations
coming friction forces acting in the non bonded crack. measured on the sensor localized at the top corner –
Under the same load the bonded cracks showed no Fig. 12) showed dynamic characteristics of the struc-
reaction (10/H and 11/H) what gives evidence of the ture. Analysis of the impulse excitation pointed that the
polymer joint strength, keeping the structure elements flexible bonded masonry damps better vibrations than
together and protecting against destruction. the undamaged stiff structure (Fig. 13). The response
of the building repaired using the FJM is almost 40%
lower in resonance than for the undamaged building
4.5 Diagnostic dynamic excitation of and favourable frequency shift is observed. It should
the structure be noticed, that such damaged building, bonded using
The tested masonry structure was dynamically diag- the flexible polymer joint, survived the horizontal har-
nosed before damage, just after damage and after monic resonance vibration measured at the top of
repair using the polymer flexible joints. As dynamic the building, of the 60 second duration and of the
exciters the Vibrosejs of 20 T mass and the modal 30 cm/s2 acceleration amplitude (Fig. 14). There were
hammer of 5 kg mass were used (Fig. 12). no additional damages on the repaired masonry after
The 16-chanells system collected data from the the vibration test what confirm efficiency of the repair
accelerometer system installed on the structure. There using the FJM.
was no test with the Vibrosejs on the damaged struc- The excitation using the modal hammer allowed us
ture because of the real possibility of collapse. The to determine the inertancy given by formula (1)
Vibrosejs excited vertical vibrations at the soil surface
at the distance of 15 m in front of the masonry.
The impact of the Vibrosejs plate at the soil sur-
face and the harmonic excitations with frequency
936
Figure 16. The fragment of polymer bonded wall during
Figure 15. The inertancy obtained for different states of the destruction (with sensors 9/H and 11/H), left after collapse.
masonry during the dynamic test using the modal hammer.
937
5 THE USE OF A POLYMER JOINT FROM THE hermetic vertical membranes stopping the horizontal
CONSERVATION POINT OF VIEW evaporation of the moisture from inside of a structure
and causing the negative side-effect in the long-therm
5.1 Advantages of polymer joints opposite to period. In such situation the drainage of the barrier has
traditional methodologies to be done. On the other hand, the polyurethane repair
joint prevent water penetration through a cracked wall.
It is obvious that the possibilities and limitations of
These advantageous properties of polyurethane mass
each innovative technique should be briefly explained.
were exploited in renovation of joints in the stone
The Flexible Joint Method is the new approach in
Monument in Latvia (Sidraba 2002). Additionally,
retrofitting of masonries and it is necessary to make a
polyurethane durability and long-term experience was
lot of experiments in laboratory and in natural scale to
evaluated in the climatic conditions of the Scandi-
ensure that the new method could safety work in vari-
navia for 15 years, performing good properties in
ous cases of monuments repair. It needs proper fitting
comparison to traditional materials (Sidraba 2002).
of polymer properties and various kinds of primers
for cooperation with structural materials. The usage
of polymer joints in historical masonries have to be
weigh out individually and also estimated and justi- 6 CONCLUSIONS
fied by calculating their effect in terms of variations
in the global behavior of the structure. The Flexible Joint Method based on polymer flexible
The application of the innovative polymer material joints is proposed as the new approach in retrofitting of
can assure survival and safe exploitation of monu- cracked historical masonries. The presented properties
ments. Good dynamic properties of proposed polymer of the flexible polymer joints, examined in laboratories
and ability to dissipation of deformation energy make and on real cracked structures, showed their effective-
this material useful to use in damaged masonries in ness and advantages in the use of it as the repair method
seismic areas. It can assure also the safeguard of the of damaged masonries. The innovative polymer mate-
formal, material and historical integrity. It is possi- rial, bonding cracked historical structures, can assure
ble, because bonding of cracks with polymer allows survival and safe exploitation of monuments and also
us to avoid unnecessary intervention and to ensure the safeguard of the formal, material and historical
acceptable safety condition. The polymer can be sim- integrity. Authors hope that the use of flexible poly-
ply applied in cracks of width 3 ÷ 50 mm and then mer joints can be accepted by conservation authorities
covered with a material acceptable from “conserva- and can be widely used as the repair methodologies in
tive” point of view, limiting in this way variations of historical constructions.
external appearance of historical structures.
The proposed method is relatively cheap and is
not time consuming. Low values of stiffness mod- REFERENCES
ules assure safe co-operation of the polymer with weak
Bonaldo, E. et al. 2005. Bond characterization between con-
materials and better “compatibility” from mechanical crete substrate and repairing SFRC using pull-off testing.
point of view comparing to stiff epoxy resin or brittle Int. Journal of Adhesion & Adhesives 25 (2005). Elsevier.
cement mortars, generating high stress concentration. Castellano, M.G. & Infanti, S. 2004. Seismic protection of
monuments by shape memory alloy devices and shock
transmitters. 4th International Seminar on Structural
5.2 Critical discussion of using the polyurethane Analysis of Historical Constructions. Padova.
mass in repair of historic buildings Ciesielski, R. et al. 2004. Application of CFRP laminates as
The polyurethane mass proposed for application in strengthening if cracked brick arches. 4th International
Seminar on Structural Analysis of Historical Construc-
historical buildings as repair (bonding) mass should
tions. Padova.
be taken individually into consideration in each case. Dantu, P. 1957 I. – Étude des contraintes dans les milieux
Presented polymer can be subjected in certain range hétérogènes application au béton. Laboratoire Central des
to requirements proposed for repair mortars by Van Ponts et Chaussées. Publication No. 57–6. Paris.
Balen et al. (2005). Jankowski, R. 2003. Nonlinear rate dependent model of
It is obvious that polyurethane as the relatively high damping rubber bearing. Bulletin of Earthquake
new bonding material does not respect the traditional Engineering 1: 397–403. Kluwer Academic Publishers.
practice of the joining of historic materials and there- Kwiecień, A. et al. 2006a. Flexible Joint Method (FJM) –
fore the criticism could be provide in regard to its a new approach to protection and repair of cracked
masonry. 1st First European Conference on Earthquake
use. It is also known that polyurethanes have her-
Engineering and Seismology (1st ECEES). Geneva.
metic properties and are a barrier for vapor transport, Kwiecień, A. et al. 2006b. Repair of cracked historical
thus detrimental effect on the modification of hygric masonry structures by use of the Flexible Joint Method
properties of masonry can take place, if it is using (FJM) – laboratory tests. Structural Analysis of Historical
unconsciously. It is especially important in case of Constructions V. New Delhi.
938
Kwiecień, A. & Zajac ˛ B. 2008. Dynamic response of Skłodowski, M. Compact diagnostic test: Outline of histori-
the cracked masonry building repaired with the flex- cal monuments testing procedure. IFRT Reports 3/2006.
ible joint method – an innovative earthquake protec- Warsaw.
tion. 7th European Conference on Structural Dynamics. Sofronie, R.A. 2004. Two techniques for repair and strength-
Southampton. ening masonry constructions. 4th International Semi-
Modena, C. 2004. Design approaches of investigations for the nar on Structural Analysis of Historical Constructions.
safety and conservation of historic buildings. 4th Inter- Padova.
national Seminar on Structural Analysis of Historical Sofronie, R.A. et al. 2005. Application of reinforcing tech-
Constructions. Padova. niques with polymer grids for masonry buildings, CAS-
Modena, C. et al. 2006. Codes of practice for architectural CADE Report No. 5.
heritage in seismic zones. StructuralAnalysis of Historical Van Balen K. et al. 2005. Introduction to requirements for
Constructions V. New Delhi. and functions and properties of repair mortars. Materials
Oñate, E. et al. 1996. Structural analysis and durability and Structures 38 (October 2005) 781–785.
assessment of historical constructions using a finite ele- Viggiani, C. 2006. Safety and use vs. integrity of historical
ment damage model. Structural Analysis of Historical constructions: conflict or synergy? Structural Analysis of
Constructions. CIMNE. Barcelona. Historical Constructions V. New Delhi.
Ortega, L.M. et al. 1996. The spires of Burgos Cathedral. Vintzileou, E. 2006. Grouting of three-leaf stone masonry:
Structural Analysis of Historical Constructions. CIMNE. types of grouts, mechanical properties of masonry before
Barcelona. and after grouting. Structural Analysis of Historical Con-
Sidraba I. 2002. New materials for conservation of stone mon- structions V. New Delhi.
uments in Latvia. New materials for safeguarding cultural
heritage, ARIADNE 10 Workshop (www.itam.cas.cz/∼
arcchip/ariadne_10.shtml).
939
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
S. Invernizzi
Dipartimento di Ingegneria Strutturale e Geotecnica, Politecnico di Torino, Torino, Italy
J.M. Tulliani
Dipartimento di Scienza dei Materiali ed Ingegneria Chimica, Politecnico di Torino, Torino, Italy
ABSTRACT: The paper presents a preliminary experimental campaign, which aims at investigating the possible
application of carbon nanotubes to the mechanical improvement of timber structures. Different wood species are
considered, as well as different solvents or resins as a dispersing media for carbon nanotubes. The experimentation
is carried out on purpose exploiting exclusively the capillary properties of wood, in view of a possible on site
application of the technique. Although still at a preliminary stage, these first results appear to be promising, and
the research on the procedure is still in progress.
941
of carbon nanotubes in the matrix is complex given characteristic when dealing with large surfaces
the widespread specific surface of the nanoparticles treated with glue, and especially in the case of wood
(1000 m2 /g or more) which tend to favour the forma- since any accumulations of humidity must be easy
tion of agglomerates. Different techniques for dispers- to disperse in order to avoid biotic degradation.
ing said materials in solvents (acetone, ethanol. . .) in
addition to ultrasounds, mechanical shaking or a com-
bination of the two techniques, have already been tried
2 MATERIALS AND METHODS
out [1–7].
The interface adhesion can be improved by chem-
The first step of the experimentation was dedicated to
ically functionalising the nanotubes surfaces; this
assess the efficiency of different impregnation tech-
generates strong covalent-type of bonding [2, 3, 6,
niques.The carbon nanotubes were dispersed by means
7]. The bonding between nanotubes and polymers
of an ultrasonic probe for different time in ethanol and
allow the strain to be transferred from one stage to
acetone. Such solvents are supposed to act as a trans-
another.
port medium to bring the carbon nanotubes directly
Amino groups have been used for this purpose and
into the channels of the wood microstructure.
double wall carbon nanotubes or DWCNT with or
Specimens with section equal to 2.5 cm × 2.5 cm
without a functional group surface can be purchased
were sunk in the suspensions for different times, as
on the market (Nanocyl, Namur, Belgium) [8].
shown in Figure 1. We decided to exploit only the cap-
A picture of said experimental difficulties is pro-
illarity natural phenomenon instead of vacuum or other
vided by the results obtained with epoxy resins where
pressure assisted impregnation techniques.This choice
by adding SWCNT in the best cases the mechani-
was adopted in view of a future onsite application in
cal resistance to bending stress is slightly increased,
existing timber structures.
although in most cases, resistance is often reduced
The following wooden species have been selected
[5]. Recently, fabrics containing up to 39% in SWCNT
for the analysis: fir, Douglas pine, oak and larch. Some
weight impregnated with epoxy resin have been pro-
samples were obtained from new timber, other from
duced; once again the results obtained fell below
eighteenth century structures.
expectations [9].
Nanocyl multiwall carbon nanotubes were used for
On the other hand, significant progress in mechan-
the process. For economical reasons, the series 7000,
ical resistance values has been made with poly-
not functionalized (Table 1), was used to set up the pro-
methyl methacrylate (PMMA), polyvynilic alcohols
cess, while series 3101 functionalized with carboxyl
and with polystyrene-based composites [5]. Promis-
groups were used for the mechanical tests (Nanocyl®-
ing results have also been achieved with polyurethane
3101 series are purified to greater than 95% carbon and
resin composites with up to 10% in weight of
then functionalized with COOH groups).
nanotubes [10].
In a second step, an epoxy resin (MAPEI epojet,
Summarizing, carbon fibre nanotubes provide a
[11]) was used for the dispersion in order to get a prod-
number of advantages:
uct for mechanical improvement that could be applied
– they are morphologically and chemically compati- on the wood surface by painting, or act as a reinforced
ble both with polymer resin, used as bonding mate- glue to connect different timber part.
rial, and with wood, they are anatomically similar to Epojet is a two component solvent-free epoxy adhe-
strong piping bonded with a thermoplastic matrix sive. The pre-measured portions (Part A = resin and
and equipped with strong dissipative capacity with Part B = hardener) must be mixed together before
regards to fracturing energy. being used. Once mixed, Epojet becomes a liquid
– nanotubes allow the polymer bonding matrix to with low viscosity very suitable for injection. The
improve it own inbuilt deformation capacities mix ratio between part A and B is 4 to 1. This
considerably thanks to the transformation of a resin has been selected because of its low Brook-
homogeneous bulk with vitreous behaviour into field viscosity (respectively 500 and 320 mPa.s for
a micro-reticule with high level of porosity and part A and B).
deformability; The efficiency of the impregnation procedure has
– the mechanical characteristics of the resin-fibre been assessed with observation at the SEM micro-
compound are considerable on account of the high scope. The overall mechanical improvement of the
specific resistance of the fibres which ensure great timber specimen is going to be evaluated from compar-
cohesive strength combined with high ductility. ison of mechanical tests performed on un-reinforced,
The combination of the two produces significant impregnated and resin painted samples. The bending
creep resistance without the composite, loosing any strength has been evaluated by the three-points bend-
deformation capacity; ing test carried on with displacement control. The span
– the tubular structure of nanofibres has great per- of the samples was 370 mm and the load was applied
meability to vapour potential: this is an important with a velocity of 12 N/s.
942
Figure 2. TEM image of Nanocyl carbon nanotubes.
Method of
Property Unit Value measurement
943
Figure 4. Load displacement diagrams for four different
unreinforced timber species.
944
Figure 8. SEM image of the top larch sample section: no Figure 10. SEM image of CNT agglomeration in the epoxy
nanotubes appear to be penetrated the wood microstructure. resin (the resin was fractured to emphasize the defects
presence).
945
when only resin is used for the coating. If CNT rein-
forced resin is used for the coating, the gain in the peak
load raises up to 42%.
On the other hand, as far as the Hemlok fir samples
is concerned, an increase of about 45% is obtained in
case of resin coating, regardless the presence of a CNT
reinforcing in the resin. The influence of the CNT rein-
forcing appears even to be slightly disadvantageous,
but this could be also due to the fact that we still have
too few results to compare.
In general, the comparison must be continued
considering more samples and more timber species.
4 CONCLUSIONS
ACKNOWLEDGEMENTS
946
at low nanotube content, Composites Science and of carbon nanotube reinforced epoxy resin composites,
Technology, 64 (2004) 2363–2371 Polymer Testing, 26 (2007) 351–360
[3] K.T. Lau, D. Lui, Effectiveness of using carbon nan- [8] http://www.nanocyl.com
otubes as nano-reinforcements for advanced composite [9] Z. Wang, Z. Liang, B. Wang, C. Zhang, L. Kramer,
structures, Letters to the Editor/Carbon, 40 (2002) Processing and property investigation of single-walled
1597–1617 carbon nanotube (SWNT) buckypaper/epoxy resin
[4] H. Miyagawa, L.T. Drzal, Thermo-physical impact matrix nanocomposites, Composites: Part A, 35 (2004)
properties of epoxy nanocomposites reinforced by 1225–1232
single-wall carbon nanotubes, Polymer, 45 (2004) [10] T.J. Webster, M.C. Waid, J.L. McKenzie, R.L. Price,
5163–5170 J.U. Ejiofor, Nano-biotechnology: carbon nanofibres
[5] K. Lau, M. Lu, C. Lam, H. Cheung, F. Sheng, as improved neural and orthopaedic implants, Nan-
H. Li, Thermal and mechanical properties of single- otechnology, 15 (2004) 48–54
walled carbon nanotube bundle-reinforced epoxy com- [11] ttp://www.mapei.it/Referenze/Multimedia/367_Epojet
posites: The role of solvent for nanotube disper- _GB.pdf
sion, Composites Science and technology, 65 (2005) [12] C. Bertolini Cestari, R. Maspoli, J. M. Tulliani,
719–725 M. Di Sivo, R. Roccati, Use of nanotechnologies in
[6] H. Miyagawa, M.J. Rich, L.T. Drzal, Thermo-physical cultural heritage for the efficiency of maintenance
properties of epoxy nanocomposites reinforced by systems in wooden built heritage: innovative technolo-
carbon nanotubes and vapour grown carbon fibers, gies for restoration, in Nanofun-Poly. 2nd International
Thermochimica Acta, 442 (2006) 67–73 Symposium on Nanostructured and Functional-Based
[7] H. Chen, O. Jacobs, W. Wu, G. Rüdiger, B. Schädel, Material and Composites, Lyon, May 2006
Effect of dispersion method on tribological properties
947
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
A. Gubana
Department of Civil Engineering, University of Udine, Italy
ABSTRACT: Cross laminated timber panels are generally used to build walls and floor slabs of wood
houses, but they have interesting mechanical characteristics to be used for strengthening ancient wood floors
in restoration interventions. In particular they can be connected to the existing wood beams to have a floor
diaphragm effect under seismic actions. In the present paper the first results of a series of tests to determine
the panel shear behavior are presented. The experimental data show high shear resistance and huge potential for
their use.
949
Figure 1. View of the panel and principal production measures.
the air impermeability. The requested length of the Table 1. Mechanical wood characteristics.
panel is reached by finger joints between the boards.
Maximum production length can grow up to 8000 mm MPa MPa
(Figs 1–2).
Compression ⊥ to the grain σc,90 = 3.23 Ec,90 = 227
Also the board height and the number of layers Tension ⊥ to the grain σt,90 = 1.49 Et,90 = 386
can vary, so that different cross section heights can Compression // to the grain σc,0 = 34.1 Ec,0 = 12148
be available.
The transversal dimension, for production technical
reasons, can reach about 1300 mm. The tests were conducted following the indications
The panels are made of spruce wood. Several tests of EN 408, EN1194 and EN 789 for compression and
were performed, at reference moisture content, to tension perpendicular to the grain and compression
check the principal mechanical characteristics. The parallel to the grain respectively.
timber is classified as GL24C, after EN 1194. Specific media data are reported in Table 1.
950
Figure 3. Overview of the shear-in plane test rig.
2.2 Description of the rig A particular device was designed to optimize the
test preparation and make easy the collocation of the
Under seismic actions the floor diaphragm is subjected
panels inside the rig.
to cyclic in plane shear tension. One of the principal
Due to the high forces necessary to reach the panel
aim of the rig design was the possibility of stressing
collapse, the length of the panels had to be reduced.
the panels with constant shear tension on the lateral
So the panels were sawn up to have a central square of
sections.
about 420–430 mm side.
This kind of rig was properly designed for shear tests
Inside of this panel portion, electronic acquisition
(Giuriani & Plizzari, 2000). It is made of an horizontal
measurement devices were placed along the diagonals
steel frame, with two lateral beams and two transversal
of an inner square of 250 mm side length, over the two
beams. A central beam is free to move longitudinally
faces of each panel (Figs 4–5).
in the centre of the frame. During each tests two panels
are stressed at the same time as described in Figure 3.
The panels lateral sections are constrained to the 2.3 Description of the tests
lateral and the central longitudinal beams. The jackets
The tests were performed using hydraulic jackets with
are placed at the end of the central movable beam,
force control.
where there is a proper contrast. The force due to
Ten couples of panels were stressed to collapse. The
pressure of the jacket against the contrast and the
first seven tests were monotonic, two tests were cyclic
transversal beam determines an axial tension action
and the last one was performed with reversed cycles
in the central beam and axial compression action in
of force. During all the tests only one of the two panels
the two lateral beams. Instability problems had to be
reached the collapse.
faced due to the high level of axial compression.
The moisture content was checked for every panel,
Two hydraulic jackets of 500 kN were used to
and the values were always around 12–13%.
perform the tests.
The transmission of the shear forces to the panels
is reached by a distributed series of dowel connectors.
3 EXPERIMENTAL DATA
This solution is similar to what can be actually built in
a building. The problem was particularly delicate, as it
3.1 Force to displacement diagrams
was necessary to avoid collapse at the interface panel-
to-rig, which was easy to happen because of high local The data direct acquired from the tests let to
pressure in timber near the dowels. draw force-to-displacement diagrams. Each test was
951
Also in this case the collapse was less fragile with
respect to timber structure general answer.
it is possible to obtain:
952
Figure 6. Force-to-displacement diagrams for Panels B under monotonic load.
953
Table 2. Maximum shear stress and elasticity modulus. The values obtained are more variable, but this is
to be expected, as the global deformation of the panel
Maximum shear Elasticity is influenced by the local deformation at the parallel
Panel stress MPa Modulus MPa shank interface, which is not glued. Also for this rea-
son the values obtained can be significantly less with
A 3.29 419
respect to gross wood section (the GL24C mean shear
B 3.22 331
C 3.14 327 modulus is 590 MPa after EN:1194:1999). In any case
D 3.29 346 they are much better with respect to a traditional wood
E 3.33 352 floor with no strengthening on the top.
F 3.15 327 In Figure 10 some views of the collapsed sections
G 3.06 332 are given, showing cracks inside of the shanks.
H 3.22 384
I 3.08 524
L 3.24 506
4 CONCLUSIONS
ACKNOWLEDGEMENTS
REFERENCES
Figure 10. View of the collapsed section of one panel (a) Ceccotti, A., Lauriola, M.P., Pinna, M. & Sandhaas,
and zoom view on the cracks (b). C. 2006. SOFIE Project – Cyclic Tests on Cross –
Laminated Wooden Panels. WCTE 2006 – 9th Inter-
national Conference on Timber Engineering, Portland,
The shear elasticity modulus were calculated taking USA.
into account the data related to 10% and 40% of the EN 408:2004. Timber structures. Structural timber and glued
shear failure strength, as indicated in most of the codes laminated timber – Determination of some physical and
(i.e. EN:789:2005). mechanical properties.
954
EN 789:2005. Timber structures. Test methods. Determina- Guggenberger, W. & Moosbrugger, T. 2006. Mechanics of
tion of mechanical properties of wood based panels. Cross Laminated Timber Plates under Uniaxial Bending.
EN 1194:1999. Timber structures. Glued laminated timber. WCTE 2006 – 9th International Conference on Timber
Strength classes and determination of some mechanical Engineering, Portland, USA.
and physical properties. Modena, C., Tempesta, P. & Tempesta F. 1998. Il recupero
Gelfi, P., Giuriani, E. & Marini, A. 2002. Stud Shear a secco di impalcati in legno. Una tecnica per solai a
Connection Design for Composite Concrete Slab and semplice e doppia orditura. L’edilizia 3/4.
Wood Beams. Journal of Structural Engineering 128(12): Moosbrugger, T., Guggenberger, W. & Bogensperger, T.
1544–1550. 2006. Cross Laminated Timber Wall Segments under
Giuriani, E. & Plizzari G. 2000. Studio sperimentale sul homogeneous Shear – with and without openings. WCTE
comportamento dei solai in legno rinforzati con lastre 2006 – 9th International Conference on Timber Engineer-
in acciaio per resistere alle azioni sismiche. V workshop ing, Portland, USA.
Italiano sulle strutture composte, Palermo. Piazza, M. & Turrini, G. 1983. Una tecnica di recupero dei
Gubana, A. 2003. An Analytical Approach to Design non solai in legno. Recuperare 5,6,7.
linear Behaviour Connections of Composite Sections.
International Journal for Restoration of Buildings and
Monuments 9(3).
955
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: This paper deals with the investigation of the mechanical behavior of lime mortar enriched with
natural pozzolana, widely used in historical masonry. The study aims to investigate the mechanical behavior of
natural pozzolana lime mortars, reproduced in laboratory by using different mix designs, as reported in historical
Italian architectural treatises, and natural pozzolana coming from Viterbo (Italy). To this purpose, the works of
the most important authors from the Ist century B.C. up to 1864 have been analyzed, in order to collect historical
mix designs of pozzolana lime mortars. Flexural and compression tests were performed following EN 1015-11
(2007). Mechanical tests were performed after 28 days and 60 days of curing. Penetration tests, not included in
European Standards but very suitable for practical applications, was also performed.
1 AIMS AND RESEARCH METHODOLOGY industrialization, still inspiring, although new materi-
als and systems were developed, the basic concept of
Historical evidence on the use of mortars to meet contemporary concrete.
several needs in architecture has existed for millen- In the past, it was empirically clear that pozzolana
nia. A systematic analysis of ancient mortars dating provides hydraulicity to lime mortar, enabling its set-
some hundred or thousand years proved that they have ting in wet conditions or underwater. Furthermore, it
resisted decay by time and other deterioration factors. was known that the addition of pozzolana was use-
Difference concerning their durability and the degree ful to increase the durability of lime mortar, allowing
of their degradation are attributed to many techno- it to overcome the need of repeated maintenance and
logical aspects that characterize each specific period, accelerating the carbonation process, as it has been
such as the binding methods used, the type and propor- also scientifically proved (Collepardi, 1993). Up to
tion of aggregates, application system as well as the now, however, such favorable properties of natural poz-
craftsmen skills. zolana have been subdued to relatively little scientific
Most of the mortars dated from the historic period investigation carried out by means of mechanical tests.
are based on the use of lime and many of them Due to the widespread use of pozzolana lime mortars in
are enriched with natural siliceous volcanic sand, historic masonry, wider scientific and technical issues
commonly called pozzolana. Referring to the Mediter- about mechanical properties of this kind of mortars
ranean building culture and practice, it is known that, would be very useful in repairing and maintenance of
since 1500 B.C., Greeks used a highly siliceous vol- historic architectural heritage.
canic Santorini earth in wall coverings (Moropoulou, As the aim of this study was to analyze the mechan-
2004). Furthermore, it is widely accepted that the ical behavior of lime mortar enriched with natural
first extensive use of pozzolana was adopted by the pozzolana, a new scientific investigation methodol-
Romans who, since the II century B.C. (at least) used ogy had to be adopted. It should be observed that it
a volcanic sand in the opus caementicium (i.e. a lime is almost impossible to extract from historical build-
mortar mixed with an aggregate of small stones or ings mortar samples without altering the mechanical
rubble), whenever possible. This material, often cov- properties of the material; moreover, the required
ered by dressed stones or bricks, represented, from the dimensions of the samples would make this kind of
foundations to the vaulting, the very core of Roman test too invasive. As a consequence, specific sam-
masonry (Lugli, 1957; Lancaster, 2005, also for fur- ples suitable for mechanical tests had to be prepared.
ther references). Pozzolana remained at the base of In this respect, the correct reproduction of the his-
building practice during the modern ages up to the toric mix designs, with their different compositions,
957
was a problem of critical importance. On this subject, authors seem to have focused on pozzolana after Vit-
actually, in absence of convincing primary archival ruvio. Seneca mentioned pozzolana (62–65 A. D.),
sources, there is no general assessment and several spending on it only few words (Seneca, 2004, III,
attempts on the characterization of historical mor- XX, 3: 285). Plinius, instead, in his Naturalis Historia
tars were recently made without historical evidence. (within 78 A. D.), wrote about pozzolana underling its
It was therefore decided to prepare the different mix ability of hardening underwater (Plinio, 1987, XXXV,
designs following the indications traced in the archi- XLVII: 1259); he also seems to be the first author
tectural treaties, printed documentary sources that can who hinted at pozzolana capacity of increasing mortar
be considered as the official expression of the building strength with the age.
culture, generally stating what, in the past, the common The worth of Vitruvius’ work increases if we con-
sense regarded to as “the rule of the art”. Although the sider that in his treat the first written indications
craftsmen skills and the circumstances often adapted for obtaining pozzolana lime mortars can be traced,
those guidelines to the case at hand, the mix designs asserting that the ideal proportion (pozzolana:lime)
described in the treaties can be taken as a reasonably was 2:1 (as volumetric ratio) “Hae autem structurae
reliable reference to reproduce historic mortars. quae in aqua sunt futurae, viderunt sic esse facien-
The works of the most important Italian authors dae uti portetur pulvis a regionibus quae sunt a Cumis
from Vitruvius (I cent. B.C.) to Giovanni Curioni continuatae ad promonturium Minervae, isque mis-
(1864) (and also some French authors whose treaties ceatur uti in mortario duo ad unum respondeant”
gained great success in the Italian architectural culture) (Vitruvio, 1997, V, II–III: 586–588). Later, many
were therefore analyzed in order to collect pozzolana authors referred to pozzolana in their architectural
lime mortars historical mix designs. Following the treaties as well, but they didn’t give practical and tech-
ratios suggested, samples were reproduced in labo- nical information on the use of this volcanic material.
ratory by using natural components only, and adding In particular, from XV century up to the end of XVIII
Pozzolana originated from the Roman volcanic district century, architects substantially referred to Vitruvius’
of the Monti Sabatini. treat adding only accessory comments. Among more
The different lime mortars were tested according than fifty treaties analyzed, there were traced only 24
to European Standard EN 1015-11, after 28 days and mix designs pozzolana lime mortars.
60 days of curing. A penetration test, not included During the XV and the XVI centuries, the only Ital-
in European Standards but very suitable for practical ian author that considered the pozzolana for practical
applications, was adopted as well. purposes was Francesco di Giorgio Martini (1480–
82), who quoted Vitruvius’ pozzolana lime mortar mix
design (Martini, 1967, I, VIII: 105) but misunderstand-
2 HISTORIC PERSPECTIVE ing his indications and inverting the pozzolana:lime
volumetric ratio (Table 1).
Much of what is known about Roman architectural cul- No pozzolana lime mortars mix designs can be
ture derives from Vitruvius, who in his De Architectura traced in the most important Italian Renaissance archi-
(I century B.C.) discussed about building knowledge tectural treaties. Alberti (1450) widely discussed about
gathering together specific information, construction lime mortar and limestones, but only hinted at poz-
guidelines and design principles probably widespread zolana (Alberti, 1989, II, XII: 86). Filarete (1460–64)
in Rome at the birth of the Empire. In the II book hinted to pozzolana too, but sending back to Vitru-
of this treaty it can be found the first written refer- vio for further information (Filerete, 1972, III, p. 67).
ence to pozzolana mortars: “Est etiam genus pulveris, Serlio (1537) spent a lot of pages of his work describ-
quod efficit naturaliter res admirandas. Nascitur in ing important roman buildings, but didn’t write about
regionibus Baianis in agris municipiorum, quae sunt materials and didn’t mention the use of pozzolana.
circa Vesuvium montem. Quod commixtum cum calce Vasari (1550), though not directly referring to vol-
et caemento non modo ceteris aedificiis praestat fir- canic material, praised the building good manufacture
mitates, sed etiam moles cum struuntur in mari, sub in the area around Pozzuoli (Vasari, 1986, XXVII: 74).
aqua solidescunt.” (Vitruvio, 1997, II, VI: 132). Vitru- Palladio (1570) hinted at pozzolana (Palladio, 1980,
vius referred to the pulvis puteolana, a sand originated I, IV, p. 15), focusing his attention on the hydraulic
from thick deposits of reactive volcanic tuffs present lime used in the Venetian region around the town
around Puteoli, the vesuvian town near Naples (today of Padova (Palladio, 1980, II, IX, p. 77) and indi-
Pozzuoli) from which the name of this material is cating, as the first among the Renaissance authors
derived (Lugli, 1957; Adam, 1989; Lancaster, 2005). examined, the quick hardening as a peculiar quality
This kind of sand, currently used in building Roman of pozzolana mortar. Cataneo (1567) was the first
practice, was also easily found in the surroundings who tried to explain the hardening process of a poz-
of Rome, which is situated between two well known zolana lime mortar (Cataneo, 1985, II, IV: 269–270)
volcanic districts. Only two other important Roman and Scamozzi (1615) dedicated an entire chapter of
958
Table 1. Pozzolana lime mortar mix designs traced in architectural treaties.
Volcanic
Reference Period Pozzolana Lime Sand Granite Pumice Breccia
his work describing pozzolana, but without giving any of the interest of the architectural culture and practice
practical indication of its use (Scamozzi, 1982, II, VII, toward it.
XXII: 236–237).
During the end of the XVIII and the first half of the
3 MATERIALS AND EXPERIMENTAL
XIX century, up to the Sixties, architects and engineers
PROGRAM
turned to focus their attention on the practical uses of
pozzolana. Some authors, like Milizia (1781) (Milizia,
3.1 Geological and chemical aspects of pozzolana
1991, III, I, IV: 37–38), Rondelet (1802) (Rondelet,
used in experiments
1834, I, I, I, III, II:125; III: 147–151; II, II: 249–250)
and Valadier (1831) (Valadier, 1992, I, II, XI: 58–59) To evaluate the natural pozzolana contribution to the
discussed about the geological and the chemical char- increase of the strength in a lime mortar, specific sam-
acteristic of the volcanic material, its geologic origin ples of mortar were prepared in laboratory adding the
and the deposits different location. On the other hand, Pozzolana rossa originated from the volcanic complex
Viviani (1830) (Viviani, 1830, VI- II: 125), Valadier of the Monti Sabatini (Rome, Italy). As it well known,
(1831), Quatremere De Quincy (1832) (Quatremere Rome is situated between two volcanic districts, the
de Quincy, 1842, II: 99–100), Cavalieri San Bertolo Monti Sabatini and the Albano Hills (Colli Albani),
(1832) (Cavalieri San Bertolo, 1845, II, IV: 41), De which have been active for most of the last millions
Cesare (1855) (De Cesare, 1855, I, V: 49), Claudel- years (Karner et al., 2001b).
Laroque (1863) (in Curioni, 1864, 1, VII: 133) and Tephra deposits from the Monti Sabatini crop out
Curioni (1864) (Curioni, 1864, 1, VIII: 134) proposed extensively to west of the Tiber River and to the north
different mix designs (Table 1). of Aniene River. Three major pyroclastic flow deposits
The growing concern with faster settings and more from the Monti Sabatini crop out extensively along the
durable mortars soon led to the development and to Tiber River Valley. They are, from the oldest to the
the spread of modern cements. Since the late XIX youngest, the Tufo Giallo della Via Tiberina, the Tufo
century, the increasing use of Portland cement dis- Rosso a Scorie Nere, and the Tufo Giallo di Sacrofano.
placed the pozzolana lime mortar and led to the end A series of ash fall (i.e. pozzolana, which was used in
959
the tests), surge and pyroclastic flow deposits separate The test specimens were prisms 160 × 40 × 40 mm;
these eruptive units (Karner et al., 2001a). three specimens were provided for each age of test.
The chemical characterization of the pozzolana Preparation and condition of storing specimens were
used for laboratory tests is shown in Table 2. chosen in compliance with EN 1015-11.
The specimens were cast in metal moulds and
eventually placed in polyethylene bags for 2 days;
3.2 Mix chosen and test procedure afterwards, the moulds were removed and the spec-
The 24 mortar compositions traced in the architectural imens remained in the bags for the following 5
treaties analyzed (I cent. B. C. – 1864) were repro- days (storage temperature of 20◦ C ± 2◦ C; relative-
duced and tested in a first testing phase. By the light humidity 95% ± 5%). After these 7 days of curing,
of this first experimental step, a second, more punc- the samples were placed in a humidity room for
tual testing phase followed, choosing only five mix 21 days (room temperature = 20◦ C ± 2◦ C; relative
designs (Table 3) (due to synthesis requirements, only humidity = 65% ± 5%).
the second testing phase results are reported and dis- Flexural strength of mortars was determined by
cussed herein, while whole set of results can be found three-point-bending tests on the prism specimens. The
in Sala, 2008). two halves derived from the flexural strength tests pro-
In reproducing the mortar specimens, both aerial vided, for each age of test, six half prisms to be used
hydrated lime and hydraulic hydrated lime were used. for the compressive strength tests.
The new European standard for limes (EN Moreover, other tests were carried out on dif-
459 2001) defines three types of limes: natural ferent specimens, named “sandwiches” (two bricks
hydraulic lime (NHL), natural hydraulic lime with and a mortar joint of 20 mm), in order to show a
added material (NHL-Z), and artificial hydraulic relationship between the compressive strength and
lime (Z). In the present study, the specimens were a probe penetration rate. The method is based on
prepared with natural moderately hydraulic lime the measurement of the penetration depth increment
NHL 3.5 (resistance after 28days: 3.5–10 MPa; of a steel probe by means of repeated blows of a
UNI EN 459-1:2001). Schmidt rebound hammer (Felicetti & Gattesco, 1998;
No indications about water quantities are given Giuriani & Gubana, 1993).
in the mortar mix designs reported in the architec- Specimens were tested after 28 days and 60 days of
tural treaties (Table 1); therefore, the percentage of curing.
water added in the mortar specimens to be tested was
determined in relation to the workability of the mixes.
Tests were carried out according to European Stan-
4 EXPERIMENTAL RESULTS
dard EN 1015-11 procedure in terms of preparation
and conditioning of mortar specimens to determine
4.1 Flexural and compressive strength
the flexural and compressive strength of mortars.
Test results, at the ages of 28 days and 60 days, are
shown in Table 4.
Table 2. Chemical composition (%) of the pozzolana used
for laboratory tests.
4.2 Penetration tests
SiO2 Al2 O3 FeO2 TiO2 CaO MgO K2 O Na2 O p.f.
Penetration tests results were obtained after 28 and
60 days of curing (Figure 1); results are reported in
48.7 17.1 7.7 0.8 9.6 4.2 8.7 2 0.1
Table 5.
Reference Period Identification Pozzolana A.H. Lime H.H. Lime Sand Water
960
Table 4. Evolution of average flexural and compressive mix design (V2), with 2:1 pozzolana:lime volumetric
strength. ratio, shows the best results.
Vitruvius’ mix and, in particular, V2 specimen
Avg flexural Avg compressive shows the highest strength increment with time; in fact,
strength [MPa] strength [MPa]
the compressive strength of 7.16 MPa after 28 days of
Reference 28d 60d Incr. 28d 60d Incr. curing, reaches 9.49 MPa after 60 days.
Tests on Ca2 (7:3 and no sand) provide lower values
V1 1.39 1.51 8.57% 4.47 5.38 20.5% of flexural and compressive strength in comparison
V2 2.12 2.66 25.4% 7.16 9.49 32.5% with V2 specimens, therefore 2:1 seems to be the best
Viv1 2.04 2.09 2.30% 6.33 7.96 25.7% volumetric ratio for mortar composed by pozzolana
Viv2 1.38 1.20 −13.0% 4.26 4.76 11.9% and hydraulic hydrated lime.
Q 1.63 1.77 8.61% 5.46 6.78 24.2% Using aerial hydrated lime, results are little differ-
Ca1 1.47 1.52 3.72% 4.85 5.21 7.43% ent. Q mix (3:1 and no sand) shows the best mechanical
Ca2 1.59 2.03 28.1% 5.83 6.73 15.5% performance. The use of a greater pozzolana percent-
Cu 0.81 1.03 26.9% 4.07 4.22 3.79%
age in the mix is evidently due to the hydrated lime
incapacity to increase strength with time. The strength
increment is therefore ensured only by pozzolana
addition.
Viv1 and Viv2 specimens have the same poz-
zolana:lime volumetric ratio (2:1) of specimens V1
and V2, but they also have a percentage of sand. The
flexural and compressive strength results lower than
Vitruvius’ one.
During the preparation of the mortar specimens it
was observed that the medium value of the volumetric
percentage of water added in the mortars, for obtain-
ing the same workability, is 20% and that the water
increases for higher quantities of pozzolana. On the
other hand, the use of sand in the mix doesn’t seem to
Figure 1. Penetration test.
influence the water quantity.
As mentioned above, in order to verify a non-
Table 5. Evolution of average penetration tests results. destructive test method for practical uses, some pen-
etration tests were performed. Experimental results,
Number of hits/cm
plotted in Figure 3 for 28 and 60 days of curing,
Reference 28d 60d show a good direct correlation between compressive
strength and penetration test results. The trend line,
V1 5.44 6.56 obtained from the test results after 28 days of curing
V2 8.00 11.11 (Figure 3), shows a good correlation between the two
Viv1 7.67 11.33 quantities (K2 = 0.8251). This evidence demonstrates
Viv2 5.56 8.56 the possibility of using penetration tests to obtain cor-
Q 6.78 10.11 rect information about historical mortars mechanical
Ca1 5.33 8.67
characteristics with a non-destructive test.
Ca2 5.89 10.78
Cu 4.00 6.00 The second diagram in Figure 3 shows the trend
line between compressive resistances and penetration
tests after 60 days of curing. The average deviation
(K2 = 0.7066) is lower than the previous one; how-
ever, the direct correlation between the two quantities
5 DISCUSSION
is clearly evident.
The diagram in Figure 1 shows the influence of aging
on the compressive strength of different pozzolana
mortars. It can be observed that, in general, mortars 6 CONCLUSIONS
with hydraulic hydrated lime (V2, Viv2, Ca2) show
higher mechanical behavior and reach higher com- The research deals with the study of historical lime
pressive resistance values in comparison with the same mortars enriched with natural pozzolana. Historic
mortars with aerial hydrated lime (V1, Viv1, Ca1). mix designs were traced in the architectural treatises,
Referring to results from mechanical tests on mortar trough the analysis of more then fifty authors from
with pozzolana and hydraulic hydrated lime,Vitruvius’ Vitruvius (I cent. B. C.) to Curioni (1864).
961
Figure 2. Influence of aging on compressive strength.
Figure 3. Trend lines. Average Compressive Strength – Penetration tests results (28 days and 60 days).
From an historical point of view, the study of – Historical analysis of architectural treaties shows a
mortars manufacture provides useful information on continuous reference to Vitruvius’ indications. Up
the building culture and the use of building materi- to the end of the XVIII century, official architectural
als, eventually helping in indicating different build- culture seemed to be concerned with the good
ing phases and in setting conservative interventions. properties of the pozzolana and of volcanic materi-
From a technological point of view, mortar identi- als but it didn’t propose new pozzolana lime mortars
fication is a fundamental step in the design of the compositions. Only at the beginning of the XIX
repair materials, taking into account the necessary century, many authors turned to examine the prac-
compatibility between the original materials and the tical uses of pozzolana, deepening its technical
new ones. properties and giving new mix designs.
An exhaustive sampling of natural pozzolana lime – An exhaustive number of mortars mix designs were
mortars was carefully accomplished in order to start selected and reproduced in laboratory in compli-
off a first study on their mechanical properties which, ance with UNI EN 1015-11. They were character-
up to now, were subdued to relatively little scien- ized in relation to their mechanical properties. The
tific investigation. Flexural and compressive tests were tests clearly show that the compressive mechanical
carried out, after 28 and 60 days of curing. strength increases with the addition of pozzolana.
