Struclural Analysis oí Historical

Constructions.
Possibilities oí Numerical and
Experimental Techniques
Structural Analysis of Historical
Constructions.
Possibilities of Numerical and
Experimental Techniques

Edited by

P. Roca, J.L. González, A.R. Marí and E. Oiíate

Universitat Politecnica de Catalunya
08034 Barcelona
Sp ain

A publication of

lnternational Center for Numerical Methods

in Engineering (CIMNE) Y f ... I I
Barcelona, Spain lDLlIlIII
1987-1997
 Structural Analysis of Historical
Constructions.
Possibilities of Numerical and
Experimental Techniques

THEORY AND ENGINEERING APPLICATIONS OF

COMPUTATIONAL METHODS

A SERIES OF HANDBOOKS
Series Editors

S.N. Atluri Georgia Institute of Technology (USA)

T.J .R. Hu ghes Standford University (USA)
H. Mang University of Vienna (Austria)
E.Oii.ate Universitat Politécnica de Catalunya (Spa in)
B. Schrefler University of Padova (ltal y)
R.L. Taylor Uni versity of Berkeley (USA)
O.C. Zienkiewicz University eollege of Swansea (UK)

Structural Analysis of Historical Constructions. Possibilities of

Numerical and Experimental Techniques
P. Roca, J.L. González, A.R. Ma ri and E. Ofiate (Eds.)

Fies! edition, January 1997

© lnternational Cenler for Numerical Methods in Engineering (CIMNE)

e/Gcan Capitán sl n, 08034 Barcelona,Spa in

Cover design by Jordi PaUi

Prinled by Artes Gráficas Torres S.A., Moralcs 17, 08029 Barcelona, Spain

Depós ito legal: B-307 18-96

ISBN: 84-81861-66-9
 PREFACE

The lnternationa! Seminar on Structural Analysis Df Historical Constructions, which

was held in Barcelona on November 8-10, 1995, gave the oportunity to presfigious
specialists to meet, to lecture about methodology and experimental and numerical
techniques, to present some emblematic practical cases, and finally to discuss about the
more challenging points which arise in the study of the structural aspects of ancien t
monuments.

This book includes the transcription af the conferences presented at the Seminar. The
text has been carefully prepared by the same authors in arder to p rovide
comprehensive technical and graphical information. Throughout the different
contributions, each Dne corresponding to a chapter of the book, the editors wish to
provide the reader with updated informatio n on the possibilities of the available
techniques for inspection, analysis and repair of historical constructions. In addítion,
an important simultaneous attempt is made to introduce the reader to the more
conceptual or methodological aspects involved in the analysis and intervenhon on
the ancient monuments.

A final chapter has been prepared inc1uding a summary of the discussions held
during the Seminar. The different opinions of some of the participants are presented
following the most relevant topics treated duríng the meeting.

It is the adequa te choice Df the methods and tooIs of analysis which shou ld lead to a
deep understanding of the structural performance of ancient constructions, thereby
allowing to define a truly effective, although fully respectfuI, strategy for intervenhon
and repair. We hope that the gathering of the opinions and the presentation of the
different practical cases selected in this book will contribute to develop the cri teria
needed to lay·out future activities in this field.

P. ROCA, J.L. GONZÁLEZ, A.R. MARí and E. ONATE

Barcelona, January 1997
This book, as part of the activities involved in the organization of the International
Seminar on Structural Analysis of Historical Constructions, has been sponsored by the
following institutions, to which the editors want to express their gratitute: Dirección
General de Promoción de la Investigación, Ministerio de Educación y Ciencia deI
Gobiemo Espaf\ol; CIRIT, Generalitat de Catalunya; Departament d'Enginyeria de la
Construcció, Universitat Politêcnica de Catalunya; Servei de Patrimoni Arquitectonic
Local de la Diputaci6 de Barcelona.
CONTENTS

The strudural analysis in lhe method af monumental restoration

A. GOllzález.

General methodology. The combined use 01 experimental and numerical techniques

inside a singlc study
G. Macell/ ... ...... ............ .. 10

Possibilities of lhe experimental techniques for lhe struclural analysis af historical

constructions
P.P. Rossi 24

Experimenta l techniques for ausculalation of historical construclions

R. Asllldillo Pa s tor . . ... .... ............ ... ... .. . 47

The testing, analysis and assessment af masonry arch bridges

T.G . Hughes . ........... . 64

Characterization of lhe mechanical behaviour of masonry

C. Molills Borrell .. ........................................................... .. 86

$Iructural aspects of lhe rehabilitation a f lhe Mexico City Calhedral

R. Meli alld A. R. Sa"chez~Ralllirez 123

The Angkor's Temple in Cambod ia : Preliminary structural s tudies

and slrategy Df intervention
G. Croei ... .......... .. ... ....... ....... .. ............. ... .. .. ... ........ . 141

The Colosseum: Safety eva luatioll and preliminary criteria Df intervention

G. Croei . 154
Analysis, diagnosis and preservation of ancienl monumenls:
The SI. Mark's Basilica in Venice
F. Mola and R. Vitaliani ..... ... . 166

51ructural analysis and durability assesmenl of hislorical construction s using a finite

elemenl damage model
E. Oiiate, A. Hanganu, A. Barba!, S. Oller, R. Vilaliani, A. 5aelta and R. Scolta . 189

Charles bridge in Prague

D. Novák and ]. Zák 225

Ancienl domes on 5pain

}.M. Iz-quierdo Berna/do de Quirós 246

Proposal for the restoration of Marcus Aurelius's column

A. Guiffre . ........... .. ............... .............................. . . 270

The spires of Burgos CathedraJ.

LM. Ortega, }. Perelli and }. Alberruche 290

Roman aqueduct of Segovia

F. Jurado jiménez ......... .. ... .. .............. .. ................ ... ................ .... ............... . 307

51ructu ral inlervention for lhe rehabilitalion of "Las Golfas de la Pedrera"

R. Bnifau Niubó .......................................... .. ................... ....................... . 341

Previous studies of the "Casa de los Sotines de Leon"

}.L. González M oreno-Navarro ....................................................................... .... .... .... . . 356

5tudies of Gaudi's "Cripta de la Colonia Güell"

P. Roca ..... 377

A summary of lhe opinions pul fornrard during the discussions

P. Roca and }.L. Cotizá/a ... ............. ... .... .................. ........... . 394
 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTlONS
P. Roca, J.L. González, A.R. Marí and E. Dilate (Eds.)
\O CIMNE, Barcelona 1996

THE STRUCTURAL ANALYSIS IN THE METHOD OF

MONUMENTAL RESTORA TION

A. González
Diputació de Barcelona
Compte d'Urgel 187
08036 BarrelOlla
Spaúl

I wiU begin my pap er with some words of thanks to the organizlI1g

com mittee of this Seminar. This is not merely a formula of courtesy in rcpl)'
to their invitation. I also intend to express my recognition that it has been
considered s uit ab le to place the contributions within in the general context
of monumental restoration from the particular viewpoint of the struct ural
analysis of hi storie buildings.

Though it may su rpri se us , in most mectings such as this it is customary

to analyze the specific aspects of a scie ntific and technical nature without
dealing with the conceptual, social and methodological framewo rk in which
the intervention s are made . I therefore hope that this desire of the organizing
committee to set the discussions within a broader context wiU enrich the
results of the Seminar.

1 wiU not speak of the techniques of structural analysis of historical buildings.

That is the task of you, the specialists. J 0111)' intend , as an architect who
coordinates the multidisciplinary teams devoted to monumental restoration ,
to provide some ideas on the objectives of this analysis, envisaged from
the viewpoint of the discipline as a whole. They are reftections borl1 from
experien ce, some of them made in the company of my brother, Jos Luis
Gonzlez (to whom I owe so much of rny concern to improve my method of
thought each day).

Monumento.} restora.tion is a young discipline, with a history of hardly twn

hundred years. They have been, however, two hundred years of intense
controversy on the theoretical aspects concerning the nature of historical
buildings or the action taken in them , both with regard to the objectives
and the means to be used.

This permanent discussion J and the changes in the socio-cultural framework

in which it is situated J have produced an evolution in both the conceptual
contents and in restoration practice, which has beel1 espcciaUy in tensive in
the last decades of our century.

One reason for this evolution is urban growth and developrnent, the extension
2 STRUCTU H/\L ANALYSIS CF HI $TO R1CAL CONST RUCT10NS

cf industrializalion and the obsolescc llce a f historical buildin g typologies,

together with t he resu lti ng confli cts over the detcri oration ofthe environment
and the historical heritage.

Anothcr rcason is tha! a more profoll ud con cept ual analysis is made of the
nature, extent a nel divcrsity af th e built hi storical heritage th at mu st be
preserved. There is even diversity in the concept af conservation according
to the different cultural and social values af each co unt ry.

Howeve r , we mu st not forgct the development af therapeu tic and analytical

techlliques lhat are applicable to the conservation af thi5 heritage.

In t his new idea af restoration one must emphasize firstly the enhan cement
of the concep! of the Monument.

In the lIew restoratio ll the Monumellt can no longer bc understood from a

partial or fragme ntary view poi nt, wh ich has 50 oftcn produced t heories and
pradices with serio us shortcomings for a pproaching in ter vention in the bu ilt
heritage.

Cert ai nly, the first con di tio n of th e monument is its nature as a hi storical
document., as a result or a sce n ario of fact s, art s, techniqu es and cultures ,
a nd therefore its nature as a report.

But togct her with this natllre as a document , one must normally consider its
vaüd ity as a li ving architectural phenomenon: its capacity to pIay an active
cultural , social, urban or regional role in the presen t, due to its capacity to be
used and due to the presence of a r tistic, sp atial, techn ical anel cOIlstructional
vallles t hat are prope r to archi tedure or engineering.

And finally, as a conseque nce of these two aspects, its nature as a meaningful
element, because it lias symboli c and emblematic values fo r its user , valu es
t.h at mu st also be decisive, like the previo us ones, for t he co nsideration of
t he objectives and strategies of conservation.

The Roman amphithealre of the city of Nmes may serve as an example of this
triple essence. It is a hi sto rical docum ent of co nsider able im portance, and at
the same time a useful sp ace for t he current population an d an emblematic
a nd representative elcment of the city.

Anoth cr con ceptual aspect t h at has suff'ereel a consid erable evolution anel that
is now fashionablc in lhe internationa l for ums of monumental resto ration is
thal of Authenticity.

Th e Letter of Veni ce of 1964 says in its preamble t hat 1I11Umanity recognizes

its joint responsibiü ty for the conse rvation of the com mon monumental
h eritage and must aspire to transmit t hese monum ental work s with aU
the wealth of t.heir aut henticity". Thirty years latcr, the very concept of
monumental authen ticity continues to be discussed.

Recently, ex perts of IC OMOS at a meeting in Nara (Jap an) t ried to fin d a

 A GONZÁLEZ I S!11Ic!ural :lIlaJys is in rnonumCJl!aJ rCSlOr.:ltlOn

definition that wouId be more respectfu l of the cultural and social valu es of
aU countries. It mu st not be forgoiten that in some easte rn count ries t he idea
of transmi ssion of a monument does not presuppose that it is untouchable ,
since in their opinion they may be destroyed and rebuilt without thereby
losing their authcnticity.

In our cu ltural context we have abo been dernanding a new definition of the
concept of authenticity for some time.

In ou r opinion, authenticity cannot be associated with the or iginality of the

m ateriaJ of which the monument is composed, but. rnust be associated with
the capacity of t-his material - regardless of its chrollology - to certify and
transmit as true (that is to say, as ((authe nti c") the sd of valu es and messages
of the monument, whether documentary, constructional, spatial, artistic or
emblematic.

The authenticity of a molding , a column or a space depends more on whether

it corresponds faithfully to the forms and intentions of its authors than
whet her the materiaIs of which these elements are composed are the original
ones or have beeo rc pl aced.

An examp le of this is t he Courtyard of the Lions of the Alhambra in Granada.

Possibly, most of t he materiaIs that we can see today in it (columns,
pavements, plaster-works, tiIes) are not of the 14th century but were added
in th e modern rcstorations.

However, if t he space corresponds to the original space co nce ived by the

Nazarene master builder, ir the babbling of the water sounds as it did then,
anel ir the beautiful G r a nada sun produces on the walls the interplay of lights
anel sh ades that bewitched our ancestors , who can deny the authenticity
of that Hispanic-Muslim architecture that we transll1it with pride (rom
generation to generation?

This enrichment and evolution of the concept of the rnonument is at the same
time having decisive consequences on the practice of restoratioll.

First ly, each day it seems more evident that. the restoratian of historical
constructions has littI e to do wit.h the restoration of other artistic heritages.
Today we can statc that ir the VerlUs of Milo were architecture... its
restoration would involve giving it back its arl11... anel poss ibly br in ging
its image up to date.

Co nsequ entl y, the inflexibility a f restoration cri teria conçcived in lhe image
af those applied in the restoration of other types of works has di sappcared.

There are no longer universal cri teria of monumental rcstoration that are
preestablis hed by schoo1s ar doctrines. They are defined in eac h case
according to the manument, its órcumstances anel the objcctives of j,he
act ion.

The resto ration af an ancien t car \ViU be very eliffere llt according to whether
the vehicle must be roadworthy. Tf it is only to be e>..hib ited, fhe recovery of
4 STRUCTURAL ANALYS1$ OF HI $TOIHCAL CONSTRUCT10NS

ih original beauty will suffice. If it bas to be roadworthYI lh e engine \ViII have

to be restored or replaced, and it will iJave to be adapt.ed to the prevailing
norms OH indicators, acoll stic signals, safety, etc.

l3ut it may occur that. lhe circumstances or the objcctives of the consc rvation
make a lho roughly differcllt treatment. advi sable.

The car in which admiraI Luis Carrero Blanco, head af the Spanish
gover mncnt. , was tllurdcred in Madrid in Oecember af 1973 was cxhibited
in a military museurn just as it. was. Because the objective af the exhibition
\Vas to show the barbarism of that event, no one thought of restoring it.

This flexibility of cri teria according to the objectives has made it necessary
to establish with maximum rigour a method that ensures the scientific
correctness of the whole processo

I \ViU refer to the mcthod which we detlned some years ago in the Local
Architectural Heritage Service of the Provincial Council of Barcelona,
because it is lhe one which I know b est. I am sure tha.t in its essential
aspect it coincides with other methods.

The basic principies 011 which our method (which we have called the
" Objective Restoration " method ) is based are the foUowing:

Firstly, the e qu itable understanding and valuation of lhe three dimensions

of the triple essence of the Monument to which I have referred above: t he
aspects of lhe monumenl as a document or a work of architecture, and the
meaningful or emblemalic aspects. Also the concept of authenticity as it has
been defined.

Secondly, thal elemental principie in all the o rders of human activity, which
is often forgot ten: the need to adapt the means to the end. The aims or
objectives, which must always res pond to the social interest, are defined
according to the triple essence of the monument, its circumstances and the
sacio-cultural and historical environment of the action. Anel the means,
which involve a dual scientific and creative asped, are chosen according to
their efficiency for achieving the objectives, taking into account lhe available
resources.

FinaUy, as a third principIe, lhe acceptance of the universality of the metho d ,

but not of the criterioll of action, which, as we have seen, must be denned in
each case.

The method b ased on these principies distinguishes three main sections

ar phases of the adion ou the monumcnt: the previous knowledge, the
int ervention itself and, between the two, a reflection and determination of
the ends, means and criteria.

The essence aí these three phases is thi s: the greater the knowledge is, the
more objectivc the reftection \ViU be and the more appropriate (that is to say,
more effedive with less errort) the intervention.
 Â. GONZÁLEZ I Sl rUCIUr::l1 analysis 111 monurnClllal rcSlOralion 5

Th e knowlcdge phase is s ubdivided into two stages. The fir st is that of

preliminary diagnosis, in which on the basis of th e preliminary analys is of the
exis ting informatioIl , the scope and depth of the secon d stage of research'and
definitive diagnosis are considered. This stagc is therefore highly important
since it wili avoid eco llolll.ic waste alld permit a greater efficiency in the
second stage.

In lhe third stagc, oncc th c monumcnt has bcen understood in depth, t he

ob jectives, the available resources and the criteria anel means to employ in
its restoration or conservation are analyzed.

The fourth and fifth stages, as s ubphases of lhe intervention phase, envi sage
the projects, the car rying out of the work anel ot her indis pensabl e tasks,
such as the promotion o f the partici pation of society through the public
elissemination of the projects.

The stage of knowledgc, which is the one that 1110st interests us here ,
contains a series of projccts and studies t hat envisage the three aspects of
the monument: the documentary aspects, the co nst ruct.ional aspects, and
the emblematic aspect s.

The first include aU the studies that pursue knowledge of the monument
as a historical document o Firstly, the archaeology projects and the other
co ncu rrent sciences (nurni smatics, ceramography, phys ical anthropology,
dendrochronology, etc.), that is to say, the hi stori ca l stu di es t hat are based
on the analysis of the materiai s of the monumellt.

Secondly, the do cumentary hi storical studies, the analysis of the history

from the written , graphic, photographic or verbal documentation on the
11l0nument. And also, the analysis t hat can be provided by the history of art
or the history of construction.

They are not , of cou rse, c10sed studies, b ut s hould be related to each other
in order to contrast t he info rmation that they ge nerate.

An importanl aspect of the stage of knowledge is the analy sis of the

mOllulllent as a material object, of its physical reality in the presen t. A facet
of this study is what we have called the " physical-co nstructive analys is" of
t he monument, of which the structural anal ys is dealt with in thi s Seminar
form s part.

Before continuing , I wish to make a termin ological definition. I use the

word architecture (und erstoo d as art or as a work resu lting from this art)
as sy nonymou s with construction, a concept that includes both bu ildin gs
and other objects of engineering. To use this synonym generally faciLitates
the exposition and evielently does Hot involve any attempt to subordinate
or di sdain any of the disciplines that share with architecture the art of
construction .

Th e ph ysical-const ructi onal analysis is und erstood as the functional analysis

anel the analysis of the state of conser vation anel safe ty of the mOllument,
which is understood as a building system , and of the various sub systems,
6 STRUCTURAL ANALYSIS OF H!STOR1CAL CONSTRUCT IONS

elements, types of masonry and material of wh ich it is made up. An essential

aspect of this physical~constructive analysis, as I said before, is the struct ural
analysis.

Having considered th is methodological framework in which I reei t he

structural analysis of historical buildings should be examined, I will now
refer to the comments that I announced at t he beginning.

F irstly, I believe it is necessary to state my conviction that in order to

effective1y approach t he physical-construction al analysis in genera l and the
structural analysis in particular, it is indispensable to envisage the physical-
material reality of the historical building with a constructional mentality that
is as dose as possible to that of the period in which it was b uilt, avoiding the
current analytical mentality that shows ao excessive tenden cy to perform a
fragmenta ry analysis of that reality.

The current building culture is unlikely to be useful for the com prehension
of most of the monuments that we restore. The mental pat terns - the
mentality - generated by industrialization, the application of p roced ures
intended for new buildings, or the separation between strudure and enclosure
wallings, between permanence and stability, and other many fragmentations
that have allowed building culture to evolve positively in Qur cent ury, h inder
an understanding of the behavioural reality of the b u ildings of the past.

This does not deny th e validity of current science a nd technology, b u t merely

the fad that they are endowed with a timeless validity that is useful for the
comprehension of the past. It is clear t h at the current methods serve as
techniques of analysis of the material microreality af the monuments, a r to
develop sophisticated techniques of consolidation and reinforcement. But the
culture in which they are located is useless in itself for an understanding of
what t hey analyze and reinforce.

An example of this is the frequent erro r of terminological interpretat ion

consisting in not differentiating the so different meanillgs t h at t he term
"st ructure" has had in the 19th and 20th centuries, or interpreting a work
of ashlar arches and pillars in terms of the behaviour of rein forced concrete.

Another fundamental asped to take iuto accoun t in order to guarantee the

efficiency of the str uctural analysis is the interrelationship of all knowledge
about the monument, whethcr of a physical.construdion al nat ure - in which
the structural analysis is included - or of a historical nature .