The main results of this study can be summarized – The mortars with hydraulic hydrated lime, proposed
in the following. by Vitruvio (V2; 2:1 pozzolana:lime volumetric
962
ratio), showed the best mechanical properties; good EN 459-1, 2002, Building lime. Definitions, specifications
performance was also shown by Quatremere de and conformity criteria
Quincy’s mix design (Q) (3:1), using aerial hydrated Felicetti R. & Gattesco N., 1998, A penetration test to study
lime. the mechanical response of mortar in ancient masonry
buildings, Materials and Structure, 31: 350–356
– In general, mortars obtained with hydraulic EN 1015-11, 2006, Methods of test for mortar for masonry.
hydrated lime show a better mechanical behavior Part 11: Determination of flexural and compressive
than those obtained with aerial hydrated lime. strength of hardened mortar
– Some further experiments were also performed in Giuriani E. & Gubana A., 1993, A penetration test to eval-
order to verify the reliability of the penetration test, uated wood decay and its application to the Loggia
a simple and non-destructive test that is signifi- monument, Materials and Structures, 26: 8–14
cant for practical use (mortar prisms are not easily Karner D.B. & Marra F., 2001a, Florindo F. & Boschi E.,
obtainable from historic buildings). The compari- Pilsed uplift estimated from terrace elevations in the coast
son between the compressive strength determined of Rome: evidence for a new phase of volcanic activity?,
Earth and Planetary Science Letters, 188: 135–148
from penetration tests and the one determined from Karner D.B., Marra F. & Renne P.R., 2001b, The history of the
compressive tests confirmed that there is a good Monti Sabatini and Alban Hills volcanoes: groundwork
correlation. Therefore, the penetration test, can be for assessing volcanic-tectonic hazards for Rome, Jour-
successfully adopted for studying the mechanical nal of Volcanology and Geothermal Research, 107, 1–3:
response of mortar joints in historical masonry and, 185–215
moreover, it can be easily used for in-situ tests. Lancaster L.C., 2005, Concrete Vaulted Construction in
Imperial Rome, New York: Cambridge University Press
Lugli G., 1988, La tecnica costruttiva romana con particolare
riguardo a Roma e Lazio, Roma: Giovanni Bardi
ACKNOWLEDGEMENTS Martini F. di Giorgio, 1967,Trattati di architettura ingegneria
e arte militare, Milano: Il Polifilo
Grateful acknowledgements go to Professor Alberto Milizia F., 1991, Principj di architettura civile,, Roma: Sapere
Clerici, University of Brescia, for the helpful discus- 2000
sions on the geological subjects; to Eng. Nicola Bettini Moropoulou A., Cakmak A., Labropoulos K.C., Van
for his contribution in editing the paper. Grieken R. & Torfs K., 2004, Accelerated microstructural
evolution of a calcium-silicate-hydrate (C-S-H) phase
The Authors would like to thank the companies
in pozzolanica pastes using fine siliceus sources: com-
T.C.S. srl (Brescia, Italy) and Cava Lapillo e Poz- parison with historic pozzolanic mortar, Cement and
zolana di Arnaldo Gentile & C. snc (Viterbo, Italy) Concrete, 34, 1: 1–6
for providing the hydraulic hydrated lime NHL 3.5 Palladio A., 1980, I quattro libri dell’architettura,, Milano:
and Pozzolana Rossa, respectively. Il Polifilo
Finally, a special acknowledgement goes to Plinio il Vecchio, 1987, Plinii Naturalis Historia, Pisa:
Mr. Domenico Caravaggi, of the Laboratory Pisa of Giardini
the University of Brescia, for his co-operation in Quatremere de Quincy A.C., 1842, Dizionario storico di
performing the experiments. architettura, Mantova: F.lli Negretti
Rondelet G., 1834, Trattato teorico e pratico dell’arte di
edificare, Mantova: Caranenti
REFERENCES Sala E., 2008, Duttilità e confinamento delle murature
storiche e caratterizzazione meccanica di malte poz-
Adam, 1989, L’arte di costruire presso i Romani: materiali e zolaniche riportate nella trattatistica d’architettura (I sec.
tecniche, Milano: Longanesi a.C.-1864), Ph.D Thesis, University of Trento, Italy
Alberti L.B., 1989, L’architettura, Milano: Il Polifilo Scamozzi V., 1982, L’idea dell’architettura universale,
Averlino A. detto il Filarete, 1972, Trattato di architettura, Bologna: Arnoldo Forni
Milano: Il Polifilo Seneca, 2004, Questioni naturali, Milano: BUR
Cataneo P. & Barozzi G. da Vignola, 1985, Trattati, Milano: Serlio S., 2001, L’architettura, Milano: Il Polifilo
Il Polifilo Valadier G., 1992, L’architettura pratica, Roma: Sapere 2000
Cavalieri San Bertolo N., 1845, Istruzioni di architettura, Vasari G., 1986, Le vite de’più eccellenti architetti, pittori, et
statica e idraulica, Mantova: Negretti scultori italiani, da Cimabue insino a’tempi nostri,Torino:
Claudel J. & Laroque L., 1850, Pratique de l’arte de Einaudi
construire, Parigi : Carilian-Goeuryet Vitruvio M.P., 1997, L’architettura di Vitruvio, Gros P. (ed.),
Collepardi M., 1993, La produzione del calcestruzzo antico Torino: Einaudi
e moderno, Proc. Convegno antico e moderno di Bres- Viviani Q., 1830, L’architettura diVitruvio tradotta in italiano
sanone, Bressanone, 6–9 July 1993: 181–192 da Quirico Viviani, Udine: Mattiuzzi
Curioni G., 1864, L’arte di fabbricare, ossia corso completo
di istituzioni teorico-pratiche, Torino: Negro
De Cesare F., 1855, La scienza dell’architettura, Napoli:
Giovanni Pellizone
EN 196-1, 1996, Methods for testing cement. Determination
of strength
963
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
J. Elsen
Research Group of Applied Geology and Mineralogy, Department of Geography–Geology, K.U.Leuven,
Heverlee, Belgium
ABSTRACT: Blended lime-cement mortars are commonly used in conservation practices even though they
may show lack of adequate strength and durability for certain cement-lime compositions. This paper focuses
on understanding the hardening reactions and their influence on the strength development, microstructure and
porosity for the cement-lime mortars in various compositions. Mortars composed of 30%, 50% and 70% cement
replacement with lime hydrate and lime putty by mass were studied. Cement hydration has been found to con-
tribute to the early stage strength development while carbonation is mostly favoured after 3 days and contributes
to the strength development until 180 days. The degree of carbonation is much more pronounced with increasing
lime content and porosity of the mortars and the reaction is still in progress at 90 days. All the blended mortars
revealed lower compressive and flexural strength than that of the reference cement mortar due to their lower
cement content and higher porosity. Long-term compressive strength development has been achieved after 180
days while flexural strength for certain mortar compositions does not increase beyond 28 days. Unlike cement
mortar, the mortars blended with lime hydrate and lime putty exhibit an elastic-plastic deformation before failure
occurs, which is preferred for repair mortars to adapt to differential settlements and to allow more deformation
under critical stresses in the masonry.
965
2006, Arandigoyen & Alvarez 2007). Apart from Table 1. Composition of the reference cement mortar, lime
cement hydration, carbonation reaction contributes to hydrate mortar and cement-lime hydrate mortars.
the strength development of the cement-lime mortars,
which is generally neglected. Therefore, in this study Cement Lime Water
special emphasis is given on understanding hardening
Mortar (g) (g) (g) (w/b)
reactions and their influence on the strength develop-
ment, porosity and microstructure of the cement-lime Cref 450 – 202.5 0.45
mortars using lime hydrate and lime putty as a partial L – 450 392 0.87
replacement of cement in varying ratios. C7L3 315 135 234 0.52
C5L5 225 225 270 0.60
C3L7 135 315 315 0.70
2 RESEARCH PROGRAMME
2.1 Materials Table 2. Composition of the lime putty mortar and cement-
lime putty mortars.
Portland cement (CEM I 52.5 N), commercial lime
hydrate and lime putty were used as binders for mor- Lime Water Total
tars. The lime hydrate is composed of 83.22% calcium Cement putty added water
hydroxide content with a Blaine specific surface area
of 19,464 cm2 /g and a mean value of its particle size Mortar (g) (g) (g) (g) w/b
distribution of 4.40 µm. The lime putty was produced
at the laboratory through slaking quicklime (0–50 mm) LP – 1125 – 675 1.50
in 1:3 quicklime-water ratio. The putty was then stored C7LP3 315 342 94.5 301.5 0.67
in hermetically sealed containers to promote its aging C5LP5 225 569.7 51.3 396 0.88
under water. The lime putty used in the mortars was C3LP7 135 797.5 – 482.5 1.07
aged for one year and contained 60% water by mass.
It is composed of 83.47% calcium hydroxide with a
Blaine specific surface area of 21,739 cm2 /g. laboratory conditions (20◦ C, 60% R.H.) for 180 days.
The progress of hardening was studied by means of
mechanical strength test and thermal analysis. The
2.2 Mortar compositions mechanical strength tests were carried out by com-
Cement was partially replaced by lime hydrate and pressive strength and three-point bending using the
lime putty in 30%, 50% and 70% by mass of the total standard mortar prisms at 3, 7, 28, 60, 90 and 180
binder (Table 1 and Table 2). Reference cement (Cref ), days of hardening (EN 196-1). For thermal analysis the
lime hydrate (L) and lime putty (LP) mortars were mortar specimens were first gently ground in order to
studied as well for the comparison of the data. The separate sand grains from the binder fraction and then
mortars were prepared using standard quartz sand with vacuum dried at 0.025 mbar. After that, the sample
maximum grain size of 2 mm (EN 196-1). Binder/sand was sieved through 112 µm sieve to obtain the binder
ratio of the mortars is 1:3 by mass. Due to higher fraction which was ground further for thermal anal-
specific surface area of calcium hydroxide particles ysis. This analysis was carried out using a Netzsch
than cement, more water was required in cement-lime STA 409 PC DSC-TGA system in static nitrogen atmo-
hydrate and cement-lime putty mortars to reach similar sphere at a temperature range between 20-1000◦ C with
workability (125–130 mm) with the reference cement a controlled heating rate 10◦ C/min. Porosity of the
mortar, which was measured by the flow table test (EN mortars at 90 days has been studied with total porosity
1015-3). The water content increased with increasing (EN 1936) and pore size distribution using Mercury
lime content in the mortar composition. Considering Intrusion Porosimetry (Micromeritics AutoPore IV
the cement-lime putty mortars, additional water was 9500. Microstructure of the mortars was studied using
required only for the C7LP3 and C5LP5 mortars as the Philips XL 30S FEG Scanning Electron Microscope
water content available in the lime putty used was ade- (SEM) coupled with X-Ray Energy Dispersive Sys-
quate to gain the required workability for the C3LP7 tem. The analysis was carried out on the fresh fractured
mortar. surfaces coated with gold after the samples were dried
at 45◦ C.
966
Table 3. Bulk density and total porosity of the mortars at
90 days.
967
Figure 4. Compressive strength development of the
mortars.
968
Figure 7. SEM micrograph of the hydrated matrix of the Figure 9. SEM micrograph of the plate-like crystals, prob-
reference cement mortar at 90 days, indicating reticular amor- ably AFm crystals, observed in the hydrated matrix of the
phous and needle-like C-S-H phase where calcium hydroxide cement-lime mortars.
crystals are locally embedded.
969
lower strength values than those of the cement-lime
hydrate mortars. Considering C3LP7 mortar having
70% cement replacement with lime putty, increase in
the volume of pores is much more pronounced than
the rest of the mortars especially for the pore diame-
ters between 10 µm and 0.6 µm as well as those less
than 0.3 µm.
30% replacement of cement with lime hydrate
has resulted in 38.4% reduction in the compressive
strength of the reference cement mortar while it is
56.3% reduction in case of replacement with lime putty
at 180 days. The reduction in the flexural strength is
23.6% for cement-lime hydrate mortar and 46.8% for
the cement-lime putty mortar. Further replacement of
the cement at 70% has led to extremely lower compres-
sive and flexural strength values. This is much more
Figure 11. A detailed SEM micrograph showing the for- pronounced for C3LP7 mortar as its flexural strength
mation of amorphous C-S-H on the edges of the calcium
hydroxide crystals.
value (0.45 N/mm2 ) is even lower than that of the LP
mortar (0.8 N/mm2 ). Therefore, the mortar composi-
tion with 30% cement and 70% lime putty does not
improve the strength of the mortar in comparison with
the LP mortar. This is not the case for C3L7 mortar
as it has higher compressive and flexural strength val-
ues than those of the L mortar and LP mortar but quite
lower than those of C5L5 and C7L3. However, its com-
pressive strength does not increase beyond 60 days as
well as its flexural strength beyond 28 days.
Thermal analysis results indicated that hardening
reactions responsible for the strength development of
the blended mortars are cement hydration and car-
bonation reaction (Figure 3). The latter is expected
due to the presence of additional calcium hydroxide
in the mortar compositions from the beginning. This
is clearly seen in the DTG thermographs with the
increase in the intensity of the calcium carbonate peak
Figure 12. SEM micrograph of the hydrated matrix of the at temperatures between 600◦ C and 800◦ C. Degree
C3L7 cement-lime hydrate mortar at 90 days. of carbonation increases in time and with increasing
calcium hydroxide content in the mortar composi-
flexural strength values for the blended mortars in tion as well as increasing porosity which favours the
comparison with the strength of the reference cement diffusion of the carbon dioxide into the core where
mortar (Figure 4 and Figure 5). Strength values calcium hydroxide is still present to carbonate (Van
become lower respectively with 30%, 50% and 70% Balen & Van Gemert 1994). Cement hydration is
replacement of cement with lime hydrate and lime mostly complete within 28 days as the intensity of
putty by mass. This is related with both decreasing the peaks of the hydrated phases does not increase
cement content and increasing water/binder ratio in beyond that period. These phases are C-S-H and AFm
the composition of these blended mortars, which has which are the same with those of the reference cement
resulted in an increase in their porosity. With decreas- mortar. The presence of these phases has been veri-
ing cement content and increasing calcium hydroxide fied by SEM analysis. Microstructure of the reference
content, an increase in the volume of pores with diam- cement mortars is composed of reticular amorphous
eters less than 1 µm is observed (Figure 1). C3L7 and needle-like C-S-H phase where calcium hydrox-
and C3LP7 mortars, which have 70% replacement ide crystals are locally embedded (Figure 7). Hydrated
of cement with lime hydrate and lime putty, revealed microstructure of the blended mortars is character-
total porosity values that are very close to the poros- ized by clusters of calcium hydroxide crystals that
ity values of the L and LP mortars respectively. The are surrounded by amorphous C-S-H phase (Figure 8)
cement-lime putty mortars revealed higher porosity and locally formed AFm crystals (Figure 9). These
than that of the cement-lime hydrate mortars for the calcium hydroxide crystals indicated corroded edges
same amount of replacement. This can explain their where amorphous C-S-H phase formed subsequently
970
(Figure 10 and Figure 11). With increasing lime con- Cement hydration contributes to the early stage
tent in the mortar composition, the microstructure strength development while carbonation is mostly
becomes more porous with less amorphous C-S-H favoured after 3 days and proceeds gradually until 180
phase and more calcium hydroxide crystals as well as days. Degree of the carbonation is much more pro-
hexagonal plates and needles (Figure 12). These plate- nounced with increasing lime content and porosity of
like and needle-like crystals are assumed to belong to the mortars. The cement-lime mortars have revealed
the AFm phase due to the peak observed at around lower compressive and flexural strength than the ref-
160◦ C in the DTG thermographs (Figure 3). erence cement mortar due to lower cement content
In the literature, hexagonal hydro-aluminate plates and higher porosity of the blended mortars. Long-term
were reported to be formed during the hydration of compressive strength development has been achieved
the tri-calcium aluminate in the presence of lime after 180 days while flexural strength for certain mor-
(Collepardi et al. 1978, Taylor 1990). Their forma- tar compositions does not increase beyond 28 days.
tion in our mortar compositions is not clear from Further replacement of the cement at 70% with lime
the DTG thermographs as the hexagonal hydro- putty does not improve the strength of the mortar
aluminates reveal dehydration peaks between 100◦ C in comparison with the LP mortar while this is not
and 200◦ C, which is the same dehydration range for the the case when lime hydrate is used. Unlike the refer-
C-S-H and AFm phases, and therefore their presence ence cement mortar, the cement-lime mortars exhibit
can be masked. an elastic-plastic deformation which enables them to
With decreasing cement content in the mortars adapt to differential settlements and more deforma-
composition less hydrated phases and more calcium tion under critical stresses in the masonry. Considering
carbonate phase are observed (Figure 3). At 3 days the mortar compositions and hardening reactions,
of hardening, hydrated phases can already be seen in compatibility of the lime-cement mortars should
the thermographs but the peak of the calcium car- be further evaluated in terms of physical, mechan-
bonate evolves in time. This shows that the cement ical and durability requirements for conservation
hydration takes place at the very early stage while the purposes.
carbonation reaction becomes effective after 3 days
and continues until 90 days. Therefore, the cement
hydration contributes to the early stage strength devel- ACKNOWLEDGEMENT
opment until 28 days. In the meantime, carbonation
reaction continues and contributes to the strength This study is part of an ongoing research project
development until 90 days. This is quite clear from the (OT/04/27 3E030765) conducted at the Department
compressive strength development of the mortars but of Civil Engineering at the K.U. Leuven. The research
the flexural strength of all cement-lime hydrate mor- grant offered to Ozlem Cizer by the university is grate-
tars and only C7LP3 mortar does not increase beyond fully acknowledged. Department of Metallurgy and
28 days. Materials Engineering (MTM) at the K.U.Leuven is
In comparison to the reference cement mortar, stiff- gratefully acknowledged for the use of the Scanning
ness decreases with increasing lime content in the com- Electron Microscope.
position of the blended mortars, which enhances their
physical and mechanical compatibility with ancient
mortars. The cement-lime mortars indicated a rea- REFERENCES
sonable plastic behaviour and more deformation than
the reference cement mortar. Similar to the self- Arandigoyen, M. & Alvarez, J.I. 2007. Pore structure and
healing ability of lime mortars, cracks occurring due mechanical properties of cement–lime mortars. Cement
and Concrete Research 37 (5): 767–775.
to the deformation can be healed by either dissolution-
Collepardi, M., Baldini, G., Pauri, M. & Corradi, M.
precipitation cycles of calcite or carbonation reaction 1978. Tricalcium aluminate hydration in the presence of
of the calcium hydroxide that is still available in lime, gypsum or sodium sulphate. Cement and Concrete
the matrix. However, the influence of formation of Research 8: 571–580.
calcium carbonate crystals on the initially hydrated EN 196-1. 1987. Methods of testing cement – Part 1:
phases and its consequence on the strength develop- Determination of strength. European Standard.
ment and durability of the blended mortars should be EN 1015-3. 1998. Methods of test for mortar for masonry –
considered well. Part 3: Determination of consistence of fresh mortar (by
flow table). European Standard.
EN 1936. 1999. Natural stone test method – Determination
of real density and apparent density and of total porosity
5 CONCLUSIONS and open porosity. European Standard.
Kuzel, H. & Pöllmann, H. 1991. Hydration of C3A in the
Hardening of the cement-lime mortars occurs as a presence of Ca(OH)2 , CaSO4 · 2H2 0 and CaCO3 . Cement
result of combined cement hydration and carbonation. and Concrete Research 21: 885–895.
971
Mosquera, M.J., Silva, B., Prieto, B. & Ruiz-Herrera, E. Van Balen, K. &Van Gemert, D. 1994. Modeling Lime Mortar
2006. Addition of cement to lime-based mortars: Effect on Carbonation Materials and Structures 27: 393–98.
pore structure and vapor transport. Cement and Concrete Van Balen, K., Hayen, R. & Van Gemert, D. 2005. Weaker can
Research 36(9): 1635–1642. be better: Learning from the past contributes to sustainable
Scrivener, K.L. & Capmas, A. 1998. Calcium Aluminate construction technology with lime. International Building
Cements. In P.C. Hewlett (ed.), Lea’s Chemistry of Cement Lime Symposium, Proc. intern. symp., Orlando, Florida,
and Concrete: 718–724. 4th Edn. London: Arnold. 9–11 March. NLA Building Lime Group.
Sebaibi, Y., Dheilly, R.M., Beaudoin, B. & Quéneudec, M. Van Hees, R.P.J., Binda, L., Papayianni, I. & Toumbakari, E.
2006. The effect of various slaked limes on the microstruc- 2006. Damage analysis as a step towards compatible repair
ture of a lime–cement–sand mortar, Cement and Concrete mortars. In C. Groot, G. Ashall & J. Hughes (eds.), Char-
Research 36(5): 971–978. acterization of old materials with respect to their repair –
Taylor, H.F.W. 1990. Cement Chemistry. London: Academic RILEM TC 167-COM: 107–152. RILEM publications
Press Limited. S.A.R.L.
972
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: Assessing the workability of mortars for restoration and more general purposes has been the
subject of many standards and scientific publications over the last decades. A series of mortars made with
different binders was brought to optimum water content for workability by an international group of masons.
Two methods are proposed to assess the static yield stress and water retention of these fresh mortars, two key
parameters of workability as defined by practitioners.Yield stress is measured with a vane test, originally designed
for clayey soil, and found to be related to specific surface area of the binder. The increase of yield stress with
resting time is quantified and the differences due to the effect of air entraining agent are demonstrated. Water
retention is characterised by the desorptivity of the mortars and measured in a vacuum suction experiment. The
setup is a modified version of an existing ASTM design. Both methods are complementary to assess the defined
workability parameters for different mortars and eventually can be an aid to design suitable mortars for specific
purposes.
973
driving mechanism, so that possible effects on yield Table 1. Description of binders.
stress are not taken into account here. The specific
apparatus used in this study was originally designed No. Description
for lab measurements of clayey soil and has recently
been proposed for use on binder pastes and mortars 1 calcic lime CL90S
2 calcic lime CL90S (EN 459-1:2001) (75%) with
(Hendrickx et al. 2006, Bauer et al. 2007).
hydraulic binder (15%) and pozzolana (10%)
The transport of water through a volume of semi- 3 calcic lime CL90S (EN 459-1:2001) with air
solid mortar slurry or paste can be described using entraining agent
the same principles as those applied for porous solids. 4 calcic lime CL90S (EN 459-1:2001) (66.7%) and
A complete modelling of flow from mortar to substrate ordinary Portland cement CEM I 42.5 R (33.3%)
using FEM could be done when the complete water 5 natural hydraulic lime NHL 5 (EN 459-1:2001)
retention curve and hygroscopic curve of both materi- 6 ordinary Portland cement CEM I 42.5 R
als are known, as well as the initial state and boundary 7 ordinary Portland cement CEM I 42.5 R with
conditions. At present, established experimental meth- admixtures: plasticizer and air entraining agent
ods do not allow the mortar to be characterised to this
extent. Alternatively, a simple sharp front model to
describe the dewatering of a slurry based on filter Table 2. Chemical properties (XRF) and specific surface
area (BET) of binders.
cake theory, is derived in (Hall and Hoff 2002) and
experimentally validated in (Collier et al. 2007). This 1 2 3 4 5 6
model presents the desorptivity R of the wet mortar at
constant capillary suction of the substrate as the main CaO (%) 96.6 75.1 96.3 75.6 51.2 63
parameter to characterise the mortar. R is determined MgO (%) 0.94 1.88 1.05 1.49 1.68 1.74
experimentally for different types of mortar using a Al2 O3 (%) 0.059 1.640 0.072 2.6 4.9 4.6
simple pressure filter used in oil industry (Carter et al. SiO2 (%) 0.144 6.2 0.19 9.6 16.6 16.7
2003). In analogy to the sorptivity of bricks, which has Fe2 O3 (%) 0.048 1.55 0.063 2.7 2.1 3.6
been accepted to be a reliable measure for its capillary CO2 (%) 2.0 10.6 2.0 3.0 19.8 2.9
absorption, desorptivity of mortar provides also more SO3 (%) 0.10 1.84 0.18 2.80 2.10 4.50
useful information than the well-known standardised SSA (m2 /g) 12.0 5.7 11.9 4.5 3.5 1.3
tests, which measure a quantity of water absorbed by
filter paper sheets (EN 413-2:2005) or filter paper
plates (EN 459-2:2001). The apparatus described in 100
60
50
40
2 MATERIALS AND METHODS
30
20
2.1 Mortar composition
10
The description of the binders are listed in Table 1, with 0
a standardised denomination is possible. The chemi- 0.01 0.1 1
Sieve aperture (mm)
cal properties measured by X-ray fluorescence (XRF)
and specific surface area (SSA) measured with the Figure 1. Grading of the aggregate (Zutendaal sand).
BET method are given in Table 2. Binder 2 and 3
are prefabricated admixtured products; binder 7 is a Table 3. Mortar composition in weight ratios.
lab mix of the ordinary Portland cement (binder 6),
plasticizer (modified stearic acid in powder) and an Binder: 1 2 3 4 5 6 7
air entraining agent (fatty acid/polyglycol solution).
The dosage was 0.4% and 0.1% of the binder weight. B:A (kg/kg) 0.13 0.20 0.13 0.21 0.21 0.25 0.25
The aggregate is a medium siliceous quarry sand 0/0.5 W:B (kg/kg) 2.02 0.90 1.48 1.00 1.13 0.93 0.79
(0/1) (EN 13139:2002) from a quarry in Zutendaal,
Belgium (Figure 1).
The compositions of the batches are given in Table 3 prefabricated blended binders (binder 2 and 3), they
by their binder to aggregate weight ratio (B:A) and are in agreement with the producer’s guidelines. W:B
water to binder weight ratio (W:B). B:A values are values have been determined for optimum workability
derived from practical experience and are comparable by a panel of six masons (Hendrickx et al. in press).
to values given in various national standards. For the Mixing is done using a Hobart planetary mixer at low
974
80
φ1 (°)
φ2 (°)
60
calibrated φ1 + φ2 (°)
torque spring
Angle (˚)
40
20
vane
0
0 1 2 3 4 5 6 7 8
Time (min)
φ1
inner scale Figure 3. Typical course of an experiment at constant
ω = 9.2◦ /min (mortar with binder 7).
knob with
carrier The assumption that shear occurs exactly in this place
ωt is a good approximation (Lidell & Boger 1996). The
homogeneity of the material close to this shear plane
is discussed further. A typical test shows three stages
φ2 (Figure 3). In the initial stage there is an increase of
outer scale φ2 due to a small repositioning of the vane and rear-
rangement of grains in the mortar. In the second stage,
Figure 2. Working principles of the soil vane test. the increase of both φ1 and φ2 are constant. In the
third stage, the spring angle reaches a maximum and
speed: 1.5 min mixing, 30 s homogenising, 2 min mix- the residual angle increases more, until the material
ing. All the components are added in the beginning: yields: this last phenomenon can not be derived from
first the water, then half of the aggregate, then the the scale reading, due to the configuration appara-
binder and at last the remaining half of the aggregate. tus. All mortars were measured with a spring constant
B = 380◦ /Nm, except binder 4, for which a vane with
B = 260◦ /Nm was used. The vane has dimensions
2.2 Yield stress experiments D = H = 45 mm. Rotation speed ω was constant and
equal to 9.2◦ /min.
Yield stress was measured with a vane apparatus for
clayey soil (ASTM D4648) (Figure 2). It measures
the maximum torque necessary for a vane (DxH), 2.3 Desorptivity experiments
immersed in a sample container with mortar, when The test setup form ASTM C91 was adapted to contain
rotation at constant is initiated. Previous research has a larger sample volume. The sample is introduced in
demonstrated that container dimensions of three times a perforated dish on top of a wetted filter paper. The
the dimensions of the vane, are sufficient to avoid any sample height is increased with respect to the standard-
influence of edge effects (Lidell & Boger 1996). The ised version from 20 to 60 mm (Figure 4). This enables
mechanism is designed in a way that the inner scale to have sufficient data points when suction is applied
indicates φ1 , the amount of deflection over the spring for identifying the initial linear regime of water loss
and thus the magnitude of the torque, and the inner with the square root of time (Hall and Hoff 2002):
scale indicates φ2 , the angle over which the vane has
rotated inside the sample. From the moment that the
mortar yields and the vane starts to move quickly, φ1
remains constant and φ2 starts to increase at a rate ωt. with i the amount of evacuated water, expressed in mm
It can be seen that φ1 = ωt − φ2 . The torque exerted water column, R the desorptivity and t time. The pres-
on the vane can be expressed as: T = B · φ1 with sure difference p1–p2 is generated by a vacuum pump,
B a constant of the calibrated spring, indicating the in stead of the pressure cell used in (Carter et al. 2003).
torque per unit of rotation. Hence the shear stress in The mass of the dish with the filter paper and sample is
a cylindrical plane circumscribing the vane can be measured at the beginning of the test, after wiping the
calculated as T /K, with the vane constant K equal to: bottom of the dish quickly with a dry cloth. The vac-
uum pressure is applied to the sample via the funnel
by turning a valve. The seal between the funnel and the
dish is made with a layer of silicone, which is applied to
975
p 150
binder 3 unsheared
1
binder 3 pre-sheared
binder 7b unsheared
4 cm
mortar sample binder 7b pre-sheared
100
2
50
0
0 20 40 60 80 100
Resting time (min)
976
105 10
100 9
95 8
Yield stress, t=30 min (Pa)
90 7
85 6
i (mm)
Binder 1
80 5
Binder 2
75 4 Binder 3
3 Binder 4
70
Binder 5
65 2 Binder 6
1 Binder 7
60
0 2 4 6 8 10 12 14
0
SSA (m²/g) 0 20 40 60 80 100
t0.5 (s0.5)
Figure 7. Yield stress measured with vane at 30 min of
resting time as a function of SSA of binder powder. Figure 8. Quantity of desorbed water as a function of time
for mortars with different binders.
(relatively low) rotation rate in this setup and that
3.5
strictly spoken, this method is neither rate controlled,
nor exactly stress controlled. The results are to be inter- 3
preted as a good basis for comparison: changing the
2.5
test parameters will influence the order of magnitude
of the absolute numbers. 2
y' (mm)
977
Table 5. Numerical results of standardised water retention tests, desorptivity measurements and
regression analysis.
Binder: 1 2 3 4 5 6 7
10 100%
8 90%
6 80%
i (mm)
4 70%
3 200 m bar
300 m bar
2 60%
400 m bar
1 500 m bar
0 50%
0 20 40 60 80 100 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Figure 10. Quantity of desorbed water as a function of time Figure 12. Water retention measured with filter paper plates
for different vacuum pressures and mortar with binder 1. as a function of calculated desorptivity.
0.1 4 CONCLUSIONS
978
The same advantages are valid for the desorptiv- and SSA on both yield stress and desorptivity, two of
ity experiment with the vacuum pump: it starts from the most important aspects in assessing the workability
a small adaptation of a standard setup which is avail- of mortar.
able in a large number of building material labs. The
actual process of suction by a porous brick resembles
more to this setup, than to the technique with a pres-
sure cell used by other researchers. The disadvantage REFERENCES
is the rather laborious test procedure and the restric- Banfill P. 2005. The rheology of fresh mortar – a review,
tion of applied suction to atmospheric pressure. The In Cincotto, M.A., Silva, D.A., Oliveira, J., Roman, H.R
result is more meaningful than the results of the tradi- (ed.), Proceedings of the 6th Brazilian and 1st Interna-
tional standardised tests with filter paper of filter paper tional Symposium on Mortar Technology, Florianopolis,
plates, because it yields a rate of water loss in stead Brazil: 73–82.
of one single point. Further more the data can be used Bauer E., De Sousa J.G.G., Guimarães E.A., and Silva
to estimate the equilibrium value at the applied vac- F.G.S. 2007. Study of the laboratory vane test on mortars,
uum pressure, which allows to characterize the water Building and environment, 42: 86–92.
Coussot P. 2005. Rheometry of pastes, suspensions and
retention curve of the mortar in the covered pressure
granular materials, Wiley, Hoboken (New Jersey).
range. The weakness remains that for high desorptiv- Green K.M., Carter M.A., Hoff W.D., Wilson M.A. 1999. The
ity mortars, a significant inclination of the curve is effects of lime and admixtures on the water-retaining prop-
observed already after the second point, which causes erties of cement mortars, Cement and concrete research
the calculated desorptivity value to be less accurate, 29: 1746–1747.
deviating towards the lower end. The measured des- Hendrickx R., Van Balen K., Van Gemert D., Minet J. 2006.
orptivity value is found to be inversely proportional Workability and rheology of lime-based mortars, 15. Kol-
to the water retention value measured by filter paper loquium und Workshop über Rheologische Messungen an
plates. This means that this simple test is a suitable mineralischen Baustoffen, Regensburg, 8–9 March 2006.
(unpublished presentation)
method to estimate desorptivity.
Hendrickx R., Minet J., Van Balen K., Van Gemert D. in press.
The observed differences between the mortars can Workability of mortars with building lime: assessment
be related to the value of the SSA and to the effect of air by a panel of masons versus lab testing, 14th Interna-
entraining agents in the mortar. SSA is inversely pro- tional brick and block masonry conference, Sydney, 16–20
portional to desorptivity, and addition of air entraining February 2008. (accepted for publishing on CD-ROM)
agent has a decreasing effect, though of less impor- Lidell, P.V. & Boger D.V. 1996. Yield stress measurements
tance. There is evidence for a negative correlation with the vane, Journal of non-newtonian fluid mechanics,
between mortar desorptivity and yield stress, through 63: 235–261.
their dependence on the SSA. A higher SSA gives Van Balen K., Papayianni I., Van Hees R., Binda L.,
Waldum A. 2005. Introduction to requirements for and
lower desorptivity and higher yield stress for mortars
functions and properties of repair mortars, Materials and
without air entraining agent. Caution is needed how- structures 38, 8: 781–785.
ever to interpret the results because of the different Wallevik J.E.H. 2003. Rheology of Particle Suspensions.
B:A values of the mortars tested. A similar trend was Fresh Concrete, Mortar and Cement Paste with Various
found for the mortars with air entraining agent. These Types of Lignosulfonates, PhD dissertation, Norwegian
findings illustrate the important effect of air content University of Science and Technology.
979
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
A. Giannantoni
Servizi di Ingegneria srl, S. Eraclio di Foligno (PG), Italy
ABSTRACT: A new technique for reinforcing rubble stone masonry walls (double and triple-leaf walls), when
it is required to keep the fair-face masonry. The reinforcement technique consists of a continuous mesh made
of tiny steel cords embedded perfectly in the mortar joints after a first repointing, and anchored to the wall by
means of galvanized steel eyebolts driven into the facing. A second repointing covers the cords and the heads of
the eyebolts completely. This leads to genuine reinforced fair-face masonry in which, as already confirmed by
the first experiments, the compression, shear and flexural strength are increased, effective transverse connection
between the facings of the masonry due to the presence of the eyebolts and also the capacity to withstand tensile
stresses. The reinforcement is non-invasive and reversible, and is aimed at integrating the masonry rather than
transforming it. It is compatible with preservation of the material of which the artefact is made and is long-lasting
in view of the materials used, which are very resistant to aggression. The analysis of this reinforcement/upgrading
work has led to the formulation of practical criteria for sizing the reinforcement and to the assessment of the
strength of the reinforced panel.
981
case of walls with a low index of voids (which The system proposed can be used both at local
is a fairly frequent occurrence). Also, it is not level, that is to say for single walls of existing build-
easy to investigate the actual spreading of the ings (and also for boundary walls such as, by way of
mortar injected into the panel. example, town walls), and at global level, that is to
C2) ferro-cement, consisting of making two thin say as a system for reinforcing a masonry construc-
reinforced concrete walls connected to one tion in its overall behaviour, with particular but not
another by means of metal connectors. This sole reference to behaviour during seismic events.
technique is often used to reinforce irregular The system can be used to treat masonry, whether
or poor-quality masonrywork and is undoubt- regular or irregular, without causing the impact and
edly very effective from the mechanical point without the invasiveness of other techniques. It is
of view, however it is very invasive. The inter- therefore particularly suitable for fair-face walls of
nal masonry is “lost” from every point of view, buildings listed due to their historical and architectural
not only because it is no longer visible but features.
above all since it is subject in time to rapid The improvements that can be achieved do not
deterioration. Another negative aspect is that consist simply of improvement of the mechanical char-
the new wall panel is far stiffer than the orig- acteristics of the wall treated in this way. Indeed, in
inal one (and this often has a negative impact addition to strengthening the wall panel, the “rein-
on the structural behaviour of the building). forced skeleton” of the continuous mesh introduced
C3) Another technique used is that of deep into the wall (and which has been given the name
repointing of the mortar joints. This consists of of “Reticolatus”) also connects the contiguous wall
stripping the joints in the masonry by removing elements to one another (adjacent orthogonal walls,
the original poor-quality mortar for a depth of horizontal elements, foundations, etc), thus forming
several centimetres (typically 6 to 8 centime- a genuine system for reinforcing the whole masonry
tres) and then repointing the joints with a good building.
quality mortar. If this rienforcement is carried The small size of the reinforcing elements and the
out on both sides of the wall facing and the ease with which they can be fitted into the mortar joints
facing is not very thick, it has a good degree also enables widespread use, thus avoiding harmful
of effectiveness. and dangerous concentrations of stress, such as those
If the masonry is made of bricks, it is that occur, for example, when metal rods are used.
also possible to insert small metal bars into
the joints (reinforced repointing). These can 2 DESCRIPTION OF THE SYSTEM
increase the strength of the panel considerably.
It is clear, however, that even minor irregular- To sum up, the system consists of a continuous mesh
ities in the fabric of the wall will lead to the made of tiny cords made of high strength steel, inserted
need to cut the bricks, which is out of keep- perfectly into the mortar joints and thus embedded
ing with the principles of preservation of old in the wall in a natural way (that is to say without
buildings. intervening on the stone parts).
In the case of irregular masonry (stones of This reinforcing technique can be used equally on
different shapes and sizes giving rise to irreg- regularly shaped walls (made of bricks or worked
ularly shaped joints) reinforced repointing is stone blocks) or irregular (rubble and barely cut stone)
not suitable, since even if the diameter of the walls (multi leaf walls). It is in any case in the latter
reinforcement rods is small, the rods cannot case that it is often more necessary to improve the
follow the irregular shapes of the joints. mechanical properties of the wall, making this tech-
C4) Another recently suggested technique con- nique more interesting.This is because the possibilities
sists basically of a system of tie-rods leading provided by other current techniques (repointing with-
out in the three orthogonal directions (CAM out reinforcements or grout injection into walls that
system – Dolce et al.). The tie-rods can also are unsuitable since they are very compact) may not
be tensioned. This technique is definitely very be very effective or may be invasive and therefore
effective in mechanical terms and improves unacceptable from the point of view of preservation
the monolithic quality and the mechanical (ferro-cement).
behaviour of the body of the wall, however The system is based on the use of materials easy to
it cannot often be proposed for old buildings, find on the market even if commonly used for other
since it has a strong impact on preservation of purposes, that is to say:
the old material.
a) High strength steel cords, which can be made from
The main aim of the technique suggested in this coils to be found on the market. The coils are about
paper is to eliminate or at least reduce the problems 30 cm wide (Fig. 1) and variable in length from
referred to above. 15 m up to 1500 m, and consist of a series of cords
982
Figure 4. Close-up of hooking system of threaded bars with
washers and nuts.
983
• insert the eyebolts or the rods, typically 4 per square
metre, arranged as regularly as possible according
to a square mesh and driving them in far enough to
involve the opposite facing, leaving the head below
the surface on which the work is being carried out
(at a depth of at least 3–4 cm);
• insert the cords into the stripped joints, passing them
trough the hooks of the eyebolts or behind the wash-
ers of the rods, proceeding in a horizontal or vertical
direction all over the facing being treated. If the
length of the single cords is insufficient, they can
be joined together with resin or, simply, overlapped
against each other for a length of about 20 cm;
• if considered necessary to further increase the
strength of the panel it is possible to apply addi-
tional cords diagonally, both in one direction and in
the other;
• repoint the joints again with mortar, so as to cover
up completely both the cords and the heads of the Figure 6. The two reinforced masonry facings enclose the
eyebolts or of the rods. inner masonry work.
Upon completing the operations described above,
a grid-like pattern (irregularly shaped) made of metal
or of composite material will have been obtained. It
will be perfectly incorporated into the wall facing but
not visible from outside, and capable of giving the
panel the mechanical characteristics (shear strength,
tensile strength and compression strength) it needs to
withstand both static and dynamic stresses.
To understand the effectiveness of this treatment, it
is possible to compare it with the ferro-cement tech-
nique. The end result of this technique is to confine
the masonry between two new thin reinforced concrete
walls (the two layers of plaster consisting of cement
mortar, reinforced with metal netting, on the two sides
of the masonry and connected to one another). With the
technique suggested here, the result is absolutely com-
parable in terms of effectiveness but the original facing
remains visible and perfectly intact, and its ability to Figure 7. Close-up of a corner area.
transpire remains unaffected. Indeed, here again there
are two new resisting walls (several centimetres thick)
connected to one another that enclose the masonry also be folded over to take in the panel at right-angles
between them (Fig. 6), but these are actually the origi- (Figs 7–8), thus connecting the vertical elements at
nal external facings which have become genuine layers right-angles to one another in a particularly effective
of reinforced masonry (reinforced wall facings) thanks manner (since they create an extensive connection).
to the grid of metal cords inserted “discreetly” but As far as concerns the bottom part of the panel (e.g.
intimately into the joints. in the foundations), the cords can be anchored effec-
Moreover, in addition to considerably increasing the tively by folding them around the pieces of stone at
tensile (and shear) strength of the masonry, this grid the bottom of the panel (Figs 7–8) or connecting them
also has the beneficial effect of confining the pieces to the foundations (if any) by means of connectors, as
of stone, that are enclosed within the circuits of the the case may be.
cords. At the top, the cords can be connected to the metal
To reinforce the whole masonry construction using or RC tie (if any), or, for example, in boundary town
the proposed system, it is possible to proceed as walls, folded over the coping so that they reach the
follows. other side of the wall.
At the ends of the wall panel, the cords can be folded As a result of these operations, each single wall was
in so that they also take in the opposite surface side of reinforced locally, and all the different elements were
the same panel. In the case of a corner panel, they can connected to one another, giving rise to a construction
984
– intended to integrate the structure without trans-
forming it;
– non-invasive;
– reversible (or at least removable);
– compliant with the principle of the “least upgrading
work”.
It is easy to identify the possible advantages from
the mechanical point of view. They concern:
1) improved mechanical characteristics, that is to say
compressive strength and shear strength, and also
flexural strength in relation to stresses on the same
and at right-angles to the plane of the masonry;
2) the ability to connect extensively any damages, the
vertical walls to one another and the vertical walls
to the horizontal elements;
3) giving the masonry the tensile strength that, in the
Figure 8. Folded and anchored cords. case of irregular masonry (where the vertical joints
are often aligned) cannot even benefit from the
in which the resisting system now consisted of a “chain effect” present in regular walls with properly
masonry reinforced to a widespread extent. staggered orthostats and joints;
To sum up, the system presented here calls for a net- 4) transverse connections between the facings of the
ting structure with flexible sides consisting of the cords masonry, since the eyebolts or the transverse rods
mentioned above, which can be positioned accord- are arranged as artificial stones capable of mak-
ing to paths corresponding to the joints between the ing the wall panel act as a monolith. They do this
elements of the wall and which can be anchored to by contrasting detachment of the facings from one
pre-established points of the latter by means of metal another and providing adequate tensile strength
eyebolts or transverse rods. Since the sides of the net- (also necessary in the presence of vertical loads
ting are flexible and pass through the hooks or the rings only and shear strength (necessary in the event of
of the transverse elements without being an actual part stresses that tend to make the wall panel tilt over,
of them, it is possible to arrange them precisely along and therefore to make the facings slide in respect
the joints (previously stripped) of the elements forming of one another).
the masonry, however they are arranged.
4 EARLY EXPERIMENTS
3 FIELDS OF APPLICATION AND In order to investigate the effectiveness of the rein-
EXPECTED BENEFITS forcement technique described above, a series of tests
were planned.
As already mentioned, the proposed technique is suit- After a description of the materials used, the first
able for treating rubble stone walls when, for reasons results of the tests are reported on below.
of preservation, the fair-face wall is required to be
retained and at the same time non-invasive reversible
4.1 Fibre characterization
and effective reinforcement is required. Typically,
therefore, some of the structural problems of old build- The fibres used during the experiment were supplied
ings and archaeological assets can be solved in a by Hardware LLC. One peculiarity of these materials
manner capable of meeting the need for the high- is their macroscopic structure. Indeed, all the fibres
est safety standards combined with the requirements are made up of high-resistance steel filaments covered
relating to protection and preservation. with a layer of brass to prevent oxidation of the metallic
The proposed technique complies with the princi- fibres. The specifications of the cords used are shown
ples underlying the protection f historical buildings, in Table 1.
since it is:
4.2 Tests with flat jacks
– “compatible” with preservation of the material of
which the building is made, as it is able to adapt In the tests with double jacks, portions of walls about
and integrate perfectly into the walls; 50 cm high were subjected to compression on a single
– “long-lasting”, since the materials used have a high vertical axis.
resistance to chemical and physical aggression and Masonry walls were tested while subjected to com-
to weathering; pression using two flat jacks (Fig. 9). During the test
985
Table 1. 3X2 cord mechanical properties. As far as concerned the normal modulus of elastic-
ity E1/3 , the results of the tests with flat jacks showed
Fibre type 3X2 cord that deep repointing of the joints on its own is not capa-
Cord diameter (mm) 0.89 ble of causing significant increases of the modulus of
Cross section area (mm2 ) 0.620 elasticity. This was probably because the mortar used
Failure tensile load (N) 1539
Young’s modulus (E) (MPa) 206842
for repointing the joints was based on hydraulic lime.
Failure stress (MPa) 2479 As far as concerns the manner in which the masonry
Elongation at failure (%) 2.1 elements between the two flat jacks failed, it was seen
that a series of vertical cracks formed between the
two flat jacks. Furthermore, there was no substiantial
14 differentiation of the type of failure between the un-
SRE 05
12
reinforced masonry (URM), the repointed masonry
SRE 04
(REP) and the masonry reinforced with metal fibres
10
(SRE). While in the cases of the unreinforced masonry
[kg/cm ]
2
REP 03
8 REP 02 and of the repointed joints the breakage occurred with
Vertical stress
6 URM 01
a small number of fairly large vertical cracks, in the
case of the masonry repointed with metal fibres a
4
larger number of smaller vertical cracks occurred. This
2 indicates an improvement in the mechanical behaviour
of the masonry.
0
0,000 0,002 0,004 0,006 0,008 0,010 The next experiments will investigate the increases
Normal strain ε
in shear strength and flexural strength. It is planned
to carry out on-site diagonal compression tests on un-
Figure 9. σ−ε diagram resulting from the tests with double
reinforced and reinforced panels, and also loading tests
flat jacks.
on panels that will first be reinforced, cut and then
Table 2. Results of the tests with flat jacks. tilted over, positioning them horizontally. They will
then be tested up to breaking point by applying vertical
Max Young’s loads.
compression modulus
Reinforcement Index stress (MPa) E1/3 (MPa)
4.3 Shear tests
Un-reinforced URM 01 0.595 480
Deep repointing REP 02 0.807 393 The diagonal test was performed on site on panels of
Deep repointing REP 03 0.857 512 1200 × 1200 mm dimension with sections of differ-
Reinforced repointing SRE 04 1.261 486 ent thickness and morphology. The load is given by
Reinforced repointing SRE 05 1.312 2416 hydraulic jacks. The test is defined by ASTM E 519-81
Standard. It is possible to calculate the characteristic
measurements were recorded the load history, maxi- strength of the masonry τk through:
mum compression stress σmax , and deflections of some
points using centesimal transducers. An equivalent
normal stiffness E1/3 was then calculated:
986
Table 3. Results of diagonal compression tests.
Shear Shear
strength modulus
Reinforcement Index τk (MPa) G1/2 (MPa)
d
5 MODELLING AND TESTS d-x x
987
barycentre of the tensite bars The sliding shear strength of the reinforced panel
P
bar originates from the combination of two resisting mech-
M t1
anisms: the shear transmitted by friction from the
masonry VRd,m and the shear consequent to the tensile
t strength of the reinforcements VRd,ts .
As an alternative, to carry out a cautious check, the
t2
sliding shear strength can be assessed conservatively
d
using the following relationship (Tassios 1988):
d-x x
0.85fmd
ftd
988
particularly suitable for fair-face walls of buildings and Building Materials, Elsevier, 16, (4), 229–239,
listed due to their historical or architectural value. 2002.
The effectiveness of the proposed technique was Corradi, M., Tedeschi, C., Binda, L. & Borri, A. Experimental
investigated by means of a series of tests with double evaluation of shear and compression strength of masonry
wall before and after reinforcement: deep repointing,
flat jacks, subjecting the masonry to vertical com- Construction and Building Materials, Elsevier (in press).
pression and diagonal compression tests (shear tests). Dolce, M. & Marnetto R. CAM technology, patent.
Based on the results obtained, it was possible to note Elgawady, M.A., Lestuzzi, P. & Badoux, M. Analytical model
an improvement of the mechanical behaviour of the for the in-plane shear behavior of URM walls retrofitted
masonry and it was noted that deep repointing of joints with FRP, Composites Science and Technology, Elsevier,
with metal fibres is capable of increasing compressive (66), 459–474, 2006.
strength significantly, even doubling it as compared Eshani, M.R. 1997. Strengthening of earthquake dam-
with the non-reinforced masonry. aged masonry structures with composite materials, Non-
metallic (FRP) Reinforcement for Concrete Structures,
Proceedings of the second international RILEM Sympo-
sium FRPRCS-2, L. Taerwe Ed., E & FN Spon, London,
REFERENCES England, 681–687.
Eurocode 6. 1995. Design of masonry structures – Part 1-
ASTM E 519-81. 1981. Standard Test Method for Diago- 1: General rules for buildings – Rules for reinforced and
nal Tension (Shear) in Masonry Assemblages, American unreinforced masonry. ENV 1996-1-1.
Society for Testing Materials. Gabor, A., Bennati, A., Jacquelin, E. & Lebon, F. 2006.
Binda, L., Modena, C., Baronio, G. & Abbaneo, S. 1997. Modelling approaches of the in-plane shear behaviour
Repair and investigation techniques for stone masonry of unreinforced and FRP strengthened masonry panels,
walls, Construction and Building Materials, Elsevier, 11, Composite Structures, Elsevier, (74), 277–288.
(3), 133–142. Tassios, P. T. 1988. Masonry mechanics , Liguori Publ.,
Cecchi, A., Milani, G. & Tralli, A. 2004. In-plane loaded Napoli, (Italian Translation).
CFRP reinforced masonry walls: Mechanical characteris- Tomazevic, M. & Alih, V. 1993. The strengthening of
tics by homogenisation procedures, Composites Science stone-masonry walls by injecting the masonry-friendly
and Technology, (64), 13–14, 2097–2112. grouts. 7th European Conf. On Earthquake Engineering,
Chiostrini, S. & Vignoli, A. 1994. In-situ determination of 1, 10–20.
the strength properties of masonry walls by destructive Turnsek, V. & Sheppard, P. F. 1980. The shear and flexural
shear and compression tests, Masonry International, 7, resistance of masonry walls, Proc. of the research conf.
(1) 87–96. on earthquake engineering., Skopje.
Corradi, M., Borri, A. & Vignoli, A. Strengthening tech- Vintzileou, E. & Tassios, T.P. 1995. Three-leaf stone masonry
niques tested on masonry structures struck by the strengthened by injecting cement grouts, Journal of Struc-
Umbrian-Marche earthquake of 1997–1998, Construction tural Engineering, 121, (5), 848–856.
989
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: Local materials are the resources that can be found readily in large quantity at a particular location
or area at a certain time. It could also be referred to as materials that can be used to fabricate a finished element.
These materials however could be abundant in some area but not available in another. The availability may largely
be dependant on geographical location of the area as well as the chemical and physical components of such
materials. The paper critically appraises some local material available in south west Nigeria. The investigation
revealed such materials as laterite, textile, bamboo, mosaic, mats stones, dye, timber, tusks, snail shell, cow dung,
cowries cane and mud. These materials are cheap relative to the imported materials from outside the country.
The neglect of these readily available materials should be discouraged. The paper also suggests practical and
innovative ways for designers, architects and manufacturers which can serve as an alternative source of material
and yet sustainable; and in the long run profitable for all concern. Proper inventory, investments, packaging and
modernization can help generate much needed foreign exchange and serve as promotion of the local culture.
Close attention must be paid to the sustainable methods and means of using these materials for the good of all.
Diagrams, pictures, plates and their application of these materials are also showcased in the paper.
991
Figure 2. Indigenous building of northern Nigerian built
with local materials.
992
Figure 3. Zulu beehive house built of rope lattice, thatch
and wooden strips. Figure 5. Ndele women at work doing decorations work and
painting of various geometrical shapes in their house. Source:
www.google.com
993
Figure 6. Locally produced sun-dried brick by Timbuktu
people. Figure 8. Common materials used in the construction
of African Traditional Buildings (bamboo and organic
materials).
994
Figure 9. Method of wall construction with mud. Figure 11. Local building at riverine area built with
wild coconut timber framework and raffia palm leave roof
covering.
995
Table 1. Types of Building examined during study survey. Table 2. Bar Chart showing the determining factors of inno-
vative building material in some selected cities of Nigeria.
30
30
25 25
20 20
15
15
10
10
5
5
0
cost avail taste durability tech-know climate
0
resd inst resort religion hotel other
Percentage Cummulative
Labels Frequency % percentage %
Architecture suffered a little setback because of “the
relative fragility of building materials and the simplic- Mud /laterite 48 60 60.0
ity of the working tools”. But it is worth of note that Timber 7 8.75 68.75
Nigerian Traditional Architecture left behind “tech- Stone 20 25 93.75
nological perfection” which is a result of a thorough Wattle 3 3.75 97.5
Bamboo 2 2.5 100.0
understanding the qualities of the materials used, and
of skill developed and improved by many generations
of builders great and small (Fadamiro, et al.1996).
Table 3b. Bar Chart interpretation of mostly used indige-
nous building materials for walls.
4 RESEARCH FINDINGS, DATA COLLECTION
AND DISCUSSIONS 60
50
The target of this data collection is to come out with 40
reasonable and reliable findings which will be of great 30
help as regards the technological influence on Nigerian
20
indigenous (traditional) building materials in terms of
innovation, availability, cost- effective, climatic suit- 10
ability, mass production, improved products, social 0
and legal acceptability, technical know-how, durabil- laterite timber stone wattle bamboo
ity as well as the occupation and economic activities,
craft activities, religious beliefs and practice, security,
compound and housing unit.
The research methodology adopted was by study- from the respondents but 10 were not retrieved. There-
ing various modern/ traditional buildings built with fore, this data analysis will make use of 80 question-
local materials found in some selected towns and cities naires retrieved as 100%. Random sampling technique
in Nigeria and the effect of innovative technology on was adopted for this project work and residents in
these materials and houseform using questionnaire. different locations were interviewed to uncover their
opinion about innovative technological influence on
4.1 Data Presentation, sample size and indigenous building materials in their modern housing
techniques design.
Table 2 corroborates the known fact that cost is a
A sum of 90 questionnaires were prepared and dis- deciding factor in the adoption of new technology and
tributed for administration at various locations in some use of innovative materials. Several of the respondents
selected towns and cities. The buildings focus are res- believe that innovative material should be made avail-
idential buildings, traditional buildings, public and able at affordable price at the open market. Table 3 a &
institutional buildings. Questionnaire of close ended b suggested that laterite/ mub block accounted for
type was used, all questionnaires were distributed and the 60% of the material used in indigenous buildings.
administered and 80 retrieved after proper follow-up Bamboo is the least used.
996
Table 4. Socio-cultural factors impacting on the adoption of holiday tax in some cases when it is highly
of innovative local material. needed.
(iii) That government should encourage the exporta-
Percentage Cummulative
Labels Frequency % percentage %
tion of the locally produced building materials
to find places in the world market and to gener-
Political/social 8 10 10 ate enough capital to enhance the development
Status of home based manufacturers.
Religion 10 12.5 22.5 (iv) That government and private bodies/ organiza-
Occupation 11 13.75 36.25 tions should be encouraged to establish indus-
Leisure activity 31 38.75 75 tries that produce local building materials.
Family 20 25 100 (v) Some planning authorities who hitherto restrict
Organisation people from using local building materials in
the cities should be made to relax their strin-
gent condition thereby encouraging developer
5 CONCLUSIONS to commence their development with the use of
local building materials.
Also, there are several existing evidences that indige- (vi) That the Government (at all levels) should
nous materials such as mud, bamboo, stone, grasses, always ensure that all related research findings
leaves, reeds, cow dungs among others were used for are implemented and not for record purposes
construction of various buildings in Nigeria. As these alone. This may be achieved by establishing
materials are readily available in abundance, the usage a well funded, research findings’ implementa-
has been on decline in some time past due to flair for tion commission, which would be responsible
imported building materials which are produced with for demonstrating such findings in the form of
disregarded for the climatic suitability of Nigeria and mass housing (construction) units.Thus, serving
high cost implication. as an architectural/structural model that would
The fact that most people believe that living in help in no small way in encouraging and pro-
brick, timber or stone houses symbolize abject poverty moting; the acceptance of the use of locally
which has created hatred for these materials. Also the sourced materials with the corresponding and
quest for modernization without considering our back- appropriate indigenous technology.
ground did not help us to look inward and develop our (vii) That public enlightenment campaign or dissem-
indigenous building materials rather than searching for ination of information should be encouraged
foreign materials. on the developed local building materials and
In view of this social impression that causes the their applications to the masses, as better sub-
decline in the utilization and with cognizance to these stitutes to the expensive and relatively scarce
materials’ numerous advantages, cost benefits, this conventional materials, possibly through the
research work provide vivid information on avail- existing rural development programme of the
able sources of building materials, their classification, Government.
properties, advantages and their uses as well as the pro- (viii) That the curriculum of technological, architec-
duction of some building components using identified ture, engineering institutions should be reviewed
local materials. to reflect the study of the application of locally
sourced building materials.
(ix) That the Government, Architects, Engineers and
5.1 Suggested recommendations
other relate professionals should lay examples
The encouragement of the use of local building mate- by building their respective houses with these
rials produced modern technology should be a col- materials as a substitute to the conventional
lective responsibility of the government, individual ones.
and private organization, for instance, the following
recommendations will be of great help:
(i) That the Government should ban or place restric- REFERENCES
tion on some imported building materials. This
will allow the growth of local building industries ALUKO, O.O. 1997. Nigerian Traditional Architecture,
A Case Study of South-west Nigeria. A Lecture Notebook,
and encourage people in using locally produced
Department of Architecture, Rufus Giwa Polytechnic,
building materials. Ondo State (Unpublished). pp. 1–10, 20–25.
(ii) That the Government should encourage the local ARAYELA, O. 1996. The Politics of Housing the Masses
building materials industries by creating con- in Nigeria – A Retrospective Introspection. A Journal
ducive atmosphere for operation right form the of the Association of Architectural Educators in Nigeria
registration process to the grant as well as granting (AARCHE). Vol. 1, No 3. pp. 29–33.
997
BASSEY, N. 1989. What are Local Building Materials? JOLAOSO, B.A. 1991. A Study of the Use of Local Build-
A Paper presented at the Mineral Seminar of the Nigerian ing Materials for Low Cost Housing in Western Nigeria.
Institute of Architects, Edo State Chapter. Unpublished, B.Sc Terminal Essay. Ahmadu Bello Uni-
BELLO, Y.O. 2001. Elements of Creativity in Traditional versity, Zaria.
Architecture; A Descriptive Study of Theory of Architec- JOLAOSO, B.A. 2001. Housing and Indigenous Building
ture. A Journal of AARCHE (Association of Architectural Technology; An Introduction. Desi-GCA Publications,
Educators in Nigeria). Vol. 1, No 6. pp. 95–13. Abeokuta, Nigeria. pp. 116, 123–131.
BOURNE, L.S. 1981. The Geography of Housing. Edward MADEDOR, A.O. & ASAOLU, E.G. 1989. Technology and
Arnold (Publishers) Limited, London. pp. 13–21. Standards for the Production of Local Building Materials
DENYER, S. 1978. African Traditional Architecture. Heine- in Nigeria. NBRRI Information Pamphlet. pp. 12–15.
mann Press, London. pp. 155–171. MADEDOR, A.O. & OMOGE, G.N. 1985. Survey of Brick-
DMOCHOSWKI, Z.R. 1987. An Introduction of Nigerian making Industries in Nigeria. A proceed from a Semi-
Traditional Architecture, Southwest and Central Nigeria. nar on Clay Bricks and Blocks: Research Development
Vol. II. Ethnolographical in Association with National and Current Practice in Nigeria; NBRRI Information
Commission of Museums and Monument. pp. 20–70, Pamphlet. pp. 7–22.
110–288. Microsoft Encarta Encyclopedia. 2005. Information on
DMOCHOWSKI, Z.R. 1990. An Introduction of Nigeria African Housing and Nigeria.
Traditional Architecture, South- Eastern Nigeria, the Ibo NBRRI 1997. Spotlight: The Nigerian Building and Research
speaking Area. Vol. III. Ethnographical in Association Institute, Lagos. In Housing Today. July/ August, Vol. I,
with National Commission of Museums and Monument. No. 1, pp. 10–17.
pp. 1–36.
FADAMIRO, J.A., & OGUNSEMI, D.R. 1996. Fundamental
of Building Design, Construction and Materials. Fancy
Publications Limited, Ile-Ife, Nigeria. p. 5.
998
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: European societies created the model for the modernization period that began in the Ottoman
state in the eighteenth century and continued during the so-called “westernization” period of the nineteenth
century. The construction of new types of buildings such as military buildings, western style palaces turned
into a reform movement which took western architecture as its model. Due to its conservative structure, the
Ottoman State could not completely subscribe to the intellectual movements in the west so it remained tied to the
industrial goods that Europe manufactured. For this, after the first quarter of the nineteenth century solid bricks,
tiles, metal beams and similar building materials were imported from countries like Britain, France, Belgium,
Italy, Germany and Austria. Many buildings of the period that remain have been registered as essential cultural
assets have faced various problems. The interventions undertaken have damaged construction techniques and
the originality of the imported materials.
999
Figure 2. Imported brick from Marseille, France
Figure 1. Traditional Ottoman brick (24 × 24 × 3.5 cm). (6.5 × 10.5 × 21.5 cm).
annuals of the nineteenth century such as “Apel Gül- that a European architect was hired for an imperial
benkian”, “P. Gülbenkian & Cie”, “S. Capou & J. project (Can, 1993). Many European architects besides
Pelletan”. G.T. Fossati, such as W.J. Smith, G. Fossati, M.A.A.
Bourgeois, G.B. Barborini and others as well as Euro-
pean educated architects of the Balyan family were
2 IMPORTED CONSTRUCTION MATERIALS
entrusted with building projects (Yergün, 2002).
These architects were instrumental for the develop-
2.1 Imported bricks
ment of imported bricks and modern building tech-
Bricks used in Ottoman architecture before the indus- niques in the Ottoman State during the 1840s as
trial age fell into three groups: “tuğla-i carşu” (full- Ottoman traditional building techniques were replaced
sized bricks), “tuğla-i miri carşu” (half-sized bricks) by European techniques and materials. The tran-
and “tuğla-i harci carşu” (outsized bricks) denot- sition from wooden buildings to modern masonry
ing the three different sizes in which bricks were techniques was achieved in secular architecture with
produced (Arseven, 1965). The dimensions of full- modern bricks through legal settlements and various
sized bricks were 24 × 24, 28 × 28, 30 × 30 cm in incentives.
length and 3, 3.5, 4 and 4.5 cm thick (Fig. 1). Since Research has shown that the bricks used in nine-
the Byzantine period, the most important produc- teenth century buildings in İstanbul generally measur-
tion sites were located on the shores of Büyükdere ing 6 − 7 × 10 − 12 × 22 − 24 cm dimensions came
and along the Golden Horn in Piripaşa and Kırkağaç from Saint-Henri and Saint André, two villages
(İnciciyan, 1976). near Marseille, France. In the mid-nineteenth cen-
With the technological advancements taking place tury these two villages were internationally known
in nineteenth century Europe, brick production as brick production centers. Bricks stamped with
develops analogously as its physical attributes are “GUICHARD FRERES St HENRY MARSEILLE”
highly resistant and all of the bricks industrially (6.5 × 10.5 × 21.5 cm) are proof of this (Fig. 2). It is
produced were of standardized compactness and den- well documented that the Çırağan Palace (1863) was
sity. Industrially produced modern bricks measur- built with bricks imported from Marseille measuring
ing 6 − 8 × 10 − 12 × 21 − 23 cm were the preferred 6.5 × 11 × 23 cm.
building material because they were inexpensive, prac- The Darülfünun (İstanbul University) dated 1845–
tical to use and fire resistant compared to wood and 1854 and Princess Adile’s Palace dated 1876 used
could support solid wall structures. bricks measuring 5 − 6 × 14 − 15 × 30 − 32 cm and
The Russian Embassy was the first building to be documentation shows that they came from Livorno,
constructed in İstanbul with modern bricks in 1838 and Italy (Fig. 3). Certainly, transporting bricks from Mar-
it was highly appreciated by the reformers of the Tanzi- seille and Livorno to İstanbul by ship was also easy
mat era. The Swiss architect G.T. Fossati was hired for and fast.
the project and after the completion of the Russian The “FRATELLI ALLATINI SALONICCO” com-
Embassy, in 1841, he was entrusted with the construc- pany was another important producer of bricks for
tion of the Military Hospital at the Imperial Gate and buildings in İstanbul was located in Salonica (now
this was the first Ottoman structure made with modern Greece) (Fig. 4). The Allatini family was of Jewish
bricks completed in 1843. The Balyan family had been origin and had emigrated from Livorno to Salonica
the palace architects until that time and it is significant in 1802 and set up the company in 1836. It was this
1000
Figure 3. Imported brick from Livorno, Italy
(5,5 × 15 × 30 cm).
1001
Figure 10. Imported glass tile.
Figure 7. Imported tile from Marseille, France.
1990). From the last quarter of the nineteenth century
on, imported metal supports of varying shapes and
sizes were used in buildings of many floors having
rooms of large dimensions in İstanbul. Examples of
these found in İstanbul were imported primarily from
England and Belgium after trade agreements were
made between these countries and the Ottoman State
in 1838.
The English engineer Sir William Fairbairn (1789–
1874) came to İstanbul in 1839 and he made various
orders for the Unkapanı bakery and flour mills. In
1841, Fairbairn made a three-story building for milling
corn entirely of cast iron and wrought iron including
Figure 8. Imported tile from Salonico, Greece. the supports, the walls and the roof structure and exhib-
ited it in his factory. This was the first building of
this type in England and became a prototype for iron
churches, houses and storage spaces.
According to Batur, this structure was disassem-
bled and brought to İstanbul in 1841 by ship. The
building measured 8.10 × 15.50 m was three stories,
only the foundations and the wall supporting the inner
machinery were masonry. The floors and the roofing
were supported by cast iron and beams (Batur & Batur,
1981). This structure no longer stands today.
Since 1855s, especially after 1870 imported metal
products from Belgium came to İstanbul and other
eastern ports by regular steam ship routes from the
Port of Anvers. The “Feshane-i Amire Factory” was
set up in order to manufacture fezes and broadcloth
Figure 9. Imported glass tile from Marseille, France. for the soldiers of the Ottoman army in the last quar-
ter of the nineteenth century. The metal roofing of the
poor. For these reasons tiles and bricks from Livorno weaving hall of the factory was supported by 274 cast
were considered better as well (Önsoy, 1988). iron columns measuring 25 cm in diameter, bearing
the stamp of “PROVIDENCE MARCHIENNE BEL-
GIQUE” (Figs 11–12). “Les Forges de la Providence in
2.3 Imported metal supports
Marchienne-au-Pont” factory was established in 1838
Iron was wrought from the fourteenth century on and and with the convergence of the Cockerill-Ougrée fac-
by the mid-eighteenth century cast iron technique was tory in 1966 became one of Belgium’s most important
used for shaping metal but after the Industrial Revolu- industries. The English industrialist, John Cockerill
tion, steel replaced iron (Ökten, 1995). In Ottoman (1790–1840) established a company in the region of
architecture, iron was not used on its own to bear Wallonia in 1817 and this company exists today as
weight but rather as a supporting component (Tanyeli, “Cockerill Maintenance&Ingénierie” since 2004.
1002
Figure 11. Imported cast-iron columns of the “Feshane-i
Amire” (Imperial Fez Factory).
1003
Figure 17. Roof structure and jack arch construction of the
“Mekteb-i Tıbbiye-i Şahane” (Military Medical School).
1004
Some of the interventions carried out with the
aim of protecting them have further damaged both
the construction technique and the original imported
materials.The original bricks are disappearing because
the walls have been opened up and additions have
been made of solid masonry. The most wide scale
incursions occur from removing the metal elements
which support the flooring and the roof structures,
and then pouring concrete flooring. The registration
of these structures is also lacking. Many of the orig-
inal elements were removed before the building was
sufficiently documented. It has been noted that when
the roofs were being repaired, even tiles that were in
good condition were replaced with new ones without
any documentation. The imported bricks and all of the
metal elements are the architectural elements that bear
Figure 19. İron pilaster stamped with “R. Ph. Waagner witness to construction techniques used at the time and
Vienne”. they need to be documented and preserved because
they reflect an era in Ottoman architecture.
Generally, the construction materials are not con-
served as required in the buildings that have been
observed. The two main causes of the decay are the
materials conditions and external causes. Due to the
conditions and locations of the materials like bricks,
tiles and metal supports can be conserved by clean-
ing methods and consolidation. If the material has lost
its feature as a structural element due to the inter-
nal causes, the material should be replaced. The new
bricks, tiles or metal elements must have the same
shape, same dimensions and same constitutional fea-
tures as to keep the harmony with the original ones.
In addition to this, the new materials should be dis-
tinguished from the originals by its texture, color
etc. according to the restoration and documentation
criterion.
REFERENCES
Arseven, C.E. 1965. Sanat Ansiklopedisi, İstanbul: Milli
Figure 20. Imported steel beams from Burbach factory. Eğitim Bakanlığı Yayınları.
Batur, A. & Batur, S. 1981. İstanbul’da XIX. Yüzyıl Sanayi
Yapılarından Fabrika-i Hümayunlar”. I. UluslararasıTürk-
İslam Bilim ve Teknoloji Tarihi Kongresi, 14–18 Eylül
3 CONCLUSION 1981: 331–342. İstanbul: İstanbul Teknik Üniversitesi
Mimarlık Fakültesi Yayını.
A forceful change occurred in design philosophy and Batur, A. 1994. Mekteb-i Tıbbiye-i Şahane Binası. Dün-
construction techniques in Ottoman architecture in den Bugüne İstanbul Ansiklopedisi. 5: 377–379. İstanbul:
the nineteenth century due to improved economic Kültür Bakanlığı ve Tarih Vakfı Ortak Yayını.
and political relations between the Ottoman State and Can, C. 1993. İstanbul’da 19. Yüzyıl Batılı ve Levanten
MimarlarınYapıları ve Koruma Sorunları. İstanbul:Yıldız
European countries and the introduction of imported
Teknik Üniversitesi Fen Bilimleri Enstitüsü Doktora Tezi
materials such as modern bricks and metal supports. (yayınlanmamış).
However, some of the buildings from this period have Can, C. & Girardelli, P. 1996. Beyoğlu’nda Bir Latin Anıtı.
collapsed without ever being thoroughly documented Yaşayan Çizgiler-Sant’Antonio Kilisesi Mimari Çizimler
from an architectural viewpoint. Many of the exist- Sergisi: 13–71. İstanbul: Yapı Kredi Kültür Merkezi.
ing structures dating to this period are on the brink of
disaster and require immediate attention even though
they have been registered as cultural assets.
1005
Çiftçi, A. 2004. 19. Yüzyılda Osmanlı Devleti’nde Askerî L’Indicateur Ottoman Annuaire-Almanach Commerce, 1881,
Mimarî ve İstanbul’da İnşa Edilen Askerî Yapılar. Ýstan- Constantinople: Cervati Frères & D. Fatzea.
bul: Yıldız Teknik Üniversitesi Fen Bilimleri Enstitüsü
Doktora Tezi (yayınlanmamış).
İnciciyan, P.Ğ. 1976. XVIII. Asırda İstanbul, İstanbul: Fetih WEB-SİTE
Cemiyeti İstanbul Enstitüsü Yayınları.
Kösebay, Y. 2007. Anadolu Demiryolu Çevresinde Gelişen
Mimari ve Korunması. İstanbul: İstanbul Teknik Üniver-
http://www.cmi.be
sitesi Fen Bilimleri Enstitüsü Doktora Tezi http://www.cockerill-sambre.com
(yayınlanmamış). http://www.crwflags.com
Kuruyazıcı, H. & Tapan, M. 1998.Sveti Stefan Bulgar Kilis- http://www.saarstahl.com
esi, Bir Yapı Monografisi, İstanbul: Yapı Kredi Yayınları. http://www.hebig.org
Mori, A. 1906. Gli Italiania Constantinopoli. Modena. http://www.lafargeroofing.co.za
Önsoy, R. 1988. Tanzimat Dönemi Osmanlı Sanayii ve Sanay- http://www.allatini.gr
ileşme Politikası, Ankara: Türkiye İş Bankası Yayınları. http://www.british history.ac.uk
Ökten, S. 1995.Yapı Mühendisliği ve Sanayi Devrimi. Ekrem http://www.answers.com
Hakkı Ayverdi Hatıra Kitabı: 127–136. İstanbul.
Ratier,Y. 1998. La Terre de Marseille: Tuiles, Briques et Car-
http://www.waagner-biro.at
reaux, Marseille: Chambre de Commerce et d’Industrie. http://www.thyssenkrupp.com
Sandalcı, M. 1997. Tuğlalar. Tombak Dergisi, 17: 50–58.
İstanbul: Horhor Yayıncılık.
Tanyeli, G. 1990. Osmanlı Mimarlığında Demirin Strük-
türel Kullanımı (15–18 y.y.). İstanbul: İstanbul Teknik
Üniversitesi Fen Bilimleri Enstitüsü Doktora Tezi
(yayınlanmamış).
Yergün, U. 2002. Batılılaşma Dönemi Mimarisinde, Yapım
Teknolojisindeki Değişim ve Gelişim. İstanbul: Yıldız
Teknik Üniversitesi Fen Bilimleri Enstitüsü Doktora Tezi
(yayınlanmamış).
Osmanlı Başkenti’nde Belçika, İlk Adımlardan Güzel Çağa
Kadar, Belçika SarayınınYüzüncüYıldönümü 1900–2000,
2001. İstanbul: Belçika Başkonsolosluğu Yayını.
1006
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
P. Casadei
Consultant Engineer, R&D
TEC.INN. S.r.l., Technical & Commercial Division, Milano, Italy
E. Agneloni
TEC.INN. S.r.l., Head Quarters, San Mariano, Perugia, Italy
ABSTRACT: Due to their light weight, high stiffness-to-weight and strength-to-weight ratios, and potentially
high resistance to environmental degradation, resulting in lower life-cycle costs, advanced composites mate-
rials, commonly known with the acronym FRP, are increasingly being considered for use in civil engineering
applications, ranging from the retrofit and rehabilitation of buildings and bridges to the restoration and strength-
ening of historical-monumental masonry structures. Thanks to the recently published Italian FRP guidelines,
CNR-DT 200/2004, by the Italian National Research Council, there has been an increasingly demand of their
implementation for general strengthening and retrofitting applications as well as seismic mitigation of structures
and infrastructures. This paper aims to provide a general overview of the Italian state of the art implementation
of FRP materials in the last 15 years on historical structures, through different case studies, and presenting the
potentials of a new family of retrofitting materials made of ultra-high tensile strength steel filaments arranged
in cords and knitted similarly to FRP sheets.
1 INTRODUCTION
1007
Figure 4. Hardwire material.
1008
Figure 5. Two types of available cords.
1009
structure following this procedure: first a static load
test is performed to evaluate the current behavior of the
structure under loads (deformations, degrees of fixity,
vibrations, strains, etc), then the FRP is installed and
finally a last static load test is performed in order to
compare the behavior of the structure before and after
strengthening. In many cases the direct comparison
of the load-vs-deflection curve before and after the
FRP installation is the key parameter that validates
the technology.
Figure 9. The masonry arch bridge of St.Peter in Cariano.
1.4 Italian situation
In Italy, as well as in many other seismic countries,
the use of advanced composite materials for strength-
ening and retrofitting existing structures has been
heavily introduced in the market by recent earthquakes
and natural disasters. In particular the earthquake that
interested the central Italian regions of Umbria and
Marche in 1997–’98 has heavily pushed engineers
and contractors on the use of the FRP technology
thanks to the aforementioned qualities. In particular, to
strengthen structures against seismic/dynamic loads,
FRPs are particularly efficient because in front of very
performing mechanical properties they have very low Figure 10. NDT testing prior and after FRP strengthening.
weights and consequently they do not add any mass
to the structure making them ideal for this type of
2 FRP APPLICATIONS ON MASONRY
retrofits.
HISTORICAL INFRASTRUCTURES
Nowadays the mainstream market of these mate-
rials in Italy is for masonry historical structures
2.1 CFRP strengthening of a masonry arch bridge
where their implementation varies from: increas-
located in San Pietro in Cariano – Verona (IT)
ing the capacity of panels, arches, or vault; wrap-
ping of columns to enhance their compressive The retrofit concerned the strengthening of a solid
strength and ductility; reducing thrust forces in thrust- brick masonry arch bridge located in San Pietro in
ing structures; transforming non-structural members Cariano (Verona). The change of the traffic loads in
into structural members by increasing their stiffness recent years and the lack of maintenance have caused
and strength; strengthening and stiffening horizon- on the bridge a remarkable crack state needed to be
tal non-thrusting structures; wrapping buildings at urgently addressed.
floors and roof locations to improve vertical and The work sequence has been as follows: first of
horizontal strength to lateral loads, typically seismic all before any FRP installation, all cracks have been
actions. filled with tixotropic mortar carefully removing any
TEC.INN. S.r.l., founded in Perugia-Italy in 1986 falling parts of masonry and leveling with mortar.Then
has been pioneer in Italy in the research, installa- the entire soffit of the arch has been hand-blasted to
tion and development of the FRP technology with prepare the substrate for an ideal bond. Once this initial
over 1000 retrofit-consolidation-strengthening appli- phase has been completed the substrate was ready for
cations all over Italy and few also abroad, vary- FRP installation.
ing from reinforced and pre-stressed concrete to CFRP sheets have been installed longitudinally and
historical-monumental masonry structures. Specialty transversally contributing to an overall increase of the
contractor on innovative materials, TEC.INN. S.r.l. arch capacity and to provide a transversal contribution
is able to offer, in Italy and worldwide, advanced of the entire structure subjected to concentrated loads.
retrofitting and strengthening solutions, using FRP Before and after strengthening, an exhaustive test-
systems and innovative materials, to public and private ing campaign has been conducted, in order to provide
owners, consulting engineering firms and contrac- all necessary information to design the CFRP retrofit
tors, starting from in-situ diagnostic phase final- solution. Firstly laboratory tests have been performed
ized to support designers of the retrofitting system, on samples for material characterization. Then on site
till the installation phase and/or knowledge trans- tests, such video endoscopy and thermography have
fer to local contractors of advanced composite FRP been conducted to evaluate the crack state and over-
technology. all structure. Once the FRP installation was completed,
1010
Figure 11. The “Maggiore Fountain” in Perugia. Figure 12. FRP strengthening of the marble panels of the
fountain.
1011
Figure 16. The Saint Biagio’s temple in Montepulciano.
1012
Figure 19. Retrofit of the masonry double curvature vaults
with unidirectional GFRP sheets.
1013
Once the strengthening of the entire vault was com-
pleted, and the entire vault was shored, the cut was
realized.At this point the strips of material reaching the
hole where anchored on it, by creating sort of stirrups
that could anchor the material and prevent any vertical
movement around the hole due to the loss-continuity
of material induced by the cut.
REFERENCES
Figure 23. SRP detail installation on the vaults.
Borri, A., Castori, G., and Grazini, A., (2007). “Cordoli
sommitali in muratura armata con SRG” XII Convegno
Nazionale L’Ingegneria Sismica in Italia (ANIDIS), Pisa,
3.5 SRP strengthening of the vaults of “Contucci Italia, 10–14 Giugno, 2007.
Palace” in Montepulciano-Siena Borri, A., Castori, G., Casadei, P., and Ebaugh, S. (2007).
“Rinforzo di Archi in Muratura con Materiali Compositi
The Contucci’s Palace is an exclusive historic private Innovativi” XII Convegno Nazionale L’Ingegneria Sismica
residence, located in the heart of Tuscany, in one of in Italia (ANIDIS), Pisa, Italia, 10–14 Giugno, 2007.
the most magnificent and most admired renaissance Borri, A., Castori, G., Casadei, P., and Ebaugh, S. (2007).
squares, Piazza Grande in Montepulciano. “Research on Composite Strengthening of Masonry
The retrofit was needed in order to consolidate the Arches” Eighth International Symposium on Fiber Rein-
vaults of the first and noble floors, necessary because forced Polymer Reinforcement of Reinforced Concrete
the owners wanted to install an elevator between the Structures (FRPRCS-8), Patras, Greece, July 16–18, 2007.
two floors, consequently cutting trough the vaults. Casadei, P., Nanni, A., Alkhrdaji, T., and Thomas, J., (2005).
“Behavior of Double-T Prestressed Beams Strengthened
In order to decide any kind of retrofit solution, a With Steel Reinforced Polymer”, Advances in Structural
finite element model was created with commercial Engineering an International Journal (ASE), Vol. 8, No. 4,
FEM routines, in order to simulate the behaviour and pp. 427–442.
stress concentrations of the two vaults before and after CNR-DT 200/2004, 2004: “Istruzioni per la Progettazione,
the cut-through. l’Esecuzione e il Controllo di Interventi di Consolida-
Once the stresses involved were known, the retrofit mento Statico mediante l’utilizzo di Compositi Fibror-
was designed in order to absorb the increment of inforzati” Published by the National Research Council
stresses generated by the cut, creating strengthened (CNR), Roma, pp. 164 (also available in english).
strips along the sides of the hole, like a sort of supports “Compositi FRP: Linee guida per il rinforzo strutturale”
published by NCT Global Media Editore, 2002, ISBN
for the vault weakened by the cut. The strengthen- 88-900892-0-2.
ing material chosen was the medium density Hardwire Hardwire LLC, 2002,“What is Hardwire,” www.hardwirellc.
3X2 sheet, impregnated with epoxy resin. The reason com, Pocomoke City, MD.
why epoxies were chosen was generated by the need TEC.INN. S.r.l., “Innovative Technologies” San Mariano-
of creating stiff strips around the hole. Perugia, Italy, www.tecinn.com.
1014
Novel conservation engineering techniques,
restoration and strengthening
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The present paper shows and discusses some aspects concerning the delamination phenomenon
of Fibre Reinforced Plastic (FRP) materials fixed on masonry support. With this aim, in the first part of the
paper the results of an experimental campaign conducted by the authors concerning the effect of the bond length
on the delamination phenomenon of CFRP glued on clay bricks are discussed. Moreover, in order to underline
further aspects concerning the decohesion mechanism of FRP-reinforcements, in the second part some numerical
analyses are also presented.
1017
Table 1a. Tests results (compressive strength).
40 mm
Dimensions (mm)
80 mm
120 mm
250 mm
160 mm
Specimen L B H Section (mm2 ) fb (MPa)
1 55 56 55 3080 41.5
2 55 55 54 3025 35.7
3 56 55 55 3080 40.6
4 55 55 55 3025 34.7
5 55 55 54 3025 36.6
120 mm 6 55 55 55 3025 41.7
legend: strain-gauge on the unbonded part
Average value 38.5
strain-gauge on the bonded part
Dimensions (mm)
1018
Table 1c. Test results (elastic modulus).
Dimensions (mm)
Dimensions (mm)
Load
Specimen Cut height (mm) (N) τb (MPa)
1019
Figure 4. (a) Decohesion mechanisms of specimens and
(b) Decohesion mechanisms of specimens.
3 NUMERICAL ANALYSES
Figure 5. Strain profiles of FRP laminates.
In order to propose a simple numerical model based
on the experimental data and able to capture the global
behaviour of CFRP-laminates glued on masonry the brick, the FRP-laminate and the FRP/brick layer
blocks, some numerical analyses have been performed. have been modelled by adopting the following finite
elements:
3.1 Modelling approach
• clay bricks: four-node quadrilateral isoparamet-
Regarding the modelling approach adopted by the ric plane stress elements (labelled in the code as
authors for simulating the response of the examined Q8MEM) based on linear interpolation and Gauss
specimens, a 2D finite element model (Figure 6) has integration.
been developed for all the examined specimens (i.e. • FRP-laminates: beam elements characterized only
varying the bond length Lb of the reinforcement) by the axial deformation (also called truss-element
using the code DIANA9.1 (2000). In the models, both and labelled in the code as L2TRU).
1020
Table 4. Material properties used for the numerical models.
3.3 Results
where: Two types of analyses have been performed: a linear-
Ea (see table 2), Eb (see table 1.c) are the elastic’s elastic analysis where all the elements are elastic and
moduli of the adhesive and the brick respectively; hi is an incremental nonlinear elastic analysis where the non
the thickness of the interface layer between the brick linearity has been adopted only for the brick elements.
and the FRP, assumed equal to 1.0 mm, as observed by The first type of analysis has been performed with
tests; Ga (see table 2), Gb (see table 4) are the shear’s the aim of capturing the elastic behaviour of speci-
moduli of the adhesive and the brick respectively. The mens. In Figure 7 have been compared the displace-
properties of the elements used in numerical model are ments of the FRP deduced by the experimental tests
summarised in Table 4. (circular symbols) and those obtained by the numer-
In order to model the nonlinear behaviour of the clay ical analyses (continue line). In particular, the figure
brick, an orthotropic elasto-plastic continuum model refers to the scheme 4 of the examined specimens con-
with the Hill (1947) yield condition and characterized sidering a low value of the applied force (assumed
by an elastic-perfectly plastic behaviour in tension and equal to 20% of the maximum force sustained by the
compression has been considered. reinforcement) in order to examine the behaviour of
1021
Figure 7. Comparison between numerical and experimental
values of FRP-displacements.
1022
From the experimental results, the influence of the
bond length on the bond strength and the delamination
mechanism has been examined in detail. In particular,
it was observed that after the detachment of the FRP
laminate only the brick support was damaged.
On the basis of the experimental evidence, a 2D
numerical model of the tested specimens has been
developed adopting an elasto-plastic behaviour for the
brick response.
The good agreement between the numerical and the
experimental results both in terms of strength and non-
linear behaviour have confirmed the reliability of the
Figure 9. Maximum force sustained by the reinforcement. proposed model.
It is clear that a more comprehensive study of the
bonding response would require the evaluation of a
fracture energy as defined in the theoretical formu-
lations reported in the codes. This means that first,
the numerical model would also provide the maximum
deformation in the brick and the post-peak behaviour,
and second, the experimental tests would be performed
by a displacement control procedure.
REFERENCES
Aiello, M.A. & Sciolti, S.M. 2005. Bond analysis of masonry
Figure 10. Bond strength variation. structures strengthened with CFRP sheets. Construction
and Building Materials. Vol.20: pp. 90–100
the bond strength linearly proportional to the bond Casareto, M., Olivieri, A., Romelli, A. & Lagomarsino,
S. 2003. Bond behaviour of FRP laminates adherent to
length of the reinforcement. From the figure it is clear
masonry. In: Proceedings of the international conference
that the variation of the bond strength with the bond advancing with composites, Milano, Italy
length is strongly far from the linear variation and is Barbi, L., Briccoli Bati, S. & Ranocchiai, G. 2004. Anal-
characterised by a smaller slope. isi sperimentale di campioni in mura-tura fasciati con
This is a further aspect that confirms the presence composito CFRP. in Atti del II Convegno Nazionale “Mec-
of different interaction mechanisms between the FRP- canica delle struttu-re in muratura rinforzate con FRP –
reinforcing and the support depending on the bond materials”, Venezia, Italy (in Italian).
length. Indeed, the increase of the bond length of the CNR-DT200/2006. 2006 Guide for the design and
FRP not only produces an increase of the zone which construction of externally bonded FRP systems for
strengthening existing structures. Materials, RC and
resists to external loads but produces a variation in
PC structures, masonry structures. National Research
terms of stress transfer between the reinforcement and Council, Rome-CNR
the support through the interface layer. National Instruments LabVIEW 2006. Graphical Develop-
ment Platform for Design, Control and Test.