As I said before, the physical·constructional analysis makes reference to the

m onument understood as a present physical reality, apparently regardless
of its historical nature. However, this analysis should be made after
the hi storical studies, since these can provide many data t h at facilitate
the comprehel1sion of that physical-construd ional reali ty. Not only the
chronological data of the constructioIl, but the data they can explain : for
exampl e, the pathogcn esis, the origin of t he d amage. The warks of st rudu ra l
anaJysis, therefore, sho uld be programmed and performed in coordination in
t he knowledge stage of the met hod.
 A. GONZÁLEZ I Structurnl ano.lysis in monumental reslorali on 7

It aIso occurs with the strllctllral intervention project that it canuot be

disconnected from the general restoration project, since this is approached
according to certain ends, rneans and criteria analyzed on the basis of the
com plex essence of the monument (as a document, as a work of architecture
and as a meaningful object) and of its overall problems.

The st ructural interventioll project shou ld HOt impose actions that may be
contrary to the objectives anel cri teria of the general ilction consiclcrccl on
the basis of the general analysis of the monument. An example: a correctin g
structural restoration of damage Of physical-material anomalies must take
into account the documentary and meaningful aspects that the externaI
manifestation of pathologies may have, which should perhaps be maintained
apparent as an illformative or even commemorative datum.

With respect to the structural intervention project, 1 believe it is suitable

to cite the recommendatiolls that arose from the discussion of the Ravello
Group, in which I have the honor of representing Spain. lt is a working group
sponsored by I COMOS and other international organizations that met last
May in that Italian town to discuss the structural aspects of restoratioll.
These are the recommendations:

"The structuraI restoration project"- says the elocument of Ravello - " must
be based on a methodology not of numerical codes but of methodological
codes."

"The project must include an expIanatory report which includes the

description of the research program , the causes of the damage and the
deterioratiou, the evaIuat ion of the safety, the comparison between possibIe
alternatives and the justification of t he options finally chosen , anel a
monitoring plan to be applied in the course of the work, indicating the
purpose of each control anel the instruments to be used. The controIs shoulel
continue after t he work has finished in order to check the results."

"Concerning the evaluati on of the safety" , continues the elocument of

Ravello , "it must be based on t h e joint consideration of the historico-critical
analysis of the construction in its context, the qualitative evaluation based
011 an examination of the building, and t he t heoretical analysis through
mathematical models. The conclusions that are obtained and that may be
incorporated in the intervention projects mllst be based on a criticai analysis
of the results of the process , without losing sight of the relative importance
of the diverse information that has been obtained."

Concerning the cri teria of the project of restoration of the structural as pects,
the document advises: " No intervention should be made until it has been
checked that the safety leveis are actually insufficient. The intervention must
be proportional to the safety objectives and therefore as minimal as possible.

As a general criterioll, the trend must be to conserve and respect the original
concept and techniques of the structural system of the monllrnent anel,
when possible, the measures taken should be reve rsible, so that they can
be rep laced by more suitable measures in the light of new knowledge".
8 STRUCTURAL ANALYS IS or HI STOR ICAL CONSTRUCTIONS

"In no case wiU measures be chosen whose result cannot be verified ar that
cou ld produce harmful secondary effects. Concerning the dilemma between
traditional techniques and new technologies ) each case mu st be cons idered
individuaU y, giving preference to the techniques that involve the greatest
res pect for the original work and that are the least aggressive ."

" All the materiais used in restoration af a st ructural type mu st be examined

in d etail, and ooe must have documented evidence not anly of their
characteristics, but also of their compatib ility with the original materiais,
in arder to avoid as (ar as possibl e harmful secondary effects. "

" Furthermorc, the difficulty af evaluating real safety levels and the possible
benefits of th e intervention may suggest an intervention in stages with lhe
subsequent adoption of cornplern entary or corredive meas ures. Thus , the
meas ures can be minimal an d , especially in the prese nce of evolutionary
phenomena, they can be adapted graduaUy according to response of the
strudure."

"In aU case~" , concludes th e document of Ravello, "upon finishing the

wo rk ~,the organi:mtion responsible for the conservation should be given the
complete docum entation and a description of the projeds and the general
lines of maintenan ce."

I wiU fini sh hy calling the attention to some points that see m to me to be

essential at this important time for the defini tive generalized io staUation of
structural allalys is in the area of monumental restoration.

The first poiot is the need for a safety code that is applicable to the
monuments a nd is different (rom that which is applicable to the design of
new buildings anel structures. In thc latter case, the safety caeffi cients must
take into account the un ccrtainty of the good execu tian of the building ar
th e projected st ru ct ure. However , in t h e analysis of the safety of historical
construction s, the checking of t heir real behavior over a long history is a most
importaot empirical e1ement that can permit t he modification of the res ult s
obtained from the t heoretical calculations.

I a lso wish to cali the attention to the need, in parallel to the advances
in techniques and methods of strudural analysis, for mcchanism s to be
established that enable easy preliminary diagnosis, with the greatest possible
precision , of wheth e r these analyses need to be applied.

An indiscriminate generalízation of stru dural analysis m et hod s call make

the restoration process m ore expensive, with t he consequent negative effeds.
Thus, the compan ies specialized in carrying out the analy sis sho uld not be
the same as those t hat prescrib e its need , something that is increasingly
co mmon, at least in our country.

F'urthermore, both the analytical methods and preliminary diagnosis

m cthods should be designed and adapted b earin g in mind their application
to rnonuments of me dium or low complexity because they are t he most
abundant. A techn ique of st ructural analysis designed solel y a ccording to
large monuments may be ineffedi ve for the whole discipline of monumental
 A. GONZÁLEZ I SlruClural analysis in monumcntal rc:.l0r.llion 9

restoratioll.

And fina lly, I put to your conside ration the advisabiüty of not encouraging
a new culture of Hstrudural restoration" as an end in itself, apparently
regardless of the essential objedives of monumental restoratioll. It is in
this sense that I c1aimcd in the Ravello Group - and now J am doing; it here
- that we should BOt speak of strudural restoratioll , but of the C:strud uraJ
aspects of restoration" .

I hopc that this Seminar wiU provide a11 the spccialists gathered here with
spccific technical knowlcdge to make their work more effective cvery day.
And J also sincerely hope that the indispensablc reflection on resloration
in general wiU aUow them to situate this particular knowledge in a wider
II1cthodological frarnework whose purpose is to guarantee the pro tectioll and
the coUective enjoymellt of the monumental heritage that we have inherited
and that we must jointly transmit to future gencrations.
 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS
P. Roca, J. L. González, A.R. Mari and E. Onate (Eds.)
e C1MNE, Barcelona 1996

GENERAL METHOOOLOGY. THE COMBINE O USE OF

BXPERIMENTAL ANO NUMERICAL TECHNIQUES INSIDE
A SINGLE STUDY
G. Macchi
U/liversity Df Pavia
Ita/y

SUMMARY.

Structural engineering is only one of the di sciplines which shou ld cooperate when
an ancient construction may show structural damages and may need a strengthening
intervention. Ali the components of the value of the building should be taken into
aceount, in an optimization procedure aimed to avoid unneeessary works and loss of
the historie and art istie value.
Historie analysis should be first , and the knowledge of the events of the
eonstruction, eollapses or restorations, is essential before proeeeding to new
interventions.
Numerieal methods are particular1y efficient means for the structural diagnosis,
but shall be supported by experimetally fou nd input data. The criticisms about lhe
use of numerical analysis is basically due 10 the suspition Ihat the physical reality and
lhe constitutive laws are nOl appropriately taken int o aceount. The sarne could apply to
physieal model s.
Nonlinear methods should be used, but even linear FEM analyses ean provi de
useful informati ons for a diagnosis.
Experimental techniques as dynamic testing and continuous monitori ng open new
routes 10 st ructural identification.

I. ROLE OF STRUCTURAL TECHNlQUES IN PRESERVATION OF

HISTORl CAL CONSTRUCTIONS.

Experimental and numerical techniques are boI h essenlial in engineering problems,

and are essenlial each other. Only in currem cases, when the mechanical modeJs are
very advanced, a reduction of experimental studies is alluwed, and lhe attained
knowledge may be applied through lhe numerical methods only.
However, lhe fa nt asl ic effi ciency reached IOday by the numerical methods is such
that one may believe they can be self-sufIicient and nol need the experimental input.
Such an absolute con fid ence in numerical methods would be wrong and give ri se 10
serious risks.
The risk of overestimati ng numerical analyses and underestimaling lhe "ontologic"
study of lhe structure [I ] is already present in normal design of new stru ctures; lhe
risk is more evidenl in the diagnosis and in lhe interventions o n hi storical
const ructions.
When dealing with hi slOrical monuments, the rol e of mechanics and generally of
engineering should be limited, and lhe engineers should accomplish such a role with
 G.MACCHI I General lllelbodology II

modesty: the problem of heritage preservation is very complex, and involves multi-
disciplinary consideralions [2]. When designing new works we should remember thal
lhe aim of the new works is not "their resistance" (11 , bul their possible use for human
needs; similarly, we should understand Ihat the preservation of a monument cannot
have lhe only aim of ils surviva!. When strengthening an hislorical construclion we
risk to damage its substance, and unfavorably affecI the Qualities themselves for which
we would like to preserve it.
Unnecessary strengthening works are not neutral: they may negative!y affec! the
historical nature of lhe building and modify its materic nature . The problem of
stability of an hislorical monumen! does no! find its optimal solution in interventions
which guaranlee lhe safety onJy: the solution should be lhe result of an optimization
process, taking into aceount ali lhe components ofthe value ofthe monumenl [3].
Due to such eonsiderations, lhe problem is panicularly difficult, and the current design
procedures for new st ruclures, as weU as the eorresponding safety crileria, cannot
apply.
Funhermore, lhe nature of lhe ancient structures is generally more complex Ihan
for the current modern struclures, 50 that the current melhods of analysis may result in
bad models of their behaviour. This is lhe reason why lhe physical sludy a:1d a
" meehanieal survey" of the eonstruction cannot be omitted, and the numerieal
analysis is only an efficient 1001 of inference of knowledge deriving from lhe physical
observation, which ofien ca nnot be totaIly included in lhe constitutive laws of lhe
materiai s.

2. CONTRJBUTION OF THE HISTORIC ANAL YSIS .

After an appropriale survey of the struclural pathoJogy, before any physical or

numerieal analysis. the methodology requires a deep hi storie survey.
The historie studies provide informations aboul lhe sponsors of lhe original
eonstruction. lhe artists succeeded in lhe produetion of lhe main valuable parts,
politieal and artistie events: sueh elemenls are the memory of lhe historical life of lhe
monumenl and are a measure ofits importance and of the need ofi ts preservation .
For a diagnosis and a possible deeision of a st ructural intervenlion, lhe historie
survey mighl provide funher informations, precious lo the struetural engineer.
If lhe building is matter of eoncem for its stability, it general1y shows evident
damages suffered in lhe past, o r past accidents are documented and previous
restoration s may be detecled. The knowledge of such events and of their date is a
preciou s element for the present structural evaluation.
The survey shou ld allow an ideal reconstruction of lhe Slruclural patlerns and of
their evolution (and sometimes superposition) which took place along lhe life of lhe
monument, and their combination in a global structural mode!.
Without an hist orie knowledge of the collapse of the vaults of the Beauvais
Cathedral in 1284, it wou ld be difficult lO understand lhe present pattern of alternated
piers and buttresses: the new piers have been introduced lo halve the o riginal span
when re-building the quadripartite vaults as sesquipartite. Without the knowledge of
the colJapse of the Jantem tower in 1573 it would be impossible to imagine the
differenl hi story of the 4 main piers, those whieh survived and those which were rebuilt
[4].
In olher cases, the change of use of the build ing has been the reason of o pening or
closure of windows, and of changes of the moduli of colonnades or arches, as in the
case of the Broletto in Pavia.
[2 STRUCTURAL ANALYS lS DF Hl STOR ICA L CONSTRUCTIONS

Fig. l . Bc.1uvais Cathcdra l. Original picrs

and ncw picrs. (Taupin)

D Fig.2. Pavia . Brolctto.

Supcrposition of severa I
Slmctural pattcrns.

Fig.) . Hislol)' of collstruction oflhe

Lcaning Towcr ofPisa (polvani)

.,.'........-...-.... ~U
~'Sí

"
t§
•

• g
"

"
... d .. .

~
, f. g
1
,' ••-

"• ~ g
Q

~
1
U
I

~
-~
e ~
, ~ g
f
~

~ g
~nn , _
~
_
 G.MACCHI / Gener.ll lnelhodology 13

Detailed infonnations on restorations, replacements of structural elements, and

strenghtening works of the past, are essential to get a complete scenario of the
pathology and proceed to new interventions. Restorations and replacements hide the
origir.al pathology and rnake difficult its understanding. In Pisa, the Tower does not
show on the plane of leaning the severe vertical crack pattern that was probably there
at the beginning of the last century, when part of the externaI rnarble facing has been
replaced ; the warning is given,on the contrary, by the cracks in the less stressed
interior facing.
On the other hand, accurate restoralÍons may be misleading in the structure
evaluation, because apparently efficient walls or colurnns may be in fact panially or
totally inactive because they could not be forced properly when placed; this is the case
ofthe medieval Maino Tower in Pavia, where a large pan of a wall was found totally
unloaded at the basis when the stresses were rneasured by fiat jacks.
Similarly, the etfective resisting role of the 180 columns of the Pisa Tower is
uncertain, because the large majority of them have been replaced at different times. In
the present phase of remedial works on the Tower, a precise documentation
(unfonunately not found) of the restorations done in 1838 by Gherardesca and in 1950
by Piero Sanpaolesi would be precious; the incornplete description of the
walerproofing works done by Rodio in 1934 led to discover during the works
unexpected connections with the foundation, and to modity the present project [5J.
The knowledge of the history of the construction works is fundamental in the case
of damages due to settlements of the foundations . In Pisa, for instance, the dates and
duration of lhe interruption of the works at the Tower are essential for the assessment
of a geotechnical history, and therefore for a calibration of a mode! of the clay and for
a numerical analysis appropriate to the prevision of the effects of any intervention
today, such as the application of counteracting moments, of pressing slabs, or an
underexcavation .
Unfonunately, very seI dom precise documents of the works are found in the
archives, and the structural aspects and details are ofien missing. Even in the best
examples, the chronicle repons the payement of the works, but not their location and
technical details which would be useful for an evaluation of lhe intervention : this is
understandable for the ancient interventions, because the structural and mechanical
features ofvaluable buildings were not considered as fundamental. Today, it is a duty
to provide as-built drawings and fuH description of the works, chains, grouts, etc .,
which will be essential to those who will work again in the future on the same
monument o

3. PHYSICAL INVESTIGA TIONS (MECHANICAL SURVEY) AND PHYSICAL

MODELS .

The practice of testing physical models of fuI! structures is today leR 10 research
programs, particularly for cyclic behaviour and large dynamic excitations (as for
seismic actions), because numerical models are far from being satisfactory.
Physical models are less competitive than numerical models in terms of cost and
time. Alberto Peano, head of the research depanrnent of ISMES , an instÍlute famous
for physical testing (as e.g . for the Milan Cathedral) believes "that appropriate results
could had been obtained by numerical models in a cheaper and faster way" [6].
Quite different seems to be the opinion of a famous architect, Raymond Lemaire :
"The ancient constructions, which are extremely complex sometimes, such as a gothic
cathedral or a Guarini's dome, cannot be correctly analysed by the classic numerical
methods" [3J .
 Max Prlnclp ..
0.5
0.45833
0.41666
toR:[IQ 05tR1R1.) ~ 1/1i!/!P-l 0.375
TCf!RE DI miA · PRII11UEW. 0.33333 '"-;
SCUPIS • SUlO PESlJ I'RIPRlO
lElt5lI1E HRllCR E lZ t.....}
0 .29166
0.25 '"Q
c
0.20833 c
0.16666
0 .125 '">r
0 .08333 >
0.04166 z
O >
~
V>
v;
O
""v;
T

Ó
'"
ri
>
r

ª
~
"
c
Q
õ
z
'"
Fig.4. Leaning Tower of Pisa. FEM anaJysis ofthe crilical zone aI lhe first loggia.
a. vertical strcsscs (conccntrated in t!te marblc facings)
b. lensile principal stresses (concentmtcd .under lhe marble facing and aI ceiling and
pavement oft!te helicoidal stair.
 G.MACC HI I G,:ncr.l.1 IllClhodology 15

Nevertheless, there are some common reasons in the apparently opposite

statements. Lemaire stresses the fact that is hardly possible to model some scarcely
known properties of the ancient mortars, therefore of the behaviour of ancient
masonries. Peano argues that for the same reason ao artificial physica! mode! is not
better than a numerical one, and the Jatter may be more appropriate if parametric
analyses are used : identification techniques allow a determination of the poorly
known mechanical parameters.
In conclusion, there is no doubt that no model, physica! or numerical, can reach a
reasonab!e credibility without some essential input data as the mechanical (and
rheologic) properties of the materiais: constitutive laws, strength in compression, in
tension and in shear, and the stochastic distribution of such properties, as well as the
(deterrninistic) variation within the structure, due e.g, to works done in different ages.
Core drillings lO! mm or 76mm of diameter, operated with usual techniques, are
quite destructive, and are ofien insufficient for direct laboratory testing, particularly
on infill materiais, when large mortar volumes are alternated with coarse slones or
randomly oriented bricks, If drilling is wet operated with diamond edged tools, the
consistency and qualily of the mortar is lost; dry operated drillings cannot always be
successful and preserve lhe entire information. However, such core drillings may be
highly useful for a physical identification ofmaterials, Ihrough BIPS color endoseopy,
dilatometrie tests and cross-hole sonic tests.
The endoscopy allows a careful sUlVey of cavities and of improperly worked joints
between stone blocks. Superposed marble blocks show usually very thin horizontal
joints; however, a regular superposition works generally on some 50 to 60mm only,
so that the transmission of lhe vertical stresses is concentrated in a thin crust only
(this is one of the reasons of lhe severe marble fracturing observed in the Milan
Cathedral, Pavia Cathedral, and Pisa Tower).
Inside the drilled holes, dilatometric tests may perform a radialloading test within
lhe masonry; if separately performed on seetions of limited length in the hole, they
may provi de preeious informations on the masonry quality.
Cross-hole sonic tests may deseribe the distribution of the mechanical stiffness of
different materiais (facings and infill) across a thick wall.
More localized is lhe information provided by lhe double flat-jacks, which in fa ct
perform the loading test of a block of masonry within the wall, without its complete
isolation, The results of such loading and unloading tests provide an authentic
measure of the elastic or histeretic behaviour of the material, and the levei of
compression at the onset of the large inelastic deformations . The limitation of this
experimental proeedure is that it can only work in a zone of 200 to 300mm from the
facing, and cannOl operate in the inner part of a wall [7].
Few "slightly destructive" teslS of such types may complement more diffused "non -
destructive" tests, as sonic tomography, radar, and thermography, therefore resulting
in a knowledge not only of some mean mechanica! properties, bul also of lhe
morphology ofthe masonry works, as it was in fact built.
The physical survey can even provide an information on the state of stress, The
well-known technique of the flat-jacks can in fact measure the existing state of stress
across horizontal joints of the masonry , on the assumption of an elastic behaviour of
the material. This proeedure, as is working under pennanent load only, was found to
be aeeurate enough to eheck numerical analyses, and to detect inactive pans of the
stru Clures. Satisfactory results and useful warnings were obtained by applications on
lhe medieval Towers in Pavia, on the Pi sa Tower, and on the St Mark Campanile in
Venice [8].
16 STRUCTURAL ANALYSIS OF II ISTOR ICAL CONSTRUCTIONS

Fig.5. Lcaning ToweT of Pisa . BIPS color endoscopy in lhe waUs of lhe sccond loggia.
Ca\'ilies in lhe infill a nd bctwccn facing a nd infill are e\'ident (up 10 27mm)

COPPI", FORI 8·5 INTERNO

""'("')
, - 1 ....1 ,-(... ) ... 1..... )

"" "" '"

'"
'M
6102
598 ~
3558
""
" """ '"
'" ""
3140

""" ''''
""" '"". ~2J9

'" '" '""" ""

""
."""
''''''' .
""'" . 0 _- 0

!::I'
!l! 3000 ~ .----
"""
''''''', ,I.
o1 02 08 09
" " "

Fig.6. Tower of Pisa. Socond loggia. Cross-holc sonic mcasuremenls 00 lhe walls.
The \"clocity is 6000 mJs in lhe ma rble fa cillgs. 3000 m/s in lhe infill .
AI Ihc righl haod sidc. sign of impcrfcc l joi nt of lhe facing.
 G.MACC HI I Gencr:ll rnethodology 17

4. NUMERlCAL ANAL YSIS .

We hope that the discussion on lhe ana!ysis of monumental constructions will

be more serene in the future, and the bloody atmosphere of a religion war will be
disso!ved .
On one hand, lhe engineers shou!d not be so naive to believe that everything is
inc1uded in the process:
Forces => Stresses :::::) Safety verification.
We have already discussed the limitations ofthe analysis in the previous chapter.
On the other hand, Architects and Art Historians should not play the role of
Aristolelians refusing to look into the GaJileo's telescope!
The line of thrust ( lhe famous Méry line) is not wrong, as being lhe simple
expression of equilibrium; it may be of great help in understanding what in France is
called the "descent ofthe loads". However, it gives only one ofthe infinite number of
paths that lhe load can follow .
Other attractive suggeslions of applications to andent constructions of the
elegant theory ofplaslicity are still raised today [9}. The theory has been developed in
lhe 30ies for slee! frames, where lhe collapse was considered to be reached in a
"ductile quasi-stable process", by formation of collapse mechanisms. The nice picture
of the mechanism of collapse of a voussoir arch published by Cou lomb in 1773
encourages to research similar mechanisms for quite more complex structures. on the
questionable basis that " if lhe designer can find a way in which the structure behaves
satisfactorily, lhen lhe structure itself certainly can".
Unfortunately most ancient structures may have local collapses, risk brittle
compression failures, may develop instabilities before lhe overall collapse.
In highly redundant structures, built with brittle materiais as most masonries are, the
equilibrium shall be supplemented by lhe compatibility of deformation, and lhe result
is therefore depending on the conslitutive laws of the materiais.
Méry is not more friendly to monuments than Cauchy, and Ihis latter may perhaps
better help in saving some of them. The numerical problem is by far more
complicated, but lhe computers can do an incredibly efficient work, and we have no
practicallimitations in the computer capacity.
However, Finite Element Analysis is smoolh only with linear e!astic constitutive
laws. Is linear analysis appropriate to ancienl constructions? Unfortunately not , in
general.
Therefore, non-linear analysis should be used : mortars have a nonlinear behaviour, and
masonries behave nonlinear also for their microcracking and macrocracking.
Some efus have been and are apostles ofnonlinear analysis (myselfincluded, since the
50ies); the algorithms are there, cemmercial computer programs are available since
longtime. However, I shall confess lhat some years ofexperience led me lO suggest, at
leasl in a first phase, and at least for brittle masonry constructions, a linear analysis.
There are basically two reasons for such apparently strange suggestion.
a) Linear analysis with tensile strength of the material may provide a reasonable
description af lhe process leading to lhe crack pattem for which we are concemed,
the pathologic state ofthe unhealthy construction. lt can also provide a ferecast ofthe
passible evelution of the damages: this is in most cases what is needed for a decision
of intervention, far before the state in which incoming failure mechanisms may hinder
lhe intervention. The case of the 8runelleschi dome in Florence has been a convincing
case [1 OJ: the undamaged linear analysis first confirmed the different nature of the
cracks of lhe even and the odd panels of the dome; furthermore, lhe following
 18 STRUCTURAL ANALYSIS DF HISTOR1CAL CONSTRUCTIONS

Torre dei Majno

Torre dell'Universitã.

nord
n ord
288(0.22) :n"II{O_3, )
q - 18.oo q-1900
1

". imJ
r- L
o- a.oo
~B (O. ~81

ovesl 1---''----1 q _ f.lIO t~)6(O_U

q.$.10
4 (000) Qves l
q _ 1 lO f.-t; (I 1' 1
'107 &O

(~
0"-' 00

.ud

.ud

Campanile dei Carmine

Tora deU'Orol ogio

nord
• o t d

tq39!1 20)
q. ) 00 21 (O 97) ?B(MI)

I I +2311I 0012)
q.12_~ q.J2,~

I q~

22BIO.99)
1'1 *1 2 .$

t 18107el
q * 1 eo

Fig.7. Pavia. Venical stress measurerncnts by flat-jacks on 4 Medieval Towcrs (ISMES 1990)
 G.MACCH [ I General rncthodoJogy 19

analysis (inc1uding in the model the first set of cracks) discovered that lhe tran sformed
statical system involves a second set of cracks, which were in fact found.
b) Nonlinear analysis is relatively long and expensive; furthermore, the results shall
be checked with great care, and numerical instabilities may appear when cracking
intervenes. The procedure should be used by highly specialized operators on ly.
However, some nonlinearities may be simulated in a rough but efficient way by a
linear analysis in which the "damaged structure" is modeled (as in the above quoted
..:ase of Flm'ence, and one of the models of the tower in Pisa). In other cases, the
effect of some elements reaching an ultimate limit state in brittle compression may be
effectively simulated by their elimination in successive runs : this procedure has been
successfulJy used for the interpretation of the collapse of the Civic Tower in Pavia
[11].
More generally speaking, even an imperfect linear FEM analysis provides precious
informations that in most cases escape to intuition and to engineering judgment ;
engineers have lhe natural tendency to think in terms of"beam behaviour", and many
structu res do not follow such a scheme. FEM analyses call the auention on non
expected behaviours. We experienced four of such cases for lhe tower in Pisa [12]:
- the maximum stress sect ion was not found o n the plane of maximum tih ;
- the role ofthe colonnades, first disregarded, finally appeared to be important;
- warping of the transversal sections gives rise to stresses much higher Ihan in the
"beam" assumption ;
- shear deformation creates an invened curvature at the top ofthe Tower.
lhe choice of some delails of lhe analysis should nol be left to lhe specialist, and
lhe user shou ld have a role, because they may affect lhe COS1 and the results. The
choice between a 2-dimensional (plane-slress, or plane-strain or axi-symmetric)
model or a 3-dimensional one involves greatly different resources; on ly the user may
decide if the simplest way can be useful or not , depending on the problems being
currently invesligaled.
lhe choice of the type of element is strictly dependent on the geometry of the
construction; lhe automalic generation of lhe mesh may be in contrast with a good
modeling when the sections of the construction to be studied are known; 3D brick
elements are very satisfactory when the state of stress is studied on horizontal or
vertical sections.
Substructuring may be a forced choice, for the complexity of the structure, but can
also be useful for a better underslanding of the behaviour of individual parts of lhe
struclure.
When the slate of stress has to be studied in more delail in a special zone of lhe
structure, a choice is possible between two alternative approaches ; lhe K approach,
by which the finess of the mesh is increased, and the P approach, by which the
polynomial order ofinterpolating displacements is increased {6].

5. FROM DYNAMIC TESTS TO lDENTlFlCATION.

A recent and very promising fie ld of interaction between physical testing and
numerical modeJling is dynamic identification.
Recent applications to lhe Medieval l owers in Pavia, to the Pisa Tower and to its
colonnades, allow an optimistic opinion on lhe capacity ofthe experimental means, and
a moderately optimistic expeclation of a practical use in the diagnosis of ancient
constructions[ 13] [14].
lhe technique of identifYing flaws by the experimental study of lhe dynamic
behaviour under small excitations was previously used on concrete dams. More
 20 ST RUCTURI\ L ANA LY$IS DF HI STO RICAL CONSTRUCTI ONS

TOJ:REDI'ISA 10. . .: DI 'ISA TURIIIWI PISA mJ:.f.lHPlSA

POLW~MOlIM..I DKLMODQ." IIOIlMA_A.I...IIJI:l,.MOOO.', rouo. MOD4I,.IIIJI:l,.MOOO.', I'OII.MA_A.l...lDa.NOOO."
~""JV. ~, "'Ib ~, UI'" ~ l lJ,It ...

Fig.S. Lcaning Towcr of Pisa . First 4 modcs of vibratioll (experimental. ISMES 1995)

~
.1 f,,-
.
,,o ,
o
<;
/ . 8
v
, / . o
"
'/ ./ 8
./
\ 1/
8
r:.
"\
g g g §,
•
Fig.9. Towcr of Pisa. Numcrical dynamic Fig. IO. Towcr of Pisa. Modcs of"ibration of a
:malysis (Un. of Pa\'ia) column. by Ladir (Un. of Pavia)
 G. MACCHI I GeneralmC!lhodology 21

pramising seem to us the applications to some kinds of monumental buildings, because

of a possible higher effect of cavities, deteriarated masonry, or defective boundary
canditions.
The dynamic response af the structure to natural excitatian of the environrnent
(naise) ar to artificial excitation (vibradynes, electrohydraulic excitatars) is a function
af the stiffness, the mass, the damping and the boundary eonditions. As a first
approximation, if the geometry is known, lhe response may be dependent on the
distribution of lhe Young's modulus within the volume of the structure and on lhe
boundary condilions. If this is the case, defective joints and defective masonries
(assimilated to a material having a low modulus) may be localized.
It is obvious that the simple knowledge of the fundamental frequency af the
strueture is of little help. The informations shall be very numeraus, several vibration
modes and the mode shapes; io fact, only when the flaws are in the maximum
curvature part of a mode shape, that mode is sensibly influeneed.
The measurements shall be done therefore on several points along the axes of the
struetural elements (in order to define the mode shapes) , and in different directions;
torsional modes appeared to be very sensitive in the analysis of lowers.
The transdueers may be af different types: aecelerometers, seismometers, whieh are
fixed 00 the structure. An altemative, very attractive technique is that of the laser
interferameter (Ladir), which praduces a signal praponianal to lhe vibration velocity
ofthe poim at which il is aimed. It allows in this way to operate without sensors on the
structure, at a distance of some 50 or 100m, and deteet vibrations of few micronlsec.
Ifthe measures are dane without artificial excitation, only with lhe natural excitation of
a moderate wind, it is possible to operate without any physical contact with the
structure: a fascinating technique!
The procedure of identification, which follows the test s, Is also very delicate; il
requires not only a numerical model (usually a FEM model), bul also an appropriate
optimization algorithm [15]. Otherwise, onIy simple cases may be studied by tentative
calibrations ofthe model.
Practical applications were done in 199000 the Medieval Towers in Pavia, and
tests are still in progress to compare the behaviour before and after streogthening.
Two different applications ofthe procedure are in progress 00 lhe Tower ofPisa.
The first one, is the identification of the mades of vibration in order to calibrate the
numerical model for the seismic verification.
The second application has the aim of defining lhe baundary conditions of the columns
on the loggias. As already menlioned, their structural role may be important, but it
eould be reduced by a lack af efficient conneetion, due to the numerous replacings.
If a columo is simply standing on the loggia, its natural frequency should be 22 Hz ;
if the bottom is built-in and the top hinged, the frequency should raise to 97 Hz: the
sensitivity to restraint conditions is therefore very high .
The measurements taken bath of the natural frequencies and of the mode shapes
indieate a ralher good efficiency of most columos.

6. FROM MONITORJNG ON UNE TO IDENTlFICATlON.

A further experimental source af knowledge of the eonstruction may derive from

monitoriog on line.
The technique consists in the installation of a number of static and dynamic sensors
an the structure, in order to continuously record settlements, displaeements,
incJinalions, apening of cracks, vibrations, etc., aod follow in thi s way lhe behaviour of
lhe structure in time The sensors measuring the "effects" are supplemented by sensors
22 STR UCT URAL A NALYS IS OF H ISTORI CA L CONSTRUCTION S

monitoring the possible "causes" , i.e. temperatures, solar radiation, water table, wind,
seismic tremors [16].
The main aim is to control continuously the " physiologic" behaviour, and get a
warning when certain established tresholds are passed, giving rise in this way to the
suspicion of a "pathologic" behaviour, or when an excessive speed of some movements
may show the danger ofa sudden loss ofsome kind ofstructural integrity.
To our knowledge, large monitoring systems are currently working on the Florence
Cathedral, St. Mark in Venice, the Pavia Cathedral and Medieval Towers, the Pisa
Tower, the Mexico City Cathedral.
Our suggestion is to undertake identification processes taking advantage of the
thousands of readings that a continuous monitoring system produces; they can lead to
a better knowledge of the structural behaviour. starting from the knowledge of the
"etfects" measured under "causes" which are known. Through the simulation of the
measured processes, numerical models may be calibrated.
Two tentative applications were performed so far, one in Pavia, the other in Pisa.
The firsl application was on the octagonal drum ofthe Pavia Cathed ral [17].
An accurate study of the cyelic thermal movements of the 8 piers supporting the drum
led to the surprising observation tha! an increase in temperat ure causes an increase of
the length of the 4 short sides (at 45°) and a decrease of length of the 4 long sides of
the irregular octagon.
On the other hand. the monitoring of cracks on the drum shows that, under the same
action, 4 vertical cracks, present nearJy at the middle of the 10ng sides, sensibly e10se
li!!
The cracks evidently pass through lhe wall s, and work as expansion joints. A
numerical mode! of the drum ineluding vertical joints at midspan of the long sides, and
the application of lhe solar radiation effect on lhe 8 sides, gave excellent agreement
with ali the experimental readings.
A second example is dealing with the Pisa Tower [18].
The monitoring system, through tiltmeters, penduli (and their te!ecoordinometers) and
wire extensometers, may provide at every time, by a geornetrical construction, the
deformed shape of lhe entire Tower. The system is highly redundant , so that the
coordinates of each point are caJculated in different ways, and the mean is retained .
The therrnal processes have been studied, and the displacement history is now known
for a day in winter and a day in summer, for sunshine or fog o It has been foun d thal
the temperature ofthe air is well correlated with the axial elongation ofthe Tower, but
the tilt is mainly due to changes in solar radiation.
We are confident that such knowledge may allow to detec! possible unfavorable
changes of structural behaviour.

7. CONCLUSIONS.

New powerful means may help the engineer when he is called to contribute to the
diagnosis of ao ancient construction and help in remedial interventions.
The numerical analysis has now great potentialities. which should be used in a
responsible way, together with lhe traditional rneans, assuring the best understanding
of the structural behaviour.
The contribution of experimental studies to the analysis is essential, lhe
construction of reliable numerical rnodels needs experimentally measured properties,
and can be calibrated on dynamic and monitoring data (identification).
 G .MACCH! I Gcneral mCl hodo!ogy 23

However, even the simple linear-elastic model can provide precious informations on
the structural behaviour.
The historie analysis can never be neglegted, and can avoid wrong interpretations
and dangerous eonclusions based on misleading appearances.

8. REFERENCES .

I. TORROJA,E. - "Raton y Ser de los tipos estmcturales" ITCC, Madrid 1960.

2. MACCHI,G. -" Ruolo della scienza nella conservazione strutturale dei monu-
menti". Prolusione ali' Anno Accademico 1995-96. Universita' di Pavia.
3. LEMAIRE,R. - " L'ingénieur et la sauvegarde du patrimoine monumental".
Stmctural Preservation ofthe Architectural Heritage. IABSE Symp. Rome. 1993 .
4. TAUPIN,J.L. - "Cathédrale de Beauvais: de J'incertitude à la décision" Ibidem.
5. POLVANI,G . et alii - "Ricerche e Studi slIlIa Torre Pendellle di Pisa ".
Istituto Geografico Militare. Firenze 1971 .
6. PEANO,A - "Model, and Assessment" . S/nte/. Preserv. IABSE Symp. Rome.1993.
7. MACCHI,G. - "Monitoring Medieval Structures in Pavia"
S/mc/llral Engineering International, 1/1993 .
8. MACCHI,G. - "Diagnosis estructural y rehabilitación de edificios históricos"
ClIademos INTEMAC n.7, 1992 .
9. HEYMAN,J. - "The Stone Skeleton" . Cambridge Univ. Press. 1995.
10. CHIARUGI,A-FANELLI,M.-GIUSEPPETTI,G - "Diagno,i, and Strengthening
ofBrunelleschi Dome" . Slruct. Presen'.IABSE Symp. Rome 1993 .
11. SACCHI LANDRIANl,G. et alii "The Collapse ofthe Civic Tower ofPavia: a
survey ofthe materiais and structure". Masonry International. n. l . 1992.
12. MACCHI,G. et alii . "Structural Assessment ofthe Leaning Tower ofPisa".
SlntCf. Preserv. IABSE Symp. Rome 1993.
13 . PAVESE,A. "Studi sul comportamento dinamico di torri in muratura e lOTo
modellazione". Tesi di Dottorato.Universila' di Pavia. 1992.
14. MACCHI,G.-PA VESE,A. " Identificazione dinamica di strutture monumentali".
Sperimentazione su strulture. Seminario lUA V Venezia. 1993 .
15 . FANELLI,M.-PAVESE,A. " Diagnosis ofMasonry Towers by Dynamic Identi-
fication" . Sfruct.Preserv. IASSE Symp. Rome, 1993 .
16. MACCHI,G. - "Monitoraggio in linea dei duomo e delle tom medioevali di Pavia"
ANAIXE. n.3. Sett. 1993.
17. RICCI,C.-MACCHI ,G.- CANTU',E.-PAVESE,A "I nterpretazione e modellazio-
ne dei comportamento strutturale di edifici monumentali:il caso dei Duomo di
Pavia". Tesi .Universila' di Pavia, 1994.
18 . MITO,S.-MACCHI,G.-CANTU',E.-PAVESE,A.- "11 monitoraggio della Torre di
Pisa: strumentazione, analisi ed elaborazione dei segnali". Tesi. Universila ' di
Pavia. 1994..
 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS
P. Roca, J.L. GonzáJez, A.R. Marí and E. Onate (Eds.)
© CiMNE, Barcelona 1996

POSSIBILITIES OF THE EXPERIMENTAL TECHNIQUES

FOR THE STRUCTURAL ANALYSIS OF HISTORICAL
CONSTRUCTIONS

P.P. Rossi
Ismes S.p.A.
Viale G. Cesare 29
24124 Bergamo, ltaly

SUMMARY

The mOSI importan t tesling lechniques used for a non-destruclivc evaluation of

the static conditions of masunry structu res are prcsenled in this paper.
TIle prcliminary phase of lhe investigalion includes a gcolllctric survcy, crack pat-
tem investigation and an historical analysis. Ancntion is then given 10 detcrmining
lhe mechanical paralllcte rs ofthe masonry struclurcs by using non-dcstruclivc or sli-
g htly-deslructive tests. Most non-destructive tesls give only qualilalive rcsults wh ich
can thcn bc uscd as "qllality indexes" of lhe masonry. Thc inrorlllation obtained by
these tCSIS are presented wi th crilical rcmarks. A reliab le evaluation of lhe mcchani -
cal paralllcters 01' lhe mason ry st rm;t ures can bc acheived only by cmploying sligh-
11 y-destnlclive lests; for Ihis purpose, emphas is is given to flat-jack lesting tcchni -
qlles for determining lhe SI ale of stress. deformability and st rength cha ractcrislics 01'
lhe slrllctures. The impol1ant role of dynamic ana lysis and mon itoring for cvalualing
the slat ic bchavior of existing masonry slruclures is also d iscusseel.