DIANA, 2000. Displacement analysis finite element
4 CONCLUSIONS software. Version 9.1, TNO-Building Division, Delf,
The Netherlands.
In this paper are discussed the results obtained by an
experimental campaign and a numerical study devoted
to analyse the delamination mechanisms of CFRP
laminates glued on clay bricks.
1023
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
N. Delinikolas
1st Ephorate of Byzantine Antiquities, Hellenic Ministry of Culture
ABSTRACT: Grouting can be a durable and mechanically efficient intervention technique, not only when
the grout composition is suitably designed, but also when the technique is correctly implemented following
an adequate methodology. Such a methodology has been developed for the in situ application of grout in the
Katholikon of Dafni Monastery in Athens. Hereby are presented the main features of this methodology. Apart
from the information regarding the masonry preparation for grouting, the necessary quality control of the grout
prepared at the worksite and the injection process, the importance of collecting specific data during the injection
process, which allow monitoring the movement of the grout into the masonry and estimating the grout volume
consumed in the masonry, is underlined.
1025
interior of the structure, and the volume consumed per and NHL5), were tested in the laboratory of DTRR,
masonry area. The need for the efficiency of the grout- and grout formulations, with or without superplastizer,
ing intervention to be controlled, after the completion were examined. The main results of the research are
of the works, by means of adequate NDT and seismic reported in Kalagri et al. (2007).
monitoring instrumentation, is also underlined. In order to determine the injectability characteris-
In the present paper the research process followed tics, the penetrability, fluidity and stability of the sus-
for the selection of the adequate grout will be first pensions were fully examined in various water/solids
briefly summarized, as it is in direct conjunction with ratios, with or without additives. The grouts were pre-
this paper, and then the most important elements of pared by using an ultrasound dispersion mixer, assisted
the applied injection grouting methodology will be by a mechanical device of low turbulence. The stan-
presented and commented upon. dardized sand column test method (NF P18-891, pr.
EN 1771), was used to check the penetrability and flu-
idity, along with the standard apparatus for testing the
fluidity (NF P18-358) and stability (NF P18-359) of
2 SELECTION OF THE GROUT COMPOSITION the suspensions. In each case, a time limit of 50 sec for
the sand column penetrability test (T36 ); an efflux time
The design of high injectability grouts was applied of 500 ml of grout (td=4.7 ) shorter than 45 sec (Marsh
on the basis of the methodology proposed by Milti- cone d = 4.7 mm fluidity test), and a maximum accept-
adou & Tassios (2006). First, were taken into account able limit of 5% for the bleeding were set initially
the performance requirements deriving from the struc- (Miltiadou 1990) for the laboratory investigations.
tural restoration study (Miltiadou et al. 2003), and The compositions presenting satisfactory injectabil-
then the following target values were set for the basic ity capacity were further tested to evaluate their
mechanical properties of the grouted masonry: ten- behavior to salt decay and estimate their mechani-
sile strength approximately double that one of the cal characteristics (compressive and flexural strength).
masonry before grouting, and compressive strength Six alternative grout formulations presenting similar
approximately equal to 3.0 MPa. injectability, were injected at low pressure into twenty
On the basis of the available literature (Vintzileou eight cylindrical specimens, simulating the infill mate-
& Tassios 1995, Tassios 2004), it was estimated that rial of three-leaf stone masonry, which were then
the compressive strength of the grout at the age of subjected to compression in different hardening ages
six months should lie between 6MPa and 10MPa; a (Kalagri et al. 2007). After comparative evaluation of
grout flexural strength of the order of 2-3MPa was the results, two grout compositions (the ternary grout
required. In addition, the physical-chemical properties and a natural hydraulic lime NHL5-based grout) ful-
of the raw materials should be selected in a way that filled simultaneously the injectability, the strength and
the durability of the structure and its precious mosaics durability requirements. Therefore, they were selected
would not be jeopardized. Finally, the grouts should to be applied to six three-leaf stone wallettes, simulat-
have high injectability capacity, so that, under low ing the masonry of the upper parts of the monument,
pressure (∼0.075 MPa), they enter and fill fine voids subjected either to compression or to diagonal com-
and cracks, with a nominal minimum width (Wnom ) pression up to their maximum resistance, before and
equal to two tenths of millimeter. after grouting (Vintzileou et al. 2006, Miltiadou et al.
According to literature (Miltiadou 1990, 2006).
Toumbakari 2002, Valluzzi 2000) and previous inves- The substantial (compressive and tensile) strength
tigations carried out by the Directorate for Technical enhancement of wallettes, the rather ductile behaviour
Research on Restoration (DTRR, Hellenic Ministry under diagonal compression (compared to that of
of Culture), two main categories of grouts could sat- masonry grouted with the ternary grout), the physico-
isfy injectability, strength and durability requirements: chemical properties that ensure a durable intervention
(i) ternary grouts composed of a low cement content and contribute to the protection of mosaics led to the
(30%), lime 25% and pozzolan 45%, and (ii) hydraulic selection of the natural hydraulic lime- based grout for
lime – based grouts. the application in the Katholikon of Dafni Monastery.
Thus, various grout mixtures, belonging to the Given that these hydraulic limes are characterized by a
above two categories, were designed and tested (in relatively high percentage of available lime, the addi-
order to assess their physical, chemical and mechani- tion of pozzolan, in adequate proportion, is expected
cal properties) at the laboratory of DTRR. The use of to have beneficial effect. Thus, in order to improve the
white Danish cement in the ternary grout was chosen, hydraulic lime based grout, the addition of fine nat-
due to its fineness, low alkali content and high sulphate ural pozzolan (dmax < 75 µm) in various proportions
resistance. As far as the hydraulic lime – based grouts was investigated. The addition of a small percentage
are concerned, various types of natural hydraulic limes of pozzolan (10%) was decided, on the basis of addi-
(classified by EN 459 as NHL2, NHL3.5, NHL3.5-Z tional data deriving from porosity measurements, salt
1026
Table 1. Composition and injectability characteristics of the
selected grout measured in the laboratory and in situ at the
first pilot preparation.
GROUT COMPOSITION
NHL5 (St Astier) 90%
Pozzolan 10%
Superplasticizer (1), (2) 1%
Water (1) 80%
GROUT PROPERTIES In lab In situ
T36 (sec) – Sand column 19–22
1.25/2.50 mm (voids
∼0.2–0.4 mm)
Bleeding <1% 1%
Apparent viscosity – td=4.5 (sec) In lab In situ
0 min after mixing 21 22
60 min after mixing (agitated) 23 25
Apparent density (gr/cm3 ) In lab In situ Figure 1. Historical pathology of the external east façade of
0 min after mixing 1.5050 1.4978 the Katholikon of Dafni Monastery. Uncolored joints still pre-
60 min after mixing (agitated) 1.4986 1.4870 serve the old Byzantine pointing mortars (Delinikolas et al.
2003).
(1) % of the solid phase of the grout.
(2) superplasticizer based on polycarboxylic ether.
1027
Katholikon of Dafni Monastery. The methodology has
been tested and further refined during the injections
of the aforementioned wallettes, and then applied to
the whole injection project for the repair of the mon-
ument’s masonries. The most important aspects of
this methodology are presented below, together with
a synthesis of indicative quality control data, concern-
ing its in situ application in the Katholikon of Dafni
Monastery.
1028
masonry structures (due to the relatively low flow rates
and pressure values).
1029
upwards, since grout should be kept under pressure
inside masonry until its hardening is completed. In
some cases, when overflow of grout had taken place at
locations that no tubes had been installed, the leakage
was stopped with the help of pozzolan powder, or an
adequate paste (pozzolan paste, clay paste or paper
pulp, etc.). Cement or lime was absolutely avoided.
After that, the masonry was immediately cleaned.
1030
Table 2. Apparent density values of grout outflows. Table 3. Evolution of strength of the in situ prepared grout.
1031
bearing mosaics, frescoes and old mortars, that have
to be preserved in situ.
The whole procedure and quality controls can be
easily applied by qualified scientific and technical per-
sonnel, and allow for corrective measures to be taken
during the project. Moreover, the collection and elab-
oration of all the proposed data during the works is
absolutely necessary for an overall assessment of the
grouting application.
This assessment should be further supported, by
means of non destructive techniques especially in case
of important historic structures.
1032
of Dafni Monastery, Phase A – Structural Survey, Analysis Mouzakis Ch., Miltiadou-Fezans A., Touliatios P.,
and remedial measures. Hellenic Ministry of Culture (in Delinikolas N., Dourakopoulos J. 2008. Earthquake
Greek). based condition monitoring of the Katholikon of Dafni
Miltiadou-FezansA.,Vintzileou E., Delinikolas N., Zaroyianni Monastery. Submitted to the 6th intern. conf. on Structural
E., Chorafa E. 2004. Pathology of the Dafni Monastery: Analysis of Historical Constructions, Bath, UK.
survey, monitoring of cracks, interpretation and numerical Tassios T.P. 2004, Rehabilitation of three-leaf masonry. In
verification. Proc. 4th inter. Sem. on Structural Analy- Evoluzione nella sperimentazione per le costruzioni, Sem-
sis of Historical Constructions, Vol. 2, Padova, 10–13 inario Internazionale 26 Settembre–3Ottobre, CIAS
November. Toumbakari E.E. 2002. Lime-Pozzolan-Cement grouts and
Miltiadou-Fezans A., Papakonstantinou E., Zambas K., their structural effects on composite masonry walls. PhD.
Panou A., Frantzikinaki K. 2005. Design and application Thesis, Department of Civil Engineering, KULeuven.
of hydraulic grouts of high injectability for the struc- Valluzzi M.-R. 2000. Comportamento meccanico di murature
tural restoration of the column drums of the Parthenon storiche consolidate con materiali e tecniche a base di
Opisthodomos., Proceedings of STREMA IX, WIT Press, calce. PhD Th., Univ. of Trieste.
pp. 461–471. Vintzileou E., Tassios T.P.1995. Three leaf stone masonry
Miltiadou-Fezans A., Tassios T.P. 2006. New rational criteria strengthened by injecting cement grouts. Journal of Struc-
for the holistic design of hydraulic grouts. 1st Conference tural Engineering, ASCE, Vol. 121, No 5.
of the Society for Research and Promotion of Scientific Vintzileou E., Miltiadou-Fezans A., Palieraki V.,
Restoration of Monuments (in Greek). Delinikolas N. 2004. The use of radar techniques and
Miltiadou-Fezans A., Vintzileou, E., Papadopoulou, E., Kala- endoscopy in investigating old masonry: the case of Dafni
gri, A. 2006. Mechanical properties of three-leaf stone Monastery. Proc. of 4th intern. sem. on Structural Analysis
masonry after grouting. Proc. 5th inter. Sem. on Structural of Historical Constructions. Modena C., Lourenco P.B.,
Analysis of Historical Constructions, New Delhi, India. Roca P. (ed.), Rotterdam, Balkema, Vol. 2, pp. 351–360.
Miltiadou-FezansA., KalagriA., Delinikolas N. 2007. Design Vintzileou E., Miltiadou-Fezans A., Vrouva A., Anagnos-
of hydraulic grout and application methodology for stone topoulou S. 2006. Mechanical properties of three-leaf
masonry structures bearing mosaics and mural paintings: stone masonry. Proc. 5th intern. Conf. on Structural
the case of the Katholikon of Dafni Monastery. Proc. int. Analysis of Historical Constructions, New Delhi, India.
symp. on Studies on Historical Heritage, Antalya, Turkey,
16–21 September, 2007.
1033
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
N. Gattesco
Department of Architectural and Urban Design, University of Trieste, Italy
L. Macorini
Department of Civil and Environmental Engineering, University of Trieste, Italy
ABSTRACT: A simple technique to increase the in-plane stiffness of ancient wooden floors is herein proposed
and discussed. The technique consists of using some nail plates for connecting longitudinally the timber boards
so as to prevent the slip among them. In such a way a stiff horizontal diaphragm can be obtained, which is
able both to restrain the masonry walls against the out-of-plane displacements caused by seismic excitation
and to transfer horizontal forces to the shear walls. Such a technique is also characterized by high reversibility
and low invasiveness, as requested by the conservation requirements for monuments and historical buildings.
Experimental tests both on connection samples and on a real size strengthened floor were carried out. The
former allowed the load-slip relationship of the nail plate to be determined, whereas the latter enabled to check
the effectiveness of the proposed technique. The results evidence good in-plane stiffness for the strengthened
floor, which is more than one order of magnitude higher than that of the unstiffend floor.
1035
Figure 1. Wooden boards connected with nail plates.
2 STIFFENING TECHNIQUE
1036
24 technique, a simple numerical investigation was car-
ried out. In the numerical analyses the in-plane stiff-
20 ness of ancient wooden floors, strengthened with
different techniques, was assessed and compared. The
floor of an existing ancient masonry building (19th
16
Shear load [kN]
1037
where ρk is the wood density in kg/m3 and d is the
nail diameter. The density of the timber of the joists
and boards was assumed equal to 400 kg/m3 . Hence
the stiffness of the linear springs, representing the
shear stiffness of two nails 3 mm diameter, is equal to
Kh = 2.4 kN/mm. The stiffness of the rotational spring,
for a nail spacing a = 150 mm, is equal to
1038
5 SHEAR TEST ON A FULL SCALE FLOOR
1039
70 60
(a)
60 50
50 40
40 30
30 20
10
10
0
0
-10
-10
-20
-20
-30
-30
-40 -40
-50 -50
3.0 m from the bottom
-60 -60
-40 -30 -20 -10 0 10 20 30 40 50 60 70 -1.2 -0.8 -0.4 0 0.4 0.8 1.2
displacement [mm] slip [mm]
60
(b)
Figure 8. Load-displacement relationship of the specimen 50
tested. 40
30
the bottom part of the specimen (S0). Moreover three
20
Shear load [kN]
1040
Table 3. Comparison between experimental and numerical for the floor stiffened with the proposed technique
results. (Section 2).
Displacement Secant stiffness
6 CONCLUSIONS
Load Experimental Numerical Experimental Numerical
(kN) (mm) (mm) (N/mm) (N/mm)
In the research work a technique to increase the in-
15 3.85 7.20 3896 plane stiffness of the wooden floors of ancient masonry
25 8.33 12.00 3000 buildings is presented and discussed. This technique
2146
40 17.20 19.20 2325 aims at preventing the longitudinal slip among the
50 25.30 24.00 1976 boards of the deck by using some nail plates. The
technique is characterized by low invasiveness and
high reversibility, which are the requirements for the
conservation of historical buildings.
(Fig. 9b), 1.0 m (Fig. 9c). Even though all the curves Some numerical simulations were carried out on
have a quite similar shape, the curve plotted in Fig- the shear behaviour of a typical wooden floor, orig-
ure 9b is steeper than the others. This reveals a greater inal and stiffened through different techniques. The
stiffness of the connection at level 2.0 m. results achieved allowed both to point out the improved
However the average stiffness differs slightly from structural performance guaranteed by the proposed
the values achieved in the push-out tests presented in technique and to compare it with some other strength-
Section 3. ening solutions. The in-plane stiffness of the floors
strengthened through the proposed technique is nearly
5.4 Comparison with the numerical results 20 times greater than that of the unstiffened floor.
The experimental test carried out on a full scale floor
The experimental test was also simulated using a com- specimen (4.0 × 4.0 m) allowed the effective shear
mercial numerical procedure for structural analyses. stiffness to be determined. Some pre-existing cracks
The specimen was modelled as a grid of perpendicu- in the boards of the floor deck caused a progressive
lar beam elements, which simulate the timber joists, stiffness reduction because of the crack propagation
the boards and the perimetric steel elements. As in Sec- by increasing the load. Moreover the cracks caused a
tion 4, the connection between boards and joists (cou- premature collapse with the occurrence of complete
ple of nails) was modelled through linear springs char- shear longitudinal splitting of the boards, while the
acterized by translational stiffness equal to 2.4 kN/mm nail plate fasteners did not show any damage.
and rotational stiffness equal to 13.5 kNm. Finally the The numerical simulation of the experimental test
connection between boards (nail plates) was modelled evidenced a good agreement with experimental results
through shear springs having stiffness equal to the till the pre-existing crack propagation became signif-
average value obtained from tests (Section 3). icant. However the in-plane secant stiffness achieved
The displacements of the nodes at the bottom of the through the experimental test is more than one order
specimen were restrained. The load was applied at the of magnitude greater than the unstiffened floor theo-
top of the sample equally distributed among the nodes retical stiffness.
of the top section. Even though these are some early results of an
The numerical simulation allowed the in-plane stiff- ongoing research project, it can be concluded that
ness of the specimen to be assessed. In Table 3 the the proposed strengthening technique allows to ade-
results of the numerical simulation are compared to quately stiffen existing wooden floors. So they can
the experimental outcomes. In particular the displace- preclude dangerous out of plane deformations of the
ments and the secant stiffness relating to some values masonry walls at floor levels when the building is
of the applied load are reported. subjected to seismic actions.
The value of the numerical stiffness is in quite
good agreement with the experimental values of the
secant stiffness. Actually, for low values of the load ACKNOWLEDGEMENTS
the experimental stiffness are appreciably greater than
the numerical one because of some friction actions The study was supported by a grant for research
in the experimental system. For loads greater than projects of the Italian Government (PRIN 2006). The
50 kN, the damage occurred in the boards (propagation nail plates used for tests were kindly offered by
of existing cracks) caused a significant progressive Wolf System, Campo di Trens, Bolzano, Italy. The
reduction of the floor in-plane stiffness. Authors wish to thank Prof. Fausto Benussi for his
The theoretical stiffness of the unstrengthened floor useful advices. The testing staff (Dr. Franco Trevisan,
is equal to 128 N/mm, which is more that one order Dr. Andrea Cernigoi, Mr. Piergiulio Corazza) is grate-
of magnitude lower than that obtained in the test fully acknowledged.
1041
REFERENCES Giuriani, E. 2004. L’organizzazione degli impalcati per gli
edifici storici. L’Edilizia, Speciale Legno Strutturale,
EN 1995-1-1, 2004 Eurocode 5. Design of Timber Struc- n. 134, 30–43.
tures. Part 1-1: General Rules and Rules for Buildings, Giuriani, E. 2006. Rinforzo di solai in legno mediante soletta
CEN European Committee for Standardization, Bruxelles. collaborante di calcestruzzo. In L. Dezi & N. Gattesco
EN 12512, 2001 Timber Structures. – Test methods. – Cyclic (eds), Strutture Composte – Nuove costruzioni, recu-
testing of joints made with mechanical fasteners. CEN pero, ponti, International Centre for Mechanical Sciences,
European Committee for Standardization, Bruxelles. CISM, Udine, Italy, ISBN 88-85137-20-2, 311–322.
Gattesco, N. 2001. Experimental Study on Different Dowel Modena, C., Valluzzi, M.R., Garbin, E. & da Porto, F.
Techniques for Shear Transfer in Wood-Concrete Com- 2004. A strengthening technique for timber floors using
posite Beams. In A. Shing (ed.), Creative Systems traditional materials. Proceedings of the Fourth Inter-
in Structural and Construction Engineering: 487–492. national Seminar on Structural Analysis of Historical
Rotterdam: Balkema. Constructions, Padova, Italy, 10–13 Nov. 2004, 911–921.
Gattesco, N. & Del Piccolo M. 1997 Shear transfer between Piazza, M. & Turrini, G. 1983. Una tecnica di recupero statico
concrete members and stone masonry walls through dei solai in legno. Recuperare, Vol. 5, 6, 7, Milano.
driven dowels. European Earthquake Engineering, No. 3, Ronca, P., Gelfi, P. & Giuriani, E. 1991. The Behavior of
3–17. a Wood-Concrete Composite Beam under Cyclic and
Gattesco, N. & Macorini, L. 2006. Strengthening and stiff- Long Term Loads. Structural Repair and Maintenance
ening ancient wooden floors with flat steel profiles. In of Historical Buildings, 2nd International Conference
P.B. Lourenco, P. Roca, C. Modena & S. Agrawal (eds.), STREMAH 91, Seville, Spain, Vol. 1, 263–275.
Structural Analysis of Historical Constructions: 405–412. Venice Charter, 1964. International Charter for the Conser-
Macmillan India ltd, New Delhi. vation and Restoration of Monuments and Sites. Second
Gattesco, N., Macorini, L. & Benussi, F. 2007. Strengthening International Congress of Architects and Technicians of
of wooden floors for seismic rehabilitation of historical Historic Monuments.
buildings through high reversibility techniques, ANIDIS
2007, XII Convegno “L’Ingegneria Sismica in Italia”,
Pisa, 10–14 June 2007.
1042
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: Local vertical prestressing is considered as strengthening measure to reduce the vulnerability of
masonry structures in case of earthquake action. The shear strength and the ductility of masonry is improved. This
is shown in static cyclic tests. Static and dynamic behaviour is investigate for different masonry walls. To avoid an
excessive amount of experimental tests, this paper is focused on numerical simulations and impacts on the results
such as boundary conditions and different means to model prestressing. Previously, brief comparison of simula-
tion techniques and material models is given. The task is mainly completed by means of a suitable material model
for macro modelling of masonry. The researched parameters which influence the numerical results are discussed.
Case studies are carried out to investigate the parameters. The importance to model the tendons, especially for
nonlinear dynamic analyses, is shown. The impact of prestressing on the dynamic behaviour is investigated.
W1 W2 W3 W4
Figure 1. Complete experimental set-up for wall 1, wall 2, wall 3 and wall 4.
1043
For all walls, two tendons (strands) have been used of cracks. 2xP0 is the sum of the initial prestressing
for vertical prestressing. The complete experimental forces of two tendons. The dead load of the wall and
set-up is depicted in Figure 1 on the left for wall 1. upper stories is expressed by G+F. In the last column,
Furthermore, another position of the tendons closer to the forces in the tendons after reaching the ultimate
the middle was investigated in wall 2. The influence of loading point (2xP0 )u are given. A significant decrease
slenderness is taken into account by means of wall 3 was observed. Moreover, it is listed where stone failure
and wall 4. With wall 4, the floor slab is supported only (partial collapse of the wall) occurred, or if maxi-
on one end. The wall properties are listed in Table 1. mal displacement of the testing equipment (vmax ) was
A horizontal static cyclic displacement was applied reached.
in the centre of the concrete slab. The crack patterns
are shown in Figure 2. The diagonal crosses are very
typical of earthquake damage. The load displacement 3 NUMERICAL SIMULATION TECHNIQUES
curves are displayed in Figure 3. Wall 2 had the most AND MATERIAL MODELS
useful behaviour. The shear capacity and the ductil-
ity are very high. The area enclosed by the hysteresis 3.1 Theoretical background
shows the energy dissipation, which is very good for In the following, it is briefly discussed which sim-
wall 2 and wall 3. Measured values like the displace- ulation techniques and material models are suitable
ment u and the horizontal load H are summarised in to investigate the dynamic behaviour of prestressed
Table 2 for all walls. The index u means the ultimate masonry. In case of cyclic and dynamic loading,
point of loading, whereas cr indicates the occurrence a degradation of stiffness and strength occur. The
dynamic behaviour is effected strongly by the stiffness
Table 1. Properties of the walls. which is an important consideration.
Usually, the accurate micro modelling is applied
Distance Distance to investigate masonry in detail. The material model
of tendon of by Oliveira (2003) includes degradation for this
Length Height Thickness to edge tendons Support modelling strategy. It is based on the plasticity theory,
m m m m m described by Lourenço (1996) and Rots (1997). Due to
the high calculation effort, the material model is very
W1 2.5 0.25 2.00 on 2 time consuming in the case of dynamic simulations
W2 2.5 2.5 0.175 0.625 1.25 sides of large structures. More suitable for such simulations
W3 2.5 0.25 0.75
is the material model of Gambarotta & Lagomarsino
W4 2.5 0.25 0.75 on 1 side
(1997), which is based on fracture mechanics and
1044
macro modelling. Also, this model is able to describe detail in Ötes & Löring (2006) as well as other liter-
the post-peak behaviour and the dynamic behaviour. ature. Two extreme cases are given. For case BC 1,
Damage modelling is used in which the inelastic the top of the wall is constrained, because of that it
strains are described by means of two internal damage stays horizontal. Mainly shear loading occurs. For case
variables which express the damage evolution in the BC 2, the top of the wall is free and can rotate. Thus,
bricks and mortar. The model of Gambarotta & Lago- the wall behaves like a cantilever, and bending loading
marsino (1997) works efficiently also in the case of occurs. Below it is designated as BC 2. Figure 4 shows
dynamic nonlinear analyses and probabilistic analyses walls with these boundary conditions and shapes.
of large scale masonry structures. In reality, the support of the wall is between these
extreme cases, depending on the behaviour of the floor
slabs. To predict the behaviour realistically, beams
4 IMPACTS ON NUMERICAL RESULTS
should be modelled, as depicted in Figure 4 in middle.
Below, the impacts are explained in more detail.
It is important to know which parameters have a
significant influence on the numerical results. The
findings of a literature review and of simulations 4.2 Case study of static analyses
made for this contribution are discussed as follows.
To gain deeper insight, a case study was performed in
Due to the great impact of boundary conditions, the
which the following parameters were varied to inves-
different conditions are separately explained in the
tigate their impacts. Four variations of slenderness
next subchapter.
S (0.5, 1, 2, 3) are made. For the different slender-
ness values, different heights of 1.25 m, 2.5 m, 5 m
and 7.5 m are obtained. The boundary conditions on
4.1 Boundary conditions of shear walls the top of the wall BC 1 (constrained), modelled by
means of a ridge L-framework, and BC 2 (free) are
The boundary conditions influence the behaviour and used. Furthermore, for BC 2 walls with tendons close
failure mechanisms strongly. This is explained in more to the middle are investigated. Two means to model
prestressing are applied (external forces and tendons),
Table 2. Loading and results of the experimental tests. illustrated in Figure 5.
Its outcome are the impacts on: the change of pre-
2xP0 G+F Hcr Ucr Hu Uu Type (2xP0)u stressing forces in the tendons, the restoring forces,
kN kN kN mm kN mm of fail. kN the rotation of the top, the shear capacity, the ductil-
ity as well as the suitability of the tendons location.
W1 360 245 +217 3 +109 17 vmax 260 The subsequent values are fixed for all variations of
−227 −105 the models: the width of the walls is 2.5 m, thick-
W2 356 245 +223 6 +89 23 Stone 128
ness 0.175 m, prestressing force per each tendon is
−220 −94
189 kN, and vertical load of upper stories is 197.3 kN.
W3 352 110 +110 7 +60 23 vmax 260
Furthermore, the material parameters are the same for
−114 −60
W4 274 110 +101 11 +34 20 Stone 140
all models as listed in Table 3.
−100 −16 Below, some important results of this case study
are briefly summarised. In general, the means to sim-
1045
ulate prestressing is important for BC 2, especially increase in the tendons on the upper corner. In the case
when the tendons are close to the edges. For BC 1 this of low walls, the top rotation leads to significant dif-
phenomenon can often be neglected. Restoring forces ferences in the prestressing forces. These differences
occur and can be simulated when the prestressing is decrease as the walls become higher as illustrated in
modelled by means of tendons. The restoring forces Figure 6. The reason is simple. An equal change of
have to be divided into horizontal and vertical compo- length of a long and short bar gives high stresses in
nents. The last one is important only for BC 2. Here, the smaller bar, but only small stresses in the longer.
the vertical movement of the corners during the top For models with tendons close to the edges, the rota-
rotation leads to changes of tendons length. The result tion of the top edge is smaller, but the tendons have to
is a variation of the prestressing forces in the tendons, be modelled to notice this effect. If only external forces
which decrease in the tendons on the lower corner and are used to model prestressing, no significant differ-
Figure 5. Investigated walls in dependency of the slenderness, boundary conditions, means to model the prestressing and
position of tendons.
1046
ence can be observed. Of course, the shear capacity and higher than 5 mm. Figure 8 depicts greater stiffness
the ductility depend decisively on the slenderness and and shear capacities for walls with BC 1 than for
boundary conditions, as already noted and described BC 2, but smaller ductilities. Especially for slender
above. walls the differences in ductility are significant. It is
To get information about the shear capacities and found in case of BC 2, when the tendons are modelled,
ductilities it is necessary to apply a horizontal loading the shear capacity is higher (Fig. 7 left) and the duc-
tility is a bit smaller. This means of modelling leads
Table 3. Material parameters used for the case study. to smaller horizontal displacement. Figure 7 left also
shows higher stiffness for walls modelled with ten-
Symbol Variable Value dons. If the tendons are placed close to the middle,
no significant difference can be observed, as shown
µ Friction coefficient 0.8 in Figure 7 right. The post-peak behaviour of mod-
σmr Tensile strength 0.15 N/mm2 els with tendons near the edges is more useful, where
of mortar joints
higher forces can be applied. Obviously, the tendons
τmr Shear strength of 0.20 N/mm2
mortar joints
carry tensile loads after tensile failure has occurred in
σbr Compressive 3.5 N/mm2 the masonry wall. If the tendons are close to the mid-
strength of masonry dle, only a small improvement regarding the post-peak
τbr Shear strength 1.5 N/mm2 behaviour can be observed. A comparison of Figure 7
of masonry left and right shows this. The findings of this case
η Poissons ratio 0.1 study regarding the impact of slenderness cannot be
ρ Density 2000 kg/m3 generalised since the width of the walls is constant.The
E Young’s modulus 2000 N/mm2 height of the wall has an important influence, because
βm Softening mortar 0.6 it is equal to the basic length of the tendons. Small
βb Softening brick 0.4 changes of length lead to large differences of the forces
cmt Inelastic deformation 1.0
parameter for mortar
inside short tendons. For long tendons, much higher
cb t Inelastic deformation 1.0 differences in length are necessary to reach significant
parameter for brick changes of such inner forces. As shown, many reasons
exist to model tendons, such as the tendon forces that
Figure 6. Forces in tendons in dependency on the horizontal top displacement of walls for BC 2 and tendons close to the
edges.
1047
change during static horizontal loading. This impact was observed in experimental tests (Budelmann et al.
is significant for BC 2 and compact walls. The shear 2004), probably caused by the reduction in height of
capacity depends considerably on the vertical loads. bed joints due to slipping in the joints. The static sim-
Another important reason for modelling the tendons, ulations show horizontal and vertical restoring forces
is the decreasing of the prestressing force in the ten- lead to smaller horizontal displacement and smaller
dons during static cyclic and seismic loading. This rotations as well as to increased stiffness (Fig. 7 left).
Figure 7. Left: Horizontal load displacement diagram for BC 2 and tendons close to the edges, right: horizontal load
displacement diagram for BC 2 and tendons close to the middle.
Figure 8. Horizontal load displacement diagram for models with tendons for BC 1 and for BC 2 for different tendon positions
(edges and middle).
1048
4.3 Case Study of nonlinear dynamic analyses vertical load of 197.3 kN. The variations ‘external
forces’ and ‘tendons’ have an additional prestress-
The increasing of stiffness, as mentioned above, affect
ing force of 189 kN per each tendon. The results
the dynamic behaviour. A small case study was per-
for the vibration behaviour regarding the horizon-
formed to investigate further. The vibration behaviour
tal top displacement are depicted in Figure 9. For
of prestressed walls modelled with external forces,
all of the wall models and load levels, the biggest
tendons, and for walls without prestressing, is cal-
value for the roof displacement is reached after 0.12 s.
culated in nonlinear dynamic simulations. A ground
The highest displacement reaches 34 mm for the non-
displacement is applied as an impulse for all men-
prestressed wall. As expected prestressing leads to a
tioned simulations below. The time for the impulse
reduction of the vibration amplitude. For the mod-
is 0.12 s to move ground and return back to the orig-
els with external forces the displacement amounts to
inal position. The three different load functions are
32 mm and for the models with tendons 32.4 mm. The
time dependent as identified in Figure 9 with dashed
observed difference between the maximum horizontal
lines. Table 4 gives a short overview of the carried
roof displacement of ‘tendons’ and ‘external forces’
out dynamic analyses, the level of the applied dis-
are smaller for lower load levels.
placement, and whether it was possible to receive
The horizontal displacement of the prestressed
convergence.
walls is insignificantly smaller than the prestressed
For these models the same values for width, thick-
walls for lower vertical load levels (5 mm and 10 mm).
ness and material parameters (Table 3) are used as
Nevertheless, the mortar damage is much less for
for the static case study above. All have a static
prestressed walls, than for the non-prestressed wall
(Fig. 10). This indicates that vertical prestressing can
be a useful strengthening measure also for dynamic
Table 4. Overview of the carried out analyses with impulse
loading. More detailed investigations are still in
loading.
progress.
Means to model prestressing The dynamic behaviour is different for all consid-
ered walls, as periods vary significantly.The walls with
Impulse External Without tendons vibrates faster. This means they are stiffer.
displacement forces Tendons prestressing Observations of experimental tests lead to a probable
reason. Ötes et al. (2002) observe that in the range
5 mm Yes Yes Yes of high horizontal loading, the stiffness of the wall
10 mm Yes Yes Yes
17 mm Yes Yes Yes
is mainly affected due to the spring properties of the
tendons after the occurrence of gaping joints. This
Figure 9. Horizontal displacement of the top time-dependent for modelling of prestressing by means of external forces,
tendons, and without prestressing in case of BC 2.
1049
Figure 10. Mortar damage for impulse loading of 10 mm, left: external forces, middle: tendons, right: without prestressing.
Figure 11. Mortar damage for impulse loading of 17 mm, left: external forces, middle: tendons, right: without prestressing.
would explain also the bigger difference between ‘ten- the walls leading to smaller stiffness, and lower fre-
dons’ and ‘external forces’ in case of higher shaking quencies. For a higher load level (horizontal ground
levels. Also, presented static simulations show that displacement of 17 mm) also the prestressed walls are
the post-peak behaviour varies with the prestressing significantly damaged (Fig. 11). Here, the difference
is modelled as ‘external forces’ or ‘tendons’ (Fig. 7 between the two means to model prestressing become
left). Modelling of tendons leads to higher resistances. larger, but not important. In all these simulations the
In general, the walls vibrates slower, when the ground mortar damage, as well as the brick damage, is a bit
acceleration is higher. A reason is higher damage of higher for tendons, than for external forces.
1050
5 CONCLUSIONS großformatigem MW mit hohem Erdbebenwiderstand,
Abschlussbericht, iBMB der TU Braunschweig, Germany
The static cyclic experimental tests on internal pre- Gambarotta, L. & Lagomarsino, S. 1997. Damage Models for
stressed shear walls indicate the functionality of the seismic response of brick masonry shear walls, Part I
and II, Earthquake Engineering and Structural Dynamics,
this method to strengthen masonry against earth-
Vol. 26, 1997: 423–462
quake loading. Useful material models were discussed Lourenço, P.B. 1996. Computational strategies for masonry
and used for numerical investigations. Many rea- structures, Ph.D. Thesis, Delft University of Technology,
sons to model the tendons were pointed out, e.g. Netherlands
the restoring force and the impact on the dynamic Oliveira, D.V. 2003. Experimental and numerical analysis
behaviour. Deeper investigations especially regarding of blocky masonry structures under cyclic loading, Ph.D.
the dynamic behaviour of prestressed walls are recom- Thesis, University of Minho, Portugal
mended. This has been done by means of first dynam- Ötes, A., Löring, S. & Elsch, B. 2002. Erhöhung der Schub-
ical nonlinear simulations. The mentioned important tragfähigkeit von KS-Wänden unter Erdbebenlasten durch
schlaf bewehrte Beton-stützen in Formsteinen bzw. durch
factors have to be considered carefully in further
Vorspann-ung der Wand, Forschungsvereinigung Kalk-
simulations. Sand e.V.
Ötes, A. & Löring, S. 2006. Zum Tragverhalten von Mauerw-
erksbauten unter Erdbebenbelastung, Bautechnik, No. 83,
ACKNOWLEDGEMENTS Heft 2, 2006: 125–138
Rots, J.G. 1997. Structural Masonry: An Experimental
We would like to thank Prof. Lagomarsino and /Numerical Basis for Practical Design Rules, Balkema,
Dr. Calderini from the Università degli Studi di Genova Rotterdam, Netherlands, ISBN 90 5410 680 8
for their permission and support in including their Verfmeltfoort, A.Th. & Raijmakers, T.M.J. 1993. Deforma-
tion controlled meso shear tests on masonry piers, Part 2,
material model into our research. Also we have to
Draft report, TU Eindhoven, dept. BKO, Netherlands.
express deepest gratitude to Prof. Bartoli from the
Università degli Studi di Firenze.
REFERENCES
Budelmann, H., Gunkler, E., Husemann, U. & Becke, A.
2004. Rationell hergestellte Wände aus vorgespanntem
1051
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
R. Keersmaekers
Department of Architecture, Provincial University College Limburg, Hasselt, Belgium
Department of Civil Engineering, Building Materials Division, Katholieke Universiteit
Leuven, Belgium
D. Posen
Department of Architecture, Provincial University College Limburg, Hasselt, Belgium
ABSTRACT: A geophysical resistivity measuring device was modified to perform automatic monitoring of
historical masonry structures before, during and after grout injection for consolidation purposes. The obtained
image is called a geo-electrical tomography. The technique was used to evaluate the deteriorated masonry of the
recently partly collapsed Maagdentoren in Zichem, Belgium. The results of these measurements are discussed.
An important improvement is the adaptation of the “cross-borehole” measurement technique for masonry. With
this technique, a constant resolution in depth is achieved. Laboratory tests, using an adaptation of the “cross-
borehole” technique, enabled to show the resistivity distribution changing during the injection of a test specimen
with a hydraulic grout.
1 INTRODUCTION GENERAL PURPOSE OF matrix, which points at the correct grout mixture and
THE RESEARCH PROJECT evaluation sequence, given a specific grouting prob-
lem and boundary conditions. This structuring forms
With a grant of the Flemish government (IWT-Institute the first part of the research project. Specific expe-
for the Promotion of Innovation by Science and Tech- riences with grout injections of a group of Belgian
nology in Flanders) and the support of a group of companies involved in restoration activities will be
Belgian companies involved in restoration activities, studied, and expert aspects will be derived out of these
a research project started in September 2005, aim- experiences.
ing to improve control methods for grout injections
in masonry. The project was developed within a part-
nership between the Provincial University College 1.2 Integration of non-destructive techniques
Limburg, Department of Architecture, Hasselt (B) and
the Katholieke Universiteit Leuven, Department of Secondly the implementation of non-destructive tech-
Civil Engineering, Building Materials Division (B). niques in the field of grout injections will be stud-
The main objectives of the project are: ied, not only for the evaluation of the quality of
the deteriorated or injected massive (masonry or
concrete) before and after injection. The aim is
1.1 Inventory of knowledge
to develop the NDT-technology that should enable
Due to the wide spectrum of possible applications of on-line surveying of grout injections. The bene-
grout injections, a lot of practical experience is present, fit would be the possibility to alter the injection
but this knowledge is partitioned, spread throughout parameters during the injection itself so that the
the building practice and not readily accessible. The desired injection result can be obtained. Geo-electrical
knowledge of injection techniques is fragmented and survey of masonry already proved to be a reli-
the structured inventory of all this knowledge is there- able non-destructive tool for this purpose (Janssens
fore very important. This classification will not only 1993; Keersmaekers 2003; Keersmaekers et al. 2004;
be the summary of existing techniques in the build- Keersmaekers et al. May 2006; Keersmaekers et al.
ing practice, but it will also be a well-argued decision Nov 2006; Marchisio et al. 2002; Van Rickstal and
1053
Vanhellemont 2002; Van Rickstal et al. 2003;
Van Rickstal et al. 2006; Venderickx 2000).
2 BASICS OF GEO-ELECTRICAL
MEASUREMENTS point, centrally located between the four electrodes
and on a depth equal to the average depth of the
In the past, geo-electrical resistivity measure- influenced zone of the potential field. Note that the
ments were already used successfully in restoration second row of measurements (with an electrode dis-
projects by the Reyntjens Laboratory (Janssens 1993; tance “2a”) has only 14 (20–2 × 3) measurements. The
Venderickx 2000). The interpretation of the geo- number of measurements decreases with augmenting
electrical measurements was based on the apparent electrode spacing (i.e. higher penetration depth). In
resistivity value. The word “apparent resistivity” is this way apparent resistivity values are gathered over
used because this value is the resistivity value that the depth of the masonry structure and a “pseudo-
would hold for a fully homogeneous and isotropic section”, which is the graphical representation of the
material. In reality this is not the case. Walls have a measured apparent resistivity values, is built.
heterogeneous resistivity distribution. Unfortunately, the 2D pseudo-section does not
In the recent past new developments were made in enable direct interpretation of the apparent resistiv-
geophysics within the field of geo-electrical survey of ity values. No conclusions about the real resistivity
soils. This was possible thanks to the growing effec- distribution in the substrate (internal structure of the
tiveness and power of numerical methods. Software masonry) can be drawn based on apparent resistiv-
was developed which can invert the measured apparent ity values. To build the real resistivity distribution
resistivity values to a distribution of absolute resistiv- it is necessary to numerically invert or transform
ity values of the subsoil. (Dey and Morrison 1979; the pseudo-section values into real resistivity val-
Geotomo 2002; Lines and Treitel 1984; Loke and ues. Forward modelling enables the calculation of the
Dahlin 1997; Loke 2002) The resistivity measurement apparent resistivity values (pseudo-section), starting
remains the same, but the sequence is different and from a given resistivity distribution. The idea is to
the number of measurements is larger than previously construct a numerical model with a resistivity distri-
common on masonry structures. bution whose calculated pseudo-section corresponds
The technique is known in geophysics as geo- with the measured pseudo-section. The section of the
electrical tomography. Figure 1 shows the necessary wall is therefore numerically divided into blocks or
measurements for a Wenner-alfa electrode-configu- elements, where every element is given a resistiv-
ration with 20 electrodes. Each of these electrodes, ity value. A least square approximation between the
mostly stainless steel nails, can be used to inject cur- calculated and measured pseudo-sections determines
rent (current electrodes C1, C2) or to measure potential how the model parameters (i.e. the resistivity values
(potential electrodes P1, P2). Conducting the first of the model blocks) must change in the next iteration,
row of measurements, a distance “a” (n = 1) between resulting in a better correlation between calculated and
C1P1, P1P2, P2C2 is kept constant, leading to 17 measured pseudo-sections.
(20–3) measurements on this particular row. Raising In practice, a measuring campaign starts by posi-
the electrode distance to “2a” (n = 2) gives a bigger tioning the electrodes. Stainless steel nails are used
influenced zone of the masonry (i.e. more profound for masonry structures. The electrodes are connected
penetration of the potential field), resulting in an to the automatic scanning module which switches
apparent resistivity value representative for a larger the nails as current or potential electrode. In this
and deeper zone of the masonry. Conventionally, this research project, the geo-electrical measuring device
apparent resistivity value is allocated to a physical Terrameter SAS1000 was used.
1054
partially placed over a visually good looking part of the
wall and a bad looking zone. This visually bad looking
zone is caused by the demolition of an historic defence
platform that was situated there (Figure 3).