I. I NTRODUCTION

Knowledge of the phys ical and mechanical characleristics 01' lhe supporting
slruclural elemenls 01' an hislOrically signiricant bui lding OI' monumenl is neces-
sary for lhe performance 01' a coherenl anel correcl analysis of the static conelilions
of Ihc st rllctu rc . T hesc sllpporting elclllen ls (pillars. walls. arches and facades) are
made lIsing materiais anel construclion techn iques that vary considerably from one
geographic arca 10 another. thus. Iheir structural characle rislics Illust bc cvaluated
case by case Ihrough a ca reflll diagnostic investigation programo
The rnc lhodology ty pically adoptcel for <lnalysing the slatic conditiol1s 01' an
cxisling masonry Slructure involves prclirninary rccogn isance surveys that includc:
a gcomctric and photographic survey; a detailed survey 01' exist ing cracks anel
other dalllage incluced to the SlrUClllre during il5 life and an analysis of lhe con-
slruction history of lhe bllilding anel ils various uses. To this inilial phase of lhe
invcsligation. which already highlighls lhe major slructura l problcms. a sc ries of
specialisecl invcstigat ions is added to determine lhe paralllcte rs sui table ro r the
definition 01' lhe slalic behaviour 01' the ed ifice.
 P. P. ROSSI I Ex pcrilllenl~1 Ic c hniqu~s

Diagnostic investigations are often conduClcd using only simple and rupid non-
destructive tcsting techniques. It must be emphasizcd, however, that the use of o nl y
non-destruct ive tests is nO[ sufficient lO sat isfacwril y resolve structural problems.
The results obtained via these tec hniques are gencra ll y qualitative in nature and give
only a preliminary evalualion of lhe mec hanica l characteristics of the masonry by
defining Iheir mechanical ·'quality indices·'. Nonethe le ss. cvcn Ihoug h these tests
cannot provide quantitativc cvaluation of the mec h~mi ca l pnrameters. lheir use is ver)'
intereslin g in Ihat they consc nt lhe acquisition of prcc io us inforlllatio n regarding the
homogeneity of the masol1ly anel the evenlual presence of local structural anomalies.
Thc only reliable mean s for lhe delennination o f lhe pararneters rclarcd 10 the
stali c behaviour of a st ruc ture is th<11 of utilising a s lig hll y- destructivc Illethod tha!
requ ire so me intcrvcntion , although small. lO lhe struclun.1l c lemc nL It is ncccssa ry
(hat Ihese intcrventi ons (eorin g and c uttin g) ha vc dimcnsions such Ihat they do not
affeet slruclural in leg rity and Ihal lhey are dcs igned such Ihal no supc rfi cia l scars
are evident al lhe end of the in vcstigatio n. Th e sli gh l a lterali o l1 oI' the huilding
mu st be permilled onl y as a tran s ito ry phase.
Diagnostic inve sti galions are often aeeompaniecl by lhe inslallatio n 0 1' in stru -
Illcntation for lhe mon itoring of deformat ional behaviour a f lhe structurc \Vitll time .
Over and above guaralltceing lhe sa fety o f the slrut·tu rc during lhe vario ll s inter-
vcntiolls. thi s slrucllIral l11o nilorin g ean also bc conside red a very reliablc investi -
galioll ll1ethod ilsc l r fo r the cvaluati on 01' lhe statie eond itio ns of the sl ruclure.

2. I'RELlMI NA RY I NVEST IGAT ION S

Before detel1l1ining lhe I11cchani cal parametcrs 01" a masonry stru cture, a broad
kno\V ledge of lhe slruc lure in qlle stion is Il cccssar)'. First. an aecuratc geolllcl ri ca l
study is needed. which c.to be ea rricd OUI by dircct tradi t io na l Illcasu rcs OI" conli -
nuoll sly by photog ranullclric tcchniques . An aeeu rate gco l11ct ri cal sludy m<.tkcs it
possi blc to idcntil"y eventual irreg ularilics (ve rti ca l clcviations. etc) by show ing
both the gcomct rical dcrects purposely int rodllecd by lhe des igner to create opl ical
cffecIs. anel lhe cffects o f bad workmanship. Durin g Ihi s sludy. the conslrllc tivc
and architectural points o f inlcre st in the rest orali oll 11111S1 be thoroll ghl y tc sted. as
wcll as alI lhe sl rllcturalmal crial s. Of parti c ular importan ce is lhe survcy 0 1' c rack
pattcllls in lhe building·s bearing struclures. A dctail cd sllrvey o f the cxten l 01' lhe
cracks anel thcir ope nings allows an early cvalll ati on of the slalie condition of a
struclure and recognil io n of poss ible causes of in :-. tab il ity. Fi g ure I sho\Vs crack
pa ltcrn s s llr"ey o n a wa ll o f the Ca th edral in Pav ia a nd o n a bcll -lowc r in
Bergamo. Normally. lhe gcolllctrical survey is e .\ tendcd 10 fOllndution slruc tures
by excavaling small exploralory shafls or by cont inll oll s core drilling.
Fina]]y. an extens ivc kn owlcdgc of the stru cture canl10 t ignore its past hi sto ry. In
fac\. it is imporlanl to rcconstruct the building phases 01' the Slructurc <lnclto co llect
info rlllatioll aboLlt the eonslrue tion techniques ll sed in every phase 01' lhe constru c-
tion. inclLlding data Oll the materiais uscd and o n subseque nl n.:info rcelllcnt work.
Thc infol"lnatiol1 on lhe construeti a n phases and use 01' lh e buileling ma)' bc 01' grcat
assista nee \Vhcn interpreting its sta lic be haviour anti whe n dcfi nin g addiliona l inve-
26 STR UCTURA L ANA LYSIS OF Il l STO RICAL CONSTRUCTlONS

Fig. I - Cmck paflem ;m'esliga!ioll

a) eXlemallmlloflllí.Calledral i" P(l\·jo 11) Hell (O iça "Campllnonc" in /J ergllmo.

slig alions 10 be in itiat cd. li may ex plain lhe connec lion bet\Vcc n diffcrc nt lypes Df
Ill asonry anel wh en cvalu ati ng lhe crac k pattcrn s. \ViII no! att ribulC lh e cause to
aC lual phenolllcna. bUllO ovcrlo ading wh ich lhe slruclure had undcrgonc in lhe pasL

3. ANALYSIS OF TH E STR UCTURAL AN D MEC HA NICAL CHARACTERI STlCS

Afte r lhe prel iminary invcsti galio n. a scco nd ph asc tcsting program is slartcd
10 dclclln inc lhe s truc wral an d mec hanical characte ris tics 0 1' lhe masonry s tructure.
TIlis ph asc is approached in mos l case s by us ing only lhe non-d cstruc tivc !cc hni -
que (w h ich rcquircs no direcI aCli on on lhe m<lsonry ). Thi s is d ll C \O lh e fa e l Ihal
lhe eq u ipme nl ll sccI fo r lhe non -des lrllcli vc Ics ls is nOI very expc nsi ve anel lhe
tes ting re lall vely si m ple. However. il IllllSI bc po in lccI o ul Ihal lhe use o f a 11 011 -
dcs tructi vc test. by itsc lf. is n01 su ffi cient fo r reso lv ing the prob lelll sm isfaclorily.
The res ults obla in ed by these les ling tec hniq ucs are generally of 11 qu al ilativc typc
and g ive on ly a pre li minary evaluati o n of the Illeehani cal c hanlC teri s ties th ro ug h
lhe de fini tion of 'qu ali ty indexes ' of lhe masonry. Eve n lhou g h l!lese les ls are una-
b le 10 suppl y lhe rnec han ical parame le rs necessary fo r a s ta tic anal ys is o fth e struc-
turc , Ihe ir use is ve ry impo rt ,m l as !hey prov ide illlportanl in fo rmal io n o n lhe
ho mogenei ly oI' lhe charaeteris li cs o f lhe rna sonry anel o n lhe prospcc live presence
of arC<lS o f an o ma lous be hav io ur.
The only re liablc \Vay lo de ter m ine lhe parmn e lers Ihar in nuencc the Illcehani -
ca l beha viou r o f lhe masonry is by s lightly -d estructi ve lcsling lec hni q ucs w h ieh
 P. P. ROSSI I EXJX:rimcnw.1 Icchniqucs 27

require driltin g borehole s and c lItling small sec lion s. These aclions mu st be of
such '1Il cntilY as nOI 10 clmnge lhe slruclure. and mu sl be slud ied in suc h a way
lhal lhe di sturbance to lhe mn sonry is temporary. AI lhe e nd o f the lests it is impor-
lélTll Ihat no visible signs of lhe \Vork remain on lhe sl rUClll re .

3.1 Non-des!ructi\'e testing !('chniques

(a) Sonic measurements

Amon g the Il on-de strucl ivc investigation Illelhods. the soni c mClhods. are withoul
doubl. lhe most widel ys pread tests uscd. TI1C tcsti ng tcc hn ique is based .in general. o n
the gcneration 01' sonic OI' ultrasoni c impu lses at a poi nt in the SUllcturc. A s ignal is
gener<ltcd by a pcrcuss ion sys telll OI' by an eleclrodynamic OI' pneunwlic transducer.
and collccted thro ugh a rece iver. wh ich em be placcd in vario us pos il ions. Elaboration
af lhe dat a consisls in measuring lhe lime lhe impulse lakcs 10 cover lhe section of
mate rial bc twcen lhe gcnerator ..md lhe receíver. and in analys ing the signal wave.
Ult ra sonic \Vaves are prcferabl y lI sed for lhe sludy o f co ntinu oll s struclllrcs.
adopti ng Illeasurcmcnl s already coded for non -destrucli vc tests 011 concrele struc tu -
res. Fo r masonry structures th at are typically unhomogeneous. it is nccessary 10 use
son ic impul scs, rcferencing lhe Icchn iqucs of lhe Si:lIllC Iypc used 011 so il and rock
Illass . SOll ic impul ses Illu st bc uscd with masonry slruc turcs beci:l usc lhe joints pre-
senl in lhe masonry \\'o utd cOl1stitutc inpcnelrablc barricrs 10 lhe ultrasoni c waves.
Thc sonic Icsts C<tn bc ca rri ed Oul lhrough lhe Illasonry by the applica ti on of an
il1lpclling force o n a surrare and rece iving a sig nal 011 lhe Oppos ile surfacc (aI an
anglc) or again. on lhe sa me surface . By placing lhe se nsors aI difTcrenl hei g hls 011
lhe sarne w.dl. it is possible (based on lhe princip ie of wave rcncclion on contac l
surfaces) 10 estimate lhe deplh of lh e foundations.
Ir Ihere are cxis ting borelloles o n the struc ture 10 be cXtllllincd. a cross- hole OI'
down -ho le sonic te st is poss ible. Fig ure 2 shows some poss ible test sche mes.

~
T

R T

T R

T
A,
c

b b

Fi.J:.2 - 5<mic (/I/(!If!/r(l.\ollic IIIc(O"urcmcnIS: differenl Icslillg schelll('J.

28 STRUCTURAL ANALYS1S OF HISTOR!CAL CONSTR UCTIONS

By using sonic tes l mClhods , lhe

following in formation can be obtained:
Mechan ical quality index' (esti m ale
af dcformabil ilY modulus);
Homogeneity o f lhe charaCle ristics af
lhe m uso nr ies which a building is
compased of;
- HOlllogenei ty oI' a single st ructural
c lemen t (st udy af an possible bag
const itu tion. lhe presence 01' tI loosc-
ncel cortical layer);
- The cffeel af grouling reinforccmenls
(Fig. 3);
- T he presence af cracks in continous
materiaIs .
1 j • 5 , - ~ • m 11 n U N MM.

Figure 4 shows sonic vc locity

mcasu re m ents taken 0 11 lhe masonry Fig. 3 - I~"ffl'( 'r of grouring 0 11 .wnic lIc/ociry:
struclUres oI' l he Abbey cf Vezzolano Comparison bctll'ccJ1rhc ralCS III ca.mred on ofri
masonry buildings hefo re anil aficr grouring.
(As!i).

(b) Sonic Tomography

T his testing technique gives a dela i-
Ice! map af lhe sonic vclocity distrihu-
tion 011 a plane scclion of lhe Slrueturc
under investigation.
Thc method eonsists of oblaining
the lime taken by -"onie impulses along
seve ral direelions whieh uniformally
cover lhe see tion unde r investigation.
The compulalion is made by using the
invcrs ion process which. starti ng from
the t ime of sonie signal propagalion.
reconS lructs lhe field velocity. TIle sec-
líon 01' lhe masonry is ma rkcd by a rcc-
tangular Il1csh grid whosc dimension is
rcla led to t he distance bctwcc n two
subscq ucnl lransmissions or receiving
poinls.
The calculat ion is carried OUI w ith
the hypothesis that (in a non-uniform
velocity rie ld ) son ic impulscs do
nol propagate in a str<light line but Fig. 4 - Mea.mrcmcnl o/Ihc sOl1ic \'c/ociry on
fo ll ow a c urvecl line caused by re frac- lhe III(1SOllry SlrUC!lfrc s o{ lhe VC::O!flI/O
lí on. Abbey (Asri ).
 P. P. ROSSI ! Experimental tcchniqucs 29

This tes lin g lechnique is widely used for testing concrete struc tures and Is used
by ISMES in thc investigation of concrete dam s to identify arcas where lhe cond i-
rion of lhe conc rete requires reinforcement. Figure 5 shows lhe results of a sonic
tomography invesligation applied on a conc rele gravil )' dam - lhe propogalion
lines are shown as well as lhe map of lhe velociries.

~H II(I"U "I'.~\I \IIIK·II)

iI"')

..
~\ (..I

..,. .. '\IU>

" ~
..... ~I ",.,

~ZfIII"""')
" , ,,",!o"
W)O • ..lOOo"1

<-'''0'

Fig. 5 - J~\""icll l rc.w/lJ oblllinedfrolll sOllic rOlllography 011 (I ("OI/crere graviry (/(1111.

A rcmarkable improvcmenl C.lO be introduced in lhe qualily and rcliabililY of

lhe rcsulls wi lh tomog raphy for attcnuation . a Icchnique whic h exa mines nOI
onl y the sonic veloci t)'. bul al so othe r c haracteris lics oI' signal transmiss ion (ampl i-
h,dL' and frc q u<.>ncy). Thi s lechni(l'lc is nmv hein g testcd. wilh so me promi s in g
rcsul ls bcing collcc tecl from lhe slud)' of conc rCle slru clurcs.
T hc so ni c lomog raphy has reccnll)' bcen used for lhe ana lys is of lhe slrucl ural
charactcrislics of SI. Mark's Basilica in Vcnicc provid in g a clear mapping of lhe
diffcrcn! veloci!)' I.ones. As an example. Fig. 6 shows the result s oblai ncd 0 11 pil-
lars 1 and 2. 1I c <ln bc clearl)' obse rved lhal pi ll ar I. which was co nso lidatccl by
grouting aboLi I lO )'cars ago. shows high velocity values, cspec ially in lhe ilHemal
zonc whi ch was invo lvcd in lhe grouting ope ration. On lhe contrar)'. lhe velocily
values obtained for pillar 2 show Ihm lhe mechanical charac lcris li cs of Ihis pillar
are ver)' poo r (in fal·t no conso lidalioll work was pcrforll1ed o n Ihis pi ll ar in lhe
p:lst) .
10 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

~1oQty m/$ _ 6000800

0&00 + 1IXXl
01000 UiOO

''''0 pillar of Sr. Mark' J /Ja,(jfiC(l in Vellice. Th e pilla/"

F ig. 6 - /(e.Hllfs of tOll1ografJlJic S/Ir1'ey ill
2 IltIs \'ery poor meclulllical clUlTacteristics I\'hile 1';1/(11' /. II'hú;h 11'{l S colIsolidared by grolltillg
30 yellr ago. sl/O\II!; high \'elociry vlIlue.s .

(c) Radar Ill vesl;g(IIioll

Thc rad ar tcsting tcchnique uses high-frequcncy electromugnetic waves (100

MHz - IG Hz) em itted through an an lcn na wilh very short impulscs (0 .5 - 5 ps)
and allows to determine locillion or scparation surrace bclwcen materiais w ith dif-
fe rent d ielectric consta nt s. The investigation is based on rcllection 01' lhe clcetro-
magnctic waves frem lhe con laC( surraces bClwccn materia is of diffcrcnt diclcct ric
constants so int erna i defccts in the Illasonry (da mp arcas. cavi ti es. presence of
metal struCl urc s. piping. nues) C<ln bc located.
Rece nt ly the radar tCl:hnique has
been used on lhe pillars Df SI. Mark 's
Bas ilica in Vcnicc (Fig. 7) in a rde r LO
determine the locatio n Df eventua l ano-
malies in the masonry ( voids. etc.).

(fi) Th ermographic llflalysis

The thermographi c <lnalysis i s

based 0 11 lhe Ihcnnal conductivity of
lhe material and may be pass ive or ac li -
ve. In lhe fo rmcr case ir analyses lhe
radiation Df lhe work during lhe cyc1e
of thermal slress due to natu ral pheno-
Illen'a (insu lation anel s ubseq uent coo-
ling). If lhe survey is aCl ive. ferccel hea -
ting of the surfaees analysed are
app lieel. The thermal radimion is collce-
Fig. 7 - Use of radar lechflique lo sIlIdy
teel by apparatu s sen s itive lO infrarcd
slruclUral clmraclcriSfics of fil e pillars of SI.
radia tio n. anct is then Irasformcd inlo /!1ark's Basilica ill Vcnice.
 P. P. ROSSI! Experimental tcchniques ]I

electric s ig nals. wh ich in tum are converted into images in differenl shades o f colour.
The rmovis ion is used 10 idenlify a reas whe re lhe stnlctures are covered by pla-
sle r Ihat hid e consl ru clion anomalies (b loc ked ope nin gs etc.). It is parti cularly in le-
resting for studies on fre scoed waUs. where it is not possible to take sampl cs o r use
lesting Icchnique s Ihal come in contacl with the fre scoed surfaces.
Olher application s are the s urvey of the prese nce of cavities. s uch as nues.
ducls. and \Va ter anel hea ting sysle m s. by disting ui s hin g lhe ;u eas of th e rma l
dispersion and Illoisturc.
It is nece ssary to point ou l Ihal lh e penelral ion dep lh of Ihis lechn ique is only a
few centimc tres so il is unable to lacate anomalies which are hidden in lhe ce ntral
pa rt of lhe masonry.

(e) RebollluJ lesls

The rcbOlmcl tcs ls are aimcd ai a qualitative eval uati on of the cO lllpressive
streng lh 01' mortal' and o f lhe s upe rfic ial slrength of slo nes O I" bric k mate ria is. Thcy
can al so prov idc informa lion o n loca l damagc oI' lhe ma tcri al. Fo r info rm atio n o n
the hardne ss of lhe sa mpling material. pcnd ullllll sclerometers (Iow energy) a r
Schm idl hallllll crs, may be uscel. Thc rcsuh oblai ncel w ith Ihi s lest can be cons ide-
rcel a 'qualil y inelex' which mu st the n be corrc latcd with re sull s o blained thl"o ugh
Illcchanica llcs ls on sa mple s.

(J) Maglletomctric ollolysis

Magnetomelry locares the presence of mClalli c c lclllcnls within lhe ma sonry

struclures . T hc tes l includcs a magnetic pro be which is pl aced o n the masonry s ur-
fa ce: thcn iron bearing materiai s a re id enlifi ed by ano malies appearing o n the
in slrumen t's magnctic field.
Thc tcchniquc was se I up 10 study Ihc reinforcc lllc nt in conc re le s truCLUres and
tran sfelTcd to mason ry bu ildi ngs. w here it is uscd la locate chains. tcns io n ba rs o r
melaI conn cc ling pins belwcc ll block$ oI' stonc.