1055
Virgins” is seriously deteriorated for a significant part 4 LABORATORY TESTS: CROSS-BOREHOLE
of its thickness. Keeping in mind the collapse on
the first of June 2006, a consolidation injection with Geo-electrical tomographies, as described above, have
an appropriate hydraulic grout is urgently needed to the disadvantage that the resolution decreases with the
preserve this monument for the next generations. penetration depth of the measurements. Nevertheless,
Figure 3. Position of the survey line on the tower; Insert: detail with in white the position of the electrodes and in black the
position of the historic defence platform, removed in an unknown past.
Figure 4. Above: position of the electrodes on the wall (white line); Below: inversion result of dipole-dipole measurements
and the location of the boreholes for the endoscopic survey.
1056
very interesting conclusions can be drawn based on is detected as a high resistivity within the graphical
these tomographies as was clearly shown in the case representation of the resistivity distribution.
study above. In the research project, another exist-
ing technique was adapted for masonry, based on a
two sided approach of the wall, whereby on both 5 LABORATORY TESTS: ON-LINE
sides electrodes are placed. In geophysics this tech- MONITORING OF A GROUT INJECTION
nique is referred to as “cross-borehole”. Theoretical
research and numerical simulations showed that this The cross-borehole technique seemed suitable for on-
approach leads to a constant resolution over the depth line follow up of hydraulic grout injections.This means
of the wall (Bernaerts and Moens, 2006; Stroobant and that it can be used during injections in a way that
Verkest, 2007). Figure 5 shows the result of a success- an image can be constructed showing the grout flow
ful cross-borehole measurement, without going into in the masonry. This would be a useful appliance to
the technical aspects. A replica of a three leaf masonry evaluate the effectiveness of the injection. Also the
wall was built in the laboratory, containing a piece of detection of grout leaching in unwanted directions
insulation simulating a void in the wall. The inversion and the incomplete filling of voids could be detected
result (Figure 5 Right) clearly shows that the insulation in real time.
To accomplish this, a customized sophisticated data
logger had to be built. This was necessary because the
system must be capable of reading, filtering and pro-
cessing the data fast enough to evaluate the injection
in real time. A Labview-based (National Instruments)
system was built using 48 analogue input channels
which can be sampled parallel with each other. The
principle is as follows. On one side of the wall 3 parallel
rows of 16 current electrodes are placed and connected
with the automatic switching module. This is the active
side of the wall. Likewise, on the backside of the wall
3 parallel rows of potential electrodes are placed and
connected to the data logger. This is called the pas-
sive side and here the response to the injected currents
is measured. In this way it is possible to measure 9
Figure 5. (a) Position of insulation as a simulation of a cross borehole tomographies in 90 to 120 seconds (see
void; (b) Position of the electrodes; (c) Inversion result of the Figure 6: Left). After inversion the nine 2D-images
cross-borehole survey, with the high resistivity at location of are combined to a single 3D-image of the resistivity
insulation block. distribution (Stroobant and Verkest, 2007).
Figure 6. (a) 9 cross-borehole tomographies used to construct a 3D image of the inversion; (b) 3 × 16 current electrodes, the
potential electrodes (3 × 16) at the backside; (c) Position of the volumes of gravel to be injected and the result of the inversion
3D-inversion result at the end of the injection; (d) Injected volumes, excavated after hardening of the grout.
1057
A laboratory test was built in order to test new a reference measurement before injection, the contrast
procedures for on-line follow up of grout injections. is caused by an external influence (the injected grout).
Three zones with high resistivity (representing dete- This definitely simplifies the interpretation, because
riorated, little cohesive masonry) were placed in a the difference in resistivity can only be caused by the
volume of sand. This was done by inserting three added grout.
volumes of tennis court gravel (different sizes) in
the sand volume (Figure 6: Middle). Gravel consists
of crushed bricks, an easily injectable material. A 6 CONCLUSIONS
reference measurement was done before the injec-
tion started. Then, every gravel volume was gradually With a grant of the Flemish Government (IWT)
injected with a hydraulic grout. Simultaneously, the set and the support of a group of Belgian companies
of 9 cross borehole tomographies was measured every involved in restoration activities, a research project
two minutes. Drawing of the relative difference charts, was started in September 2005, aiming to control
which gives graphic representation of the difference in grout injections in masonry in a more reliable way.
resistivity between two subsequent 3D-tomographies, One of the objectives was the implementation of non-
clearly shows the gradual filling of the different gravel destructive techniques in the field of grout injections.
volumes. The good electrical conductivity of the grout The project showed the feasibility of performing geo-
leads to local resistivity drops creating a big contrast electrical tomographies (one side approach and two
with the former high resistivity of the gravel. This sided approach by means of cross-borehole) for the
provides an excellent condition to detect the injected evaluation of the quality of deteriorated or injected
area on a relative difference chart. Figure 6 (Middle masonry before and after injection. An adapted new
right) gives the 3D relative difference chart between NDT-technology has been developed for on-line sur-
the last measurement (at the end of the injection of the veying of grout injections.The benefit is the possibility
three volumes) and the reference measurement before to adapt the injection parameters during the injection
injection. Notice how clearly visible the injected zones so that the desired injection result can be obtained.
appear on the image. The three volumes were exca-
vated after hardening of the grout (Figure 6: Right).
A good correlation between these volumes and the ACKNOWLEDGMENTS
inversion result was observed.
Drawing of the relative difference charts, which The authors thank the master thesis students
gives graphic representation of the difference in K. Bernaerts, M. Moens, D. Verkest and W. Stroobant,
resistivity between two subsequent 3D-tomographies, for their contribution to the discussed research pro-
clearly shows the gradual filling of the different gravel gram and their work in the laboratory.
volumes. The good electrical conductivity of the grout The grant of the Institute for the Promotion of Inno-
leads to local resistivity drops creating a big contrast vation by Science and Technology in Flanders (IWT)
with the former high resistivity of the gravel. This and the support of the Belgian companies to the project
provides an excellent condition to detect the injected are gratefully acknowledged.
area on a relative difference chart. Figure 6 (Middle
right) gives the 3D relative difference chart between
the last measurement (at the end of the injection of the REFERENCES
three volumes) and the reference measurement before
injection. Notice how clearly visible the injected zones Bernaerts K., Moens M., 2006 Onderzoek van geo-
elektrische meettechnieken voor niet destructief onder-
appear on the image. The three volumes were exca-
zoek van historische metselwerkmassieven. Master
vated after hardening of the grout (Figure 6: Right). Thesis. KULeuven. Department of Civil Engineering.
A good correlation between these volumes and the Leuven
inversion result was observed. Dey A., Morrison H.F., 1979a, Resistivity modelling for
The above technique seems very suitable for on- arbitrary shaped two-dimensional structures. Geophysical
line follow up of grout injections. Conventional geo- Prospecting 27.
electrical tomographies (like in the case of the Tower Goetomo., 2002. Manual of the software packages
of the Virgins, or cross-borehole), used to determine Res2Dmod and Res2Dinv version 3.5. Download from
the internal structure of the masonry, can not always be http://www.geoelectrical.com/.
Janssens H. 1993., Geo-elektrische kontrole van consolidatie-
interpreted unambiguously.Therefore, the conclusions
injectie bij metselwerk. PhD thesis. KULeuven.
based on conventional topographies should always be Keersmaekers R. 2003. De geo-elektrische methode toegepast
verified with destructive tests, like endoscopic survey op metselwerkstructuren: implementatie van recente
or core drilling. This disadvantage disappears when the ontwikkelingen. Master thesis. KULeuven. Department of
technique is used for on-line follow up of grout injec- Civil Engineering. Download http://www.kuleuven.ac.be/
tions. When composing a difference chart, relative to bwk/materials/Publications/.
1058
Keersmaekers R., Van Rickstal F., Van Gemert D., Geo- applied on the cathedral of Lucca. Proceedings of the 8th
electrical techniques as a non-destructive appliance for Meeting EEGS – ES. Aveiro.
restoration purposes. Proceedings of 4th international Oost T., Geyskens L., 1982. Overzicht van de opgravingen
seminar on structural analysis of historical constructions – 1981–1982 aan de Maagdentoren te Zichem.
2004. Padova. Italy. Stroobant W., Verkest D., 2007. Verfijning en afijking van
Keersmaekers R., Schueremans L., Van Rickstal F., Van geo-elektrische kaarten voor NDT van monumentaal met-
Gemert D., 2006. Development of an appropriate grout selwerk. Master Thesis. KULeuven. Department of Civil
for the consolidation of the column foundations in Our Engineering. Leuven.
Lady’s Basilica at Tongeren (Belgium). Proceedings of Van Rickstal F., Vanhellemont Y., 2002. Niet-destructief
the 8th CANMET/ACI international conference on recent opzoeken van scheuren in metselwerk en pleisters.
advances in concrete technology –Montreal. Canada. Technieken en case studies. Wetenschappelijk-technische
Keersmaekers R., Schueremans L., Van Rickstal F., Van groep voor aanbevelingen inzake bouwrenovatie en mon-
Gemert D., Knapen M., Posen D., 2006. NDT-control of umentenzorg. Studiedag. Academiezaal Sint-Truiden.
injection of an appropriate grout mixture for the consolida- Van Rickstal F., Keersmaekers R., Van Gemert D., 2003.
tion of the columns foundations of Our Lady’s Basilica at Geo-electrical investigation of masonry walls: develop-
Tongeren (Belgium). Proceedings of the 5th international ments and case studies. Department of Civil Engineering.
seminar on structural analysis of historical constructions – KULeuven. Belgium. Proceedings of the 6th meeting
New Delhi. India. of the Materials Science and Restoration Society –
Lines L.R., Treitel A., 1984. A review of least-squares Karlsruhe. Germany.
inversion and its application to geophysical problems. Van Rickstal F., Keersmaekers R., Van Gemert D., 2006.
Geophysical Prospecting 32. Consolidation of ancient masonry. assessment. develop-
Loke M.H., Dahlin T., 1997. A combined Gauss-Newton ments in grouting. 22. Internationales WTA Kolloquium
inversion method for the interpretation of apparent resis- Bauinstandsetzen-Bauphysik. HAWK. Hildesheim. WTA
tivity pseudo-sections. Proceedings of the 3rd meeting of Almanach ‘Restoration and Building-Physics’.
the Environmental and Engineering Geophysics Society – ED. J. Gaenssmantel. ISBN 3-937066-05-5.
European Section. Aarhus. Denmark. Venderickx K., Evaluatie van geo-elektrische metingen op
Marchisio M. et al., 2002. Non-destructive testing on metselwerkstructuren. PhD thesis. KULeuven juni 2000.
masonry structures: a series of different methodologies
1059
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M. Mezzi
Department of Civil and Environmental Engineering, University of Perugia, Perugia, Italy
M. Marzullo
Studio Arch. Marzullo, Terracina, Italy
G. Valletta
Department of Construction, University of the Republic of Uruguay, Montevideo, Uruguay
ABSTRACT: The project concerns the restoration of a ruined conventual building of the 18th century in
Southern Italy, aimed at the insertion of a school complex. A new project was carried out providing for the
building of new steel and wooden light structures inserted within the fields defined by the in-plan alignments
of the existing walls. The steel structure foundations allow for a minimum interference with the expected
underground finding of graves and burials. The masonry walls will be strengthened by mortar injection and tied
by the insertion of twin steel beams at the floor levels, joined across the wall thickness by means of passing-
through injected bars. A particular solution has been conceived based on the synergetic interaction between
new and existing structure. Special connection systems between new structures and old masonry walls allow
for a triple function: avoiding the interaction in the vertical load bearing capacity, avoiding collapse mechanism
orthogonal to the wall plane through the horizontal bracing effect of the steel frame, using the horizontal load
bearing capacity of the retrofitted and strengthened walls in their plane. The adopted structural and constructive
solutions are presented. The restoration works are being currently carried out.
1061
not have univocal pre-defined solutions, but adapts
itself, case by case, recognising the knowledge (data
from archives, analyses of the materials in situ, direct
investigations) and the special and unique nature of
the single building.
1062
Figure 3. South view of the complex.
1063
di Diano” National Park. Within this project a num-
ber of different initiatives are provided, supporting a
unique territorial complex in which the protection of
the environment and the social and economic devel-
opment of the activities related to the development
of the park will harmonically coexist. Two areas were
pointed out for pursuing the project goals: one, related
to the urban activities, consists of an existing build-
ing previously hosting the Rangers’ barracks; another
one consists of the ruins of the so called Monastery
of Salemme - Santa Maria delle Stelle, located at the
slopes of the park’s mountains. Figure 8. Initial project. Plan of buildings B and C.
The first initiative of the project to be put into effect
concerns the construction of the Centre for the Envi-
ronmental Education, composed by a school of jour-
nalism and by a number of information and learning
activities, in the ancient Monastery of Salemme. The
recovery of the ruins will allow for an information desk
of the visitors, a didactic approach before the visits,
a formation school of environmental journalism. The
project solution hypothesizes a use of the Monastery
even apart from the ordinary institutional activities:
the didactic laboratory will allow holiday visits to the
complex also to the inhabitants of the territory around, Figure 9. Initial project. Plan of building A.
a small restaurant will allow a more comfortable stay-
ing. Moreover the complex will include a new building
(called C) devoted to a conference hall, of about 100
places, promoting the fruition of the building over the
use as school and information centre.
1064
by the superintendence authority on the historical retrofitted and strengthened walls in their plane. Suit-
heritage. able technological solutions have been adopted allow-
ing for the compatibility with the existing materials
while evidencing the separation with the existent.
5 CONCEPT SOLUTION OF THE FINAL
RETROFITTING DESIGN
6 DETAILS OF THE DESIGNED SOLUTION
A completely new concept solution has been con-
ceived. The building restoration provides for a number
6.1 Consolidation and strengthening of stone
of works allowing for obtaining a significant increase
masonry
of the static and seismic safety of the building con-
sidering the present situation of failure and decay. The following works are provided for the consolida-
The typical provided works for each class of statical tion of the existing stone masonry walls and their
problems are presented in a following paragraph. strengthening.
The seismic enhancement of the existing structures Demolishing of unstable portions. The situation
has taken into account both the Codice dei Beni Cul- concerns some vertical cantilevered walls and some
turali (2005) and the Linee Guida (2006), that is the unstable architraves.
most recent Italian guidelines concerning the preser- Deep refilling of mortar joints. This work concerns
vation of the cultural heritage, with particular attention all the walls in a generalised way and consists of taking
to the constructions located in seismic areas. The pri- away and restoring the mortar of the external portion
mary goal offered by the provided works consists of of the joints among the stone blocks.
the reduction or elimination of the main vulnerability Injection of the elevation walls. All the walls are
causes and can be resumed in the following items: concerned by a lack of link and compactness in the
depth. To overcome the lack of mortar, a generalized
• increasing the strength of the existing stone masonry
injection with fluid no-shrinkage lime mortar is pro-
walls;
vided. Ultrasonic tests will be carried out before and
• tying the building in correspondence to the levels
after the injection for evaluating its effectiveness in
of intermediate floors and roof;
increasing the compactness.
• conferring a suitable flexural strength to the
Rebuilding/Adjusting the top portions and ruins.
architraves.
The top portions are rebuilt with masonry having
The seismic enhancement is completed thanks to the same characteristics (blocks and texture) of the
the interaction of the enhanced old structures with the existing one. The insertion of reinforcing steel bars
new ones that guarantees: is provided, so that the top portions will include a
reinforced masonry tying beams.
• the behaviour of the floors as horizontal diaphragms;
Drilled steel connectors. Many L (angular) and
• the sustain against the overturning mechanisms (1st
T (intermediate) connections among walls are com-
mode mechanisms in general);
pletely or partially lacking in continuity and do not
• the reduction of the wall length included between
give an effective solidarity and connection. In all these
successive orthogonal walls;
cases is provided the tying of the orthogonal walls
• the effective connections between walls and floors.
by means of bars inserted within holes drilled in the
A particular solution has been conceived based on masonry and injected with the same mortar used for
the synergetic interaction between new and existing the masonry injection.
structure. The new project was carried out provid- Existing foundation strengthening. The existing
ing for the building of new steel and wooden light masonry foundations will be strengthened by means
structures to be inserted within the fields defined by of deep injection, slightly reinforced, performed in the
the in-plan alignments of the existing walls. The steel foundation and bottom part of the masonry (Figure 11).
structure foundations allow for a minimum interfer- Top tying beam in reinforced masonry. The actual
ence with the expected underground finding of graves situation is characterised by the lacking of longitudinal
and burials. The masonry walls will be strengthened link at the top of the masonry walls. Moreover, the
by mortar injection and by the insertion of twin steel most part of the walls are completely lacking of the top
beams at the floor levels, across the wall thickness portion, already collapsed.A reinforced masonry tying
by means of passing-through injected bars. Special beams is provided at the top of all the walls (Figure 12).
connection systems between new structures and old Intermediate steel tying beams. At the levels of the
masonry walls allow for a triple function: to avoid intermediate floors of the internal new structure lon-
the interaction in the vertical load bearing capacity, gitudinal steel tying beams will be built consisting of
to avoid collapse mechanism orthogonal to the wall two U-shaped steel beams, coupled through the wall
plane through the horizontal bracing effect of the steel thickness by means of anchoring bars injected in holes
frame, to use the horizontal load bearing capacity of the drilled in the masonry (Figure 13).
1065
A multiple structural effect results: relevant width. Different kinds of work are provided
for the different situations:
• obtaining a longitudinal continuity of the walls at the
intermediate level (at the walls’ top, the reinforced • removing and repositioning of blocks along the
masonry tying beam is provided); cracks;
• creating a connection system with the new structure • mortar injection and tying effect given by the tying
for the mutual bracing against horizontal actions; beams;
• tying effect at the floor level contributing to equili- • connection by means of slightly reinforced injected
brate some static horizontal actions (e.g. the pushes holes.
from the arches).
Re-making up architraves. Almost all the open-
Closing of cracks. Many cracks are present, some- ing are lacking of their architraves. They will be
times passing through and in some cases having re-made up according to a flexural resistant arrange-
ment consisting of steel beam covered by wood boards
underneath.
Re-making up failed openings. In some cases the
openings, not only are lacking in architraves, but are
failed along their total perimeter. They will be restored
rebuilding the stone masonry of the horizontal and
vertical sides.
Closing niches and flues. The walls integrity will
be pursued by the filling of the voids by effec-
tively interconnected stone masonry having the same
characteristics of the existing ones.
Re-opening pre-existing openings. Many old open-
ings are present, closed with non-continuised infill
masonry tables. If the opening shall be re-opened, the
infill will be removed and the opening will be sub-
Figure 11. Consolidation of foundations.
jected to the works provided for the failed ones. If the
opening shall remain closed, the same works provided
for the crack closing will be applied.
Arches of the western porch and relevant ones.
Figures 15 and 16 respectively show the situation of the
porch of building A and that one of an internal arch. It
is assumed that the insertion of the intermediate tying
beams gives to the masonry the capacity to sustain the
horizontal pushing forces of the arches. In some cases
specific tying bars are provided at the base of the arch.
Figure 13. Intermediate steel tying beams. Figure 14. Solution for the new architraves.
1066
destination, decay status, other witnesses not yet inves- both those already pointed out in the old chapel and
tigated, the new structures were designed as made of the other ones still hidden.
steel frames with wooden horizontal structures. The framed structure consists of columns and beams
This typology, which guarantees punctual founda- having standard I-shaped section. The floors are made
tions, longer spans and more “connections” with the of wood with single-aligned or double-aligned beams.
“existing”, allows to reduce the risk to compromise the The horizontal bracing is given by crossed tying bars
historical witnesses. In each internal area of the exist- located in all the rectangles of the structural grid.
ing building, individuated in-plan by perimeter walls, Figure 17 show a plan of the building A with the inter-
only four base supports are present (at the corners), so nal structures. Figure 18 show a section of the same
allowing for the preservation of the ancient structures, building with relevant details.
1067
Figure 18. Section and details of building A.
ACKNOWLEDGEMENTS
REFERENCES
The Venice Charter “International charter for the conserva-
tion and restoration of monuments and sites”. 1964. 2nd
Figure 20. Transfer mechanism of the seismic forces to the International Congress of Architects and Technicians of
masonry. Historic Monuments. Venice.
The Athens Charter for the Restoration of Historic Monu-
ments. 1931. 1st International Congress of Architects and
Technicians of Historic Monuments. Athens.
7 CONCLUSIONS
Codice dei beni culturali e del paesaggio. 2005. D.L.
22/1/2004, n. 42. G.U. n. 45 24/2/2004 (in Italian).
A restoration project aimed at the re-use of an old Linee Guida per la valutazione e riduzione del rischio sis-
monastery, at present in a ruined condition, has been mico del patrimonio culturale con riferimento alle norme
carried out providing for new technological solutions tecniche per le costruzioni. 2006. Dip. della Protezione
allowing for the preservation of the existing structures Civile e il Dip. per i Beni Culturali e Paesaggistici (in
even decayed. Italian).
1068
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Sandeep Sikka
Superstructures (Architects + Engineers), New York, USA
ABSTRACT: The areas of Spiti and Kinnaur district in the Western Himalayan region have experienced some
significantly strong earthquakes in the last few decades. Recent seismic tremors have caused some serious struc-
tural damage to the buildings in the region especially to the un-reinforced historic adobe structures. The deforma-
tion and bulges resulted in the adobe structures, as a resistance to the seismic vibrations, are the vulnerable areas,
most likely to collapse in an event of any future horizontal force. These historic structures, which were originally
designed for arid climate, are now facing problem with increased precipitation and regular rainfall for the last few
years. The paper puts forward various repair techniques carried out on the site to consolidate load-bearing walls
with cracks and research data on upgraded component used for the protection or repair work for these building.
1 INTRODUCTION
1069
four temples in elevation rise to a height of about and extreme climatic conditions and natural disas-
5 to 6 meters from their respective entrance lev- ters. The walls of these historic structures are made of
els on the exterior. The two lower temples Lotsawa (Adobe) a sun-dried large sized mud brick laid in mud
Lhakhang dedicated to the great translator Rinchen- mortar with the foundation in rubble stone masonry,
bzangpo and Karjung Lhakhang ‘the white temple’ generally rests on a stable solid ground1 . The thick-
face east lie on a relatively flat ground are mentioned ness of the walls varies from 2½ feet to about three
in this paper as above ground structures. The other feet. Vertical measurements shows that sometimes the
two temple Gongma Lhakhang and the Gya-pag-pai outer faces of the walls are slanted so that the wall
Lhakhang facing west are mentioned as below ground thickness is wider at the base and gradually tapers to
or semi-subterranean structures because the ground the top, providing extra stability to these tall and flex-
level outside these temples is much higher than the ible structures. Due to the cold climatic conditions for
interior floor level.This historic temple complex which most time of the year, the openings in the walls of these
houses invaluable artwork, murals, polychrome clay temples are minimum possible and contribute to less
sculptures and decorative wooden ceilings, survived than 5% of the total wall surface in a rectangular space
for more than 800 years, is showing alarming signs of and are located at the center of the wall. Besides the
deterioration. wooden doorframes such punctures in the walls are
NakoTemple Complex was nominated and included reinforced with thick vertical and horizontal tensile
in the 100 most endangered sites of the world by WMF wooden members connected to each other with flex-
in 2001. Since then efforts have been made every year ible joint. Series of wooden lintels laid next to each
under Nako Preservation Project [2, 3] to protect and other along the thickness of the walls are anchored
stabilize the structure and its artwork. deep and extend into the masonry on both sides of the
In the year 2006 and 2007 Institute of Conser- openings like additional horizontal tie members. The
vation and Restoration, University of Applied Arts, load bearing walls of the temples are reinforced with
Vienna,Austria carried out a conservation and research a wooden framework of horizontal wall ties, with a
campaign [4] at the temple complex with limited cross section of 6 to 8 in width and 4 to 5 in height,
funds from the University and Austrian Development as ring beams around the building, installed externally
Agency (ADA). The conservation project team mostly and internally flushed with the surface of the wall.
focused on the following tasks. The ring beams tie the entire structure together with
each beam running at a distance of approximately 1½
2.1 Art work conservation or 2 meters to the other. The exterior horizontal wall
ties were joined to each other at the corner externally
Task 1: The restoration of damaged clay sculptures with wooden vertical ties, now missing in most of the
inside the Lotsawa Lhakhang, temples. The arrangement thus prevents any outward
Task 2: Stabilization and cleaning of wall paintings movement during any seismic vibration and these wall
Task 3: Stabilization and cleaning decorative ties, along with the wooden lintels disrupt structural
wooden ceiling panels inside the Gongma cracks that could otherwise extend the full height of
Lhakhang. the wall eventually causing total collapse. Strength of
the adobe and the mortar joints vary all along the wall
2.2 Structural conservation hence differential loading. The horizontal ties there-
fore redistribute the load evenly through out the wall.
Task 4: Structural assessment and stabilization of load
In addition to these horizontal members, the load-
bearing wall at Gya-pag-pai Lhakhang and Karjung
bearing walls of the temples in Nako are reinforced at
Lhakhang.
the corners with buttresses at the corners. These but-
This paper is concerned with the structural issues
tresses were added later to support the masonry, after
and describes the research and repairs for structural
the building was struck by an earthquake separation of
conservation work carried out on the two temples
corner walls. The buttresses are random rubble stones
(Gya-pag-pai Lhakhang – Semi subterranean struc-
stacked one over the other against the corners of the
ture and Karjung Lhakhang – Above ground structure)
building forming a pyramid. The roofs of the temples
to accomplish Task 4.
are flat and are made of mud laid in various layers and
compacted. About 7 inches of compacted mud rest on
2 inches thick rectangular wooden panels or a mesh of
3 ARCHITECTURE OF BUDDHIST TEMPLE
willow twigs, with a layer of local shrubs or birch bark
COMPLEX AT NAKO: MATERIALS AND
sandwiched between the two for waterproofing. These
METHOD OF CONSTRUCTION
1070
are in turn supported over wooden rafters and beams,
which are further supported directly on load bearing
mud walls and wooden columns.
4 STRUCTURAL ASSESSMENT
1071
Figure 3. Thrust created by the expanding ground due to moisture gain next to the wall results in a horizontal crack.
1072
Figure 4. The samples were tested for their Flexural
Strength through an apparatus designed specifically for test-
ing the adobe in the field; using compressed earth block units Figure 5. Results of strength test conducted on four samples
of 7 kg. of adobe on site.
1073
present in the soil sample. To increase the binding
properties and inter-particular binding strength of the
local soil it is required that their elementary proper-
ties are modified so that their particles remain firmly
united and unaffected by moisture variation. Addition
of lime will certainly change its chemical properties
in which the clay of the soil reacts with lime to form
a relatively strong binder. This hypothesis was then
tested on site at Nako temple complex with other
locally available stabilizers into different plaster sam-
ples. The samples were then monitored for two weeks
and finally tested and documented for their durability
against water pressure.
Figure 6. Result of spray erosion test conducted on six
stabilized adobe samples.
5.2.1 Composition and application of plaster
samples
It was decided that the plaster samples should be
applied and tested on the ancillary structure of the
Temple Complex before any trials with the historic
fabric.
The wall surface selected for application was even,
completely dry and faced no problem of seepage or
dampness. The surface received adequate sunlight,
natural rain and wind. It was also exposed to the tem-
perature variations throughout the day. The first step
towards surface preparation for the application of the
plaster included scraping with a wire brush to create
undulations on the surface, so that the new plaster can
adhere to the wall. The following step involved wet-
ting of the wall surface properly with water spray at a
very low pressure before application. This was done to
avoid suction of water into the wall from the new plas-
ter mix. The various plaster mixes were then prepared
in a mixing tray with a trowel taking into consideration
Figure 7. Particle size analysis of local soil and local clay. the percentage of water the minimum of which should
be able to stick to trowel and spread evenly on the wall
surface. Fifteen samples of the plaster mixes (<6 mm
stabilizers. It was also studied that the additives for thick) with different local materials were then applied
stabilization of plaster cause no adverse effects on the each on Feet Square of area of the wall. These fif-
historic structure as they render a coat of water proof- teen plaster samples [7] were monitored for two weeks
ing layer externally and at the same time keep the walls and then tried against the wet erosion test. The water
porous. through a spray gun of pressure 50 Kpa was thrown
These mixes were then tested on site for their perfor- directly at the centre of each sample at a distance of
mance against general weather conditions and water. 600 mm. After starting from the bottommost plaster
The mix finally designed should also impart strength sample; each sample was exposed to equal water pres-
by consolidation to the otherwise fragile surface of the sure. It was observed that the spray jet used in this
historic monasteries. test has flow rates much higher than the rain obtained
To determine the exact mix of the stabilizer the first in this region. The process of erosion was carefully
step undertaken was to understand the composition of documented for the failure of plaster samples against
the local soil and clay and their behaviour. continuous spray for three minutes. It was observed
The data obtained from the laboratory analysis of that the finer particles (comprising the soil) washed
the local soil samples helped in understanding the out of the plaster mixes creating initial pits on the
physical nature and behaviour of the existing soil and surface and finally resulting in absolute failure. The
the local clay in the Nako Temple Complex, Kinnaur. plaster samples with lime as a binder and a stabilizer
The following results thus obtained can be read. could resist water pressure for a longer time. Perme-
The plasticity index indicated a low cohesive ability of the water into the plaster was also tested on
strength when plotted against the quantity of clay site, simply by digging the wet plaster and measuring
1074
Figure 8. Fifteen plaster samples tested on-site for water
erosion.
1075
Figure 11. Perforated clay pipe laid at the bottom of the drain sloped away from the site.
above the pipe. The drain was then covered with stones
of descending size till the ground level for effective
evaporation. The drainage had outlet points for effec-
tive cleaning and the slope of the drain was designed
as per levels of the surrounding site to effectively drain
water away from the building complex. This state of
the art inception of the drainage system was tested and
monitored for a year for its effectiveness by checking
the moisture levels in the ground outside the building
and also inside the masonry. Repairs to the horizontal
cracks will proceed as the moisture subsides.
5.4 Stitching of diagonal and vertical cracks in
above ground structures
The cracks and crevasses developed in theKarjung
Lhakhang are the areas of the potential weakness in
the load bearing walls need to be repaired and sta-
bilized to provide structural continuity. The crevasses
are mostly along the cracks and in such case stitching
would be a sympathetic alternative to strengthen and
retain most of the original masonry.
Stitching adobe buildings have been developed, dis-
cussed and published before by many professionals.
Techniques of stitching a crack in the load bearing
wall with mural paintings on the inner face in a highly
seismic zone is certainly a challenge and have not been
tried or been discussed before.
As part of the research and development of a sus- Figure 12. The structural cracks have been raked and the
tainable strategy for repair at Nako, it was decided to masonry is reinforced again with thin slate tiles.
1076
do an example as trial with local material and tech- adobe blocks, tested through onsite research, discussed
niques that could be monitored for its effectiveness on above, will be used for structural repairs to the outer
the bases of its performance. Severely weather portion Wythe of the Karjung Lhakhang’s load bearing wall.
of the load-bearing wall at the rear of this above ground Finally drain pipes and gutters were designed and
structure was selected for making repairs. The surface installed at the roof level of all the temples. The corru-
of the wall which is about 2 –6 thick was chased to gated sheets covering the roof were fitted with gutters
a depth of about 1 –3 into the wall, with a pointed molded with flat galvanized sheets. It is proposed to
Slater Hammer about 10 to 20 cms on both side of the remove the cement flooring inside the temple to relive
cracks on the weathered surface to form a space for the masonry and flooring from trapped moisture.
resting the stitch material. Generally masons or con-
servators tend to widen the crack to effectively rest the
repair material. The wider cracks require substantial 6 CONCLUSION
material to hold on to the parent masonry to provide
enough flexibility as well as a strong adhesion to react There are no ready-made answers, no ideal material
appropriately to any building movement. and no ideal solution to the complex problems these
Hence, widening of crack was avoided and the stitch earth structures are facing. For sustainable repair and
material was inserted into the masonry by making conservation of these historical construction there is a
small slits into the adobe blocks in zigzag manner need to understand the structure, materials and causes
alternately to bridge the crack to achieve extra strength. of damage in order to employ the appropriate remedia-
Study of ancient adobe blocks at Nako temple tion. With limited budget, inaccessibility to laboratory
complex indicate the use of varied size of course and treatment materials and distance from the civiliza-
aggregates and slate stones to gain extra reinforce- tion, in the middle of the high Himalayan Mountains
ment and strength. Thin Slate tiles are locally available, more than 3500 feet above sea level, it was essential
lightweight and have been use extensively for the con- to find locally available materials and develop sus-
struction of traditional houses. It is well known, learnt tainable repair and maintenance techniques to through
and documented by several locals and professional in on site research which can be carried out as an when
the Himalayan region that if the slates tiles are well laid required by the local craftsmen to stabilize 800 year
horizontally in layers inside a masonry wall, performs old structures against further seismic vibrations and
well during an earthquake. According to the survey changing climate.
conducted in 2003 the stone masonry structures with
such horizontal band of slate tiles were affected less
REFERENCES
during earthquakes of 1975 in the region as compared
to similar stone masonry structures without such band.
1. Luczanits, C. 2004. Buddhist sculptures in clay.
Slates tile and local fiber reinforced lime-stabilized
Early Western Himalayan Art, late 10th to early 13th
mud-mortar as filler was therefore selected as stitch
centuries. Chicago: Serindia Publications.
material for cracks.The tiles were laid into the slits hor-
2. Khosla, R. 2003–2005. Report on Nako Preservation
izontally and voids in-between were then filled with
Project, Phases-I, II, III and IV. RKDS: New Delhi
mortar. This was repeated up to the top of the crack.
(Unpublished).
Addition of fibers in the mortar mix reduced shrink-
3. http://athene.gis.univie.ac.at/project/nako
age and created better cohesion. The stitched portion
4. http://www.dieangewandte.at/konservierung/stories/
of the wall was then rendered with lime stabilized mud
story Reader$66
plaster discussed above. A stabilized soil with lime and
5. Sikka, S. 2003. Research report on Conservation of
jaggery grout previously tested in the lab for its per-
historic earthen structures in the Western Himalayas-
formance was injected from the inner painted surface
Museum of archaeology and anthropology. Univer-
to fill up rest of the void along the remaining width of
sity of Cambridge: England (Unpublished).
the wall.
6. Sikka, S. 2002. Conservation of earth structures in the
Western Himalayas- MA Dissertation Bournemouth
5.5 Removal and replacement of outer Wythe
University: England (Unpublished).
of wall
7. Sikka, S. 2003. A research on Stabilization of Plaster
With diminishing knowledge of traditional mud bricks Samples for conservation of historic earthen struc-
and repair techniques and sudden earthquake of tures in the Western Himalayas. Terra 2003, 9th inter-
1975 resulted in some serious interventions in terms national conference on the study and conservation of
of masonry repairs. Most of the upper courses of earthen architecture. Preprints Yazd.
the masonry near the roof, externally, have been 8. Bogin, S. (2005). A technical study of the early
replaced with big stone blocks that have eventually Buddhist wall paintings at Nako, Himachal Pradesh,
detached or are budging outwards due to excessive India. Zeitschrift für Kunsttechnologie und Kon-
load on the walls. Replacement of stone with improved servierung. 19(2): 199–229.
1077
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Alan J. Lutenegger
Department of Civil & Environmental Engineering, University of Massachusetts, Amherst, Ma. USA
James H. Kemper
Hubbell Power Systems, Centralia, Mo. USA
ABSTRACT: The structural integrity of historic buildings is often compromised over the life of the structure as
a result of foundation deformations over time. Foundation deformations can be created by a number of problems
related to the subsurface conditions, groundwater removal, softening of the bearing soils, or simply from long-
term soil consolidation. Often times, such deformations produce large and usually differential settlements that
lead to a condition of imminent collapse or unsafe conditions and place the structure in jeopardy. The preservation
of these structures requires innovative foundation repair approaches that will not unnecessarily disrespect the
historic nature of the structure. This paper describes the preservation of several historic structures by foundation
underpinning using helical Screw-Pile foundations. Five case histories are presented where helical Screw-Piles
were used to preserve the foundations of structures ranging in age and use from the stone gate house of an 11th
century castle to a late 19th century university brick classroom. The cases illustrate the economic and effective
use of helical Screw-Piles to retrofit the foundations with minimal intrusion to the structures. For each case, a
brief historical review of the structure is presented and a summary of the foundation problems and site conditions
are given along with a description of the Screw-Piles selected for the repair work.
1079
Figure 1. Typical geometry of Screw Pile used by Mitchell
(1836–1880).
Figure 2. Configuration of modern Screw-Piles: (a) single
helix; (b) multi-helix; (c) grouted shaft.
approach for engineers who are faced with such
problems. for offshore and onshore lighthouses, ocean front
piers, bridges, slope stabilization and underpinning.
Duckham (1889) described the use of Screw-Pile foun-
2 SCREW-PILE FOUNDATIONS dations to underpin the town hall at Great Yarmouth
which had undergone several inches of settlement.
Screw-Pile foundations are a product of mid 19th Since the early 1950s, the Screw-Pile has taken on
century Victorian engineering. Most historian give a different form, to solve a different kind of problem.
credit to the use of Screw-Piles as viable founda- A.B. Chance recognized that a helical screw element
tions to Alexander Mitchell, a Scottish self-taught could be used to provide tension support for electrical
engineer and contractor who initially developed iron transmission tower and other tall structures and devel-
Screw-Piles for use as foundations for offshore light- oped a screw anchor industry that still is in use today.
house around the coasts of England. Mitchell (1848) Since about the 1970s however, Screw-Pile founda-
described the mechanics of the functionality of Screw- tions have returned to applications similar to those
Piles as: in which they were originally developed by Mitchell
and others. Modern Screw-Piles have a variety of
“Whether this broad spiral flange, or “Ground
geometries and applications and include both single
Screw,” as it may be termed, be applied …to sup-
helical and multi-helix configurations on both square
port a superincumbent weight, or be employed
shafts and round shafts. With the advent of modern
…to resist an upward strain, its holding power
hydraulics in the construction industry which largely
entirely depends upon the area of its disc, the
took off after the 2nd World War, Screw-Piles can now
nature of the ground into which it is inserted,
be installed with ease using a variety of commonly
and the depth to which it is forced beneath the
available construction equipment.
surface.”
Modern Screw-Piles have a variety of names, such
Figure 1 shows the geometry of a typical Screw- as Helical Pier, Screw Anchor, or Helical Pile but they
Pile used by Mitchell at the Maplin Sands Lighthouse are all generally fabricated as one of three general con-
in the Thames estuary. All of the Screw-Piles used by figurations as shown in Figure 2. They may consist of
Mitchell were fabricated by a combination of cast-iron a single helical element, a multi-helix lead section or
plates and wrought-iron shafts, both of which could a helical lead section with a grout column. The cen-
be readily manufactured in England during this time tral shaft may be square or round. If a grout column is
period. In nearly every case of Screw-Pile installation used, an appropriate amount of time is allowed for the
before about 1900, the Screw-piles were installed by grout to achieve sufficient strength before load trans-
manual labor using a large capstan attached to the top fer. Grout columns may be used to provide additional
of the shaft and turned by as many as 40 men. No verti- load capacity, working in combination with the lead
cal load was applied to assist the installation, the pitch helical section, or they may be used to simply trans-
of the screw blade was sufficient to produce advance fer the load to the lead helical section, as in the case
of the crew with rotation of the shaft. of short Screw-Piles end-bearing on rock or other hard
During the mid to late 19th century, Screw-Piles material. The grout column typically consists of a Port-
were used throughout the world to provide support land cement based neat grout or some similar material.
1080
structural connection between the two and provide for
load transfer.
If appropriate, given the condition of the structure
and the degree of distress, a small amount of lifting
may also be performed using hydraulic jacking sys-
tem to provide vertical adjustment to the structure and
preloading of the screw-pile. This load is then locked
off and the final position of the foundation bracket
is achieved. If needed, the existing foundation may
also be reinforced using either thread bars or grouted
bars inserted into the existing foundation blocks or a
small section of reinforced concrete may be cast in
place and tied to the existing foundation. The under-
pinning sequence is illustrated in Figure 3. Note that
Figure 3 illustrates that the Screw-Pile can include a
grout column around the central shaft or it may be
installed without a grout column, depending on the
soil conditions and the design of the pile for bearing
capacity.
4 CASE HISTORIES
1081
Table 1. Summary of case histories.
Case Age of
no. Structure Location structure
1082
Figure 6. Heavy duty foundation bracket attached to exist- Figure 7. Foundation bracket and leveling jack used at
ing foundation. Forest Lodge.
1083
beneath the bell tower. A skid loader was used to install
the Screw-Piles on the outer perimeter while a portable
hand held hydraulic torque head was used inside,
where the work space was limited. Screw-Piles were
attached to a heavy duty foundation bracket attached
to the existing masonry rubble foundation which was
then encased in a new cast-in-place concrete block to
tie the existing foundation to the Screw-Piles.
Table 2 gives a summary of the different configu-
rations of Screw-Piles used for each of the cases.
1084
Table 2. Screw-Piles used in case histories.
1 Grouted shaft triple-helix (8/10/12) Klinger & Associates, P.C., Hannibal, Mo. Tri-State Concrete Correction, Co.
2 Square shaft double- and triple-helix Klinger & Associates, P.C., Hannibal, Mo. Tri-State Concrete Correction, Co.
3 Square shaft triple-helix (8/10/12) Meyer Borgman & Johnson, Inc. Atlas Foundation Co.
4 Grouted shaft double-helix (10/12) Patrick Parson Ltd. Subsurface Technologies, Ltd.
5 Square shaft double-helix Structural Engineering Association, Kidwell Construction and
Kansas City, Mo. Pro Services.
be installed with small light duty construction equip- steer or mini excavator is about 1 hour for a 10 m
ment such as a rubber track mini excavator or skid length of pile. Only a small amount of additional time
steer or portable hand held equipment. is needed when using a grouted shaft Screw-Pile.
1085
Supply/Atlas Foundation Company; and Duncan Crocker, E.E., 2005. Non-invasive Underpinning Technolo-
Robson, Subsurface Technologies Ltd, Alnwick, gies in Historic Settings. Proceedings of the 4th Inter-
Northumberland. national Seminar on Structural Analysis of Historical
Constructions: 821–828.
Drooff, E.R., Furth, A.J., and Scarborough, J.A., 1995.
REFERENCES Jet Grouting to Support Historic Structures. Foundation
Upgrading and Repair for Infrastructure Improvement,
Avellan, K., and Lange, K., 1997. St.-John’s Chirch, ASCE: 42–55.
Tartu, Estonia – Underpinning of Tower. Geotechnical Engi- Duckham, F.E., 1889. Underpinning Great Yarmouth Town
neering for the Preservation of Monuments and Historic Hall. Minutes of the Proceedings of the Institution of Civil
Sites: 521–525. Engineers, 98: 372–374.
Burland, J.B., Jamiolkowski, M., and Viggiani, C., 2003. Mitchell, A., 1848. On Submarine Foundations; particularly
The Stabilization of the Leaning Tower of Pisa. Soils and Screw-Pile and Moorings, Civil Engineer and Architects
Foundations, 43 (5): 63–80. Journal, 12: 108–149.
Chartres, F.R.D., 1997. Monitoring the Installation of Piles Niehoff, J.W., 2004. Stabilization of a Historic Struc-
Beneath the Round Tower, Windsor Castle, U.K. Geotech- ture Founded on an Unstable Mesa. Proceedings of
nical Engineering for the Preservation of Monuments and the 5th International Conference on Case Histories in
Historic Sites: 349–356. Geotechnical Engineering, paper no. 1.05, 6 pp.
Ciancia, A.J., Biesiadecki, G.L., and Ladd, B.E., 2006. Rodriguez Ortiz, J,M., and Monteverde, P.R., 1997. A Review
Challenges of Underpinning Two Historic Landmark of Recent Spanish Interventions in Monuments. Geotech-
Buildings. Practice Periodical on Structural Design and nical Engineering for the Preservation of Monuments and
Construction, ASCE, 11(3): 142–148. Historic Sites: 499–508.
1086
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: This case-study concerns an old industrial reinforced concrete shell structure at the port of
Antwerp. The building serves as a pilot for stowing of shipped steel and wooden products. A fork-lift hit one of
the supporting columns. The impact of the collision caused the concrete to crush and the internal reinforcement
to shift considerably. Some of the internal reinforcement yielded or broke. Temporary supports had to be placed
to avoid collapse of the total structure. The stress situation is analyzed. A FEM model was built to gain insight
into the stress distribution in the shell structure. In comparison with the undamaged situation, membrane forces,
shear forces and moments did strongly increase which explains the actual damage. A solution is worked out to
repair and strengthen the damaged part of the structure. The FEM-analysis demonstrates that a compression arch
will arise in the top of the barrel shell to span the displaced support. This requires a tension member at the bottom
side of the concrete barrel. Therefore, externally bonded reinforcement is applied to increase the tensile capacity
of the perimeter beam. The web reinforcement has yielded or is broken so that it is replaced by external CFRP
reinforcement. After strengthening the structure, a monitoring system is installed. Strain gauges are glued on
several laminates. These measurements give feed back on the structural behaviour of the strengthened structure
and the applicability of the FEM used. This case-study illustrates the load distribution mechanisms in the shell
structure, the assessment of the actual condition and an appropriate intervention based on external reinforcement.
1 INTRODUCTION
1087
Figure 2. Damaged column, December 2004.
1088
3.1.2 Internal reinforcement
The reinforcement of the concrete shell exists of main
steel bars and a web. The location of the bars (φ15 mm)
correspond with the position on plan. The distance
between the bars of the web (φ5 mm) is 150 mm in
the longitudinal direction and 200 mm in the other.
Close to the column, the mean crack opening is
lb = 15.9 mm. The strain in the steel bars can be esti-
mated by this elongation. Taking the anchorage length
as the original length, the strain is equal to Equation 5.
The anchorage length can be calculated according to
Eurocode 2, Equation 6.
with
The Young’s modulus can be determined using µi shape factor, equal to 1.08
formula 4 and equals 35.9 kN/mm2 : sk characteristic snow load in Antwerp
Ce exposure coefficient, equal to 1.0
Ct thermal coefficient, equal to 1.0.
1089
Figure 8. Calculation of the membrane forces in the Y-axis
Figure 7. Schematic view of reinforced concrete shell
in the shell before the column settlement.
construction.
3.3 FEM-analysis
A FEM-model is built to analyse the stress distribution
in the shell structure (Ansys 2005). As the damaged
column is located in the central part of the building,
edge effects can be neglected. Only the two neighbor-
ing shells are included in the FEM-model, the damaged
column being located at the outer left part, Figure 7.
Symmetry boundary conditions are applied to the left
and right edge of the model.
The shell structure is modeled with the 8-node
quadrilateral SHELL93 shell element, which has 6 Figure 9. Original reinforcement located at high stress
degrees of freedom at each node: translation in the zones.
X, Y and Z direction and rotations around the X,
Y and Z-axis. The beams around the roof lights are
modeled by means of 2-node BEAM4 and BEAM44 The original reinforcement plan (1955) show that
elements, which have also 6 degrees of freedom at the internal main reinforcement is bundled in the mid
each node. These node displacements are translation span of the beam and spreads out to the supports. This
in three directions and rotation around the three axes. main reinforcement take care of the tensile stresses in
Membrane forces, shear forces and moments are the shell, Figure 9.
calculated (Figeys and Van Gemert 2006). The original In comparison with the undamaged situation, cal-
situation is studied, Figure 7. Tensile forces in longi- culations show that the membrane forces, shear
tudinal direction are present at mid span of the edge forces and moments strongly increase at the imposed
beam (with a maximum of 289 kN/m) and in the shell deformations. In particular, the membrane tensile
near the central column (150 kN/m), Figure 8. forces in the longitudinal direction of the barrel
1090
Figure 10. Calculation of the membrane forces in theY-axis Figure 11. Jack up of the shell structure.
in the shell after the column settlement.
with
TSd Design value of tensile force in the bottom of
shells (longitudinal axis, Figures 8 and 10) increased the barrel at one side
extremely. L Span without damaged column, 30.0 m
This caused the actual damage state and, more- f Height of the shell, 2.95 m
over, changed the structural system.The damaged shell p Load per meter along the longitudinal axis,
seems to be supported by the two outer columns only. that equals:
High tensile stresses (with a peak of 4820 kN/m2 )
appear in the edge beam at the missing support (i.e.
damaged column). In the top of the shell compressive
stresses in the longitudinal direction are introduced
by the arch action. The available reinforcement in the θ Half of the opening angle of the shell, 43◦
edge beam at the support is insufficient, which led to R Radius of the shell trajectory, 10.9 m
the observed cracking of the concrete. d Mean thickness of the shell, 0.10 m
ρd Design value of load, including proper weight,
wind loading, snow loading 40.76 kN/m3 .
4 THERAPY The tensile force present at the beam for each half
shell equals TSd = 1285 kN . Taking into account the
A method to repair and strengthen the damaged part original reinforcement (9 × φ15 mm, TRd = 325 kN),
of the structure was studied and executed. The FEM- five CFRP laminates 100 × 1.2 mm2 are additionally
analysis demonstrates that, due to the vertical settle- needed to take of this tensile force, Equation 13.
ment of the support, a compressive arch developed in A safety factor γM ,CFRP of 1.5 is applied.
the shell, that spans the displaced support. In order
to restore as much as possible the initial stress-strain
situation, the structure is jacked up, Figure 11.
However, it is not possible to neutralize the total
vertical displacement. The plastic deformation of the
internal reinforcement can not be recovered and chip-
ping off of concrete would arise at excessive lifting. Also the internal reinforcement mesh near the col-
An additional tension member at the bottom edge of umn is damaged: yielding and cracking are observed.
the concrete barrel should take up the induced tensile However, because of the crack formation, the calcu-
forces. For that purpose, externally bonded reinforce- lated stress peaks at the imposed deformations are
ment is applied to increase the tensile capacity of the reduced. Therefore, it is safe to take the stress level of
longitudinal beam. the undamaged situation into account. The stress level,
The tensile force in the edge beam can be calcu- calculated with the FEM-model, is compared with the
lated as: amount of undamaged reinforcement. In the longitu-
dinal as well as in the transversal direction additional
reinforcement is needed.
The tensile force in the longitudinal direction equals
about 150 kN/m near the column and decreases to
1091
Figure 12. Proposed strengthening of the shell structure.
1092
Figure 16. Location of the different strain gauge.
1 −24.4
2 −36.5
3 65.8
4 12.8
5 −73.0
6 14.5
7 −78.5
8 −12.0
9 19.7
10 11.7 Figure 17. Stress level on different location derived from
11 11.4 measured strains.
12 −40.3
+: compression; −: tension.
Location 8 and 9 are situated at the upper part of
the shell, Figure 16. The tensile stresses decrease into
situation, only the mean density acting on the struc- the upper part to change in small compressive stresses
ture (ρd = 22.68 kN /m3 instead of 40.76 kN /m3 ), the at the top, as observed at location 9.
stress level will be reduced to about 90 N/mm2 . These Three strain gauges measure the strain in the trans-
tensile stresses can be observed in these locations verse direction. According to the calculations only
(5 and 7), except in location 6. small tensile stresses are observed, Figure 18.
Strain gauges 1, 2, 3 and 4 are located further on the
laminates. In these zone the tensile stress decreases.
Except for location 3, the tensile stress levels in these 6 CONCLUSIONS
locations decreased. Immediately after removing the
supports, a relatively large compressive stress is found Externally bonded CFRP-laminates are used to
in location 3. This location is situated at the bottom strengthen and repair a damaged shell structure. FEM
CFRP laminate, two meters from the concrete column. analyses are used to determine stresses and strains in
The measuring point is close to the injected crack with both the undamaged and damaged shell. This FEM-
the largest opening of about 20 mm. After removing model allowed to identify the formation of a com-
of the supports, the original situation is approached pressive arch in the damaged shell structure and to
in which compressive stresses will appear in the bot- determine the required additional reinforcement, to be
tom of the edge beam in the zone of the crack. These applied by means of externally bonded CFRP lami-
stresses will compress the epoxy resin in the crack. The nates on the edge beam and on the shell structure in
laminate is compressed as well, which can be observed the vicinity of the damaged column. Monitoring of
in the compressive stress curves. stresses in the laminates during different phases of the
1093
ACKNOWLEDGEMENT
REFERENCES
Ansys, I. (2005). Ansys version 10.0.
Brosens, K. (2001). Anchorage of externally bonded steel
plates and CFRP laminates for strengthening of concrete
element. Ph. D. thesis, Katholieke Universiteit Leuven,
Leuven.
CEN pr EN 1542 (1998). Products and systems for the
protection and repair of concrete structures – test methods-
measurements of bond strength by pull-off.
Eurocode 0 (2002). Basis of design: Design principles.
Eurocode 1 (1995). Basis of design and actions on structures-
part 2–4: Actions on structures – wind actions.
Eurocode 1 (1999). Basis of design and actions on structures-
part 1 – 1: Actions on structures – densities, self-weight
and imposed loads.
Eurocode 1 (2003). Eurocode 1 – basis of design and actions
on structures – part 1–3: General actions – snow loads.
Eurocode 2 (1999). Design of concrete constructions – part
1–1 (in dutch).
Figeys, W. and D. Van Gemert (2006). Report d/00413/05
dd. 13-03-2006. shell repair. hybrid solution (in Dutch).
Technical report, Triconsult N.V.
ICOMOS (2003). Recommendations for the analysis, con-
0: before removing of the supports A: after removing of the servation and structural restoration of architectural her-
supports itage. International Scientific Committee for Analysis and
Restoration of Structures of Architectural Heritage.
Figure 18. Stress level on different location derived from Schueremans, L. (2003). Evaluation of bearing capacity and
measured strains. design of consolidation and strengthening by means of
injections. In WTATagung, Leuven.
repair showed the effectiveness of the repair procedure.
As it was possible to jack up the shell and to restore the
support, stress levels as predicted for the undamaged
situation are observed.
1094
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
Claudia Cancino
Getty Conservation Institute
ABSTRACT: This paper presents the main results obtained during an ongoing experimental test program
performed to explore the possibility of repairing structural cracks on adobe walls through the injection of mud-
based grouts. The structural objective was to attempt to recover the original strength of severely cracked adobe
masonry elements by repairing them with mud-based grout injections. Masonry strength was first estimated
with indirect tension tests of adobe sandwiches and later measured with diagonal compression tests of adobe
masonry small walls. It was decided to study the use of grouts prepared with sieved soil alone or stabilized with
lime, cement, and gypsum. The test results showed that grout injections based on the original soil used to build
the walls are effective in restoring completely the original strength of adobe walls which had endured significant
seismic damage. It seems, therefore, that a new criterion for the conservation of earthen buildings in seismic
areas could be formulated, using compatible materials for the restitution of the original seismic strength of the
earthen structural elements.
1095
Figure 1. Device developed to perform indirect tension test
on adobe sandwiches.
1096
Figure 5. Results of indirect tension tests. Tension stress in
the mortar-adobe interface in sandwiches with earthen mor-
tar with 25% moisture. The thinner mortars tend to improve
tension strength at the interface. The curve corresponds to a
linear adjustment using the least squares method.
1097
Table 1. Strength recovery in repaired wall sections with
grouts of sifted soil.
1098
Table 2. Strength restoration in wall specimens repaired the cracked adobe walls. Good quality of the injection
with grouts of sifted soil stabilized with gypsum in 10%, process is crucial to achieving effective repair.
20% and 30% proportions. Stabilizing mud mortars by adding cement or lime
did not produce good results. Gypsum was the best
Force of fracture (kN)
Gypsum Wall % Strength
of the stabilizers studied. However, its use is ques-
content specimen Initial Repaired recovery tionable, since gypsum increases the difficulty of the
repair.
10% MDNR-8 7.62 6.59 86% A thinner mortar increased adobe masonry strength.
MDNR-9 9.11 10.25 113% This implies that by grouting thinner cracks, a more
MDNR-11 7.40 3.59 49 % effective repair may be obtained.
MDNR-12 7.37 5.96 81% Further research is necessary on repairing thinner
20 % MDNR-7 9.46 12.79 135% structural cracks and thicker walls, in order to cover
MDNR-13 5.55 11.21 202% all the cases presented in historical adobe monuments.
MDNR-15 10.44 6.37 61% New types of grouts and more efficient injection
30 % MDNR-10 5.43 9.27 171%
techniques must continue to be explored. Emphasis
MDNR-14 8.32 8.63 104%
MDNR-20 13.22 15.14 115% should be placed on the use of natural and processed
additives through the study of their mechanical and
physical-chemical properties.
which reduces the drying shrinkage of the earthen
mortar, and therefore reduces the micro-cracking and ACKNOWLEDGEMENTS
increases the strength. The increased volume of the
mortars stabilized with gypsum has been confirmed by A previous version of this paper has been presented,
measuring the volumetric shrinkage of 20% gypsum in Spanish, at the Adobe USA2007 conference, held
pastes (Mattone et al. 2005). in El Rito, New Mexico in May 2007. The authors are
The strength of the sandwiches prepared with sta- grateful to the Academic Research Directorate of the
bilized soil with lime or cement, in 5%, 7% and 10% PUCP for providing funding, without which it would
percentages, was generally lower than that obtained have been impossible to carry out this project. They
with sifted soil. Therefore, only adobe wall specimens would also like to thank the staff of the Laboratory
repaired with injection of grouts of sifted soil and sifted of Structures of the Department of Engineering, espe-
soil stabilized with gypsum were subject to diagonal cially its director, Eng. Gladys Villa-Garcia; Dr. Stefan
compression tests. Simon, for his ideas and suggestions; and Mary Hardy,
Soil from the PUCP was used in manufacturing and Project Specialist of the Getty Conservation Institute,
repairing the diagonal compression specimens. The for their efforts, recommendations and information in
cracked specimens were repaired with grouts of sifted support of the working group.
soil with 10%, 20% and 30% gypsum. The percentage
of moisture used, from 35% to 40%, is the minimum
needed to practice injection of the grout. Injections of REFERENCES
grouts stabilized with gypsum were more difficult to
do than those of only soil, because the gypsum makes Mattone R, Pasero G, Rivotti A, and Tosco V. 2005. Uso de
the grouts harden faster. Table 2 shows the degree of productos naturales para mejorar el comportamiento al
strength restoration obtained. agua de revoques a base de tierra. IV Seminario Ibero-
Americano de Construcción de Tierra. Portugal.
The results show that in order to increase the Tolles L, Kimbro E, Webster F, and Ginell W. 2000. Seis-
strength of the adobe masonry walls repaired with soil, mic Stabilization of Historic Adobe Structures. Getty
it is necessary to add gypsum in percentages between Conservation Institute Scientific Program Reports. Los
20% and 30%. However, it is possible that the relative Angeles.
difficulty of the process of repair using grouts with Tolles L, Kimbro E, and Ginell W. 2003. Planning and
gypsum could affect the quality of the repair. Engineering Guidelines for the Seismic Retrofitting of
Historical Adobe Structures. Getty Conservation Institute
Scientific Program Reports. Los Angeles.
4 CONCLUSIONS AND RECOMMENDATIONS Vargas J. 1979. Albañilería de Adobe con Variaciones de
Mortero. Pontificia Universidad Católica del Perú, Publi-
The proposal to repair structural cracks in historic cación DI-79-02. Lima, Perú.
Vargas J, Bariola J, Blondet M, and Mehta P K. 1986. “Seismic
adobe buildings appears to be a viable alternative, Strength of Adobe Masonry”. Materials and Structures,
based on criteria of resistance, to return the building Vol. 19, No 112, July–August.
to its original, uncracked condition. WJE Engineers, Architects, Material Scientists. 2004. Ran-
Repair by injection of grouts made of soil without cho San Andres Castro Adobe Rehabilitation. California
stabilizers was effective for recovering the strength of State Parks.
1099
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
D.F. Henriques
Instituto Superior de Engenharia de Lisboa, Lisboa, Portugal
Laboratório Nacional de Engenharia Civil, Lisboa, Portugal
L. Nunes
Laboratório Nacional de Engenharia Civil, Lisboa, Portugal
J. de Brito
Instituto Superior Técnico, Lisboa, Portugal
ABSTRACT: This paper concerns the study of biocides application in old timber structures of maritime pine
(Pinus pinaster Ait.), previously impregnated with other products. A method was developed in laboratory to
determine in situ the penetration depth of a product applied superficially. As initial treatment, three traditional
products for sawn timber for buildings were used and, for new treatment, two newer, more environmentally benign
products were used. Their ability to penetrate the pre-treated surfaces was evaluated after 1, 2 and 3 applications
at 24 hours intervals and the results obtained are presented. Finally, the applicability of the developed test to the
in-situ evaluation of timber structures is also discussed.
1101
preservative (LOSP) using white spirit as the solvent (Pinus pinaster, Ait.) and for maritime pine treated
carrier to deliver the actives into timber and arsenic with either an oil carrier formulation or two typical
copper combinations such as chromate copper arsenate LOSP formulations, one that has disappeared from
(CCA). The later was not considered relevant under the the market due to the active ingredients restrictions
present study as it usually performs well when applied already mentioned and one that has been marketed for
in roof timber structures and re-treatments are seldom a very long time.
needed. Two products were chosen for the re-treatment, one
This paper describes the test developed and presents of them a water borne product (boron based) with bet-
the results of its calibration for untreated maritime pine ter toxicological and environmental characteristics and
a second product (LOSP) of common use at the present
time, easily obtained for the curative and preventive
treatment of timber structures. Their ability to pene-
trate the pre-treated surfaces was evaluated after 1, 2
and 3 applications at 24 hours intervals and the results
obtained are presented.
These products and their sequence of application
are represented schematically in Figure 1. In Table 1
the active ingredients and the solvents of each of the
products applied are presented.
2 WOOD TREATMENT
1102
Creosote remained until the 30’s of last century biocide product, when applied on timber from an
as the only recognized timber preservative product. ancient building treated with a non-documented prod-
In 1931 pentachlorophenol was patented and in 1933 uct difficult to characterize without extensive and
the first product containing copper salts, chromium eventually expensive analysis.
and arsenates, commonly known as CCA, followed
suit. The domination of these three substances was 3.2 Development
only slightly shaken by the introduction in the 50’s
of boron in treatments by immersion – diffusion and For laboratorial simulation of real situations, con-
especially of products in light organic solvent, some trolled quantities of the products chosen for the initial
of which were 10 years later top sellers, mostly in the treatment were applied in a first step and later on new
market segment so-called “general public” due to the products of prevention and/or treatment were similarly
versatility and ease of application (Nunes, 2001). applied.
The last two decades of the XX century dramatically To quantify in situ the penetration depth, a method
changed the panorama of an area that seemed to be was devised, very little intrusive and easy to apply to
consolidated, environmental concerns, disposal issues any building timber structure. It consists on drilling
and general public perceptions questioned the use of holes parallel to the face under treatment, at distances
traditional active substances (Goodell et al., 2003). from it successively bigger, in which cotton swabs tes-
Research related with the search of new active raw timonies are inserted before the application of the new
materials for timber treatment has been a field of treatment product. The testimonies absorb the liquid
intense work in the last years. Copper (e.g. alkaline impregnated into the timber when it reaches the depth
copper quat and copper azole) stood out as fungi- that corresponds to its position.
cide agent, applied in formulations without chromium
and arsenate, both in organic solvent and in aqueous
basis. Products based on copper are therefore posi- 4 MATERIALS AND METHODS
tioned as the predominant solution in the near future
(Humar et al., 2007) for the preventive long term treat- 4.1 Test specimens
ment of wood. However, rot fungi that are resistant to
For this purpose, 50 × 7 × 5 cm specimens were used,
copper have been known for some time and most of
cut from dry maritime pine (Pinus pinaster Ait.),
the environmental concerns that were expressed for
with various ring dispositions, from pith and sap,
other active ingredients may also be called against
with some knots and some sapstain. The aim was
copper use.
simulate non-exhaustively the diverse real situations
The use of boron should also be considered when-
that can be found in ancient buildings timber struc-
ever the risk of leaching is not a conditioning factor,
tures. These specimens were divided in four groups,
since it possesses good fungicide and insecticide char-
to three of which a pre-treatment was applied. The
acteristics (Nunes, 1997). Laboratory and field tests
remaining one, without treatment, was used for control
have demonstrated that timber treatments with boron
purposes.
are very effective in the prevention of brown and white
The preparation of the test specimens occurred five
rot as well as insect attack (Jorge et al., 2004) and
and a half months after the application of the pre-
wood modification together with boron might be able
treatment and according to the procedure described
to help the recognized problem of leaching (Kartal
next. Four holes (diameter = 0,45 cm) were drilled
et al., 2007).
parallel to the face under treatment and at distances
New multi-components biocide systems to inhibit
successively greater from it (0,5, 1,0, 1,5, and 2,0 cm),
the growth of moulds, rot fungi and termites are
as shown in Figures 3 and 4. Six plastic rings were
also under development and show promising results
then glued over the larger face that had fewer defects,
(Clausen & Yang, 2007).
as represented in Figures 2 and 4, to limit the applica-
The use of natural products, such as rosin deriva-
tion of the new product. A transparent epoxy adhesive
tives and other extractives and modified wood, are
was used, uniformly distributed in the border and
also possibilities to take into account (Rowell, 2006;
around the ring, in order to confine the absorption of
Savluchinske-Feio et al., 2007).
the product to the area inside the ring. The compat-
ibility between the glue and the solvents was tested
experimentally.
3 STUDY DESCRIPTION
After the preparation, the specimens were left to
stabilize for 1 week in a conditioned chamber with
3.1 Objective
constant relative humidity (65 ± 5%) and temperature
The objective of the present study is the develop- (20 ± 2◦ C). The moisture content just prior to the tests
ment of a method to be used in situ to determine was as presented in Table 2. This value was obtained
the impregnation depth achieved by a new generation using a needle humidimeter.
1103
Table 3. Average absorption of initial treatment.
A 225 ± 37
B 218 ± 47
C 246 ± 22
Figure 2. Dimensional representation of the test specimen
and location of the rings for the introduction of the new
products.
Table 4. Densities and quantities of new treatment products
applied.
Quantity applied
Figure 4. Detail of a specimen prior to being tested. To verify the method under development two products
(D and E) were used, as described above, as well as
Table 2. Test specimens and moisture content before the their solvents, water and white spirit.
new treatment. To make the presence of the products and solvents
more easily recognizable a dye was added before treat-
Test specimens and moisture content ment. After various preliminary selection tests, it was
decided to use Neutral Red as a marker for water and
0A 12.5% 1D 14.0% 2A 12.3% 3D 14.0% the boron-based product. For organic solvents, the dark
0B 14.0% 1G 12.5% 2D 13.2% 3E 14.2%
0D 13.3% 1H 11.5% 2H 12.0% 3H 12.5%
blue dye Ceresblau was selected. The concentrations,
0E 12.2% 1I 14.2% 2I 13.0% 3J 12.5% measured in mass, were as follows:
– Water + 0.5% Neutral Red;
– Water + 5.0% Product D + 0.5% Neutral Red;
4.2 Initial treatment products – White spirit + 0.5% Ceresblau;
As referred three types of products were used for the – Product D + 0.5% Ceresblau.
initial treatment, considered representative to simulate With the help of a pipette, 1 ml of product was
real situations. Products A and C are both light organic applied over the area defined by the interior of the pre-
solvent preservatives (LOSP) and product B is an oil- glued rings, measuring 22 ± 0.8 cm2 . The density of
borne product obtained from an auto repair-shop and is each product and the quantity introduced are presented
a mix of various used motor oils, whose composition in Table 4.
1104
Untreated wood
2,50
1,00 3° application
0,50
0,00
Water White Spirit Product D Product E
New treatment
(a)
Initial treatment with Product A - LOSP
2,50
1,00 3° application
0,50
Figure 5. Sequence of actions relative to the second appli-
cation of colored white spirit over product B: (a) application 0,00
Water White Spirit Product D Product E
of new product; (b) covering with a membrane; (c) after 24 h,
New treatment
visual checking that the product was totally absorbed by the
(b)
timber; (d) testimonies reading.
Initial treatment with Product B - Oil
2,50
4.4 Methodology
Penetration medium (cm)
2,00
The new treatment products were applied in three 1st application
1,50
coats with 24 h intervals between them. Before the 2° application
3° application
application, the testimonies were positioned within 1,00
2° application
removed and the respective presence or absence of
1,00 3° application
color registered.
In Figure 5 (from a. to d.) the sequence of non- 0,50
1105
• Also, for the organic solvents alone each application
seems to increase the depth at which the product
penetrates (around 0,5 cm);
• The number of applications does not seem to espe-
cially influence the final result, for water and the
water borne product.
The applicability of the developed test to the in-
situ evaluation of timber structures was not yet fully
studied but the results obtained are quite promising as
a good picture of the distribution of the products inside
the wood was obtained just by the insertion of easy to
obtain cotton swabs in pre-bored small holes.
Figure 7. Results obtained using the destructive analysis for
the various situations studied: (a) water over pre-treatment
with product B; (b) white spirit over pre-treatment with prod- ACKNOWLEDGMENT
uct C; (c) product D over pre-treatment with product C; (d)
product E over pre-treatment with product C. The authors acknowledge the contribution of Sofia
Simões to the work presented.
1106
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
J.F. Norden
Gibble Norden Champion Brown, Consulting Engineers, Inc. Old Saybrook, USA
ABSTRACT: The Wauregan Hotel (c. 1855) in Norwich, CT, USA was restored in 2005–6 and adaptively
reused for new affordable housing and retail. The structure was unique with sloping timber joists to accommodate
early indoor baths with full mortar setting beds. Floors were leveled and strengthened by inserting engineered
wood members (Laminated Veneer Lumber). The exterior walls contained narrow fireboxes and stove flues where
central boilers could not go at that time. These flues needed to be grouted solid as they represented weakened areas
in these bearing and shear walls. Special construction shoring was designed to allow environmental abatement
while providing safe working floors for the selective demolition and abatement crews. Modern Codes (IBC
2003) required a full seismic upgrade and an interactive structural analysis was performed using 1850’s brick
bearing walls in conjunction with interspersed reinforced concrete block walls. Special diaphragm to shear wall
ties were devised and over 1500 installed.
1 INTRODUCTION were remodeled about the same time. About 1950, the
ornate Victorian window trim was removed to “mod-
The Wauregan Hotel was constructed in downtown ernize” the building during the post WWII modern
Norwich, Connecticut, from 1853 to 1855. Local busi- image era in the USA. Construction of the interstate
nessmen, manufacturers, and railroad and steamboat highway system, and motels and hotels by the high-
interests formed a corporation for the purpose of cre- way exits, was a factor in the decline of the Wauregan
ating a first-class hotel, designed to accommodate as a hotel in the second half of the 20th century. By
travelers between New York and Boston, and those the 1980’s, the building was deteriorating. A fire in
visiting Norwich on business or pleasure. The upper 1989 destroyed the boiler room, in a rear addition, and
two floors of the five story building were rented to the Wauregan was condemned. In 1999, the City of
permanent, year round residents. Norwich announced its intention to demolish the hotel.
The president of the corporation was William Buck- The Connecticut Historical Commission, in a decision
ingham, later Republican governor of Connecticut on December 1, 2001, voted to oppose the demoli-
during the U.S. Civil War. Local architect Evan Bur- tion, as the Wauregan was a contributing structure
dick designed the building in the Italianate style. to the Downtown Norwich National Register Historic
The hotel contained rooms for public meetings. Plan- District. The entire complex became a major restora-
ning meetings for commemorative events, such as the tion and development project by Becker & Becker
200th anniversary of the founding of Norwich, polit- Associates and has transformed the building into 70
ical meeting, and public receptions were held at the apartment units with retail space on the first floor
hotel. The most famous guest was Abraham Lincoln, (Plummer, D. 2007).
on March 9, 1860. Lincoln spoke at the Norwich city Significant structural challenges had to be met due
hall that evening, and later held forth in the hotel to the devastating water and environmental damage
parlor. Other distinguished visitors included President sustained during the vacancy period. Modern Code
Ulysses S. Grant and Booker T. Washington, in whose requirements came into play adding complexity to the
honor receptions were given. Gypsy Rose Lee stayed safe structural design required for these upgrades. This
at the Wauregan while giving performances at the paper will highlight some of these unique challenges;
summer stock theater in town. their solution allowing this historic structure to be
The Clarendon Annex, with a second floor dining saved and progressively transformed to a downtown
room/ballroom, was added as a north, integral wing, focal piece and impetus for future urban development
in 1906. The hotel lobby, entrance, and storefronts in Norwich.
1107
2 GENERAL DESIGN REQUIREMENTS coefficient, R, for URM walls is 1.5 as opposed to a
higher value of R = 5.5 for special reinforced masonry
Generally, the building was going from a commercial shear walls which we used in numerous locations as
hotel use to a new multi-family residential use. This replacement walls. As can be seen in the base seismic
change in consort with the major level of renovation shear (V) equation.
triggered the requirement for a structural upgrade to V = 1.25DS W/R, a smaller R value produces a large,
current code standards. Structurally, this meant com- more conservative seismic design base shear. Even so,
plying with Chapter 16 “Structural Design” of the 2003 the high wind loads offset by the lower floor mass (W)
International Building Code (IBC). Both gravity floor produced a higher wind base shear for design purposes.
and roof loads needed to meet the current code require-
ments as well as an analysis and upgrade for lateral
loads caused by either wind or earthquake.
3 STRUCTURAL CHALLENGES
The building’s structural systems were comprised of
sawn timber joists of local pine species for both the low
Due to water damage and severe winter climate of New
sloping roofs and interior floors.These were bearing on
England, the building’s unoccupied years and lack of
the exterior, multi-wyeth brick walls and interior stud
any maintenance caused areas of major deterioration.
corridor or room demising wall partitions. Original
In particular, one area was in extremely bad condition
wood lath and lime plaster was the predominant wall
and collapse in 2004 was imminent. The Wauregan’s
finish material even on the exterior walls which were
rear, reentrant corner five store rear walls were in dan-
wood furred and then plastered.
ger of collapsing. The roof’s valley terminated at the
Gravity live loads for hotel use and modern apart-
corner and water cascaded down these corner walls
ment use were almost equivalent, and most of the
causing excessive damage. In order to save the floor
existing floor joists which were not water damaged
framing in these walls, special two-way heavy timber
were found to be capable of sustaining the necessary
shoring was designed and erected in a tiered fashion.
superimposed loads required for structural upgrade.
This was completed in September of 2004 and over
Where members had minimal damage, new “sister”
the upcoming winter the walls did collapse, exposing
timbers could be added and fastened to the existing
the buildings interior to the weather. The floors were
timber to provide the necessary reinforcing. In the
saved, however, due to the prior shoring being timely
case of major damage of floor timbers and joists, new
installed (See Fig. 1).
members were inserted to replace these excessively
This shoring remained in place until the floor could
damaged ones. Stronger, modern laminated veneer
be adequately reinforced and the new exterior wall
lumber (LVL) members or engineered I joists were
chosen as their depth could replicate the original
timbers and provide added strength and serviceability.
The building’s lateral loads were originally taken
by the mass brick exterior walls as a “box system”
with some interior masonry walls being used to act
as bearing and shearwalls. These augmented the exte-
rior walls and added strength due to the building’s ell
shape. The design basic wind speed of 120 mph for
Norwich produced the greater building design lateral
loading as opposed to the seismic lateral requirement.
Connecticut is in a moderate seismic zone, with the
State’s ground accelerations per Code analysis pro-
ducing less forces in the building’s lateral resisting
structural elements. Two factors contributed to this
result. First, the building’s height (5 stories for the
Wauregan and 6 stories for the Clarendon Annex) pre-
sented a substantial area for wind loads, and second,
the light weight wood floor did not contribute sub-
stantial mass in the earthquake’s lateral load formulas.
The wind’s total base shear was thus in excess of that
produced by seismic analysis. It is worth noting here
that due to the re-use of the building’s numerous orig-
inal unreinforced masonry brick walls (URM’s), the
most conservative seismic design factors were used
in the analysis. The seismic response modification Figure 1. Collapsed corner walls.
1108
reconstructed. Where possible, new reinforced con- Demolition. This of course uncovered numerous struc-
crete or hollow, grouted reinforced concrete masonry tural issues including faults in the major shearwalls
units (CMU) were installed for the necessary shear- due to old fireplaces and, more notably, old unlined
wall strengthening at the building’s lower levels (See flues. These flues were voids in the building shear-
Fig. 2). The rear of the building was being covered walls that were unanticipated during the earlier design
with an Exterior Insulated and Finish System (EIFS) phase. These voids had to be grouted from above, but
and stuccoed a brick color to match the existing fin- the old soot/tar inhibited the bond of grout to brick.
ished brick. This was a cost savings effort applied to a To bond these two together to act as a single shear-
building’s rarely viewed exterior walls. wall unit, new 9 mm Helifix stainless steel self drilling
Again, excessive water had damaged much of the rod anchors were inserted through the brick and void
building’s original wood deck floor including sup- to provide a “shear-lug” anchor system. These were
porting joists. As this original timber deck would not
provide the necessary floor diaphragm to resist loads
due to wind forces, it required replacement or reinforc-
ing. A new, lightweight flowable gypsum floor fill was
specified to be used in floor areas where the old sags or
deflection exceeded a value of 1 in. (2.5 cm). In order
to achieve the required diaphragm design values and
provide a sufficient substrate to receive the gypsum
floor fill, a new layer of 3/4 in. (1.9 cm) plywood was
applied over the original decking. Where the original
decking was rotted and removed, new in-kind decking
was placed prior to the new plywood installation (See
Fig. 3).
As necessitated per Code, prescriptive for seismic
requirements, all floor diaphragms must be attached
to transfer their in plane loads to the building’s shear-
walls. Generally, 1850’s construction was very weak
on this requirement and no original connections were Figure 3. Damaged floor decking.
found. With all new plywood, this diaphragm was then
attached to the brick or CMU shear walls with clip
angles, screws and epoxy inserts to the masonry. Some
1500 such clips were required to provide this vital
positive attachment (See Fig. 4).
The project developer chose not to have the plas-
ter finishes removed prior to bidding the project due
to initial lack of project funding at that point. This
left the finish removal as part of the General Con-
tractor’s job during construction, Phase I – Selective
1109
Figure 6. Trussed ballroom joists.
Figure 5. Flue/grout anchor detail.
1110
5 CONCLUSIONS
1111
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
K. Papadopoulos
Technical Team for the Restoration of the Temple of Apollo Epikourios, Bassai, Greece
ABSTRACT: One of the main objectives of the ongoing restoration programme of the classical Greek temple of
Apollo Epikourios is the repair and strengthening of its foundation. In this paper the restoration of the foundation
in the north side of the temple (where the work has been completed) is presented. In that part of the monument,
foundation failure was caused, mainly, from the existence of a clay layer, beneath the foundation top course
(the euthynteria). Consequently, the restoration involved removing the clay material, and repositioning it at a
later stage, after stabilizing it with lime and cement. Thus, a durable base was created for the resetting of the
restored euthynteria that followed. The work included, also, the built of small, non-visible walls in the north-west
corner, for retaining the subeuthynteria course. The restoration of the temple north foundation was carried out
with respect to the ancient constructions, and did not alter significantly the original structural character of the
monument.
1113
Figure 3. The euthynteria stone-blocks in the north side of
Figure 2. The foundation in the east side of the temple (cross the temple (isometric sketch).
section very close to the north side, Svolopoulos 1995).
1114
Figure 6. Sketch presenting, in exaggeration, the defor-
mations in the foundation of the temple (Papantonopoulos
1995).
Figure 5. The stones inside the bedrock cavity and the
reused, reshaped block in the north limit of the cavity (view
from SE).
1115
It should be noted that, although small creeping set-
tlements are certain to have take place in some parts
of the bedrock during the monument’s lifetime, they
are evidences which denote that no significant local
collapse in the bedrock has occurred. The evidences
are the existence of the group of stones inside the cav-
ity and the fact that the altitudes of the top surfaces
of the subeuthynteria stones are close to the altitude
of the top course of the western ancient retaining wall
and to the altitude of the bedrock in the central part of
Figure 8. The large corner subeuthynteria stone (right) and the north side of the building.
its side stone to the east (left). View from north. It is obvi- Based on the above assumptions, the course of
ous that the stones have settled, inclined toward the slope in events that led to the damages of the temple in its north
the west, and that their cracks are at the contact points with
side can be described as follows: The north façade,
the rock.
as the whole building, remained intact for many cen-
turies, until the collapse of its roof, which, probably,
took place due to the biological decay of the wooden
supporting elements (historical data report that the
temple was whole with none noticeable deforma-
tion, six centuries after its construction!). The absence
of cover allowed rainwater inside the building, onto
the blocks of the steps. As a consequence, the rain-
water chemical action caused material deterioration,
diminishing the surfaces of the blocks, opening the
gaps in their interfaces. This fact permitted rainwa-
ter to penetrate further, until, eventually, it was able
to reach the clay layer beneath euthynteria. Then,
rainwater started to wash away clay material and to
make easier the creeping flow of clay toward lower
Figure 9. Left: The numerical model. Right: North (up) and parts of the bedrock, causing: (i) reduction of vary-
south (down) views of the model lowermost part. ing magnitude to the clay layer thickness; (ii) sunk
of the subfounadtion stones; and (iii) slide of the
subeuthynteria course. Those damaging for the temple
(ii) The stones based on thick clay have settled signif- foundation slow natural phenomena accelerated when
icantly, inclined in every case toward the side where pillagers, searching for the metal clamps, removed the
the below clay layer was thicker. This was very clear intercolumnar slabs and broke the façade blocks of
in the euthynteria blocks placed above the bedrock the steps.
cavity (Figs 7, 9 right). (iii) The blocks based partly The deformations in the supporting system of
on bedrock and partly on clay were fractured across the euthynteria generated non-uniform settlements
the base condition differentiation line, with their frag- and irregular displacements of its blocks. (However,
ments based on clay suffering the larger settlements. because almost all the blocks remain fully based on
The latter was quite obvious in the reused block placed clay, they did not fracture). Naturally, the euthynteria
beside the group of subfoundation stones (Fig. 7), subsidence led to simultaneous deformations in the
and in the north stones of the subeuthynteria course blocks of the steps, which caused loss of full contact
(Fig. 8). in the blocks interfaces and subsequent modification
Thus, the data strongly indicate that, during the cen- of the blocks loading condition from compression to
turies, a gradual reduction of the clay layer led to the tension. The latter alteration induced material failure
settlements of the foundation stone members. The rea- in the weakness points of the blocks; therefore, fur-
sons for the clay layer diminishing were the washing ther settlements occurred in the steps, causing columns
away of material from rainwater (which started pass- inclinations and entablature deformation. The first
ing into the temple, after the collapse of the roof), cracks in the blocks of the steps led to rearrangement
and the creeping flow of clay toward lower parts of of their contacts and consequential creation of new
the bedrock, under the loads of the building and with point or line loadings, which caused new fractures and
the decisive cooperation of rainwater. The gradual clay so on.
escape was toward the lowermost parts of the cavity It is worth noticing that as long as the temple
(Fig. 7), through the fissures and the discontinuities remained intact, earthquakes did not damage it signif-
of the bedrock, and down the slope in the west side icantly. Only after human intervention (the breakage
(Fig. 8). of the blocks and the looting of the clamps by the
1116
pillagers), strong earthquakes probably disintegrated are cracking in tension and crushing in compression.
further the building, by opening the blocks joints and The properties adopted in the analysis were based
their fragments gaps or/and by creating instantaneous on experimental study about the temple limestone
or residual stresses concentrations that led to new material (Papantonopoulos 1995) and are for the mod-
blocks fractures. ulus of elasticity 80.1 GPa, for the Poisson’s ratio 0.35,
for the tensile strength 1.4 MPa and for the compres-
sive strength 38.5 MPa. The material behaviour of
the rest blocks was simulated using a linear model,
4 NUMERICAL REPRODUCTION OF THE
with modulus of elasticity 80.1 GPa and Poisson’s
DAMAGES IN A PART OF THE TEMPLE
ratio 0.35.
The interactive behaviour of the model members
4.1 The numerical model
normal to their interfaces was determined with the use
In order to further enforce the assessment that foun- of a ‘hard’contact model that allows when two surfaces
dation failure is responsible for most of the damages are in contact, any pressure to be transmitted between
in the north side of the temple of Apollo Epikourios, them and, when the surfaces separate, it reduces the
a numerical analysis was conducted with the aim to contact pressure to zero. In the tangential direction of
reproduce the damages in part of the monument, as the interfaces, a classical friction model was used, with
a consequence of the euthynteria deformations. For friction coefficient of 0.75 (value derived from tests,
that purpose, with the use of the software Abaqus, a Papantonopoulos 1995).
3-dimentional model was created, which included the
second from the east column of the north side, and
the blocks of the steps and of the euthynteria that are 4.2 The uncertainties in the analysis
beneath the column (Fig. 9 left). In other words, in
It should be noted that the numerical analysis contains
the analysis it was simulated the part of the temple
uncertainties related to the simplifying assumptions
which is over the cavity of the bedrock, without the
made in the simulating process, as well to the complex
entablature.
parameters that influence the dynamic behaviour of
The model’s geometry definition was based on
the temple. The main points that render the numerical
actual measurements of the temple stone blocks. The
predictions as a rough estimation of the behaviour of
model was formatted in such a way so that all the
the simulated temple part are:
blocks (except those of the euthynteria) were simu-
The limestone of the monument is anisotropic, due
lated in the state and position that they were in the
to the layers and the discontinuities of its structure;
initial, undamaged condition of the temple; meaning
instead, in the analysis it was simulated as a isotropic,
as distinct members, in full contact, with no relative
continuous material. The state of the temple when
displacement, and in vertical or horizontal positions.
foundation failure occurred and cracks appeared in the
On the contrary, the euthynteria blocks were sim-
blocks, is unknown. Hence, it is uncertain if the model
ulated inclined and displaced (Fig. 9 right), so as
should have included stone members from the entab-
to represent the documented in-situ settled positions
lature, for more accurate simulation of the dead loads
of the corresponding temple blocks. The aim was to
that caused material failures. Furthermore, dynamic
include in the analysis the deformations of the tem-
actions due to seismic events are not included in the
ple foundation, for finding out what damages will
analysis; although those actions may have partly con-
appear in the superstructure, as a consequence of the
tributed to the monument geometrical deformations or
deformed foundation. In order to succeed that, the
material failures. Finally, elements representatives of
euthynteria blocks were fully constrained, while the
the metal clamps in the blocks of the first two steps
rest of the model was free to deform nonlinearly in
were not included in the numerical model.
any direction under its own weight.