3.2 Slighlly.dcstructivc tests

The non -de struc tive te sting tec hniqu es de sc ribeel in lhe prev ioll s parag raph s
arc nOI s uffic ient to detcrmin e th e paramelers nccessary to cva luale lhe static con-
dition o I' a sl!"uC lUre anel 10 dcs ig n reinfo rce lllcllt \Vork. Thc ev aluation of these
parame lc rs is poss ible o nl y by lI si ng spcc ial lllcchanica l tes ls w hich requirc. unfor-
11Inatcly. sma ll opcralion s on the masonry. 11 is Ileccssary fo r this \Vork to be ,IS tit-
tle as poss iblc :Ind visiblc only during tcsl in g. Moreovc r lhe tesls Illust be s imple
anel !"inished quic kl y to reduce lhe cos ts o I' the in vest igat ion and to cntirely res ta re
lhe structu rc in the s hortest po.ssib le time.
Rese arc h imo Ihese lesli ng mcthods have rcce nll y secn g rcat eleve lo pmcnts in
Ital y due lO lhe slrong be lei f thal the y represem the o nl y way ar o u!" present di spo-
sa l to eva llla tc tlle strllctllral anel mec hanical propcrties o f a rnasonry Sl ruc lUre.
 STRUCTURA L ANALYS IS OF 1I1STORICAL CONSTR UCT 10NS

((I) Coring leclllliques

To understand lhe SlJ11Clur;11 propcl1ies of me di lTercnt tyTX=s of masonry of which a

building is composcd. il is illl lX)11ant lo core small diameler borcholes taking samplcs in
lhe mosl reprcscntativc poims oI' lhe slJ1lclurc. lllis olXration bccoll1cS indispensablc in lhe
ver)' frequcnl case when lhe masonry consisl~ of two surface layCr.'i in regular bond wim
intemal il1l!gu l.u· packing. Coring mllSI bc done wim a rOlar)' sa.w lIsi ng a dialllolllJ clIuing
edgc. By lIsing a ve ly lighl. Imncly perforalion looL cori ngs can bc oblaineel from inacces-
siblc arcas. Th is coring operalion alio\\' samples to be CX1.ractcd flúlll lhe material on which
laboraloly tesls can Cc madc: th is is p;u1icul arly imporlant to anal ysc thc chcmic,t1-physical
and Illcchanical characteristics of bricks. stones and 1ll011nrs. The borcholes can men be
lISCO for adclilional investigalions (v ideo camem sUIVey. dilalOlllelJ'ic tests. soni c Illensurc-
Illenls) which help 10 delinc the slruclural and Illechanical prolJCI1ies 01' lhe masonry.
ClllC cori ng Icchnique is (\Iso ve ry important 10 cvaluale me char'.'IClcriSlics 01' lhe foun-
dation 1ll<L'>Onries. Fig. 8 shows lhe chilling cquipment which was lIscd 10 in vcsligatc the
founc!;uions of SI. Mark 's Basilica in Vcnice. ln lhe S<lIllC figure me stl1lclural schellle of
lhe foundation masomy is shown.

lliJ Bnck masonry .-'

·-''''r ·
:: ~
~

,
1- slo~e masoory

"" i , l
' ''' I

'" I
'''' ~ .- =-
Wood plate . ~1 jL ~
~' :\ '; ;,;
Wood piles P]l\:,;',1~, ;:, }V:,:
---
Fig. li . AI/aly.\·i.f of IlIe Slrll('fllral clwra('lerislics 01 lhe fútll/dmiml flwsollry (~r
SI. M ar/;. s
U(/.I'ilim ill Velllcc: a) dl'llltllg eqllipllll'm b' sfmclllral SC!/{'II/C (~fl"(' ,ffl/llldaliolllll/l.wnry.

(b) /JorellOle vitleo Sllrl'e)'s

A smal l co lou r vidco C;'II11cra Illay be inserted inl o lhe borehol e allow in g a
dClailed stud y of both lhe I'ront anel sides af lhe ho le. Thc resu lts 01' this slucl y Illay
bc rccordeel anel archivccl for I'urthcr analysis after lhe bo reho lcs ha ve bccn scalcd.
The informat ion obtained by thi s survey includc:
the struc tu ral cha ractcri stics oI' lhe masonry.
lhe mcasuremenls 01' lhc inlclllal ca vities or lhe ma sanry.
lhe analys is 01' lhe prop<lgation 01' illlcl1lal cracks anel Illcasurcllll..:nls 01' lhe ir opcnings.
 P. P. ROSSI I Expcril11cntal tcdllliqw:s

Figure 9 shows an examplc of lhe video camera survcy carried out on lhe foun-
dation masonry o f St. Mark's Bell Tower in Vc ni cc.

Fig. I} - Vidco ("(II/JC/"{/ .\lm ·(\' CMI"ict! OI/I O/r lhe lowrdal;ol/ lIJa10wy oI SI. M{/J"k'~ Bell Tmrel" in Ve/JiCl':
(/) {aleral r;clI oll/)e .{/II/(I(·c ollhc Im/"c}jo{c b) n'fOnslrunioll oI IlJe 5{)(!âflll'l1 by fOmpll/el" progmmm.
 STRUCTURAL ANALYS IS DF I-IISTORICAL CONSTRUCTIONS

(c) FIaI jack lesls

An in lc rcsli ng lesling lechniqu e based on lhe use o f fla! jacks. w as develo ped
ai lhe ISMES labor:Jtory abou l 15 years ago for lhe ana lys is o I' lhe mechan ical cha-
racteristi cs of ex islin g masonry SlruClures. Thc firsl applicati ons o f thi s lechn ique
o n some h isto ri cal mo nurne nls c1early showed il s gre<l l potenti al. lt appe,lred lo bc
lhe o n ly way 10 achievc rel iab lc inf01l11at ion o n the mai n mcchan ica l charac leri -
sl ics of a masonry strllc lll rc (ic . dc fo rm abi lity. s lrenglh . s tale o f sl ress) .
This very simplc lechn ique. which is carried o ul by int rod uc in g a Ih in f1at -jack
in to lhe mortar layer. is only sl igh lly des trllcli vc. A fter lhe te si is complelcd. lhe
n at-jack can e:'ls ily bc rCllloved and lhe morta r laycr rcsto rcd to its orig ina l condi-
ti o n. The h igh re liabilil y of lhe test is rc lated to Ihc llndi sturbed cond ilio ns of lhe
sampl e on wh ich the mechani ca l characte ris tics are dctc rmin cd . and to Ihc largc
arca of meas urc menl whose behav io r is rcprese nled.
The les ling Icc hn iquc has bcc n o rgan izcd to give rc liable ans\Vcrs fo r the fol-
low ing:
- Illeasurc me nt oI' lhe statc 01' strcss
- detcllll in ali on o f deformab il ity and slrcngth charac te ri st ics
- detemlin ali on of lhe shear s lrcng lh a long lhe monar laye rs.

M ea\"urcm enf aOlte sWf e ar Hr ess

Thc de lerminat ion of lhe Slale of Slress is based on lhe slress re lease causcd by a
plan e cU ll ing normal to lhe surface of lh e wa ll. Fig. 10 shows lhe differenl phases o f
lhe ICS I. Two re rerence poi Tll s are install ed on lh e waU surface and lhe inilial di stan -
cc (di) bc lwccn lhe two points is mcasured. A cu l perpend icular to the \Va li surface
is Ihc n made and lhe slress rclcase is delermin ed by a pa rti al closi ng of lhe cu u ing.
the d istance (d) a fte r lhe cul bc ing d < d i. A th in flal-jac k is placcd ins idc lhe cul.
and lhe pressurc is graduall y incrcased lo cancel th e previolls ly measurccl conver-
gcncy.

a c

J
.,
J ,

Fi!:. 10 - Scheme 01 the flm ·j(lck rcsfing plua-cs on brick lI/t1sollry.

 P. P. ROSS I ! Expclimcnlal lcclmiqucs 35

In lhis condition, lhe prcssure (p) inside lhe jack is cquat to lhe pre-exisling slatc of
stress in a direclion nornlat 10 lhe pt:Ule of lhe cut. Thc value obrained m USl be corrected
by a cocfficient which dcpends on lhe ratio bctween lhe Oat-jack surface and on the rigi-
dily of lhe weJded boundal)'.

The value of lhe sla le of slress (a) in lhe lesling poilll is g iven by:

a = p" Km" Ka "

where:
P ,.. oi l pressure
KIll - jack conslanl which mu sl be detcrmined in the laborarory
Ka - AJ/ Ac (ralio belwccn lhe surrare oI' lhe jad: and lhe surrace 01' lhe Ctu)

In a briek m<l sonry, lhe plane eul ean be casily made in lhe morta l' layer
belwecn IwO laycrs 01' brick s by ovcrlapp ing ho les rnade wilh a hand 1001. In (h is
lypc of masonry a rcctan g ltlar Oal -jack is useel (40 x 20 em.). Sllla ll cr jacks are
also uscd for rncasuring lhe stale 01' slress 011 structural elements sucll as arches.
pilla rs, and vaults. A ditTerent cuuing tcchniquc is lIsed in lhe case of a stone
masonry wilh ve ry thin mo nar laycrs. The cul is madc by a sleel d isco wilh a dia-
1110l1d cUHing edgc. and lhe Oa l-jack has lhe salllc shapc (circu lar scgmcnl with
Icngth 32 e l11 .. dCplh 12 e m. and thickncss ~ 111111.) The diffe renl phascs 01' Ihi s
tcsting tcehnique are show ll in Figure li. It Illu sl be poinled oul Ihallhe very lillli-
ted Ill ickncss 01' this kind of jack rcquircel lhe solu lion of de licale problems in ils
c1csign anel construc lion phases .

F (~. 11 - SchclIIC oflhcj701-j(lck 1('.~liI!R phascs OI! stOI/C lIIasol!ry.

Fig. 12 shows lhe ditlercnl shares o f Ilal-jacks useel to measure lhe slate of
slress o n bri ck and slone Illasonries.
,6 STRUCTU RA L ANA LY$ I$ OF I I I STO I~ I CAL CONSTRUCT 10NS

"
"
"
Spess 4mm ,

w
{ ~ ','

(1+
, T,,
, -- ~
,
,- -
,, ,,
', f,,
, '
, -- ~
J
'" "" '"
f igo 12 . Shapes ( 111 si:cs 01 lhe flar-jackJ u,W:d fo r lhe II/COJl/remCnlS oflhe silo! o!sfre.u:
a) b) c) brick m(lsonries, d) e)J) SfO n e m llSO l/rieJ.

Wi th lhe aid 0 1' lhe naH acks ty pcs

e) an d f) it is poss ible to anal ize vc ry
irregular and non-homogencous ston e
m<lsonry. as shown in F ig. 13.

An cxa mplc o f the rcsults obtaincd

by tl a t-jac k tcs l is show n in Figure 14,
wh ic h ill us tra te s lhe rcsul ts o f lhe mca -
s urcme nl S tak cn on lhe pil1ars and late -
r.1I wa ll s o f lhe Cathcdra l in Orvicto.

Fig. IJ - TcsllI'ith ajlflt-jack 01 lhe I)pe e) on a

\'ery lI'eek (lnd irregular s/Onl! mason,y.

"
•
! ~. J~'
11 "
11 " • •

Fig. /4· Srress I'lIlues (MI'(J' mcn.mred byflat-jack te.S/ing ()fI fhe nlllSOllry s/r uclllres O/lhe Cmhetlm! in Orrie/o.
 P. P. ROSSI I E."IXllI11Cl1wl (çdllliques 37

Determinat/o" o(C/eforlllabilitv anc/ strell{{th char({cteristics

In a homogcncoLls isotropic material. the tcst prcviously dcscribed can also be

used to determine it5 eleformability characteristics. In the case of masonry. which is
a l1ighly anisotropic material. it ls advisable to inlroduce some changes in the testing
Icchniqllc. For Ihi.s purpose. a seconel clltting is madc. parallcl lO lhe first one. anel a
second jack j" inq"rted. ai a di"lanee of aboLI I 50 em from lhe OlheI' The 1wo jack"
dclimit. therefore. a masonry samplc oI' appreciablc size 10 which they apply a
llniaxial cOlllprcssion stress (Fig. 15(1). Several Illcasurement bases for rernovable
mcchanical strain-gallgc. installcel on lhe s<.lmple free face. nmke il possible to
oblain a full picturc 01' axial ancllransversal deforlllalion ofthe samp le . In Ihi s \\Iay a
uniaxial cOlllpression test is ca!Tiecl out on an unclislurbed sample 01' large arca.

,[

'[

Fig. 15 . Tesl \\·írll 111·0 paraI/cf j1al-jacks 10 determine deforlllabiliry c/wracrerisl;cs:

(I)schefl/e of tlle tcst b) e.H/mplc: of (/ deformabililY Icsr pCfformed OI/ lhe cXlernalll"af! ofthe
Cllllrch 01 "Sal! Frofl("('sco·· in Arc::o.

Several loading cycles are performed ~t gradua ll)' incredsing stress levei s in
order to determine lhe deronnabililY modulus 01' the masonry in ils load in g and
unloading phascs. Fig. ISb shows a vicw oI' lhe test applicd to a stone Illasonry.
Thc loading tcsl above can also bc usedlO evaluate lhe comprcss ivc sln: ng lh 01' 11l~
rnasonry. The load is increased until lhe rirsl cracks in lhe brich: app~ar. Ihl'n Ih~
strenglh limit of lhe masonry can bc approxilllated by eXlrapo lal ing til\:' Slres'·
slrain cu rve . Thc clTecl 01' the hllcral confinelllent 0 1' the sampk may bc I:llú.'n in lo
aceount by calibratioll lests done in lhe laboratory.
It must bc nOlcd IhaL whcn nearing failurc concl ili ons SOTlle crad.:s appcar in
lhe brick. but tllC damage suffe recl by lhe masonry is quilc ncgligiblc anel can hc
r\:paircd casily. Fig. 16 sho\Vs <In cxample or lhe resulls 0 1' a I\:'sl ll1;tdc OI] Ih\.'
:u; STR UCT URAL A NALY SJS QF HJ STO HJ CA L CON STR UCT IONS

" a (MPa)

30
~ ~

"-- ,. /
\ ',I " í
,
--t, I :: -.-
,' +++-+------.,
:
:
Eo,

f
: · +T · i
-- ----\ - - f- :l ___=.Ev___ j:
- ,, 06,
/ I
' 00 - '00 - 200 - 300

Fig . /6 - 7)'pi("lIl Slress-Sl rain curve o /)rained by .flm-jack l esl up !O failu/"c. ol/d vicH' of lhe
'esli,,!; {Join, (/pa "u: 1('SI.

masonry w<llls of the C loislers 01' SI. Eustorgio in Milan. Thc typical slrcss -strain
curve up to failllre is presented anel lhe testing point afieI' th!..! tcs I is shown. 1I can
be observed Ihat the damage undcrgone by lhe mi.lsonry in quite neg ligible.

Recently lhe Icsl ing techniqucs havc bcen dcvclopcel in orelcr 10 allow Iheir exe-
cul ion also in po in ts where Ihc conel ilions of acccss are vcry dilTicult (i.c. the high
bui ldings anel lhe lowcrs). A speciallraining 10 lhe technicians. combined with clil1l-
bing Icch ni ques. 10 elay allows 10 carry oul ali lhe invcsligations, including coring
anel f1al-jack test. on lhe hig h st rllctures without the aid of scaffoldings (Fig . 17).

Fig. / 7 - Flal -jack /es/ 01/ high s/rue/ure.\" H'i/llOu//lte aid oi scaf(o ldillgs:
a ) Bcll TOI\"cro/SI. S/cphcn in Vcniee b ) TOH'er "Fra cca/"O" in "(II·ia.
 P. P. ROSSI I Expcnmenlal lechniqucs :>9

Shear "SI alollg (1 mortar la ver

The nat·jack testing lechnique (combi-

ned wilh a hydraulic jack) can also be used
lo delennine the slrcnglh charaClerislics of
lhe moT1ar be lwt:en lhe brick layers.
[
A brick is ex traclcd from lhe cenler af
lhe masonry sample delimileel by IwO Oal-
jacks anel a hydraulic jack of lhe same L
sizc is inscI1cd in ils place for lhe applicll-
tion of shcar force (Fi g. 18). Thc lesling
lechnique dClCrmines the pcak and res i-
dual shear slrcnglh of lhe mortar Iaycrs.
. By doing. seve ral tes!s wilh differe11l
values of lhe slress in a direcli on perpen-
dicular to Lhe join!. lhe friction angle anel
lhe masonry cohcsion can bc dClennincel.
Th is tes! is imponanl fo r lhe sludy o f
masonry re sislance 10 sc ismic move menl.
AI lhe enel of lhe leSI lhe jacks are ex- Fig. /8 - Scheme of a shear test alollg tlle
tr.lcted anel lhe masomy is rcslored to its ori- mortar /(/)'e ,.s: tlle 1I0rll1a/ .n/"e.u ;.{ applied by
ginaI stale. TI1Cbrick rcmoved for lhe impla- fll"Q flllf·jacks 01/{1 tlle sllcar stre.u i.5 llPl'lied
/Jya hydraulicjoc/.:.
cement of lhe hydraulic jack is replaced, and
lhe spacc bclwecn lhe bricks are fillcd with mortal" of lhe same colar as lhe original.
Al i of lhe Oal- jack tes tin g phases have bcen calibralcd aI lhe bboralOry by a wide
se ries of Ic sls on very large masonry samplcs. In Fi g. 19 lhe loadi ng frame and the
mcasuring equipmel11 used for lhe cal ibration lesls are show ll. lt is ncccssary to point
oul lhe high rcliabilit y of thcsc te sts and the greal importan<:e 01' lhe re!; ult s. Thcir u ~c
is highl y recollll1lendcd cvc n ir small CUIS on lhe masonry surrace are rcquircd.

fi!:. 19 - Vieu· of '''l' loadillg and meml/ring eqllipmcl/l fo r lhe calibrar;oll ofJlllf-jacJ.: leSIS.
40 STRUCTURAL ANALYS1S DF HISTQR1 C,\L CONSTRUCT10NS

Boreho/e di/atometer

Using lhe tcsts with parallel fia!

jacks onc C'Ln only determine lhe defor-
nwbi lily chanlClcri stics oI' lhe supe rfi ·
c ial la yel' 01' masonry. In arder to
acqu irc inrormation on lhe deformabi-
li1y charactcristics 01' lhe internai
mi.lsonry i1 bccomes Ilcccssary to ca rl'y
au l dilatomctric tests using boreholcs
madc by coring. A spccia l probc aboli!
25 em long applics uniform hyd rostatic
prcssurc on lhe borchole surface. and
lhe I11casurClllcnt af lhe consequcnt
dcfor!ll.ltion determines lhe modulus oI' "

dcrormability (Fig. 20). As lhe pOrlio n

of Illasonry lIscá for this les! is very Fig. 20 . Seh!'''''' (~r bor('//o/t· diloroll/crric fesr
limited. lhe valucs obtaincd by lhe dila- r!tar d('/cnll/lll'.\ fhe dljO/'II/(/hili/y dU/racreri-
.I'/io ol//u: .\'/lIfáce (/1/{/ i"IIC1' IlI)'en,
tOlllc tric tes! <.Ire less rcprcsc nlal ivc Ihan
thosc obtained by the flat-jack tests.
llli s lesting lec hniquc. howevcr. is undoubtedly usel"ul as it determines lhe ratio of
deformabitily of the intcmat masollt)' to lhe OlHeI' tayer. Ditalolllctric lesting is abo
important for Icsling lhe dcformabitity charactcrislics of lhe I"oundation slnlClures.
Figure 2 ] sllows the lesling scheme used on lhe fOllndation ring o f Ihc Towcr of Pisa.
Several dilatollleu'ic tcsls were catTicd Oul in radial lx>rcholcs logclher wilh sonic logs
anel c ross-holc llIeaSUrcmcnlS. TIlc valucs 01" defol'lll<tbility lllodlllus are shown in lhe
hislOgralll in Figure 21. It was rcvcalcd lha! lhe roundations. which in lhe pasl had
undcrgonc strcngthcning by grouting. stil] hacl good lllcc hanical charactcri slics throu-
gl1011t ils Ihickn~ss. lllcrefore. lhe grouti ng proccdllrc is belicvcd 10 havc Ixcn ell'cclive.