Gravity loading was the only load applied to the
model; it was ramped up over 0.3 sec, and then it was
4.3 Numerical results – conclusion
maintained constant. The dynamic analysis total time
(1.5 sec) was sufficient for the full appearance of the The numerical analysis resulted, as it was expected,
damages in the model. For simplicity purposes, the both significant displacements and rotations of all the
model discretization took place using 8-nodes hexa- model members and material failures in the blocks
hedra elements that resulted to the cross sections of of the steps. The deformed model is presented in
the column shaft to be polygonal, instead of circular Figure 10. It is quite obvious that the steps have set-
with 20 flutes, as in reality. tled significantly, and that the column, carried along
The material behaviour of the blocks of the three by the steps, has subsided and inclined toward SE,
steps was simulated using a non-linear model, suit- that is toward the position of the bedrock cavity in the
able for quasi-brittle materials which elastic behaviour monument (Fig. 7). Specifically, the analysis predicted
is isotropic and linear and its main failure mechanisms maximum column deviation from the vertical position
1117
5 THE BASIC MATERIAL USED IN THE
RESTORATION OF THE FOUNDATION
1118
of the monument), 3.0% lime, 7.5% cement, and 6 THE RESTORATION OF THE
14.3% water. The basic experimental results about the FOUNDATION
mechanical properties of the stabilized clay mixture
which, eventually, was used in the temple restoration 6.1 Consolidation of the subfoundation stones
work are the following: The material developed, at and the subeuthynteria course
the age of 28 days, compressive strength 2.156 MPa
For the consolidation of the subfoundation stones
(average value from testing 8 cylinder specimens,
found in the north side of Apollo Epikourios temple,
d/h = 7/14 [cm]) and, at the age of 63 days, 3.204 MPa
the five scattered stones and the top stone of the
(from tests on 3 cylinder specimens); its indirect ten-
group inside the cavity were removed and structurally
sile strength was also measured at the age of 28 and
restored; meaning the fragments of the stones were
63 days, and it was found 0.341 MPa and 0.651 MPa,
connected using threaded titanium bars and white
respectively (the latter values were deduced from
cement paste, according to the technique developed
‘Brazilian tests’ on 8 and 2 specimens).
in the restoration work of the Acropolis monuments
(Zambas 1988, Vintzileou & Papadopoulos 2001).
Then, the stones were conserved and repositioned.
5.3 Test implementation In the cases where the stones did not fully based on
Before the use of stabilized clay in the restoration work bedrock, the void between the stones and the rock
of the temple, its behaviour was tested under real load- was filled with stabilized clay. The stones were reposi-
ing conditions in a structure that simulated a small part tioned horizontally, with altitudes resulted from obser-
of the building foundation. More specifically, a layer vations about the altitude variation of the proximate
of stabilized clay, 20 cm thick, was constructed on the bedrock surfaces.
bedrock by the monument’s east side. On the stabilized The consolidation of the group of stones inside the
layer two stones were laid, within the first hour from bedrock cavity was accomplished with the construc-
the layer construction. tion of a wall in the remaining empty cavity space.
The stones were in shape and in size similar to the The wall was built with stones from the local area and
temple euthynteria blocks (Fig. 12 left). On top of the stabilized clay as binding material.
stones, large blocks were placed (Fig. 12 rigth) for fur- For the consolidation of the subeuthynteria course
ther loading of the stabilized layer. The loading was in the north-west corner of the monument, trenches
applied for over a year, gradually increasing, achiev- 1.0–1.5 m wide were excavated across the north and
ing at its peak, 90 kPa compressive stress at the layer south limits of the course, both in its external and inter-
(value larger than the loading which the stabilized layer nal sides. It should be noted that for the excavation of
will bear in the restored state of the monument); during the west external trench a part of the artificial fill, con-
that time, periodic altitudes measurements were taken, sisting of clay and stones of irregular shape and size,
which showed that the stabilized layer settled, about, had to be disturbed; however, the trench width was kept
1 mm. the minimum necessary, and the removed stones were
The diminutive layer settlement, in combination replaced in nearby positions. The trenches extended
with observations made during the preparation of the down to the bedrock, and underneath the course for a
stabilized material and the laying of the stones (such width of 0.50 m.
as they were no voids between the stones and the layer, In the trenches, walls were built to support and
Fig. 12 left), led to the overall conclusion that stabi- retain the course (Fig. 13). The walls were built with
lized clay exhibited satisfactory behaviour in its test some of the stones removed from the ancient structure,
implementation and it can be used in the restoration
work of the temple.
Figure 12. Left: The two stones placed on the stabilized clay Figure 13. The consolidation of the subeuthynteria course.
layer. Rigth: Loads upon the stabilized layer. Cross section parallel to the transverse axe of the temple.
1119
so that the joints of the stabilized layer are not in the
same plane with the vertical joints of the blocks.
For the restoration of the foundation in the north side
of the temple 45 euthynteria blocks were restored and
repositioned (Fig. 14). The restoration of the temple
north foundation was completed with the reconnection
of the euthynteria façade blocks using new titanium
clamps. The new clamps are designed to improve the
monument seismic behaviour, without inducing blocks
fractures (Papadopoulos 2007).
7 CONCLUSIONS
Figure 14. The restored foundation in the north side of the
temple, from NW.
Although Apollo Epikourios temple is one of the
best preserved buildings of classical antiquity, it faces
stones from the local area distinguished by indica-
many worrying structural problems, one of which is
tive sign, and stabilized clay as binding material. The
foundation failure. In the north side of the temple
remained empty spaces of the trenches were filled with
foundation deformations were caused, mainly, from
clay and the rest removed stones.
the existence of a clay layer beneath the foundation
top course (the euthynteria). New data, revealed in the
6.2 Resetting of the euthynteria blocks
course of the dismantling (for restoration purposes) of
The resetting of the euthynteria blocks in the north side the temple north side and subsequent damages diag-
of the monument followed the consolidation of the nosis further enforced the aforementioned assessment,
subfoundation stones and the subeuthynteria course confirming the necessity of the large-scale interven-
and took place simultaneously with the construction tion. Consequently, the restoration of the monument
of the stabilized clay layer, which was interjected foundation involved removing the clay material and
between the blocks and the bedrock. All the blocks repositioning it at a later stage, after stabilizing it with
were structurally restored and conserved before they lime and cement. Thus, a strong and durable base was
were repositioned. created for the resetting of the structurally restored and
First repositioned block was the eleventh from east conserved euthynteria that followed. The exact com-
of the façade row (EU-N11).This block is based almost position of the stabilized material was derived from an
entirely on bedrock and it is from the euthynteria part experimental investigation and, prior to its implemen-
which presented the minimum deformations. Thus, the tation on the monument, its behaviour was checked
position and the altitude of the block before the inter- under real loading conditions, in a structure that sim-
vention were considered as the least deviated from ulated a small part of the foundation. The restoration
the initial ones; therefore, block EU-N11 was selected work in the north foundation of the temple included,
as the key stone of the resetting. After block EU- also, consolidation of the subfoundation stones and the
N11 the resetting continued towards the two ends of subeuthynteria course in the north-west corner, which
the north side, simultaneously. All the blocks were was carried out with the construction of supporting
repositioned in horizontal positions, with top face alti- and retaining walls, using stabilized clay and stones
tude 1130.119 m (value equal to the altitude of block from the local area.
EU-N11). The restoration of the temple north foundation was
A brief description of the blocks resetting procedure carried out with respect to the ancient constructions,
is presented below: Initially, the block is placed on and did not alter significantly the original structural
wooden wedges in its four corners, fact which allows character of the monument. In order to attain the latter
microadjustments of the block position; then, after objective, the restoration study was undertaken from
the approximate finding of the block resetting posi- an approach based on the one hand, on an attempt to
tion a load is placed temporary on top of the block, understand the monument’s structural behaviour and
for securing its position. Afterwards, stabilized clay its repair needs, and on the other hand, on numerical
is inserted in the gap between the block bottom face and experimental analyses. Furthermore, although the
and the bedrock, and it is compacted. Finally, the last study and its consequent intervention practices were
corrections of the block position are made, using topo- aiming at the restoration of a small part of the tem-
graphical instruments for the position checking. The ple foundation, they can also be applied when the
next day the wedges and the load are removed and the intervention expands to the remaining parts of the
resetting continues to the adjacent block. Care is taken monument.
1120
REFERENCES Preservation of the Temple of Apollo Epikourios (in
Greek, with English summary).
Kolias, S. 2003. Stabilization of the clay layer in the foun- Svolopoulos, D. 1995. Temple of Apollo Epikourios. Archi-
dation of Apollo Epikourios temple. Athens: National tectural study. Athens: Committee for the Preservation of
Technical University of Athens (in Greek). the Temple of Apollo Epikourios (in Greek, with English
Papadopoulos, K. & Karahalios, A. 2003. The material for summary).
the restoration of the foundation in the north pteron of Vintzileou, E. & Papadopoulos, K. 2001. Dowel action of tita-
Apollo Epikourios temple. Athens: Committee for the nium bars connecting marble elements. In R. Eligehausen
Preservation of the Temple of Apollo Epikourios (in (ed.), Connection between steel and concrete; Proc.
Greek). intern. symp., Stuttgart, 10–12 September 2001.Bagneux:
Papadopoulos, K. 2007. The restoration study of the con- RILEM.
nections between the stone-blocks in the steps of the Zambas, C. 1988. Principals for the structural restoration of
temple of Apollo Epikourios. In P.B. Lourenço, P. Roca the Acropolis monuments. In P. Marinos and G.C. Koukis
C. Modena & S. Agrawal (eds), Structural analysis of his- (eds), The engineering geology of ancient works, mon-
torical constructions; Proc. 5th intern. symp., New Delhi, uments and historical sites; Proc. intern. symp., Athens,
6–8 November 2006. New Delhi: Macmillan. 19–12 September 1988. Rotterdam: Balkema.
Papantonopoulos, C. 1995. Temple of Apollo Epikourios.
Structural restoration study. Athens: Committee for the
1121
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
M. Mariani
Studio per le Ricerche Applicate – Arch. Ing. Massimo Mariani, Perugia, Italy
ABSTRACT: Between the years 1990 and 2001, restoration and strengthening operations were carried out on
the Temple in two distinct phases: firstly, in order to face the historical foundational difficulties of this edifice,
which had begun as far back as at the completion of its construction (spread roughly over a century, from 1508
to 1606); secondly, in order to repair the damages to its higher structure caused by the earthquakes in Umbria
and Marche, which had begun in 1997. After an ample description of the building, the geology of the Todi Hill
is defined, followed by the diagnostic and analytical phases of the project (1985–1987), the project itself (1987)
and the restoration and strengthening operations (1990). Subsequently, we describe the preliminary studies, the
project and the conservation and repair work carried out on the monument after it was damaged by earthquakes
beginning in 1997.
1 INTRODUCTION
1123
reference about his presence, have come to the con-
clusion that it is the fruit of a worthy conception of
little-known architects and builders.
De Angelis D’Ossat, contrary to this theory, reports
that Bramante accepted the project commission and
carried it out, but, “being unable to go there person-
ally... sent Ventura Vitoni da Pistoia in his place”.
Todi tradition, as early as 1574, and as testified by
the Apostolic Visitor Pietro Camaiani when the works
on the Temple were not yet finished, also documents
that the Consolazione Temple “was designed by Bra-
mante”, and this conviction remained constant until
1872, when Adamo Rossi discovered the name of Cola
di Matteuccio da Caprarola in the construction records,
whereas he never found the name of Bramante during
his research.
Cola appears in 1499 next to Antonio da Sangallo
the Elder, working for Pope Alexander VI as a papal
fortress builder. In Todi, in the construction of the
Consolazione, he appears from 1508 in the records
as a magistro contractor, a construction conductor a
cotimo or coptimarius or murator, although from May
1509 until the end of his involvement in 1515, he is
called architettore. Cola was also contractor of the Figure 2. North-South section with identification of the
papal fortress in Civitacastellana designed by Antonio main vertical crack.
da Sangallo the Elder, who was most likely assisted by
Bramante as sottoarchitettore for Pope Alexander VI.
Reports from those times also tell of the collapse of • 1524/1563 – Filippo di Salvi da Meli, architect and
a cross vault in the Foligno Duomo, built by Cola as sculptor;
contractor, for which Bramante’s advice was sought • 1565 – Giovandomenico Berzugli da Carrara, archi-
first. Cola da Caprarola is therefore an architect- tect;
contractor, certainly connected to the Roman Sangallo • 1565 – Jacopo Barozzi da Vignola, architect;
and Bramante circle. • 1584 – Guglielmo Portoghese, architect;
This seems to confirm Jurgen Zanker’s theory, • 1584/1587 – Valentino Martelli, architect;
which vehemently rules out any intervention by • 1584/1594/1597 – Ippolito Scalza di Orvieto.
Bramante, assuming that “the Consolazione archi-
Even from the first decade of the 17th century, just
tecture is the result of various consecutive projects,
after its building was completed, up until today, the
which were drawn up or modified during construction,
stability of the Temple’s structure has often been put
depending on the opinion of the various architects who
to the test and has revealed its hardship in cracks and
followed each other.” And he concludes: “there is no
structural element collapses, which were mainly due to
single architect thanks to whom the Consolazione was
differential foundation subsiding or to seismic activity,
built, but there are several of them.”
as was the case during the most recent earthquake in
It is indeed documented that from 1508 – the year
1997 (Fig. 2).
in which the works began – onwards, several architects
In 1638, the tie-beam which closed the eastern apse
took part in the construction in a way that was anything
broke; the same tie-beam broke again another two
but marginal; and it could not have been otherwise
times in the following centuries, and in 1670 an inves-
because the construction was finished in 1606, that
tigation was carried out, which consisted of installing
is to say almost a century later. From the documents
topographic reference points with the intention of mea-
collected, here is the list of the architects who took part
suring the relative movement between the construction
in the construction of the Temple:
and the natural slope; this verification was necessary
because a part of the monument is built on settled
• 1508 – Cola da Caprarola, architect; ground built with the aim of creating a square facing
• 1509 – Gabriele di Giovanni da Como, construction the valley.
master; After subsequent interventions and continual
• 1515 – Giovandomenico da Pavia, architect; episodes causing instability leading to the complete
• 1518 – Baldassarre Peruzzi, architect; exposure of the foundations of the southern apse,
• 1516/1525 – Ambrogio da Milano, architect; which had already undergone under-foundation work
1124
in 1792, a semicircular gravity wall, designed by the
engineer Luigi Poletti, architect for the Papal State,
was built to the South towards the valley between 1836
and 1860, aimed at contrasting the thrust of the square’s
protuberance.
In 1926, following serious damage to the drum, the
eastern apse underwent consolidation work, consisting
mainly of creating an under-foundation to support the
southeastern pilaster.
In 1953 a Technical-Scientific Committee was set
up which had the task of examining the episodes of
instability which had caused a large crack to open up
along the median of the semi-cupola above the eastern
section, and the consequent damage to the external Figure 3. Southern gravity wall project (Luigi Poletti 1836).
walls.
The Committee had the task of suggesting pos- of about 25 square meters. The apses and the pilasters
sible remedies to finally put an end to the ongoing are built at different depths; the former, about 20 cm
dynamism. wider on each side compared to the higher section, are
The resulting consolidation work mainly consisted 2 m below the ground level, whereas the latter are 4 m
of the injection of a cement-like mixture into the foun- below (without considering the under-foundation work
dations, with the intention of making them monolithic. carried out on the north-eastern pilaster at a depth of
16.50 m, and on the eastern apse).
The weight of the structure is about 17,000 tons,
3 THE BUILDING composed of 6250 tons for the columns and the pen-
dentives, 1870 tons for each polygonal apse, 2030 tons
The church, which has a Greek cross plan, stands on for the altar’s circular apse, 1740 tons for the drum and
a structural system resting on four large pilasters at 1370 tons for the large cupola and the smaller cupola.
the top of a square, on the sides of which the apses The building transmits, on average, 0.95 MPa from
are placed: three of these are polygonal and one, the the pilasters and 0.43 MPa from the apse walls to the
one in the north, is semicircular with its centre slightly foundation grounds.
moved towards the geometric centre of the church. The pilasters and the external walls are composed of
The square formation, brought about by the limestone block masonry held together by lime mortar
pilasters, is also found in the terrace which can be on the outside, and masonry with not regular filler on
reached by a stairway set in the shaft to the left of the the inside.
choir stalls.
The pilasters, by means of large arches, sustain the
drum and the cupola; the latter has a small lantern 4 GEOLOGY OF THE TODI HILL
above it, also covered by a cupola. Twenty double
embrasure windows, all surmounted by a tympanum, The geologic formation of the Todi hill is due to
open up between one pilaster and another. This same the sedimentation of lacustrine Pliopleistocene ori-
scheme is repeated, with the same number of square gin in the ancient Tiber Basin, which extended to
windows, beneath the impost of the apsidal cupola. the Umbrian Valley, the Tiber Valley and other Minor
The drum contains a sequence of alternating windows Valleys. The hill, originating at the south-western
and niches. edge of this basin, is made up of distinct lithological
The Church’s masonry consists of excellent qual- sequences, at the lower level, in two complexes:
ity calcareous stone originating from the Titignano
• basic clayey complex: grey-blue silty-sandy clays,
quarry, near Todi, and partly also from the demolition
locally marly, with an average slope of a few degrees
of the medieval fortress nearby.
towards north-east;
The Temple is set in a wide square, almost com-
• peak conglomerate complex: shingle and sand with
pletely made with an embankment supported by a
a slimy matrix and lenticular position.
gravity wall to the south, built, as already mentioned,
by the engineer Luigi Poletti between 1836 and 1860 On average the soil is plastic (Ip = 20% ÷ 30%) and
(Fig. 3). strongly over-consolidated.
The building is about 50 m high and seen on a loca- These materials have irregular surfaces of tectonic
tion plan, it can be contained within a circumference origin and deposits of thin layers of sand.
of a diameter of about 53 m. The apse walls are about The piezometric level of the aquifer layer inside the
2 m thick, and the pilasters’ imprint covers a surface town as well as on the hillside, in a modified and
1125
disrupted surface environment, is not always paral- It may thus be deduced that the presence of ver-
lel to the external soil profile. This variability is due tical and sub-vertical discontinuities (cracks) in the
to the existence of soils with different positions and soil structure represents a dominant element for the
different permeability levels: high permeability in the spreading of the swelling and for the lessening of its
detritus, in the sand levels and in the conglomerates mechanical capacities.
(K = 10−2 ÷ 10−3 cm s−1 ) and low permeability in the The swelling obtained in the laboratory is quite rep-
silt (K = 10−4 cm sec−1 ) and, even more so, in the clays resentative of the actual swelling that occurs on site,
(K = 10−5 ÷ 10−7 cm sec−1 ). and it is accompanied by a considerable reduction of
The undrained (cu ) and drained (c ) cohesion lies the effective cohesion, mostly due to the increase in
within the following values: water content.
Taking into account the fact that clay cohesion
(and silt cohesion) depends on particle cementing,
on electrostatic forces which they exchange amongst
each other individually, and on the binding between
This geologically young sedimentary composition,
atoms, one becomes even more convinced that, sur-
which has always been influenced by morphogenetic
passing a certain distance limit between one particle
factors, is prone to continuous modifications due to
and another during the swelling of the composition, a
natural deterioration, constantly increased by damag-
polarity inversion sets in and repulsive forces begin to
ing anthropic contributions.
develop which cause, as already mentioned, a lessen-
The consolidation interventions, financed by subsi-
ing of the cohesion. In summary, the persistence of the
dies deriving from the Law 545/87 and the Regional
over consolidated clays and silts in Todi in a “wet-dry”
Council Resolution 548/98, were carried out on all the
situation due to the alternation of the stratum because
hill’s slopes subject to landslides threatening the town
of season variations, causes the constant lessening of
boundaries, which have always been prone to upheaval,
their mechanical properties, and in particular of the
nowadays still visible in the deficits and cracks in some
effective cohesion in the wet-dry area.
of its structural parts. The landslides are mainly due
The Umbrians, the Etruscans and the Romans set-
to two predominant factors: the continuous incision
tled in this beautiful but unsafe – because of possible
in the ditches and the alteration of the clays and silts
landslides and earthquakes – landscape, and with time
below the stratum. The incision in the ditches is due
they enlarged the town, adapting it to the hill’s config-
to the dynamic action of the meteoric waters and the
uration and expanding it with terracing that not only
chemical aggression of substances found in the sew-
filled furrows and ditches, but also acted as a boundary
ers which deteriorates the organic components of the
wall.
clays and silts. The continuous deepening of these inci-
This settlement, as also happened on other hills of
sions on the hill extends to bordering areas, causing
sedimentary origin of the Tiber basin (Perugia, Mon-
the retrogressive loss towards the hill of the hillside
tone, Ilci, Monte Castello di Vibio, and so on), was
stability.
certainly advantaged by the easy drawing of water from
It was only in the 1980s that the reasons for the con-
the stratum which was not very deep, thanks to wells
stant deterioration of the clays and silts of the Todi hill
and drainage shafts.
below the stratum and the lessening of their mechani-
cal capacities were revealed in laboratory experiments.
Research in the last twenty years has led to new knowl-
edge about these soils with regard to the stability of the 5 DIAGNOSTIC PHASE
slopes and in particular to the swelling and alteration
processes of the surface layers. An initial study and research phase aimed at examining
It is precisely the Todi clay that was the sub- the Temple’s stability was carried out by the Turin Poly-
ject of an experimental study programme undertaken technic in the context of the agreement entered into
at the geotechnical laboratory of the Structural and with the Umbrian Monuments and Fine Arts Office
Geotechnical Engineering Department of La Sapienza No 1072 of 1985, rep. 2961.
University in Rome (Calabresi, Esu, Pane, Scarpelli, It lasted over two years and included the carrying
Rampello). out of topographic and geo-mechanical measurements,
From the analysis performed on undisturbed soil drilling, laboratory tests, numerical analysis and the
samples immersed in the same water as the stratum setting up of instruments to measure the gradient of
on site, it appeared that the swelling of the material the slope to check the static behaviour of the struc-
was independent from the history of the applied loads, tural elements and their interaction with the foundation
which occurs “in more time than the consolidation pro- ground.
cess, considering the same tension level”, and that the The researchers who undertook the study were Prof.
inter-particle binding intensity in the Todi clay is not Bruno Astori and Prof. Roberto Chiabrando for the
very strong. topographic part and Prof. Gian Paolo Giani for the
1126
geotechnical part. Later on, in 1990, the consolida- as proof of the recurrence of the instability episodes
tion of the Consolazione Temple was planned as the concerning the elevation structures, the foundations
funds connected to the Special Act for the Todi Hill and southern side protuberance on which it partially
and the Orvieto Rock were made available (Act No rests, which led the researchers to set up a programme
545/87). On that occasion, further geognostic inves- of experimental measurements, which in essence were:
tigations were carried out which completed the ones
carried out by the Turin Polytechnic, and inclinometers • the monitoring of the movements of the structures
were installed on the square and just below the square’s and the terrain of the foundation;
support wall, on which a lengthy monitoring was car- • the definition of the loads affecting the pilasters and
ried out, the results of which were always connected the apse walls;
to the atmospheric precipitations occurring. • the variation in the opening of the existing cracks;
• the mechanical characterization of the materials
making up the structures and the terrain.
5.1 Geognostic investigations and geotechnical
laboratory tests (Turin Polytechnic) Static analyses were also carried using numerical
modelling, aimed at reconstructing the history of the
Three geognostic drillings were carried out in addition foundation’s terrain and at calculating the differential
to those which had already been done in the inves- subsiding and the rotations of the pilasters’ foundation
tigations for the “Definitive consolidation of the Todi base, in order to be able to trace the reason behind the
Hill and Orvieto Rock” (Law 280/78) and to those pre- drops measured. Tests were also carried out to check
liminary to the undersigned’s project for the Temple’s the stability of the whole southern side of the Todi
consolidation. Hill, which has always been subject to landslides due
The geotechnical laboratory investigations carried to erosion at the foot of the Naia Torrent slope, at the
out by the Turin Polytechnic consisted in: granulom- Arnada Torrent inlet.
etry, Atterberg limits, humidity and volume weight, The topographic survey carried out by the Turin
edometric tests, three-axial, isotropically stabilised Polytechnic mainly concerned altrimetic monitoring,
and undrained tests. with the creation of a wire-mesh installed around the
The geotechnical parameters obtained were: Church. Twelve points (which became fourteen in the
summer of 1987) were fixed to monitor the Church’s
• OCR consolidation degree; possible differential movements in relation to the sur-
• RR recompression ratio; rounding slope, as well as the differential movements
• CR pressure ratio; of the various parts making up the structure Particu-
• SR drain ratio; lar attention was given to the southern area, where a
• Ed load deformability module; benchmark was fixed near the support wall below the
• Es drain deformability module; Temple.
•
σvp maximum and effective pressure that has Another three benchmarks were placed on the
strained the land in its history; northern side (originally there was only one, but as
• σvo present effective vertical pressure; anomalies in the behaviour of the wall were noted,
• K compression module (load and drain); another two benchmarks were added). All the external
• p average effective pressure. points were linked to each other by means of a closed
polygon hinged to the same point where it converged
Further investigations, preliminary to the planning and the high precision geometric levelling closed ring
of the consolidation of the depth upheaval, in addi- was bound, inside the Church.
tion to the thorough research made by the Turin Another fundamental operation concerned the mea-
Polytechnic, made it possible to identify a shifting suring of the pilasters’ drop variation using the four
landslide with its detaching edge immediately above brackets cemented from 1953 (for the same purpose),
the temple, involving Poletti’s wall which supported placed at a height of approx. 12 m above the pilaster
the square obtained from an embankment, and the base. A wire was fixed to each of these, with a 5 kg
entire structure. weight at its extremity immersed in a basin full of oil
The modelling of the upheaval, based on the inves- to reduce oscillations (Fig. 4).
tigations, apart from small differences, coincided with The large crack on the drum and on the semi-
the schematization of the past. cupola of the eastern apse, which also cut through the
portal, was put under control by means of a milles-
imal crack measuring device. During the two years
5.2 Topographic precision monitoring (Turin
of Research, seven measurement campaigns were car-
Polytechnic 1985–1987)
ried out in addition to the initial one (December
It was precisely the analysis of the documents regard- 1985), which served to determine the initial reference
ing the Temple’s history from the 17th century to date, measurements.
1127
Figure 4. Historic development of the drops.
1128
movement without discontinuity. In the calculation,
the unstable terrain was modelled using flat elements
with 3 or 4 nodes with two leeway degrees for each
junction (translation according to Y and Z) and a level
state of deformation.
The mesh of the elements in the proximity of the
estimated position of the contrast bulkhead was inten-
tionally thickened to collect the values of the tensions
acting on it with greater precision. The model was
bound along the sliding area and to the area which
was in contact with the bulkhead of piles. The weight
which the Temple puts on the ground in upheaval was
not taken into account, because at that stage it had
already been decided to carry out an underpinning with
small diameter piles over the entire foundation area. A Figure 6. Planimetry of the operation on the upheaval. Bulk-
supposition, also used for calculating the micropiles, heads with large diameter piles and depth reinforcement with
was that the whole weight of the building be entirely small diameter piles.
transferred to the ground below the sliding area. The
comparison between the results of the finite element
analysis on the consolidated system and those of the
stability test has shown a substantial qualitative and
quantitative coherence of the two methods used.
6 CONSOLIDATION INTERVENTIONS
1129
Figure 10. Detail of the internal binding.
1130
sulphate-resistant hydraulic mortar sediment, without
alkaline water-soluble salts and with a minimal com-
pression resistance Rc ≥ 10 MPa.
6.5 Conclusions
The structural consolidation methods thought up to
save this important example of Italian Renaissance
Figure 11. Tensioner apparatus for the external “B” band.
architecture, which has always been under threat from
The sketch shows also the junction element.
landslides and earthquakes, are two-fold. The two
approaches used belong to scientific fields that are
band’s connection elements for the restitution of con- close to each other but distinct: geo-technology and
tinuity at the tensioners’ level could be soldered when seismology. Specific and evolved technology belong-
the thermal dilatation had already occurred. ing to the two sectors has been applied, especially
Even in this case, the spaces between the masonry that which is in no way invasive, and that respects the
and the band, once positioned, were sealed with magnificent building in question.
1131
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
J.D. Miller
MA CEng MICE MIStructE Membre CNISF, Technical Director, Gifford
ABSTRACT: This paper considers the necessary pre-requisites for a successful conservation project in terms of
the profile of the consultancy team, its relationship to the client and curatorial representatives and the functioning
of the contract for execution of the works. It also draws out a number of examples of interest from the project at
Westminster Hall:
• Issues of document research and archiving
• The nature of the initial comprehensive diagnostic investigations
• The safe application of heavy ground engineering techniques in the context of materials of very high
archaeological significance
• The conservation philosophies applied to the work
• The conclusions of the post-project review.
1133
Of particular interest to us is the history of the An OJEU notice was published by the Parliamentary
floor of the Hall. It appears that the original medieval Works Services Directorate towards the end of 2004
masons took steps to address the poor ground they and Gifford were appointed to carry out the commis-
found when they laid the earth floor by capping the sion, commencing in January 2005. The brief invited
ground with a layer of compacted clay. Richard II laid best conservation practice because it required diagno-
a Purbeck stone floor. sis of the cause and options to be presented, not just
Then between 1834–7 Robert Smirke, architect a solution. The brief also included new railings to the
for the British Museum date and Somerset House, steps and a new plant space beneath the steps but these
undertook extensive excavations and archaeological aspects are not discussed in this paper.
recording of the ground and earlier floor levels as part
of his restoration of the Hall and the building of the
3 RESEARCH
floor we see largely unchanged today. He built it using
some 500 Crosland HillYork flagstones each weighing
From the outset, considerable time was spent by
around 750 kg, supported, as we confirmed during this
the project engineer and archaeologist in the early
project, on approximately two-courses of brick sleeper
stages of the project researching the archives with the
walls around the edges of each flag. These walls were
Palace Archivist, and the Institution of Civil Engi-
supported on a 400 mm thick mass lime concrete slab.
neers, amongst other records. These provided initial
Sir Charles Barry, with input from Pugin, won an
background knowledge of the site and history of
architectural competition for designs to rebuild the
Westminster Hall and then, as the project progressed,
Houses of Parliament. His vision was for Westminster
the archaeological team was able to search for specific
Hall to become the entrance to the Houses of Parlia-
information to inform, interpret and explain findings
ment. In 1852 he transformed the Hall by removing
from the site work. So the project gradually added to
the arched window from the south wall of the Hall to
the body of information about the site. Archaeological
create an opening to the new St Stephens porch. He
research examined the site from times when inter-tidal
added a magnificent processional accent steps from
sediments were deposited, research continued through
Westminster Hall to St Stephens Porch thus creating a
the history of the construction methods and materials
staged setting and continuous movement of people in
to the social and political use of the Hall, all of which
the previous static layout of the Hall. He constructed
were pertinent when piecing together the value and
the steps by supporting Hopton Wood stone treads on
significance of the Hall and interpreting geotechnical
a series of closely spaced brick walls and piers with
data.
arched openings to allow access across the under the
Research also yielded invaluable details of set-
stair space.
tlement monitoring work from the 1940’s, drawings
Westminster Hall is Grade I listed and part of the
showing some construction details of the nineteenth
UNESCO Westminster World Heritage Site.
century steps and proposed repairs in the 1960’s. As
consultants we were in the rare position of having
2 BRIEF movement monitoring of sorts for a period of some 60
years, about half the life of the present steps. Amongst
The substantial structural movement apparent in other information, the early research gave four sets of
Smirke’s floor and Barry’s staircase are the subject of data on floor settlement over a total period of around
this paper. 60 years. It was clear from this that the movement of
However, this project was undertaken in parallel the steps showed no signs of stopping, with an addi-
with a separate project to build a new Visitors’ Recep- tional 16 mm of settlement in the centre of the floor
tion Building. Visitors were to, in future, enter the being recorded between 1992 and 2004 (Figure 2).
Palace of Westminster through the North doors of
Mid height of Lower Steps
Westminster Hall, walk through the Hall, up the impos- 3.700
ing South Steps and into St Stephen’s porch and the
Houses of Parliament, a route that recognizes Barry’s 3.650
the South steps and in the floor close to the steps. 0.000 5.000 10.000 15.000 20.000 25.000
The brief for this work was to diagnose the cause of Distance fromeastedge of hall (m)
the settlement to the floor and steps, consider options
for repair and recommend treatment for the problem. Figure 2. Deflected profile of the steps from East to West.
1134
This historical monitoring was assessed and suggested the flags were carefully cut free using hand tools and
historical and ongoing settlement of approximately where necessary, after careful reflection a small disc
1 mm a year. cutter. It was found important to completely free the
joints otherwise spalling started to occur at the edge of
the flag. A single stone flag was successfully freed and
4 INVESTIGATIONS
relaid in the middle of the area of settlement, reveal-
ing Smirke’s construction of dwarf sleeper walls and
However, although one or two boreholes were avail-
oversite concrete slab below (Figure 3).
able for areas close by, very little engineering data
Once it had been established that the flags could
or calibrated properties on the ground conditions was
safely be lifted, ten flag stones were selected at loca-
uncovered by research for the locations beneath the
tions across the Hall, concentrating on transects across
steps and floor. It was clear that to both correctly diag-
the areas of settlement but with one chosen at the
nose the movement and to support any discussions
North end, remote from the steps. At each flag loca-
with the curatorial bodies with regard to intervention,
tion two cores were taken through the 400 mm concrete
a major ground investigation needed to be carried out,
oversite slab to enable Dynamic Probe Penetrometer
and that the success of this would be key to the quality
testing and allow the retrieval of MOSTAP thin-walled
of decisions made thereafter.
samples for visual analysis and archaeological record-
It was also important to gain as much as infor-
ing. The Penetrometer was mounted on a portable rig
mation as we could through non destructive testing,
specially made for the project and surcharged with 4
so a two stage approach was developed; firstly com-
tonnes of dismountable kentledge, which permitted
missioning a suite of geophysical tests to determine
penetrometer readings to be taken down to a depth
the floor construction and as much about the ground
of around 7 m, at which point the kentledge lifted on
below as we could, across the whole of Westminster
the underlying material, presumed to be stiff clays and
Hall and secondly targeted lifting of flags to confirm
gravels. Ground water level was monitored and found
the construction and intrusive geotechnical boreholes.
to vary by just 55 mm compared with tidal variation
A suite of non destructive tests comprising geophys-
or between 4 and 6 m. Samples of the stone flags were
ical radar, microgravity and resistivity was commis-
taken for petrographical analysis to identify the quarry
sioned. Whilst the specification and execution of such
from which the flags had been won. Interestingly the
non-intrusive investigations was an essential part of
the initial work in the context of the Hall, at the time
they were interpreted they were of marginal value in
defining sub-floor and deeper features of stratigraphy.
With developments in processing of geophysical test-
ing there may still be more information to be gleaned
from these results.
Faced with the value of the historic fabric within the
Hall it quickly became apparent that any work, includ-
ing investigations, would require significant intrusion
into the building. Thus the design of such investiga-
tions would require considerable thought. To make
an appropriate diagnosis of the causes of settlement
and to design an appropriate solution we needed t
a robust theorem as to the cause of settlement, as
there was much speculation based on fragments of
evidence. It was essential to have a comprehensive
reliable geotechnical data for diagnosis.
Fundamental to this was the assumption that data
could actually be obtained. Any investigation would
require the successful lifting and relaying of floor flags
within the Hall, something that no-one working on the
staff in the Palace had witnessed before. The 1.6 m
square flags have tight or minimal joints and there
were concerns that lifting individual slabs might not
be achieved without major dismantling from the edges
of the Hall, or that flags might be dropped, cracked or
not be re-laid satisfactorily Equipment was therefore
assembled for a trial lift one weekend in the pres-
ence of the Conservation Architect. The joints between Figure 3. Trial lifting of the stone flags.
1135
using established conversions with design parame-
ters being determined in-situ and correlated to shear
strength using established conversions. The samples
were examined by geotechnical engineers and archae-
ologists together, collaborating to enhance their under-
standing and produce a single interpretation of the
ground: there is only one reality, not a ‘geotechnical’
interpretation and an ‘archaeological’ interpretation.
The investigation revealed that overlying the London
Clay formation is a layer of Terrace Gravels, then a suc-
cession of soft and loose alluvial deposits, then made
ground that formed the dry land for Westminster Hall
and then a sequence of earth floors.
The results showed that material at a depth of
between roughly 1 m and 3 m below the floor at the
base of the steps comprised very weak alluvial clays
Figure 4. Dynamic Probe and MOSTAP sampling rig. with some organic content and varying in thickness. It
is this local area of weak clay alluvium that is gradually
stone masons’ visual identification of the stone source compressing, and causing settlement, the problem is
was confirmed by the laboratory testing, so it is always exacerbated by the additional load due to a large stone
worth asking Samples of the concrete oversite slab candelabra support structure of considerable weight.
were also taken and tested for strength and composi- Further to the North, this material became firmer,
tion. Work was carried out at weekends through late giving way to terrace silty sands and sandy gravels
May and June 2005 whilst the Houses were not sitting, (Figure 5).
at a rate of one or two locations per day (Figure 4). The diagnosis was, therefore, that the floor and steps
Under the steps it was observed that in places the were likely to be settling because of the weak and
brick support had settled and so the steps were in organic materials present locally below, with constant
effect hanging. Beneath the lower flights of the steps internal shearing of the soil resulting in continuous
headroom was very restricted and it was not possi- movement.
ble to undertake the dynamic probing and thin walled
sampling. Instead we examined some pre-existing 6 DESIGN APPROACH AND CONSTRAINTS
excavations in the concrete slab. These revealed infor-
mation about the steps foundations and some stone Thus it had been established that the basis of the prob-
fragments of archaeological interest. We also carried lem was essentially a ground engineering issue with
out PANDA tests, which in essence con sisted of a possibly very significant archaeological impact, and so
man striking a probe into the ground and measuring strong representation was made by specialists within
physical resistance. the company in both these disciplines.
Archaeological attendance on the investigations Alongside the conservation structural engineer and
yielded some exciting finds in both building recording project director these formed a tight group of four indi-
and examination of the recovered cores. The lifting of viduals who remained close to the project throughout.
slabs showed evidence of the trackway support of the The project director also adopted the separate role of
mobile access scaffolding used by Sir Frank Baines in lead consultant. The internal team structure for the
the repairs of the roof in the 1920’s, with a fragment project was a key factor in delivering what we believe
of newspaper dating from 1922 being found. The soil to be a highly successful project.
cores showed evidence of a rammed chalk floor in the Once the cause of settlement was understood we
Hall, the stratification giving some indication that this held team workshops to brainstorm different options
had been built up over centuries, possibly as previous for remedial solutions. This approach meant that
settlement had taken place. we were working as one multi-disciplinary team.
Each discipline, structural engineer, archaeologist and
geotechnical engineer had a good understanding of the
5 DIAGNOSIS OF SETTLEMENT problem, the brief and the site and an appreciation of
the other disciplines, each was able to offer possible
The quality of data obtained from this investigation solutions and assess the advantages and disadvantage
was consistently high, allowing the classification of of schemes. This offered an integrated approach that
soils and the soil strength and stiffness to be measured was very different from three specialists each work-
directly in-situ with the probe and correlated through ing to a limited brief or a solution produced by one
the testing shear strength of recovered soil samples discipline to be commented on by the others.
1136
Figure 5. Geotechnical section showing layer of very weak alluvial clays.
The key issues and priorities of each aspect not make the situation worse. Although there would
project discipline were captured, including operational be no impact on the below ground archaeological
requirements peculiar to the Houses of Parliament remains, the movement would not be arrested and
and noted project constraints against which we could the brief would not be met.
assess the suitability of possible solutions. 2. Dismantle and re-level the lower steps and floor
without any foundation strengthening, in effect
correcting the superstructure or symptoms with-
7 OPTIONS AND DESIGN OF PERMANENT
out addressing the substructure and cause. This
WORKS
approach has the advantage of appearing to solve
the problem by lifting and packing stones to create
The full Gifford project team met to consider the
a level finish with out interfering with the below
data and final options. Any option appraisal for con-
ground archaeological potential. If the movement
servation works should start with the option and
continues at between 1–1.5 mm a year, the flags
implications of doing nothing; of leaving the build-
could be lifted say every 70 years and an acceptable
ing as found. Although the brief had indicated that
level finish maintained. Indeed, in 70 years time
works were anticipated, this was the assumed starting
new technologies could make other approaches
point for the scheme to develop the correct conserva-
viable. However, it does not address the mecha-
tion approach. We also needed to assess the necessary
nism of settlement. In effect this is the approach
extent of the repair works. We considered that it was
that appears to have been taken throughout the life
not necessary to carry out major intervention to all
of the Hall to date; there is stratigraphic evidence
flights of steps, just the lower flight, where settlement
of the earth floors being re-laid and documentary
was worse, and we selected an area where settlement
evidence that a number of stone floors have been
of the floor was greater than 20 mm, in both cases our
re-laid, presumably because they were no longer
intent was to minimise the interventions.
satisfactory. In the past, however, there has not been
Different options were considered for the correction
the option to address the underlying cause of settle-
of the steps and for the floor, but these can be broadly
ment and every time the flags are lifted and re-laid,
categorized in to the following approaches each with
there is a risk of damage.
their advantages and disadvantages:
3. Dismantle and re-level the lower steps and floor,
1. Do nothing and monitor further. and strengthen the ground beneath the floor and
It was evident that the movement of steps and steps to permanently arrest settlement. Within
floor was continuing and not showing signs of this approach a number of different underpin-
abating and that further settlement would exacer- ning or ground strengthening options were consid-
bate uneven floor surfaces (a real concern given ered. Under the steps these included mini piling
the future increase in footfall), and also result in or trench underpinning. Under the floor ground
an increasing risk of damage to the elements of strengthening by compaction grouting or soil frac-
the building fabric, such as the candelabra support ture grouting were considered. These solutions
structure and railings. ‘Do nothing’ can only be addressed the mechanism of movement but clearly
judged to be sound conservation practice if it does had more impact on the archaeological remains, in
1137
In addition, the oversite slab was strengthened.
Under the steps, new reinforced ground beams above
the piles were cast within the depth of the existing
oversite slab, reinforced with stainless steel bars to
give a design life of nominally 500 years (comparable
with cathedrals) to support the brick crosswalls. The
steps were rebuilt on new crosswalls. The oversite slab
beneath the floor flags and dwarf walls was reinforced
with a grid of stainless steel bars, in effect it was post
reinforced to spread the load to the point supports at
the grouting locations The area of strengthened floor
was dowelled into the area of unstrengthened floor to
ensure continuity across the floor.
1138
in the drill, slowing the progress of drilling. Once this
became evident, an early review of the design of the
strengthening of the slab was undertaken and as a con-
sequence, instead of drilling and coring stainless steel
bars and grouting them in place the method of strength-
ening the concrete was changed. Instead slots were
chased in to the slab and bars laid into the slots and
grouted in place.
The intention was to keep as much of the sleeper
walls as possible, however sections needed to be cut
out in order to insert the reinforcement bars. The walls
were rebuilt in concrete for speed and ease but also to
clearly differentiate the modern work.
1139
number of flagstones to undertake geotechnical environment using carefully chosen engineering tech-
investigation. nology and employing the least intrusive methods
4. A good brief is very helpful, in this case requiring possible, but nevertheless methods that are of our
cause, options and solution to be presented. time. The interventions were based on a very good
5. The design team worked as a tripartite multi- understanding of the causes of damage. We have under-
disciplinary team. Each discipline shared in the taken extensive recording of structures and have taken
aspirations of the project goals and there was a the opportunity to gain a greater understanding of
strong desire to understand and to compromise but Westminster Hall than existed before the project.
also to stand firm when required. There was mutual
respect across the team and healthy argument. Client :
6. During the design process there were a number of Parliamentary Works Services Directorate
previews, reviews and combined site visits. Com- Lead Consultant, Structural Conservation Engineer,
munication was face to face not through emails and Archaeological Consultant:
minutes. Gifford
7. There was realistic fee for the work. Contractor:
8. The subcontractors were very good and their ideas Verry Construction Limited
fed into design development. Ultimately, however Stonework Subcontractor: Stonewest Limited
tight the specification and supervision, the qual- Ground Engineering Subcontractor: Keller Founda-
ity of the finished stonework is done to the master tions Limited
mason executing the works. Archaeological Contractor: Museum of London
Archaeological Service.
10 CONCLUSIONS
1140
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The Feyzullah Efendi Madrasah, built in Fatih, Istanbul in 1700, is a small complex with a
library, school, fountain for ablutions and two other fountains. In the 1999 earthquake, the madrasah was severely
damaged and cracks appeared in its foundation, main walls and domes. The restoration of the building was carried
out between 2000 and 2007, the ground under the building was firmed up, foundations, domes and walls were
consolidated, the cracks in the walls and domes were stiffened using an injection method. This complex was
used first as a madrasah / faculty and then as a library. In this paper, the damage caused by the earthquake and
the problems of the new function given to the building is investigated. As for the restoration, the repair methods
to be applied were examined while estimating to what extent the historical value of the building was preserved.
2 ARCHITECTURAL CHARACTERISTICS
1141
Figure 2. A general view of the Millet Library (Archive
of the Protection Council of the Natural and Cultural
Properties).
1142
Figure 8. A view of the courtyard, “sadirvan” and madrasah,
2007.
1143
Figure 10. The consolidation method of bar-mat on the
dome, main building, restoration stage.
1144
Figure 11. The application of the raft foundation for
madrasah.
1145
Figure 13. The application of NPU 120 profiles of the
madrasah walls, restoration stage.
Figure 15. Laying of the lead boards on the dome, restora-
tion stage (Archive of The Protection Council of the Natural
and Cultural Properties).
1146
made and it was seen as appropriate that the building
should continue its library function that it has fulfilled
since 1916. The main library building was separated
from the collection of historical handwritten books
and resident research experts; the madrasah rooms
have been designated for administration and as a book
depot; and the portico section for reference readers
and visiting researchers. The facades that overlook the
courtyard of the portico section of madrasah and the
iwan in the main building were enclosed in order to
gain enclosed space and because of the difficulty of
walking in the half-enclosed areas they were enclosed
with glass panes in aluminum frames. In addition the
portico section was sectioned off with glass framed
Figure 16. Detail of a main motif, before painting repair with aluminum again to create the reading room. Due
(Archive of the İstanbul Survey and Monuments Directorate).
to these changes the main form of the madrasah with
the closed space/ the room-half open space / the portico
in the main building, traditional methods and mate- and courtyard/open space relationship was spoiled.
rials were used and, in the madrasah building, steel Using the madrasah rooms for a book depot is also not
and concrete were used. The new materials used were the correct approach to maintaining the spatial quality
concealed within the bearer walls and the plaster. It of the original layout.
could not be escaped that the structural fabric of the In conclusion the new function given to the building
building that had been damaged in earthquakes, and is a burden on it and detracts from the spatial integrity
still could be, had to be strengthened using materials of the historic building. In this building the museum
and methods. In addition, concealing the interven- function is a usage that is more appropriate and does
tion provided integrity to the restoration. However, not spoil the basic form. However, despite these issues,
in restorations of this type, priority has to be given what is pleasing is that the window borders are partially
to methods that will result in the least damage to the concealed in the back of the arch and the tie rods,
building. In strengthening the madrasah building, the using glass panes in the undivided and allowing these
dimensions and places of the materials that were to additions to be distinguished and removed.
make up the metal construction did not clarify the rela-
tions between the main walls. Also, the approach paid
insufficient attention to protecting its inherent value, REFERENCES
such as authenticity and historical evidence. The repair
of the main building was carried out with greater care Ahunbay, Z. 1994. “Feyzullah Efendi Medresesi”, İstanbul
through the pressure of the Protection Commission Ansiklopedisi. Tarih Vakfı Publication, Vol.3: 308–309.
Ahunbay, Z. & Diren, Ş. 2001. The Report on the repair of
and the decorations on the inner spaces in particu- Faih Millet Library, 16.05.2001.
lar were protected as a valuable historical document Akcan, S., et. al. 2001. The Determination Report of Ankara
for the structure and the paintings were repaired with Survey and Monuments Directorate, 29. 01. 2001.
painstaking care. Dwight, H.G. 1915. Constantinople Old and New, Long-
Protecting historical buildings and giving them a mans, Green & Co., London, 69–70.
new function is a complex challenge. The new func- Eriç, M. et al. 2002. The Technical Report of Millet Library
tion must not destroy the mass and spatial integrity of Mimar Sinan University, 08.07.2002.
of the old structure. The arrangements to be made Goodwin, G. 1971.A History of OttomanArchitecture, United
and the new materials to be used must meet mod- States of America – Baltimore, 1971.
Kütükoğlu, S.M. 1978. Darü’l-Hilafeti’l-Aliyye Medresesi
ern standards. The new function must protect and ve Kuruluş Arefesinde İstanbul Medreseleri, İstanbul,
enrich the old building’s spatial quality. There is no İstanbul Üniversitesi Edebiyat Fakültesi Publication
role in the modern educational system for the Feyzul- 157–158.
lah Efendi Medrese’s educational function from the Kütükoğlu, S.M. 2000. XX.Asra Erişen İstanbul Medreseleri:
Ottoman period. Therefore a change in function was Ankara Türk Tarih Kurumu Publication, 264–265.
1147
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The Katholikon of Dafni Monastery, one of the most important monuments of the middle-
byzantine period situated near Athens, has suffered severe damages due to the 1999 earthquake. Several studies
have been engaged, in order to collect all relevant data of the internal masonry structure and that of the
mosaics substrata by applying non-destructive techniques, such as Ground Penetrating Radar (GPR) and Seismic
techniques including tomographies. In some areas these techniques have been used both before and after grouting.
1149
measurements, conducted first on a test wall and after-
wards on the Katholikon itself, are presented and
commented upon.
Investigations have been also realized to control the
grouting effect into the masonry walls, using two seis-
mic techniques. The objective herein was to propose a
simple survey methodology which gives information
for the mechanical characterization of the materials
inside the structure.This approach includes some sonic
2D travel time tomographies that present the advantage
to describe a section of the structure, and may illus-
trate the nature and the geometry of its interior (stone
blocks, fillings, . . . ). A first campaign using this tech-
nique has been undertaken in non injected and semi-
injected areas, as well as in injected ones. A second
campaign is still in progress comprising the repeti-
tion of the tomographies after the completion of the Figure 1. Manual-sonic map of “Saint-Nicolas, constructed
injections. Furthermore, these few tomographies per- by the mosaic’s conservators”. Damaged zones are in gray.
mit the understanding and calibration of the behaviour
of exhaustive Vp transmission measurements that can
The objective of the current study is to correlate
be practiced over the major part of the structure.
this radar gray-scale maps to a level of detachment
or heterogeneity by comparing them to the manual-
sonic maps and the pathology ones, constructed by the
2 RADAR TECHNIQUES Conservators of the mosaics. Although, it has to be
noted that this comparison has certain limitations, due
2.1 Mosaics mapping mainly to the fact that GPR maps are constructed based
In 2005, an exhaustive radar survey has been realized on a multi-valued gray-scale, while the manual-sonic
on the fifty main mosaics of the Katholikon of Dafni maps consist of two values (bad or good), deriving
Monastery. The principle of the radar technique, which from subjective human-expert senses.
is particularly adapted to masonry structures (Maier- Apart from the aforementioned maps, all the infor-
hofer & Leipold 2001, Valle et al. 1999), is based mation selected by the Conservators concerning the
on the transmission of an electromagnetic pulse, at nature of the substrata mortar has been taken into
very high frequencies, through the examined struc- account in order to investigate any possible existence
ture. Successive echoes caused by reflections at the of GPR data-similarities among the mosaics exhibiting
inner layered structures are recorded through time sig- the same substrata. Four main types of mortars were
nals and eventually the juxtaposition of these signals identified: (i) the original lime based mortar (Byzan-
enables the construction of a time–section (GPR pro- tine period), preserved only in very few mosaics, (ii)
file), which can be related to the internal geometry of the Novo-type mortar (late 19th century interven-
the investigated structure. tions), having as binder natural hydraulic lime, used
For this specific application a 1.5GHz central- in the vast majority of the mosaics, (iii) the cement-
frequency antenna has been used. The applied proce- based mortars (1950’s interventions), used in a small
dure has already been tested by Côte et al. (2004). percentage of the mosaics and (iv) the cement-lime
For reasons of comparison the selected parameters based mortars (1960’s interventions), found as well
remained constant during the execution of the radar in a small percentage of the mosaics (Chryssopoulos
survey. Thanks to specific parallel radar profiles, maps et al. 2003). The coexistence of all types of mortars is
can be constructed from chosen trenches of the struc- possible in cases of great size mosaics, due to previous
ture. The gray-color scale of these maps is related to the interventions.
amplitude of the echoes in that trench. In this specific The two following examples show the significance
case, maps were constructed, representing the energy of such non-destructive technique, which can be auto-
of the first part of the radar surface wave. mated in order to obtain comparable information. For
As the wavelengths of the radar technique are about the mosaic “Saint-Nicolas” (Figs 1–2), radar results
7 cm in the mosaics, echoes from the substrata are indicate that the head and the left shoulder and arm
mixed with the surface echo, therefore the resolution is correspond to a sound area, confirmed also by the
not enough to distinguish one from the other. As con- manual-sonic map. The rest of the mosaic is either
sequence, the radar parameter was fixed (
t ∼ 1 ns) completely detached or/and the nature of mortar used
in order to map the first few centimeters including the for the substrata is utterly different from the one of the
mosaic itself and its substrata. sound area. This result is also confirmed by the fact
1150
Figure 2. Sub-surface radar map of “Saint-Nicolas”. The
damaged zones are in dark.
1151
lime, is used in the most of the surface the results
are more complicated, than the ones presented by the
manual sonic maps.
1152
Figure 9. Subtraction of profiles on the “Juda’s Betrayal”
mosaic after the 4th injection.
Figure 8. Amplitudes of the surface echoes, obtained from
successive radar profiles at GPR path B, after each elemen-
tary injection.
1153
Thus it is not evident that the grout will really pass
behind the area that is monitored.
3 SONIC TECHNIQUES
1154
the reconstruction algorithm converges to a three-leaf
medium (again without a priori information). This
time, the inner filling materials are associated with
very low velocities (500 m/s).
The comparison of these two cases permits to
conclude that the grouting survey of such masonry
structures may be achieved by the observation of the
velocities of their inner part. Adapted tomographical
algorithms are able to reconstruct the inner geometry
of such structures.
These tomographies can be realized over some
particular areas, while a simple punctual time trans-
mission set can consider the whole structure by a
convenient scanning.
Figure 13. Tomographical reconstruction in a grouted area. 3.2 Sonic transmission testing
Some simple sonic transmission tests have been real-
ized at Dafni Monastery. The acquisition devices are
quite simpler than the ones applied to tomographical
measurements. A couple of accelerometers are linked
to an acquisition card or to a simple oscilloscope. A
hammer equipped with a shock accelerometer consti-
tutes the source, whereas a second accelerometer is the
receiver. The measurement principle is based on posi-
tioning the accelerometers face to face on both sides
of the wall and knocking the wall in the vicinity of one
of the accelerometers, which is used as a trigger. The
first arrival time can be then picked on the signal from
the other sensor. The whole set is then displaced and a
scan of wall or of a part of wall can be performed.
Figure 14. Tomographical reconstruction in a non grouted The example illustrated in Figure 15 corresponds
area.
to a set of measurements conducted over a vertical
process. The real edges of the wall are figured by the line along a wall which has been grouted up to the
black lines. limit indicated by the dotted line. Beyond this line, the
This image illustrates quite clearly the inner behavior of the sonic measurements changes radically.
building-up of this wall. Three leaves are observed. Two other measurement sets (over horizontal lines)
On both external parts of the wall the high veloci- are plotted in Figure 16. They have been obtained at
ties are associated to stone masonry, while the lower two levels of the same wall. The upper level has not
velocities at inner part are associated to the infill mate- yet been grouted. The difference in the behavior of the
rial. It is quite interesting to notice that the geometry mean transmission velocities is quite obvious.
of the stone facings is well presented. The thickness Such measurements are relatively easy to be per-
of the stones can be estimated. The linear arrangement formed and interpreted. Large parts of structures can
of the inner face of these facing stones is obtained be then observed. Some indicators can be deduced for
without any geometrical control of the solution, nei- the qualification of the homogeneity of the structure
ther with any a priori model. The mean velocity of the before any treatment. In case of complex behavior,
stones is about 4200 m/s while the mean velocity of a complementary tomographical arrangement can be
the filling materials is about 1500 m/s. This part of the performed. After grouting, the same measurements
wall has already been grouted. can be repeated in order to evaluate the efficiency of
The tomography illustrated in Figure 14 has been the intervention.
realized in a non grouted part of the structure. The
process is identical to the one of the previous case. 4 CONCLUSIONS
Here the investigated length of the wall is about 2 m
for a thickness of 0.6 m. This time, the pickings of first High frequency Ground Penetrating Radar (GPR)
arrivals for each signal are much more difficult. mapping of more than 50 mosaics, related to their
The attenuation of the signals is very strong. The very near bearing-structure, has been realized, in
early beginning of each signal is quite soft and then order to locate doubtful zones including delamina-
the accuracy of the picking decreases. Nevertheless, tions, changes of structures (i.e. recent repairs) or
1155
Directorate for the Conservation of Ancient and Mod-
ern Monuments and the 1st Ephorate of Byzantine
Antiquities of the Hellenic Ministry of Culture, that
has assisted the in situ measurements is gratefully
acknowledged. The project has been realized in the
framework of a research program realized and co-
financed by the Directorate for Technical Research on
Restoration of the Hellenic Ministry of Culture and
the Laboratoire Central des Ponts et Chaussées.
REFERENCES
Abraham, O., Ben Slimane, K., Côte Ph. 1998. Seismic
tomography: factoring anisotropy into iterative geomet-
Figure 15. Sonic transmission velocities across a wall ric reconstruction algorithms. Int. J. Rock Mech. Mining
partially grouted. Sci. 35 (1), 31–41.
Côte, Ph., Abraham, O. 1995. Seismic Tomography in Civil
Engineering. NDT-CE, Berlin, 459–466.
Côte, Ph., Dérobert, X., Miltiadou-Fezans,A., Delinikolas, N.
2004. Mosaic-grouting monitoring by ground-penetrating
radar. Proc. Int. Sem. on Struct. Anal. of Hist. Const. Proc.,
Padova (IT), nov 2004.
Chryssopoulos, D., Anamaterou, L., Georganis, F. 2003.
Documentation study for the mosaics of the Katholikon of
Dafni’s Monastery after the 1999 earthquake. Directorate
for the Conservation of Ancient Monuments, Hellenic
Ministry of Culture (in Greek).
Delinikolas, N., Miltiadou-Fezans, A., Chorafa E.,
Zaroyianni, E. 2003. Restoration Study of the Katholikon
of Dafni Monastery, Phase A-Architectural and Historical
Survey, Hellenic Ministry of Culture (in Greek).
Gilbert, P. 1972. Iterative methods for the three dimensional
reconstruction of an object from projections. J. Theor.
Figure 16. Sonic transmission velocities across a similar
Biol., 105–117.
wall grouted (red) or not yet grouted (purple).
MacCann, D.M., Jackson, P.D., Fenning, P.J. 1988. Compar-
buried heterogeneities. The results of this work in con- ison of the seismic and ground probing radar methods in
junction with the manual sonic classification realized geological surveying. IEE Proc. 4 , 380–391, part F.
by the competent conservators were very useful for a Maierhofer, C., Leipold, S. 2001. Radar investigation of
masonry structures. NDT&E Int., 34: 139–147.
detailed evaluation of mosaic substrata. Miltiadou-Fezans, A., Delinikolas, N., Chorafa, E.,
Sonic tomographies permit the understanding and Zaroyianni, E., 2003. Restoration Study of the Katholikon
calibration of the behavior of exhaustive Vp transmis- of Dafni Monastery, Phase A Structural Survey, Analysis
sion measurements that can be practiced over some and remedial measures, Hellenic Ministry of Culture (in
areas of the structure. These investigations are com- Greek).
plementary to simple transmission investigations that Miltiadou-Fezans, A., Vintzileou, E., Delinicolas, N.,
need simply to locate corresponding measurement Zaroyianni, E., Chorafa E. 2004. Pathology of Dafni
grids on both sides of the wall. At each node, the Monastery: Survey, monitoring of cracks, Interpretation
transmission time is measured. The map of these mean and numerical verification, Proc. Int. Sem. on Struct.Anal.
of Hist. Const. Proc., Padova, Nov 2004.
velocities for an entire wall may be constructed before Spathis, A.T., Blair, D.P., Grant, J.R. 1983. Seismic pulse
and after a grouting process. Their comparison leads assessment of the changing rock mass condition induced
to a verification of the homogeneity of the wall after by mining. Int. J. Rock Mech. Mining Sci. 22 (5) , 303–312.
the grouting process. Valle, S., Zanzi, L., Rocca, F. 1999. Radar tomography for
The co-evaluation of NDT results can be a useful NDT: comparison of techniques. Journ. Appl. Geophys.,
tool for the assessment of the efficiency of the grouting 41: 259–269.
intervention. Vintzileou, E., Miltiadou-Fezans, A., Palieraki, V.,
Delinikolas, N. 2004. The use of radar techniques and
endoscopy in investigating old masonry: The case of Dafni
ACKNOWLEDGMENT Monastery. In Modena C., Lourenco P.B., Roca P.(ed.)
Proceedings of the 4th International Seminar on Struc-
The scientific and technical personnel of the Direc- tural Analysis of Historical Constructions., Rotterdam,
torate for Technical Research on Restoration, of The Balkema, Vol. 2, pp. 351–360
1156
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
F.M. Mazzolani
Department of Structural Engineering, University of Naples “Federico II”, Naples, Italy
ABSTRACT: The PROHITECH research project is framed within the INCO thematic areas, devoted to “Pro-
tection and conservation of cultural heritage” in the Mediterranean area. The main subject of the research is
the seismic protection of historical and monumental buildings, and the main objective consists in developing
sustainable methodologies for the use of Reversible Mixed Technologies (RMTs) in the seismic protection of
the existing constructions. RMTs exploit the peculiarities of innovative materials and special devices, allowing
ease of removal when necessary. Furthermore, an optimization of the global behaviour under seismic actions is
achieved by the combined use of different materials and techniques. The endpoint of the research is a proposal
of codification for the use of such technologies in the seismic protection of existing constructions.
1157
the necessity to ensure seismic protection against col-
lapse also in case of destroying events. In this view, it
can be properly framed within the advanced concept
of Performance Based Design (PBD). As well known,
the PBD is a new way to approach the structural design
against seismic actions, having the purpose to ensure a
proper degree of structural reliability under any speci-
fied working conditions, including both serviceability
and ultimate limit states.
Till now, the PBD has been applied to new structures
only, which can be easily designed complying with
relevant behavioural thresholds set by PBD it-self. No
Figure 1. The PROHITECH partner countries. applications exist in the field of existing constructions,
yet. In particular, neither criteria nor methodologies
The partner Countries are: Algeria (AL), Belgium are available for achieving a satisfying design level
(B), Egypt (EG), Macedonia (MK), Greece (GR), against strong intensity earthquakes. This is indirectly
Israel (ISR), Italy (I), Morocco (M), Portugal (P), confirmed by most of national seismic codifications,
Romania (RO), Slovenia (SL), Turkey (TR). which, as a matter of fact, allow to avoid a rigorous
In the following the partner institutions and the seismic retrofit in case of historical constructions. This
relative responsible persons are indicated. UNINA: approach, of course, tends to preserve the monumen-
University of Naples “Federico II”-Engineering Fac- tal value of the construction, but at the same time is
ulty (F.M. Mazzolani, project general coordinator); not adequate to protect against severe earthquakes. It
B: University of Liège (J. Jaspart); MK: Univer- is evident how this aspect deserves great attention not
sity of Skopje (K. Gramatikov); GR: Technical only in the perspective of saving human lives, but also
University of Athens (I. Vayas); NA-ARC: Univer- at the light of preserving inestimable buildings from
sity of Naples “Federico II”-Architecture Faculty destruction. The use of innovative materials and Mixed
(R. Landolfo); P: Technical University of Lisbon Technologies is the most appropriate answer for ensur-
(L. Calado); ROPUT: Polytechnical University of ing an adequate performance, and hence the Structural
Timisoara (D. Dubina); ROTUB: Technical Univer- Integrity, under strong seismic actions.
sity of Bucharest (D. Lungu); SL: University of RMTs are based on the integration of structural
Ljubljana (D. Beg); TR: Boğziçi University of Istanbul members of different materials and/or construction
(G. Altay Askar); ISR: Technion Israel Institute of methods into a single constructional organism. The
Technology, Haifa (A.V. Rutenberg); EG: Engineer- basic feature of RMTs is that their application should
ing Centre for Archeology and Environment, Faculty be always completely recoverable, that is reversible,
of Engineering, Cairo University (M. El Zahabi); M: if required. This is considered as an essential design
Moroccan National Scientific and Technical Research requirement in order to prevent historical and monu-
Centre, Rabat (A. Iben Brahim); SUN: Second Uni- mental buildings from unsuitable rehabilitation oper-
versity of Naples (A. Mandara); AL: University of ations. The main aim of RMTs is the best exploitation
Science and Technology “H. Boumedien” of Algier, of material and technology features, in order to opti-
Civil Engineering Faculty (M. Chemrouk); UNICH: mize the structural behaviour under any condition,
University of Chieti/Pescara (G. De Matteis, project including very severe limit states produced by strong
technical coordinator). seismic actions. This practice, initially concerned with
new technologically advanced buildings, is now being
3 THE BASIC ISSUES looked up with increasing interest also in the field
of structural rehabilitation, due to the large possibili-
Within the technical field of seismic rehabilitation, ties of structural optimization and, hence, performance
two aspects are receiving an increasing attention by maximization, both achieved thanks to mixed tech-
engineers and researchers, namely: nologies. In few words, the use of RMTs would
– Preservation of Structural Integrity of existing involve the best exploitation of each material and/or
buildings under severe/exceptional seismic actions; technology used in the intervention, providing the
– Improvement of building seismic performance by best performance from both technical and economical
means of RMTs. point of view.
Both these aspects are closely interrelated each
other, in the sense that the application of RMTs is, 4 THE RESEARCH WORKPLAN
in some cases, the only tool to achieve a satisfying
level of Structural Integrity under severe earthquake It is planned to achieve the above objectives through
actions. The concept of Structural Integrity relies on the creation of twelve scientific workpackages dealing
1158
Figure 2. Workplan of the PROHITECH project.
1159
Chapter 3 (author A. Iben Brahim, WP3 leader), systems are underlined. Moreover a general classifi-
is devoted to the seismic vulnerability assessment cation of the innovative devices for seismic protection
of historical buildings in the PROHITECH Coun- is briefly presented.
tries. In particular, the current approaches to damage The contents of Chapter 2 (authors L. Calado,
assessment and the definition of the vulnerabilities of J.M. Proença, P. Skuber and M. Esposto) mainly de-
structural types are presented, together with the anal- rive from the WP5 final report. Innovative materials
ysis of seismic risk related to the historical building are presented considering their basic principles, struc-
heritage of the Mediterranean area, on the basis of the tural features, fields of application, experimental tests
damage data reported in Chapter 2. and numerical models, design criteria and codification
In Chapter 4 (authors A. Mandara, WP1 leader, and issues, structural applications and economic aspects.
F.M. Mazzolani) the existing techniques used in the The Chapter is concluded by a comparison and eval-
PROHITECH Countries for the protection of build- uation among the considered materials, based on the
ings against the seismic action, with reference to the output of WP6 final report.
different structural types, are described. Chapter 3 (authors L. Calado, A. Panão and
Finally, Chapter 5 (authors D. Lungu, WP4 leader, L. Pavlovčič) deals with the innovative devices which
and C. Arion) illustrates the intervention strategies for can be adopted for mixed reversible technologies. The
seismic protection of buildings, with special attention structure of the Chapter is based on WP5 final report,
towards both the technological and policy aspects of while the contents come from both WP5 and WP6 final
the problem. reports. Also in this case, every device is presented by
considering the aspects mentioned in Chapter 2.
5.2 Project deliverables D1-to-D4 Chapter 4 (authors D. Beg and G. De Matteis) is
devoted to the seismic protection systems. Its con-tents
Besides the main deliverable D-I, the output of Part R1 come from WP6 final report and deal with the seismic
is completed by four project deliverables, as previously protection systems for masonry buildings, reinforced
mentioned. Each project deliverable, whose contents concrete frames, brick or stone masonry structures,
come from the partner contributions, has been edited like towers, domes and vaults, and stone temples.
by the leader of the respective workpackage, and is At the end of the Chapter a general over-view on
made of two parts. The first part is the text of the the seismic protection with active control system is
deliverable, written by the WP leader on the basis of the presented.
partner contributions. The second part is an appendix
collecting all the contribution documents written by 5.4 Project deliverables D5-to-D7
the partners.
Project deliverable D1 (Mandara 2005) is an The project deliverables D5, D6 and D7 are related to
overview of the traditional technological systems used the activities developed within WP5 and WP6.
in the Euro-Mediterranean area for the rehabilitation Project deliverable D5 (Calado 2006) is focused
of historical buildings. Project deliverable D2 (Altay on innovative materials, including new metals and
Askar 2005) deals with the assessment of the seismic metal-based intervention techniques, in order to select
damage in historical constructions. Project deliverable suitable materials for creating both strengthening sys-
D3 (Iben Brahim 2005) is focused on the risk and tems and special devices aimed at the optimization of
vulnerability assessment of the building historical her- the structural behaviour.
itage. Project deliverable D4 (Lungu 2005) deals with Project deliverables D6 and D7 (Beg 2006), as
the intervention strategies for the seismic protection specific output of WP6, contemporary represent the
of historical buildings. complement and completion of the work performed
in WP4 and WP5, by providing the information nec-
essary to the proper use of innovative materials and
5.3 Second main deliverable mixed technologies in strengthening interventions, as
The main deliverable D-II deals with the elements well as the definition of special systems for seismic
required for a correct choice of both materials and protection to be applied to existing buildings.
technologies to be used in seismic rehabilitation
(Mazzolani et al. 2006). It merges the outcome of the
6 ACTIVITY IN PROGRESS
activity of WP5 (Innovative Materials and Techniques
for Seismic Protection, leader L. Calado) and WP6
6.1 General
(Set-up of Advanced Reversible Mixed Technologies
for Seismic Protection, leader D. Beg). The activities actually in progress are those related to
D-II is composed of four Chapters. Chapter 1 the project Parts R3 and R4 (WP7-to-WP12). Part R3,
(authors F.M. Mazzolani and L. Calado) is an introduc- finalized to the set-up of adequate calculation methods
tion to the RMTs, where the main aspects of seismic for RMTs used in seismic protection, deals with the
protection based on the use of special devices and experimental and numerical analyses which represent
1160
Figure 4. R.c. building for full scale tests in Naples (Italy).
1161
Figure 10. FEM model of Fossanova Gothic Abbey.
1162
case of monumental constructions, a feasibility study
will be performed to assess the applicability of var-
ious innovative solutions to different types of build-
ings. Together with the considerations based on the
structural performance, additional architectural and
historical–cultural aspects will be accounted for in the
context of the possible interventions.
Within WP10, the validation criteria for restoration
interventions are synthesized in the project deliverable
D12, while project deliverable D13 deals with the per-
formance assessment of the new technologies, which
Figure 13. FEM model of Mustafa Pasha Mosque.
are compared to the traditional ones.
The knowledge gained within the project will be
The results of Part R3 will compose the third main
conveniently applied to some selected study cases
deliverable, which will be organized in three main
(WP11), consisting in historical buildings belong-
parts: (1) Experimental activity (referring to the dif-
ing to the heritage of Mediterranean Countries, in
ferent types of experiments); (2) Numerical activity
order to perform feasibility analyses of seismic pro-
(dealing with the simulation of experimental tests
tection interventions by means of RMTs. The selected
by numerical models and comparisons of results);
study cases are indicated hereafter. The Mustafa Pasha
(3) Calculation models (referred to the identification
Mosque of Skopje and the St. Nikola Church in Psacha,
of simplified models to be used as design tools in
both in Macedonia, as well as the Gothic Abbey in
structural restoration).
Fossanova (Italy) and the Greek Temple in Athens, cor-
The output of WP7 and WP8 will be collected in D8
respond to the project large scale models (WP7). The
and D9 project deliverables, respectively. The output
Royal Palace of Naples and the Gallery “Umberto I”,
of WP9 will lead to two project deliverables: D10,
both in Naples (Italy), are characterized by wooden
dealing with analytical models for special materials
and iron structures, respectively. The Koletti building
and devices for the seismic structural control, and D11,
in Athens (Greece) is the only r.c. structure consid-
dealing with simplified models for the seismic analysis
ered, while the Beylerbeyi Palace in Istanbul (Turkey)
of historical constructions.
is interesting for studying the interaction between tim-
ber and masonry elements. The Medina of Salé in
6.3 Part R4: Set-up of codification rules
Morocco has been also selected for considering the
The activity related to Part R4 will be completed at the complex behaviour of undiversified building blocks.
end of the project. The results of the work performed The design of relevant application solutions based
during the whole project period will be finalized to on RMTs is also foreseen together with the critical
codification proposals. Contemporary, the selected evaluation of collected examples of real restoration
innovative materials and techniques, the developed cases, belonging to the current experience in each
reversible mixed technologies, together with the calcu- partner Country. This activity is framed in WP11, lead-
lation methods, will be subjected to extensive valida- ing to the preparation of the corresponding project
tion studies. The data obtained by WP5 and WP6 will deliverable (D14). The interconnection between the
be analysed and compared with traditional materials study cases and the development of design guidelines
and technologies. Firstly, a cost-to-benefit evaluation (WP12) is evident, since the first step can be consid-
will be performed to outline the scope and the effec- ered as a benchmark for the set-up of the codification
tiveness of new materials and/or technologies. As a rules. As a result of the WP12 activity, two project
second step, the performance of the proposed solu- deliverables will be realized: a proposal of codification
tions will be assessed in terms of strength, including rules for the design of seismic protection interventions
low-cycle fatigue resistance, ductility and dissipation by using RMTs (D15); a manual for the actual imple-
capacity compared with conventional solutions. The mentation of the procedures proposed and developed
results from the experimental (WP7) and numeri- within the project (D16).
cal (WP8) analyses will be extensively used. Special The final document of the whole project will com-
conclusions will be drawn on the applicability of dif- prise the fourth main deliverable, downstream of the
ferent innovative solutions depending on the specific fourth year activity, containing a set of recommen-
conditions and design requirements. dations elaborated on the basis of all collected data
Special attention will be paid to the validation and results. The proposal will comply with the most
of the proposed calculation models on the basis of up-to-dated codification issues in the field of seismic
experimental and numerical results, so as to prove design, say the Performance Based Design, and will
their reliability in the perspective of their inclusion share the same global layout, language and philosophy
in the Design Guidelines (WP12). For the specific as Structural Eurocodes issued by CEN.
1163
7 EXPECTED RESULTS Beg, D. (ed) 2006. PROHITECH project deliverable D6-7
Development of reinforcement procedures for structural
The PROHITECH project is an important opportunity elements based on the use of reversible mixed technolo-
to develop knowledge and technology in the field of the gies, Set-up of seismic protection systems based on the
dissipation of seismic input energy.
seismic protection of the Euro-Mediterranean cultural
Calado, L. (ed) 2006. PROHITECH project deliverable D5
heritage (Mazzolani 2007a, b). The innovative char- Identification of innovative materials and special devices
acter of the technical solutions proposed for seismic to be used for reversible mixed technologies in structural
retrofitting is mainly based on the concept of RMTs. rehabilitation.
The main expected results of the research activity are Iben Brahim, A. (ed) 2005. PROHITECH project deliverable
the following. D3 Assessment of seismic risk maps and evaluation of
The basis for the assessment of an up-to-dated state- seismic vulnerability of historical building heritage in the
of-the-art concerning advanced systems of seismic Mediterranean area.
protection of existing constructions will be set-up, Krstevska, L., Taskov, Lj., Gramatikov, K., Landolfo, R.,
Mammana, O., Portioli, F., Mazzolani, F.M. 2007. Experi-
in harmony with the specific demand of all Euro-
mental and numerical investigations on the Mustafa Pasha
Mediterranean Countries for a more comprehensive Mosque large scale model. COST C26 Workshop, Prague,
framing of anti-seismic rehabilitation. Czech Republic.
Both conscience and knowledge about “new” mate- Krstevska, L., Taskov, Lj., Gramatikov, K., Landolfo, R.,
rials and technologies as a suitable alternative to Mammana, O., Portioli, F., Mazzolani, F.M. 2008. Shak-
“traditional” solutions will increase, since the last ones ing table tests on the large scale model of Mustafa
are proved to be often inadequate to provide a satisfy- Pasha Mosque without and with FRP. Proc. of SAHC08
ing seismic performance, in particular when applied Conference, Bath, UK.
to historical and monumental constructions. Landolfo, R., Portioli, F., Mammana, O., Mazzolani, F.M.
2007. Finite element and limit analysis of the large scale
The adoption of materials and systems which
model of Mustafa Pasha Mosque in Skopje strength-
are reversible, recyclable, environmentally friendly, ened with FRP. Proc. of APFIS 2007 Conf., Hong Kong,
and economically sustainable will be supported. The China.
present degree of knowledge on the application Lungu, D. (ed) 2005. PROHITECH project deliverable D4
of these materials and systems is not particularly Definition of methodologies for seismic up-grading of
advanced, neither codified in any form, hence it is constructions based on both strengthening of structural
expected that the research activity carried out within elements and control of the seismic response.
PROHI-TECH project will yield significant innova- Mandara, A. (ed) 2005. PROHITECH project deliverable D1
tion in seismic protection practice. Overview of traditional technological systems adopted for
seismic rehabilitation of historical buildings in European
New and up-to-dated information on the problem
and Mediterranean Countries.
of seismic protection will be disseminated, thanks Mazzolani, F.M. 2005a. Earthquake protection of historical
to the participation of acknowledged institutions, be- buildings by reversible mixed technologies: the PROHI-
longing to both Europe and Mediterranean basin, all of TECH project (oral presentation). Symp. on Seismic prot.
them widely experienced in the field of seismic design of the cultural heritage by innovative techniques, Rome,
and with an ongoing significant research activity in Italy.
such area. Mazzolani, F.M. 2005b. Earthquake protection of historical
Young engineers and architects, as well as buildings by reversible mixed technologies: the PROHI-
researchers involved in seismic design, will have the TECH project. Proc. of the Symposium on Damage and
repair of historical and monumental buildings, Venice,
opportunity for a qualified training and research activ-
Italy.
ity, aimed at an enrichment of existing skills in the field Mazzolani, F.M. 2005c. Earthquake protection of historical
of structural engineering. buildings by reversible mixed technologies: the PROHI-
Design and constructional rules for interventions TECH project. 2nd H & mH International Conference
based on advanced and innovative technologies will be on “Vulnerability of 20th Century Cultural Heritage to
set-up. This is expected to recall to a greater interest Hazards and Prevention Measures”, KOS, Greece.
of both construction industry and practicing engineers, Mazzolani, F.M. 2006a. Earthquake protection of histori-
so as to have a remarkable impact on the current anti- cal buildings by reversible mixed technologies. Proc. of
seismic rehabilitation practice. STESSA 2006 Conference, Yokohama, Japan.
Mazzolani, F.M. 2006b. Earthquake protection of historical
Information on the ongoing activity is available in
buildings by reversible mixed technologies: the PROHI-
the web site www.prohitech.unina.it. TECH project. Proc. of the 7th European Conference
”SAUVEUR” Safeguarded Cultural Heritage, Prague,
Czech Republic.
REFERENCES Mazzolani, F.M. 2006c. The research project PROHITECH:
state of advancement. In The Construction Aspect of Built
Altay Askar, G. (ed) 2005. PROHITECH project deliverable Heritage Protection, Dubrovnik, Croatia.
D2 Assessment of earthquake-induced structural damage Mazzolani, F.M. 2007a. Earthquake protection of historical
in historical buildings of the Mediterranean area. buildings. Invited lecture, Reluis workshop, Salerno, Italy.
1164
Mazzolani, F.M. 2007b. Earthquake protection of historical on 14th–15th July 2005; 1st editorial meeting on 29th
buildings. COST C26 Workshop, Prague, Czech Republic. August 2005 in Ohrid (Macedonia) and 1st Seminar on
Mazzolani, F.M., De Matteis, G., Mandara, A., 30th August 2005; WP5-to-WP9 joint meeting in Naples
Altay Askar, G., Lungu, D. (eds) 2005a. PROHITECH (Italy) on 6th–7th October 2005; 3rd general meeting
first main deliverable D-I Assessment of intervention in Crete (Greece) on 24th–25th November 2005; WP8
strategies for the seismic protection of historical building special meeting in Naples (Italy) on 23rd–24th February
heritage in the Mediterranean basin. 2006; WP5-WP6 joint meeting in Ljubljana (Slovenia)
Mazzolani, F.M., Gramatikov, K., Dumova-Jovanoska, E., on 9th–11th March 2006; 4th general meeting in Istanbul
Milutinovic, Z., Taskov, L. 2005b. Earthquake protec- (Turkey) on 6th–7th April 2006 and 2nd Seminar on 8th
tion of historical buildings by reversible mixed tech- April 2006; WP11 “Iron meeting” in Naples (Italy) on 8th
nologies: the PROHITECH project. MASE Macedonian June 2006; WP11 “Rabat meeting” in Rabat (Morocco)
Association of Structural Engineers. on 9th–11th July 2006; 5th general meeting in Poiana-
Mazzolani, F.M., De Matteis, G., Calado, L. Beg, D. Brasov (Romania) on 18th–19th September 2006; WP7
(eds) 2006. PROHITECH second main deliverable meetings in Skopje (Macedonia) on 13th–14th November
D-II Reversible mixed technologies for seismic protection: 2006; WP8 meeting in Timisoara (Romania) on 8th–
guide to material and technology selection. 9th December 2006; WP10-WP11 meeting in Athens
Out-put of the PROHITECH project meetings: 1st general (Greece) on 8th–9th March 2007; 6th general meeting in
meeting in Anacapri (Italy) on 5th–6th November 2004; Liège (Belgium) on 26th–27th April 2007; WP10-WP12
2nd general meeting in Azores (Portugal) on 2nd–3rd meeting in Haifa (Israel) on 5th–6th September 2007.
June 2005; WP1-to-WP4 joint meeting in Naples (Italy)
1165
Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)
© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5
ABSTRACT: The Column of Marcian, or Kıztaşı, is one of the magnificent monuments that were constructed
in the name of Roman Emperor Marcian in Istanbul. It was built in his honor around 450–452. The monument
consists of a square marble plinth block seated on a three-stepped platform and marble blocks on a single granite
column topped with a marble column capital. Through the intervening years, the Marcian Column has been
damaged by earthquakes, fires and repairs. With the 1999 earthquake, the column became slanted on its axis;
the iron rings on the bases and those on the body itself were dislocated; the capital was deformed and hairline
cracks were found in the column’s body. In 2005, the Marcian Column was restored by the Istanbul Metropolitan
Municipality. This article will examine the old and new restorations and will utilize contemporary restoration
criteria to evaluate the last intervention.
1167
Figure 1. The Column of Marcianus/Kıztaşı, at the begin-
ning of the 20th century (Eken, 1992).
1168
Figure 4. Kıztaşı, west façade of the column plinth after
the restoration, 2007.
1169
4.1 Ottoman period repairs
In the photographs of the Marcian Monument taken at
the beginning of the 20th century, it is seen that the
monument had had hoops put on it at various levels.
This repair must have been carried out on the parts of
the monument that were damaged in the 1894 earth-
quake and the 1908 Fatih fire. In this photograph,
the marble stepped platform is still at street level. On
the lower plinth of the column, there are steel hoops
at the column bracelet and on the body. The lower
plinth has been encircled with four hoops, the bracelet
with two and the column body with three. One cannot
understand, however, if the metal cramp and chains
that are on the column today–but are not seen in the
photograph–were put on during this repair or in repairs
carried out later. The photograph also does not show
the statue that is said to have been on the uppermost
of the monument, so it could not have been still extant
at this date.
1170
to the technical reports that had been prepared by – The writing in the inscription section should be re-
experts. mounted on the marble column using bronze letters.
– The rusted and disintegrated metal parts must be
replaced with materials that are bronze or rustproof.
5 THE DAMAGE TO KIZTAŞI PRIOR TO THE – The capital with its eagle figures has to be fixed
2005 RESTORATION into place or the original should be removed to a
museum for protection and a copy placed on top
In the survey measurements carried out for the repair of the column. In the case that a new capital and
it was determined that the monument was not in an new upper plinth are used they should be produced
upright position. It had begun listing, 7 cm to the south by pouring a mixture of white aggregate and white
and 16 cm to the east from its axis (Akşit & Akşit, cement strengthened with fiber mesh into special
2001a). It was determined that, because there was moulds.
an asymmetric distribution of the load created due – The monument inclination should be measured
to the disintegration of the marble upper plinth on periodically for 12 months to determine whether
which undoubtedly the statue rested, the column was there is an increase in incline from its axis. If
in weak condition vis-à-vis lateral movements and it the inclination increases, action has to be taken
had become dangerous for the vicinity (Babuş, 2001). accordingly.
Also, the rusting and decomposing of the cramps and – In order for the repair work to be done in connec-
chains on the capital were the reason for the cracks tion with Kıztaşı, first the monument’s resistance
and broken off parts in the marble. The hoops at the against lateral forces has to be increased and a sup-
level of the lower plinth had dispersed because the bolts portive skeleton must be established with a trans-
had snapped off. Aside from these structural problems, verse support against it from the ground (Akşit &
other problems were also noted: the dirtying of the Akşit, 2001b).
Kıztaşı granite body by time and atmospheric condi-
tions, the cleavages on the north façade and formation
of a crust, corrosion and stain on the metal parts, cracks 6.2 Repair methods recommended in the
on the body, the damage on the relief decorations and restoration project
the breaks in the corners of the eagle motifs. In addi- – Cleaning must be carried out without damaging the
tion, the cement mortar that had been used in previous stone in order to remove both the dirty layer on the
repairs had produced salt and caused the materials to façade and the cement mortar supplements. Squirt-
behave differently. ing water on the monument will soften the dirt and
then a brush with a plastic claw brush and water
should be used to eliminate the dirt. For dirt that
6 2005 RESTORATION PROJECT DECISIONS cannot be removed, a mixture of mashed paper and
10% ammonia carbonate should be applied onto
In the restoration project, the suggestions given by the surface of the stone. This should be covered
the Istanbul Restoration and Conservation Center were with plastic sheeting and allowed to wait for 2–3
adhered to for cleaning, repairing and protecting the hours and then the process should be repeated until
stone and metal materials in the Kıztaşı (Gökce, the desired result is attained. During this work care
2002). must be taken that metal parts do not come into
contact with the solution.
– Filling for the ditch and cavity stone surfaces: Mor-
6.1 Basic principles approved of for the restoration
tar including lime and mortar that suits the aspect
project
of the stone must be used. In joint fills the filling
– Today the monument is located at a crossroads with material used in prior repairs that include cement
heavy traffic flow and must be partially protected must be carefully removed and replaced with mor-
from the traffic by being isolated within a circular tar that is a mixture of hydraulic lime mortar and
square. marble powder.
– The stepped platform of the monument that remains – Filling cracks: epoxy that is not affected by water
underground must be uncovered with an excavation and is especially designed for bonding (Araldit
supervised by the Istanbul Archaeological Muse- AY103 – Hardaner 956) ha