Õ 5
] 4 r-- -
~" 2

Dcform:Lbi l llY mndulu,

Fig, 2/ - UJl' of fhe borcl/Ole di/a/ol/ll'TCI'for lIflllly:.illg lhe dt'fomwbilily c/wmcrcrisricJ 01 lhe
IO/fl/darioll oi rhe Tmrel' of Jli.\'lI . HiJrogram olrlle delof'llwbiliry 1II0(/IIIi,
 p, P. ROSSI I Expcrimcnlal Icdmiqucs -lI

4, NUMERICAL MODELLlNG

A significant contribution to the knowlcdge of the sta lic cond ilions of a struc-
ture i5 provided by the mathcmat ical model wh ich ulilise ali lhe dala oblai ned
throug h in -situ and 1<Iboratory in ves ligations and lhe inFormation com in g from lhe
moni toring sys lem. As concerns lhe mechanical pa ramclcrs of lhe masonry. lhe
in put dala for Ihc ma thematica l model arc obruincd by nat-jacks dcformability
tcsts and the va lida lion oF the model is carried o ul through lhe comparison
bc t\Vccn lhe ca lcu latcd va llles 01' lhe stale of stress anel Ihose measured in si tu by
flat -jack tcsts. Diffe rent loacling co nditi ons are
applied lo lhe mode!: (dead load. thermal effecl. dilTcrcntial settlcmcnts of lhe
foundalio n struc lllfes anel dynamic loadings) and lhe dcfollni.llion behaviour of lhe
structurc is analyzed wilh grea t attention in order lO oblained a bettcr underslan-
ding and a mo re meaningflll inle rpre lati oll of lhe daI a provided by lhe l1lonitoring
systelll.
In Fig. 22 a view o f the malhcmalical model of SI. Mark\ Basilica in Vcn ice. is
presented. The high degrce of complex]ly 01' lhe slruclUre suggcslcd lhe OppOrl U-
nil y 01' co nsiderin g lhe whole Basilica as lhe slIm oI' sllilab lc subst ructurcs. It sec-
I11cd advisable 10 devide lhe slruClU rc inlo 9 subsl rllctures in order to analyze sepa-
ratcly lhe pillars. lhe lateral walls and lhe domcs: lhe boundary condilions of e'lch
singlc subs lruclUre n::prodlKc con·ectly lhe slilTness 01' lhe adjacenl su hslrUClures .
111C cOl11plexi ly of Ihis l110dcl is clcarly shoWI1 by the very high numbcr of degrec s
of frcedom (about 250.000).

,,
í: -,
r

F(~. n . N/lmericalmodcllil/g 01 Sr. Mark·.1 IJasi!im il/ \ '('IÚ("('.

42 STRUCTURAL ANA LYSIS DF I-IISTORICAL CON$TRUCT IONS

5. DYNAMIC ANALYSIS

Thc in-silll tcstin g using d ynamic

rnclhods ca n bc co nsi dc rcd a reliablc
in strull1cn l oI' a non - dcslru ctive lypC.
lo verify lh e s lructura l bchavior and
intcgri ty of il building.
Dynamic invcstigations can bc ca r-
rieel OUI according lO lhe following 1'1'0-
ccdu rcs:
a ) Through lhe anal ys is af data
g.tth erc d as a rcspo nsc to d ynam ic
loacl ... conti nu ollsly i mposcd on lhe
structurc (urban anel rai l road Irarrie .
bcll s ringin g. CIC.) o r from irregular
aClions. A s eislll o lllctric net work i5
PiS. 23 . D)'namic ono!y.ú s in {/ ('o nter (~f Ih"
ins lalled in dilTc rc nt pari S Df lhe struc-
"1'('1'1//(' di Cal'(fca!((I" iI, RI/m e. lo lIS.\CH lhe
rure anel lhe s ig nals are ana lyscd in c:{ft' C( oi Ih(' /'{)mllralfic 011 (he .1'/1'11('(11,.('.
tcrms af amplitud e a nel fre qucncy
conlcnL
Through s pcc lr;'11 <ln al ys is tcchniqucs il i5 II1 CH possiblc 10 cva luarc lh e dy namic
Illodel paramclcrs. As an cX<lmplc, Fig ure 2:\ s how s lhe se islllomc lers inslalled in a
co mer or lhe "Tc rme di Carac;:llla" in ROIllC, 10 assess lhe responsc bchav io ur o r
lhe masonry slructurc 10 lhe urban lrarrie.

b) Subjce ling lhe cons lrucl io ll to low intc ll s il y forced vibration Ic sls (so as 10
produce vibra ti onal leveis Ihat do not afrect lhe slrue tural inl eg rily) anel rccording
the sys lcm responsc in Ic nns of displaccmen ls. ve loci t ies and accclcrat ion s.
Thc forced v ibrat ions are in duccd by vibrod ynes and lhe respo nse mens ured by
scis1l1ollle trie scnsors. Thi s sccond k ind o r ana lys is allows lhe iclclllifkalion 01' lhe
dynamic behaviolll' 01' lhe structu rc th rough lhe l!ValUalion 01' ils moelal p;lnlmete rs
(natural frequc ne ics. modal shapc s. damping rali os). The kn ow ledge 01' t!lese para-
Illc Lers allows the computat ion of lhe s lructura l rcs po ll sc \O an y Iypc 01' dynamic
Joad wirh known cha racle ris tics. and in pa ni cula r for evaluilling lhe seism ic vu lne -
rab iJily o f lhe masonry cons lruclion. Forccd vib ration ICSIS repcatcd ovcr a kng th
0 1' lime, all ow l h e cval uali o n 01' po ss ihlc m od a I param c l er v ariali o n s.
Thesc varialions ma y bc assoc ialcd w il h 1ll0dific;:H ions in lhe s lnu: tural inlcgrily.
en ablin g lhe aClual s tru c tural d cg r ad :lli o n or the Jlla so nry Clcl1len lS lo bl'
quanlified .
II must bc s lressecl lha! lhe inforlll<llion co ll ected not on ly qu antifil's the s lruelura!
degrada lion 01' lhe hui ldin gs or parIs 0 1' th!.!lll. hu l Ihey ar!.! al so illlportanl in tht'
plannin g stagc 01' any s lruclU ral ada plaljOJlS, whe n c hoos in g lhe Iypc ~llld sií't, oI'
lhe appropr iat e st rengthcnin g \Vorks.
Thc rorccd vihralion le chniquc lIsi ng a vihrod yne \Vas rcccnlly tlst'd hy ISrvl ES
lo analyse the <.; Iru cltlral conditioll ar seve ral towcrs in lhe 10\\'11 0 1' Pa vi:1 Fig. l -L
 P. P. ROSSI f Expcriment:tltcchniques 43

6. MONITORING

Ins lal lillg measu ring ins lrumenls 10

monilOr lhe slruc tural behavi o ur of <l
bllilding can be cons id e red a reliablc
Illelhod fo r lhe cvaluali on 01' lhe s talic
condition 01' lhe slrll ctll!"C . Thi s inve sli -
ga ti ve lec hniqll c is gaining popu laril y
beca use, bcsidcs s llppl y in g info nnali on
on lhe stil lic con eliti o ns 01' lhe builelin g.
it is cons iele rccl lhe o nly wi.ly 10 g uaran-
ICC lhe s afc ly of lhe slru c ture bcforc.
c1urin g anel afIeI" lhe co n sol icl,lIion
wor\.:.

Th e pr incipa l fe<.lIU reS w h ic h are

Illonito rcd are lhe follow in g:
- opc n ings Df lhe l1lain crac \.: s in
Inasonry st ruclllres:
ab so lutc anel relalive h orizon lal
Fig. 2-1 . FOl'("ed \'ihrmioll f e.l·f wr,.ied m a hy (I
1ll0VCIllcnt s 01' vertical st ructurcs:
\·jf)/"{)(JYII(' 011 file f 011"(''' "F/"(I("("(Jro" i"I'm·;a.
rotation oI' ve rti ca l anel horizontal
struc turc s:
intclllal anel ex lema l tc mpcralllrc:
bcha v iour 01' so il an d roc \.: fOUllebti o ll s:

The evOlulion 01' the dcl"o rmation. in I"un ction 01' time anel Icmpcrature. allows
onc 10 sc paralC lhe the nnal erfect frolll lhe c1cforma ti on duc 10 Olhe I' ca uscs (foun-
elali on SC \llelllcnt. slruc lural 1110dificalions. new forces induced on the structure.
aging o f lhc Jllaterial s). Measllrin g lhe opellillgs of cracks is by far lhe m ost wide -
sprcad Illclhod. The easicst mcasu ring lechn iquc is lIsing a rClllovablc 1l1cchanical
CX lcnsO lllctel" which is positiollcd on 1\\'0 sma lllll cl al platcs pos il ionceI 011 opposile
.'l ides of the c rad: .
To illlprove Ihc I"eliabi li ty oI' lhe Illcasun.:menls. fi:\ccl cXlcnSOlllc lcrs can be
used tha l are p rovidcd with c lcct rlc transducI.' rs cOllnccted 10 <In .tUI0I11;111C dala
collcction sys lelll . In Figu re 25 an eXlenso rneter. recenll y installccI by ISMES in a
pillar of SI. Marl(s Basil ic a in Venice is showll.
To Illcasurc absol ute horizon lallllovcments 01' ve rtical clelllcllls. a fixed pcnclu -
lum is uscel \Vith a Illc .. tsu rin g systcm baseei o n a Ic lccoordinomett:r.
Less c:\pc ns ivc anel easie r to in s la ll. are lhe inst ru lllcnts for Illc ..tsuring relalive
movelllc nl s o I' vertical s lrUClurcs. In Ihis casc. a long base cx lenSOIllC" lcr is used
whic h is equippcd wit h an inv:!r w ire h:cpl in ten sion hy li wcighl. Thc movelllcn ts
Df lhe wcig ht are easily mca s ured by rncan s oI' clcclric Iransducers. Figure 26
sho\Vs lhe inslallalion 01' Ihesc long base CX lcn SO lllelers 011 the intc rnal wall 0 1' lhe
Pi sa lowcr.
44 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

Fig . 25 . E.nellJOllleler for lIIeasurillg fite Fig. 26 - Itulallarion of fong-base exlensome-

o{Jening (~r ti ('mck ill (} pillars of SI. Mark' s fers 01/ lhe inlemal wal! of lhe Pisa Imver.
in VCllicl'.

Fig. 27 shows lhe diagrams of lhe relalive d isp lacemen ls belween lhe pill ars Df
SI. Mark's Basilica in Ven icc. measured by long-base extensometer in a period of
about ~ ycars . In the figure lhe diagra m Df lhe Icmperature of lhe air inside lhe
Basilica (T4) is also shown.

[0 i ~jllllllllll

"8
'<

Fig. 27 . Diagrollls of lhe rdalive displacement.\· befllleen lhe pil/ars of Sf. Mm"k's Basilica
(Vellicc) 1IIf.'(/.\·/lred by 10llg-ha.l"e exlensomcfen.
 P. P ROSSI! Experimental techniques 45

Rotation changes of vertical ar hori-

zontal elements can be measured by
installing either fixed or removable cli-
nometers.
In constructions of great historical
or monumental interest. the instruments
are usually connected to automati c col-
lecting and recording systems which
can quickly indiGlte possible ilTCgulari-
ties in lhe slructura[ behavior.
In Fig. 28 lhe automatic data acqui-
sition and rcco rding system in stalled in
lhe Cathedral in Milan in arder 10 con-
traI the effect 01' the co nsolidation
\vorks caJTied out on lhe pillars.

Fig. 28 - Auromaric dara acqu;siriol1 and

r eco rding .çy.Hem insralled ro l1Ioniror r!te
cOlIsolida/íon w{jrk 0/1 lhe pilla r s oI r!t e
Carhedral i/1 Miftlll.
46 STRUCTURAL ANALY$I$ DF HISTORICAL CONSTRUCTIONS

REFERENCES

BONALD I r .. JURINA L.. ROSSI p.r.. ~/l1d(/gilti sperimcl1tafi e II/lnterid/e sui dissesri (lei
Pa/a:.:o (leI/C/ Regiom' di Mi/ol1o n , CO ll1uni caz ionc prcscntata ai Congresso Naz iona1c d i
GCOlccnica. Fircnzc. Ouobrc 1980. lsrncs BulI. n. 156.

SACClIl LANDR IANI G. RICC ION I R.. "CompOrlflflll'lltu s/(Ilico l' sismico de/li: s/mllu r e mura·
ril''', Clup. Milanu (19~2).

ROSSI. P.P. "AI/a/ysis o/lIIl'chanical chamc{eristiD' 01 bridi l/I(wJ/lry teste,! by IIICW/S 01 1101/"
dC.\'ll'Uctil'c il/ silU lesls". óth In!. 13rid Masonry Cunferente - ROlllc 19x2.

ROSSI P.P. "Determ inariOIl experimental lic.r carr/c/(',;sriqllc II/l'('(/ lIiqu e des IIllIçOl/l1erie"
Rcslauralion dcs Ouvragcs CI dcs Structurcs - Pont ct Chaussccs· Paris 19X3.

IlINDA. L, ROSSI. r.p .. SACC HI. G. ~ Ditlf:llosfie (/Iwlysis of 1I/(I.\'O/l"Y bui!tJillSs" IABSE
Symposium - Vcnicc 19S3.

Bi\RLA. G .. ROSSI, p,r. ~St,-C.fS lI/etlSllrClI/cm ill tUII/lelli"ing" Inl. Symp. Fieltl Mcasurcmcm in
GculIlcçhanirs - Zuriçh 1983.

ROSSI. r.r. "FIm jack res/ for fite tmalysü of mecllllll;callx'JWl'iollr of brick II/asonry struCfures H

71h 1nl. Brick Masonry Confcrcnrc - Melbourne 1985.

ROSSI. p.r. "Recel/l tlel'elopmclIf 0/ lhe flat-jack fc.\"t OI! II/(lsollry Sfru ClUres" U.S.A. - ITALY
work~h()p on cvalu.:u ion anti rClroi!1 of masonry structurc. Augu),t 1987.

BLASI. C.. ROSSI. p.r. "Indagini sullc strullun: 1Il1l1"(11'ic dei rell/llio cosid('f/(J di !<ml/O/O IId Foro
RO/l/ano" Inl. Confcr. Non Dcslructive Icsls. Perugi:l. April. 19M8.

GIANI. G.P .. DEL GRECO, O., ASTOR I. B. - ROSSI. p.r. "illla/ysis mui COII/ro/ offlll' JllIfic hd,(/-
I'iour vf file COlIso/a:.iol/e C/JUreh in Todi, l laly" Inl. Symposiulll "The üIgilU'aillg G('%gy of Aliciem
lVorh. MOl/llllu:llfa! (IIu/llisrorica ! Sires". Athcns. Scptembcr 1988.

A5TOR I. B. - GIANI. G.P, - ROSSI. p.r. "CVnfro!lIIl'asllr('IIIl'1If tIIlr! sIIIbilil)' {/}/{}Iysi~ of file
Calhel/ml 0IOn·iefo. lta/y" Int. Symposium HTh e Engill('ering G('()/ogy of Anóelllent Works. MVllllfIIC/!/S
1II11/lfiSlOricol Sires" Alhcns Scptcrnbcr 1988.

RICCIONI. R.. ROSS I P.P. "ResllIllro c(lilj:.io c fIIOl1/lIlIel1lole, Diagnosi (' con,m/il/lIl1/elllo" Editricc
il Cigno. 031ilco G:llilci. ROfll:l 19R9.

ROSSI. P.P, "Nol/ deSl1'l4ctil'C el"(/Iulltioll of medullliCl/l clwmctcri.wics olll/osollry buildi/lgJ".

COII!('/"l'I1Cl' VII "No/l f)('.\"Iruoil'e El'lIlllarion ofCil'il Sll'tle/ur('s muI Matcrwls", Unlv.:rslty 01 Color. ulo.
l3oult!cr, USA Octobcr [989.

BLASI c.. CONFORTO M.L. . ROSSI P.P. "A l1afysis oi file ml/ica{ behlll'iour oI 1111' COSlanlil1o's
(ll"ch in RO/1/('" Inl. Tcçh n. Confcrcncc Structura[ "CVlls(' I"\'a/ion of Srol1e Masr!!!ry" A!hcns 19X9.

MODENA. C.. ROSS I. P.P. "In .\"itll illl'esligatiolls, slru{"/um/ (lI1a/ysis (1/1(/ stl"clIgtltellillg of a slrllle
mosonry bd/ /Ou'a," Il n1. Congrcs~ 011 Rcsloration of lhe ArchitcCtlH:l1 Hcriwgc aotl Building. Canarias.
July 1992.

ROSSI r.p .. VENTRUCC I F.. VIO E.. "S/m etI/m/ A s~·e.UlIIl'l1I of S/. Mark's Basilica, Venicc".
IA135E Symposiulll "S/rIlCfllml/'reSl'IWllion oflhe Archirecfuml I/ pri/(lge" Roma 1993.
 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTlONS
P. Roca, J.L González, A.R. Marí and E. Dilate (Eds.)
e ClMNE, Barcelona 1996

EXPERIMENTAL TECHNIQUES FOR AUSCULTATION OF

HISTORICAL CONSTRUCTIONS

R. Astudillo Pastor
Millisterio de Obras Públicas, Trml sporte y Medio Ambiellte
CEDEX: Laboratorio Central de Estrllcttlras y Mat eriales
Alfofl sO XII, 11°3
28014 Madrid, Spaitl

SUMMARY

The development of numerical models for the analysis of monumental or historical

constructions from lhe point of view of their structural behavior musl include, for a
minimal reliability ofthe theoretical model, a complete study ofthe real properties ofthe
materiais and a realistic evaluation ofthe shapes, sizes and connection conditi ons ofthe
structural e lements of lhe construct ian. A variety af techniques for charactcrizing the
physieal, ehemieal and mcehanical properties of materiais, as we ll as for the experimenta l
analysis of lhe statieal and dynamical behavior of lhe strueture are shortl y described.

I.-INTROD UCT ION

The analysis of the structura l bchavior af a historie canstruct ion rcquires, in a genera l
situation, lhe elaboration of a theorctical structural model based on c1assical or numerical
methods, lhat it allow lo explain lhe resistant state and the conditions of stabi lity of the
complete building and each one oftheir parts.

The development ofthese thcoretical mode!s always requi res a knowledge ofthe geomCLry
ofthe structure and of the characteristies ofthe materiais, as for their nalure, conservation
and mechanieal properties.

The refinement or validation ofthe thearetical models usually requirc the implementation
of slalic or dynamic lests that allow la compare the foreseen output with the real one and
to correct in lhe proper way the madel ofbehavior.

Finally lhe observation ofthe evolution ofthe slructure requires Lhe direcl measurement
along the time of some parameters thal charactcri ze its behavior and lhat, after
comparation with the results of a theoreticaly contrasled model, allow to know the safety
and funclio nality cond itions ofthe structure.

So, lhe experimental techniques for auscu ltación of a slructure are fundamental in order
Lo establi sh ilS behav ior along its life.

These experimental techniques cou ld resume in:

Techniques to define the geometry ofthe construction

Teehniques 10 define the nature and charaeteristic ofthe materiais
Tcchniques in order to evaluate the resistant behavior
48 ,STRucrURAL ANA LYSI S OF HI STORICAL CONSTRUCT IONS

2.- TECHNIQUES FOR CHARACTERIZATION OF STRUCTURAL

GEOMETRY.INSPECTlON

2.1.- Geornctry

Thc analysis of a monumental conslruction musl bcgin with ao exact knowlcdge af lhe
geometry: fonns, dimcnsinn and masses. These data wil l permit lhe calculation of the dead
load as well as lhe resistant sections af lhe structural elements.

Thc exact informalion ofthe geomctry will Icad to good drawings wherc ali lhe damages
cao be referred, allowing to lechnicians af differcnt and specific fields la collahorate in
lhe auscultation project wilhout visit ing lhe construction .

T hc cxacl dirncnsions and geometry af lhe struclure are someti mcs nccessary for
comparison with prcvious ar latlcr geometries, in arder to anali zc lhe cvo lution.

Thc geometric definition ofa struclure could be undertaken with di fTe renl tcchniques and
lhe uti lizatio n of Olle or olher will rely on the Iypology, di mension of lhe slructure and
depth of lhe auscultación process:

- Basic methods: meter. mclric ribbon, plumb linc, etc.

- Topographical methods.
- FOlogram etry

2.2.- Inspcction

The evaluation and control of the behavior of a hi slorical construction always requi re a
previous knowledge and a posterior pursuil o f ali kind of damages lhal could ex isl. The
knowledge of lhe state of damage will be fundamenta l in order to lhe development of a
numeri cal medel ofthe structureand for lhe interpretalion ofthe measured data in lhe case
of the implementation of a continuous monitoring systern.

The inspection \Vi II discover lhe several damages: cracks, degradation of materiais,
efl orcscences, humidities, corrosion, spall ing, excessive dcfomlations, co ll apscs, etc.

lhe necessary inslrUlllentalion for lhe inspect ion \V iII be, on one hand, lha! to allow lhe
access to the zoncs lo observe, and on lhe other, difTe rent oplical instrumenls
(photography, TV , video) for recording lhe damage.

3.- TECHNIQUES FOR C HARACTE RIZATION OF MATERIALS

3.1.- C h~tracteriz:ltion of the nature of materiais

In order to know lhe slructura l behavior of a construction and in order to develop

numerical models with enough reliability for its 3na lysis, it musI bc known:
 R. ASTUDILLO PASTOR I Experimental lechniques 49

Which are lhe different materiais involved ?

Which are the physical, chemical and mechanical properties of such

materiais?

In arder la sa lve lhe firsl questian: which are the materiais, they could be used methods
like:

Cuts and openings in order lo reach zones inside lhe structure .

Geophysical melhods in arder to detennine layers af different materiais.
Perforations in o rder to take oul cores af materiais and to introduce
endoscopy devices.
Magnetic methods in arder to determine reinforcements.
Thermography
Gammagraphy ar X rays.

3.2.- Characterizatian af materiais prapcrtics

In a general approach for every material in the structure the fallowing propcrties mus! be
known,:

- Chemical compos ition and physical structure ofthe material.

- Conscrvation of lhe material.
- Resistance to compression , tension or shear.
- Dcformability of lhe material under lhe aclion af the laads.

For the knowlcdge of lhe previaus properties il is necessary to carry Ou! a serie af
experimental tests and analysis af diverse nature, perfamlcd "i n situ" ar in laboratory. on
partions af materiais extracted from lhe building.

For lhe developrnent of mode ls of behavior are espec iall y important lhe mechanic
properties: tensian and compression strength and capacity of defonnation under the act ion
ofthe loads.

3.3.- Non dcstructivc Icsling

Sometimes, for some materiais like steel , concrete or brick, it is possible to ge! data of
Iheir mechanical , physical and chemical praperties, and of lhe levei of deteriorati on. by
lechniques that onl y slightly damage the structure. These lechniques (NDT) are speciall y
convcnient in lhe auscultalian ofvalious historical conslructions where the integrity of the
construction is fundamental.

The lechniques af ultrasanic waves, rebound index , penetrating liquids, potentials 01'
corrosión, etc, are utilized frcquent ly for il.
50 STRUCTURAL ANALYSIS OF H ISTOR ICAL CONSTRUCTIONS

4.- C HARACTERI ZATl ON OF ST RUCTURA L BEHAV IOR: EX PE RI MENTAL

ANA LYS IS

The Experimental Analysis is a group oft echni ques lha! permit to know, in an instant Df
lhe lire ora construclion, wh ich is ilS structural behavior. according to lhe actions 00 it,
or how is lhe evolution Df lhe behavior of lhe structure and orthe materiais with time.

The {wo kinds ofproccsses of the Experimental Analysis are Ihen:

Short tenn, static Of dynamic Icsts in order to charactcrize lhe behavior ofthe
structurc from lhe poiol of view cf strenglh Of dynamical bchav ior.

Imp lcmentation cf systcrns for lhe measurement and analysis of loog term
cvol ution of lhe structurc.

In general, lhe methodology for lhe characterization of lhe structural behavior will consist
in lhe application of some deternlined actions on the structure, accordi ngly to lhal lhe
structure will respond in some \Vay. The measurement of magnitudes related to lhe appl ied
actions, to the response of lhe slructure and to lhe conditions of lhe environmcnt will
permi l, with the contrast with a thcoretical model , to analyze lhe resistant behavior of the
structure .

The tests lO curry out in order to analyzc lhe structural bchavior could be grouped in:

Static tests: load lests, tcnsion re1ease tests (dilatometer, flatjack)

Dynamical tests: characterization and dynamic auscultación
Long term mca suremenl

4.1.- Sta tic lesls

Two usual statical tests arc lhe load lests and the direct determination of lensions by
means of lhe tcchnique of nat jack or boreho le dilatomctcr.

a) Load tcsts

A load Icst on a conslruction (bridge dcck, noor, covering slructure. elc) pcrmits, in
genera l, lhe experimenta l confirmation 01' Iha1 the structurc behavcs in bend in g in
accordance with a predctcrmined model o that is to say. that lhe real structure has the
stifTness estimated in accordance with the chamcteristics af the materiais and lhe geometry
involved.

In the case of constructions wilh historical ar artistical character, circumstances appear

Ihat cond itian these tests compared with other more general structures. One of these
circumstanccs is that many of these buildings are erected with materia is likc stone or
ceramic, and therefore prcvails the work in compression (arches. pill ars and walls) more
than in bcnding (beams and sIabs). That introduces a difficulty for load tests becausc such
experiment could imply very important loads if a certain levei 01' deformati on is nceded
 R. ASTUD IL LO PASTOR I Experimental tcchniqucs 51

in order to get precise measures. The load tests could result useful ncvertheless in lhe case
af anal ysis of f100r struclures, especialJ y when lhe change a f use of a bui ld ing is
contemplatcd, prctcnd ing an utilizatian lhat impl icates zones of concentration ofpeople
(allditories. classrooms) or zones for storage ofmaterials (files and libraries).

Thc reali zation of a load lest implicates lhe placemcnt of distribulcd load usuall y
matcriali zed hy means ofsacks. pallets. blocks oromore frequenll y, w(ller reci pi enl s.

The measurement ofthe response ofthe structure will rely 011 the tesC the structure. and
af the dcpth of the assess ment to be dane. USllall y will be measured the respon se in
di sp[acernenls and rotations and, for more detailed analysis, the response in deformations.

b) Flat jack tesls and borchole dilatomcter rncthod

The n at jack tests permit:

Thc measurcment of the compressive strcss in a po int of an structure

The dcformation properties of a part of the structurc (\\'all, column)

It is indicated for vertical stresses and for brick and organized stone bl ocs struclures and
it will bc specifically expla ined in other sesians ofthis Seminar. as well as the boreho le
dilatometer method.

4.2.- Dynamic tcsts

The comp lete charactcrization af the behavior of a structure has to include not only th e
stati cal aspect but the dynamic behavior also. The more large span. higb ri se, slender
struclure and strong wind or seismic si tuations. the more importance of such analysis.

Nevertheless. the fact that monumental conslruct ions are not, in general, long span or
slendcr structures. the damage caused in many of lhcm by seismic or trame vibrations, or
the use fulness ofthc experimental analysis based in the dynamical auscultation technique,
reveal lhat lhe dy namical bchavior is an aspect that havc to bc frccuently considered III
a auscultation processo

Two kinds of dinamical analysis can be considered:

Measurement ofthe parameters that charactcrizc the dynamiea l response of

a structural element: Frequenc ics, modal sbapes, damping.

Dynamical auscultation technique

In the dynamic teSI ofa structure three different aspects must be considered ;
52 STRUCTURAL ANA LY SIS OF HISTORICAL CONSTRUCTIONS

- A dynamic excitation af the structure, that is, an actian that enters the strueture in a
state af vibratian and that cauld be caused by lhe awn methad af testing, ar by the
envi ranment of the strueture: traftie, wind, micro-earthquakes, etc.

- A measurement and registration 01' some physical magnitudes as functions of the time,
and that represent the response of lhe structure to the dynamic excitatian.

- An analysis or the signals registered in arder to get lhe parameters that characterize the
dy nam ic behaviar of the structure.

4.2. 1.- Excitation systems

Althaugh more systems exist, three of lhe more utilized are:

a) frequency swccp

It consisl in lhe application on lhe structure af a force lhat varies sinusoidally and whose
fTequcncy sweeps to cover the range af frequenc ies to analyze. Analysis af resonances in
lhe respanse permits lhe determination 01' modal frequencics .

The exciters far these tesls are constitutcd by one a r several masses that oscillate with
frequency and amplitude controll able. The devices are electrohydraulic a r electrodynam ic.

b) Excitation for free vibration

- Imposition or ao initial displacement

It consists in the application of a load lhat causes an initial defonnation (weight fixed to
the structure, tcnsioned cable) and that it is instantaneously relaxed.

- I mposition or an initial spced

Also known as excitation by impact. lt is the usual system in the tests ofmodal analysis.
In these are obtained lhe func tions 01' lransference between an impulsive excitation
applied on the structure and the response in a series ofpoints.

c) Transitory excitation

In the case 01' structures under the action of dynamic loads, thesc loads eou ld be lhe
excitat ian. It is an adequate method when it is desirable to measure leveis ofvibrat ion or
vibrat ion transmiss ion.

4.2.2.- Dynamical auscultation

It is attraetivc the idea af finding, in general, an auscultation method for structures that
permito by means of the measurement ar any of thei r global characteristics, lhe
dete rminalion of their Slate of degradation and the evolution aceording to time af the
res istant eondition.
 R. A$TUD ILLO PASTOR I Experimental teehniques 53

The technique of the dynamic auscultation consists, in essence, in determining the

existence, localization and evolution of damage in a structural element by means of the
comparison ofthe dynamic response ofsuch element in diverse moments ofits life.
The dynamic characteristics that use to be analyzed are the self-frequencies ofthe very
first modes ofvibration, especially the fundamental , and the modal shapes.

In monumental buildings dynamical tests use to have different applications:

The experimental characterization of the dynamic behavior of the structure

helps in the definition ofthe numerical modeJ for both, static and dynamic
analysis, because it permits a better knowledge of masses, stiffncss, bearing
condition, interactions, effec! of damage in structural behavior, ..

The differcnces in dynamic response of repetitive or identical structural

clements leads to the identification of damages or structural singularities in
some cases.

The dynamical auscultation helps to discover diffcrcnccs in structura[

behavior with time.

4.3.- Long term mcasurcmcnt

One of the more frequent applications of the experimental analysis Icchniques to

monumental constructions is the measuremcnt along the time af the evolution of lhe
structural characteristics of the structure, that is, lhe periodic measurement of magnitudes
like:

The movemcnts in the foundation

The movements in the superstructure: longitudinal and angular (geomctry)
The variation of width and length of cracks
The evolution of the state of strain in particular zones
The evolution ofthe characteristics ofthe environment: temperature, humidity,
sun radiation, etc

The analysis ofthe magnitudes measured witl allow to define lhe behavior of a structure
and the adjustment of mathematical modets lhat permit to analyze the several components
ofthe structure response (seasonal, random, etc) and to establish ranges ofstable behavior
for certain variables as function of values rclated in general to the environrnent
(tempcrature, water levei, conditions ofwind, foundation movement ,etc).

TIle long teml measurement implicates. besides an instrumentation thal will bc the more
autornated possible, a ccntralized systcm of measure that permit lhe swept of readings of
ali the instruments and to organize the data in files for their management, if this is not
performed in real time.
 STRucrURA L ANALYS IS OF HISTORICAL CONSTR UCTIONS

5.- INSTRUMENTATION

The magnitudes to measure during the realization of a test will reJy on the depth and the
objectives ofthe structural analysis to carry ouL They, in lhe most general case, wilI be:

t) Magnitudcs related to the structural response

• absolute vertical and horizontal displacements

· rel ative displaccments bctwcen eJements (i.e. stone blocks)
· stra ins in criticai zones
- presence, length and variation of thickness of cracks
· rolations
- movements in joints
• settlements and movement s in foundations
- accelerat ions, veJocities

2) Magnitudes relatcd to the environrnent:

- temperature
- sun radiation
· humidity
· ground waler levei
- wind (di rection and speed)

The instrumentation is diverse and a complete description leaves oulside of the field of
th is Seminar, only some of the more frequently used instrumentati on will be shortl y
described :

5.1.- Measurcrncnt of dcformations

The measurement of dcfonnations in a slruclure is lhe slandard melhodology in order to

dcduce, according with the relationships stress/strain oflhe material. lhe slress state in lhe
measurin g zone .

The extensometers utilized for historical structures auscultatian should gather some
speeific requirement s, especially rcferring to the base ofmeasuring, since this should be
large enough for minimize the local Ilo·homogenity af materiais or fabrics. There are
several types:

a) meehanical
b) vibrating wire
c) resistance strain·gauges
d) Adaplation of displacement transduccrs (i nductives. capacitives ... )

In the experimental analysis applied to monumental conslruct ions are frequently used the
mechanical, vibrating wire, and inductive (LYDT) types. The small base a f Illeasuring of
lhe resistance strain·gauges and thei r bad long term stabi lity does not make Ihem
especially useful in this application.
 R. ASTUDILLO PASTOR I Experiment31 tcchniques 55

5.2.- Measuremcnt of movements

Th e measure af movemenls in a paim af a structure always requircs lhe existence af a

reference poiOI, this will bc a fixed poiol in lhe sUIToundings, in lhe case of lhe
measurcmeot of absoJute movements, or il will be another poiot of thc slructure, fixcd ar
not, in lhe case of relative movements. The fixed point of reference for absolute
displacemenl use to be the graund in zones where movements caused by lhe tran smi ss ion
of reactions of lhe canslruction ar for olher causes are nol foresee n .

5.2.1 .- General movements

Survcying

Survey iog melhods are based on lhe delennination ofthe exac l position of criti cai points
of lhe structure and of lhe sUITounding ground along lhe time. The differences between
Iwo readings pennil lhe measurement, in general , ofthe movemenlS in the three axes of
lhe space, although some of techniques are lim ited lo lhe measurement of vertical
(leveling). ar ho ri zontal (coli malión) movements.

Because lhe small values of lhe movemenls in this t)'pe of structures lhe methodolog)'
shou ld look afier lhe precision in ali their slages.

Thcsc methads dcmand the placement of firm and slable bases for lhe instal lation of lhe
apparatuses and the situation of lhe reading points materia li zed on the faci ngs of lhe
slructure and on lhe gro und .

Mechanical and el~ctrical dc\'iccs

There are man )' instrumenlS for measuring displacemenls. ali of them have la pari s: one
fi xed ar connec led with lhe abso lute reference of lhe measure and olher con necled to the
slruclure in lhe paim where lhe displacemenl has lo be measurcd.

Apart from lhe mechanica l devices. lhe electrica l ones (Iransd ucers) are based in:

- vibroting wire: similar to the devices used far measuring deformat ions. in thi s case
wilh mechanical ampli fi carion systems in a rder to increase the measuring range.

_ indllclive Iransducers (L VDT): based on lhe variation of lhe magnelic

characle risl ics of an eJec lrical circuil in func tian af the displacement of a mClallic
nucJeus.

5.2.2.- Horizontal movemcnts: pcndulum

Thc bas is a I' tbe pendulum is the implemenlation af an ve rtica l axi s for measuring
horizontal movements of di verse points of lhe slruclure rcferrcd to it.
56 STR UCTURAL ANALY$IS OF HISTOR1 CAL CO NSTR UCT10N S

Two modes exisl af measuring : diTecI

pendulum and invertcd pendulum .
-I- The diTecI pcndulum consists af a sleel
wire fixed in lhe upper part in a poinl af
lhe structure (figure 1) and tensioned
and maintained vertical with a wcight in
lhe battam. The horizontal movcments
ofthe line so derined are de rei atives to
lhe upper anchorage poin! 01' pendulum.

In different heights ofthc wire systems

01' reference fixed to the structure allow
-I- to measure lhe two horizontal
components oftheir movemenl reJerrcd
I to wire, getting in [his way lhe rei ative
horizontal movemenl af with lhe
anchorage poin!.

In lhe case af lhe inverted pendulum lhe

anchor poinl is in an inferior zonc in lhe
cnd af a drill Ihat should penetrale in the
foundation unlil a zone in lhe one which
-I- Ihey are not foreseen movements. The
lhread puts on in tension by means of
lhe placement in lhe other end of a nOa!
wilh capacity in order lo produce a
enough traction in the wire. The
Fig. 1.- Dircet pcndulum
measuremenl is carried oul in a simi lar
way to lhe case of the direct pendulum.
In this case, nevertheless, lhe movemenl measured is absolute.

For the mcasurement ofthe componenls ofthe movemcnt exist diverse instruments of
optic type or eJectrician: Planchettes, optical coordinometers (figure 2),
telecoordinometers: optoelectronic or electromagnetic based, (figure 3).

The systems of direct and optic reading proportion a resolution of 0,1mm, whi le lhc
optoe\ectronic syslems reaches resolutions of 0,01 mm.

The optica l and mechanica l systems are less expensive but lhey do nOI all ow lhe
automation of the readings.
 R_ ASTUD!LLO PASTOR I Experimenta! techniques 57

Fig. 2.- Optical coordinometer (Mattcini)

58 STRUCTURAL ANALYSIS OF HISTOR ICAL CONSTRUCTIONS

Fig. 3.- Tclccoordinornctcr (ISMES)

5.2.3.- MovcIIIcnts in joints

:IDd cracks

Thc comparali vc movcmcnts

betwccn lhe Iwo blocks lha!
co nstitulc a jo int can bc
Illcasurcd with severa l oI' Ihl.:
syslCllls for mcasufln g
di splacCIllcnts. plac ing lhe fixcd
..
"
el clllcnl oI' lhe devi cc in one 01'
,.' ~
" lhe blocks and lhe mobile
{\. , .
l' ' .-
1 c1cmcnt in lhe othc r.

Fig. 4.- .Joiot movcmcnt Whcn lhe onl y magnitude lo

mcasu rc 15 l he opcnin g Of
clos in g 01' lhe joinl. il is cnough to provi de an appawtus with its base 01' lllL"usuring
perpend icular lo lhe plan orthcjoinL Whcn il is wan lcd to know also lhe movcmcnl in lhe
d ircction ofthcjoint or crack il is nCl:cssary carry ou! a confíguration in triang lc as inl he
fi gure 4.
 R. ASTUD iLLO PASTOR I Experimental techniques 59

Thc devices based in mechanical extensometers are easy to install and economic but they
involve the access to the measuring zone every data reading. They do not permit the
automatization of readings.

The devices bascd in a transduccr (an clectrical magnitude varies accordi ng the physical
phenomena) permit the impl ernentation of aulomatic measuring and data acquisiti on
systems.

5.3. ~ Mcasurcrncnt of rotations

Tllc rotation movemenl s. mainly those rderred to horizontal axes. are measurcd wi th
instnuncnls denominatcd clinometers. The clinomeler measures lhe ungle formed by lhe
body Df the appawlus with a hori zontal plan. Therc is a wide range of clinomete rs
accordingly to the physical laws Ihey are based on.

Bubblc c1inomctcrs

Thcy are constituted by a Ihorical bubble water levei 01' great scnsibility Illounled on a
base. A micromelcr measures the inclination bclween succcssive rcadings.

Thc main di sadvanlage of lhe bubblc clinomclers is the requiremenl fo r accessi ng the
measuring point for its reading and lha! i1 is nol possiblc lhe autornation Df the
measurelllenl.

Clinorncters bascd on rcsistancc straill g~lgCS

The)' are bascd in the measurcmen t 01' lhe dclormation oI' un elcmenl (ll1eaSllremCl11
carricd oul by mcans 0 1' strain gagcs) produced by the inclinalion (lf lhe apparatus.

C linornctcrs bascd on inductin transduccrs: p:lrallclogr:lm .:tnd pcmlulum sysh.'ms

p(lrallelo~ral1l -\yslems: Thc instrurncnt is bused on lhe arnpldication of lhe rotalÍotl by

mcans 01' a parallclogram 01' crosscd threads and countcrwcight. Illovcmcnt thnl is dctec tcd
by an inductive lransduct'r. Thcy pcnnil lhe reading by lllcans 01" an optical scalc and lhe
automaled measurClllcnt to distnnce.

Pelldil/IlII1.\)'S/C/IIS: Thc)' are bascd in a small pendullllll whose movcmenl is dctectcd w ith
an indllctivt.: transducer. There arc somc devices {har mount pendulums in Iwo orthogo nal
dircctions.

C linornctcrs b':lsed on Scrvo-accc!crornctcrs

Thcy are based in lhe considemtion of gravily accderation as refe rence. Upon varying lhe
:mglc of lhe aClive axis 01' the nccclerol11der with rcf~rence to the vertical axis il is
produccd a variation 01' the acceleralion proporlional to a Irigonolllclry rUllction oI' the
ang!e.
60 STRUCTURAL ANALYSIS DF HI STDR ICA L CDNSTRUCT IDNS

S.4.~ Measurcrncnt of tempcraturcs

For the measurement oftemperatures different types oftransducers ex ist:

• Thennocouples
• Resistance thcnnometers
• Thennistors

The attached table picks up the range of app lication and the advantages and
inconveniences of lhe three methods for measuring temperatures.

Thermocouplc Rcsistance Thermistor

RANG E (0 C) -200 a +200 -200 a 500 (PTI OO) -60 a 400

advantages Selfpowered Stability High levei

Simple Accuracy outpU!
Slrong Good lineaJity Fast
Cheap measurement
Wide range

Disadvantagcs No lineality Expensive No lineality

Low OUIPU! levei Slow rcspo nse Limited range
Reference necded Power requirement Fragility
Less stability Small variation Df R Power
Poor sensib ility Four wires requi rernent
$clf. he:lting Self.heating

5.5.- Acceleration

In general, the basis of the transduccrs of aeeeleration (accelerometers) co nsists of

measuring the accelcration in an indirect way: measuring the forces of incrt ia that appear
in a small mass that is part ofthe instrument. Forces ofinertia caused upon locating lhe
apparatus in a paint ofthe structure subjected to accclerations in the active measuring axis
ofthe instrumcnt.

lhe scheme of ali acceJerometers consists in a mass, a spring clem ent (proport ional
displacernent to the force) and an clement ofviseous damping (proportional spced to lhe
force). The fundamental frequeney oflhat syslern should be severallirncs lhe maximal
frequency of interest included in lhe vibralory response that is to measu re .

lhcre are accelerometers based on resistance strain-gagcs, LVOT, piczocleclric cffeet,

servocontrolled, ..
 R. ASTUDILLO PASTOR I Experimental Icchniqucs 61

6.- DATA ADQUISITlON

The experimental analysis of the structural behavior of an historical construct ion

implicates, as it was previously described, lhe realization of the measurement of several
magnitudes using a wide range of instrumentation. This variety in the instrumentation
type, the high number of measure points and the access difficulty to many of them, makes
that the excl usive util ization of manual methods for reading instruments and for analyzing
lhe data cou ld comp li cale lhe investigation.

In lhe recenl decade the techniques of auscultation, but specially the techniques of data
acquisition and management of information, have evolved to the utilization of instruments
lhal pcrmit:

• Fast and reliable rcading

• Remote reading
• Integration in a system ofautomatic reading
• Transmission of the data to distance
• Storage of1he infonnation in databases 1ha1 allow their analysis with different
applications

Logicall y the more or less importance of lhe data acquisition implementation depends on
the importance of the construction to control, on lhe period for continuous auscultation,
on lhe kind of magnitudes to measure, on lhe number and acccssibility of measuring
points and on the consequences of the ruin of the construction.

6.1.- Automation of instrumentation

The automation of lhe auscultation of a structure requires as a basic condition lhat lhe
instrumentat ion for measuring the different magnitudes involved were oftransducer type.
This not will always be possible because, apart from economical considerations in some
cases, lhere are measurements difficlllt or even impossible for being inclllded in an
automatic system , for example, the corresponding to lhe topographical melhods.

In Olher cases, for the measuring lhe same magni tude there are bOlh, manual instrumenls
and transducers, these last will be necessary in the case of a complete automat ion.

The usual situation is thatthe auscultation system is hybrid, with a serie of measurements
inc1uded in an alllomated system. and lhat will allow the observation of the behavior of
the structure in rea l time, and therefore to establish the necessary wanns in case of
anomalous behavior, and beside lhal a serie of non-alllomati zed mcasurings lhat will
supplemenl lhe infonnalion.
62 STR UCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

6.2.- D.lta p r occss

In general, lhe initial data obtained as lhe d ifecl rcuding oI' an illstrumcnt, automatic or
manual, has to be submitted lo more Of less comp[cx transformations (Iinearizat ioll .
application of translormation constan! Of function, calibration. crror corrcctions .... ) in
ordcr to lranslalc it to lhe correc! measurcmcnt af un spcc ilic magnit ude. in a ccrtain point
ofthe structurc nnd in a ccrtain instan! oflimc.

As lhe volume of in [ormation lha! could gcncratc a pcrmanent s)'stcm for auscultation oI'
a construction can bc extremely high, it is fundamental lhe application 01' a serics 01' data
rcduct ions so IhaL wit houtlack ofthc information conta incd in lhe data lha! is uscfu l for
the interprctalion oI' the bchavior of the slruclure. lhe vo lume oI' dal<l to transm ito slorage
or process, could be diminished.

The firsl stages in the ana!ysis ofthe data are:

T ra nslat ion to physic al ma gn itudes

The direct reading oI' some instrumenl can be voltagcs. currcnt intesities. number
of divisiolis. etc. and they have lo be converted. by means oI' the adequate
formulae. to the physical corrcsponding magnitude . This conversion will be
manuaL in some cases. or automatic. wil h a process more or less sophisticated .

In this phasc lhe opportune corrcclions of d irect readings could a lso bc made :
lemperaturc ctTccts. for cxamplc .

• Elimin ation of er r ors

In lhe direct readings evidence of errors could exisl and Ihey have to be discarded
before processing the in lormation. lt is frequent that infonnution related wíth the
behavior of some part of a structure is determincd by severa l measurcmenls lhat
offcr some redundancy lhat permi ts. joined with symmctri cal effccts or similar
slructural situations in other momcnts. to dctemúne ir a data is clearly erroncOllS
or doubtful.

• St:ltistical reduction

In many cases the data Illcasurcd in diverse points or in differcnt momcnts permit
a first statistical rcduction substitut ing a group of rcadings for its avcragc and
standard dcsviation.
 R. ASTUDILLO PASTOR I Expcnmcmal techniqucs 63

I!EFIi:R ENCIi:S

ASTUDILLO PASTOR, R., GARCIA RUIZ. I). "EI ensayo de gnto plano COIll{l
téc nica para la auscull ac ión in situ de obras de fabrica". Ingc nieria C ivil. n 98, 1995.

i\LV i\REZ DE 13UERGO I3i\LLESTER, M., GONZi\LEZ LlMON, '1'.,

"Reslaurm:ión de cdi licios monumenta les". Min isterio de Obras públi!.:i.ls, T mns porte y
Urbani smo . Monografia C EDEX. 1994 .

ALONSO E., GENS, A .. h[nslrUlllentación de obras". Uni vers idad Po[ilcc nica d~
Cata[uiia . [989 .

FERRARII)i\ rA SANO. c.. "Stalic prcscrvnl ion of lhc I)uomo oI' Milan ". Proccs.
Intc rnaliona[ Assoc ialion for Shells and Spatial Slructurcs (lA SS) Sy mposi um . Mil ;mo
1995.
 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS
P. Roca, J.L. González, A.R. Marí and E. Onate (Eds.)
© CIMNE, Barcelona 1996

THE TESTING, ANALYSIS AND ASSESSMENT OF

MASONRY ARCH BRIDGES

T.H. Hughes
Cardiff School of Engineering
Queens Buildings, Cardiff Universify of Wales
C"d;ff, CF2 lXH. U.K.

NOTATION

d arch ring thiekness

f depth of fill over areh crown
PAL provisional axle load
r areh rise
s arch span
w loaded length
o: tensile strength of unit
d. tensile strength of mortar
o: eompressive strength of unir
if. compressive strength of mortar
0"1 ,0"2 principIe stresses (radial and ei rcumfrential direction)

ABSTRACT

Masonry areh bridges exist in varying numbers in most eountries in the world.
A full understanding of these eomplex struetures is important for two principal
reasons, they continue to rorm a significant element of many countries transport
infrastructure and because there are a large number of historieally important
structures. In general the work in the UK undertaken in Tecent years has been driven
by the eeonomic need to maintain the infrastructure. In addition there is a move to
construct gravity arch structures using new materiais in an attempt to replicate lhe
longevity clearly demonstrated by these, many, aneient structures.

This paper eontains a slate of lhe art report on the work undertaken during the
las! decade on masonry arch bridges. The Teview concentrates on the work
undertaken in lhe UK since Ihis forms lhe bulk of the Tcsearch buI does include
contributions from other parts of the world. The paper contains details of the (a)
experimental studies undertaken aI both small and medium model scale as well as on
real structures, (b) analysis us ing plastic methods and finite element analysis and (c)
some consideralion of lhe use of lhe analysis lools for load assessment.

Keywords
Analysis, Arehes, Bridges, Masonry, Mechanism, Finile Ele ment
 T. H. HUG HES I Masonry arch bridges 65

1. lNTRODUCT10N

Masonry arch bridges

represent the oldest form of
crossing still of economic
significance. In some, even "
developed, countries they sti II
form the principIe element of 30

the bridge stock . The resu lts

of a samp!e sllrvey llndertaken
in the UK for lhe Departmenl
of Transport [ I ] are presenled BAITISH AA, IL wo.TERw.o.yS BOA,F10 GOVEANMENT

in Figure J which shows lhe _ MASONAY t@Sj ME TAL O AE INFORCED CONCAETE

percentage type of bridge
structures and their ownership Figu re I Distribulion of bridge type and ownership
It is apparent that in lhe UK in lhe U K[I]
masonry arches sti!! eonstinlte
a significant infraslrllctllre il1veSl lllcnt. The survey also conlained details of lhe
dislriblllioll of lhe spans of lhe masonry arches, this infonnalion is conlained in
Fi gure 2. AI the smaller end the difference belween a bridge and a culvert beeomes
somewhat academic . T he maximul11 spall in lhe UK is actllally 66 metres.

In addition lO lhe economic arglllllcnl l!lere exists, in the UK, a large number
of listed, that is hislorically prolecleel, arch bridges bllild by renowned Engineers
inc1uding Smeaton , Rennie, Telford an d Brunel, these are in addition to earlier
mediaeval examples. Thcse struc tures frequcntly conlinuing to flllfil both the
ecollomic alld historie roles .

There is little
eurren! distinetion between
the cons iderati on given to
" arches constrllcted of brick
and block masonry, both
normally being cove red by
the lerm masonry. The
significan! (eatures of lhe
traditional arch are shown
in Figure 3. Tlle principal
elemenlS are lhe
\2 .5 ablllments, lhe arch ring ,
" " SPAN
lhe spandrel walls
Figure 2 Di stribulion of m<tsonry arch bridge spans in (including lhe parapet
[l1e UK[ 1] walls) and lhe fill material.
The wingwal ls (nol shown) <Ire also significam . Tllere is also frequently additional
masonry between lhe arch ring <lnu lhe fill. This can be i n lhe fonn of hallnching
above lhe ablltments or additional arch rings of less competent masonry . Oceasionally
there are interna! spandrcls, induded both 10 rcduce weight and to provide additional
strength. Voicled cylinders are also sometimes included above lhe abutments to
 66 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

reduce weight in certain geometries these may continue through lhe spandrel and thus
be visibJe ar lhe ends may be c!oseu by lhe spandrel wall. In addition for muItispan
arches lhe piers themselves are significam.

There are tWQ

feCen! books dealing
exclusively with
masonry arches lhe
first contains a
significanl elemenl Df , .
lhe international
historical background Spandrd Wall

to these stnlcwres[2]
and lhe second a
recent state Df lhe art
review cf currenl UK
research and
pracli sef3]. In
addition a recent
international
conference on arch
bridges was devoted
aI mos! exclusively to Figure J PrincipIe reatures af a rnasonry arch bridge
masonry bridges[4].

2. TESTING PROGRAMMES

The testing programmes can bc convenien tly divided into lhe fuH sca le testing
af real arch struClllres and lhe laboratory lllodeHing aI smaUer scalc. The cancept of
scale is inlcresting bccause lhe dominance of gravity forces in lhe strength of these
arch struclures Illeans Ihat small !1lodels are nOI sl11all scale Illodels of larger structures
but are in fael full SC<1le !1lodels 01' sl11{!ll bridges. Some of lhe larger laboralory
models have however becn 01' an equivalen! size 10 tlle small er fllll scale structllres.

2.1 FuI! Scale Site Testing

The UK full scale testing programme was becn organised and financed by the
TranspOr! Research Laboratory for lhe Departlllenl of Transport. The prograrnrne
starled in 1985 and involved the lesting 10 deslruction af a number of disused bridges.
To date 1i masonry arch bridges have been tested incJuding 2 full scale laboratory
SlrtlClures.

The programme was intended 10 cover lhe full range of arch materiais,
including block concrete, brickwork oI' different numbcrs of rings, dressed stone and
random rubble, lhe full range of geolllelric shapes incJueling segmental and elliptical
of different spanlrise ratios anel span /deplh mlias, af different crossing orientation
including square and skew spans, 01' single anel Illulti-span as well as of different
conclition including both badly crackeel anel distorted ZlS weU as brand new structures.
 T H_ HUGHES / Masonry arch bridges 67

It is clear that 11 is insufficient for a full parametric study of each parameter but it
was felt important to ensure ali features were included in at least one test bridge.
Table 1 details the range of the geometric parameler variation tested to date.

Geometric Property Range of Values

rise to span ratio 0.16 < ris < 0.50
ring Ihickness 10 spll.n rat io .034 < d/s < .094
soil to ring thickness ratio 0.28 < fld < 2.15
loaded length 0.3m < ·w < .75m
skew O deg < angle < 29 deg
Table I Range of geornetnc pararneter values from ful! scale tests

lt is clear from Table 1 that there are is a large range of basic geometric
shapes covered by the tests but the fuH range of the real structures is considerably
larger. This is important in identifying the principIe structural actions to be
considered since these are largely determined by these geometric parameters . For
example the tests have shown the significance of the restraining action provided by the
fill material in high rise to span ratio arches and for shallow arches failure modes
involving elastic snap-through have occurred.

The static load tests involved loading in lhe vicinity of the 1/4 to 1/3 point of
the arch span with the load being applied througb a spreader beam of length, usually,
750 mm across the fuH width of the bridge between parapels . The loading beam was
used to ensure that there was not a local soil bearing failllre beneath the applied load.
The loading system normally reacted against ground anchors previously drilled
through the arch. Attempts were made to isolated these anchors from the arch using
ducting although Ihere is same evidence of t!lese providing significant longitudinal
restraint against the normal sway failllre mode.

The instrumentation included in the tests were primarily remote sensing or

surface mounted to avoid undue disturbance of the fill material and the loss of
expensive instrumentatian at collapse. In arle case the actlJal thickness of lhe arch
ring is questionable becallse af lhe random nature of lhe rubble and its total
disintegration at final collapse coupled with this need lO avoid Irial pits prior to
loading. For lhe ful! scale "modeI" lests there has been the inclusian of pressure
gauges in lhe backfill. These pressure gauges have become mare praminent as lhe
recognition af lhe importance of the fiH behaviour has become clearer.

Fuller details of the background \O the tests is contained elsewhere [31 togelher
with a full list of references of lhe specific publications covering the individual teslS.

2.2 Small Scale Laboratory Test ing

As identified above the cancept of physical scale modelling af lllasanry arches

is problematic. Many madels are referred 10 as scale madels of larger protolype
structures whilst lhe structures and models are, in fact, simply geomctrically simi lar.
Indeed in some cases 1/4 or 1/2 scale brick units have been used lO maintain the
geometric similarity. In trulh many of the real effects identilied, whether the effect 01'
68 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

backfill or Df lhe crushing strength Df lhe arch ring materia l, are nol properly
modelled aI ali , with lhe results only giving qualitative and nol quantitative
asscssments Df lhe erfeelS. For this reason these models should be classcd by span
ralhe r lhan by scalco

No attempt is made here to detail a11 lhe Jaborato ry tests undertaken. The
paramcters investigated in lhe laboratory modelling and the general ou tcomes , where
available, and how Ihey may efreel lhe numerical mode!ling, to be conside red laleT,
are included below. The following erfeels have been identified/i nvestigated using
laboratory tests cf between 300mm-6000mm span.

1) Spandrel wall SUPOQrt The action Df spandrel walls has becn shown lO
significan tly enhance lhe failure load of ncw model arches [5]. Unfortunately for
general assessments their contribuiion is rarely conside rcd bccause of the difficulty in
ensuring combined action. In lhe real slructures lhey have both fallen orf at an early
loading stage or remained intact wi th lhe arch failing between them. For mult ispan
arches thei r contribution is more assured because rotation of the pier requires spandrel
movement.

2) Arch nn!! seoarallon Invesligation of lhe condition of, particula rl y, brickwork

arches has shown lhat on a number of occasions the individual brickwork courses that
form the main arch ring have become separated from each olher. This can oeeur for a
number of rcasons and becausc arch brickwork is gcnerally streleher bondcd, Ihat is
with no headed eourses eonneeting lhe separale layers, it is surprising lhat it is nOI
more eornrnon. The lack of headers, always olherwise included in multi·course walls,
may be associaled with historie experienee but is more likely to be simply related to
lhe difricully in matehing lhe differcnl courses. 1I is no! uneommon in tunnels for a
row of headers lO be provided where lhe eourses gel baek in line. This slructural
weakness has becn simulated in laboralory st ruelures by using a sand layer between
eou rses ralhe r Ihan lhe normal morlar [6}. In these situations lhe results indicate a
signi ficant rcduetion in strength when eompared to idcnt ical tests with a monar layer.
Recent full scale patch load tests, not to failure, on an arch with severe ring
separation where attempts were made to invesligate lhe movcment belween layers, has
shown a different result with liule or no discernible difference in the vertical
Illovements of the different layers unde r load. Ring separation was also present in the
fllll scale leSIS and does not pa rticularly seem 10 have effected lhe nurncrical model
prediclions made, none of which have atlemplcd 10 model ring sepa ration.

3) Arch rino crushing There is littlc informat ioll of the effect of varialions in lhe
crush ing sl renglh of masonry although some early work has been undertaken loadi ng
areh rings without backfill [7]. Therc is ll\Jmerical analysis evidence of its effeel buI
this is di scllssed laler.

4) Cyclic loading Some reeent work has bcen undertaken investigaling the effecI
of repeated cyclic loading [8]. Thcsc loads havc been applicd c\ose to lhe quartcr
poim and after some initial damage the archcs appcar to !lave settled into a stable
condition. The arches have then becn loaded to failure and have shown little effeet on
the ultimate load.
 T. H. HUGHES I Ma<;onry arch bridges 69

5) Fundamental freguencies The application of a small impulse load. via a

hammer, to produce a vibration response, at various stages of applied load, is
eurrently being used in an attempt to quantify the effeet of "damage" on lhe natural
frequencies of arehes. This work in volves medium span models and full scale
monitoring [9].

6) Ablltment movement Most nllmerieal analysis assume a rigid ablltment.

Reeent moves to develop new construetiol1 teehniques for masonry arehes together
with the need for suffieient headroom has identified flat arehes on synthetiea1iy
reinforeed soil foundatians as being econamically viable. 011 shallow arehes lhe
effee! af ablltment movemenl is considered crucial and some limited testing has been
undertaken [lOJ.

7) Serviceability Experimentation on serviceability has only properly been

considered via the cyelie loading referred to above (4). As a result of the load
deflection curves from a large number of tests it has been assllmed, for the UK
standards, that the limit for serviceability approximately corresponds to one hal f of
the ultimate load [lI]. This is considered to eorrespand approximately with the limit
of the elast ie respo nse and with the visual and audib!e onset of damage. This is a
very subjective view as seve ra1 arches have shown almost no evidence of damage up
to the maximum load and have then demonstrated significan! softening to failure.

8) StrengthcninO" There are a range of strengthening techniques used in the UK

under experimental investigation incl lld ing increasing the ring th ickness by saddling
[1 2] or by spraying th e int rados (13], bath witil minima l concrete reinfarcement.
80th have demonstratcd significant increases in strengt h over both lhe damaged slate
and the new original state. In addition madel tests using both retrofit reinforcernent
[14J alld spandrcl wall tics have bcen undertaken.

9) rvlulti-snan Limited labaratory studies of the effeels of adjaeent spans on the

strength of arch bridges have been undertaken [15]. The lests undertaken on three
span arches have shown a signifieant decrease in strength, when compared to lhe
identical single span equivalen t, the intcrmediate supporting piers were however quite
slender.

To date little of lhe results 01' Ihis model1ing work has found its way inte the current
UK assesSlllent practise [11] but it has helped identify effects and Illllch af it is still
vcry rcccnt and ongoing.

2.:1 Acceler;lIed Gravity Laboratory TestinO"

To date ali the model tests have been undertaken aI unit gravity althOllgh work
is just slarting in lhe U K on lhe testing aI accelerated gravily using a geotechnicaJ
centrifuge [16] . The dominance of gravity, both in the stability af the structllres and
the slresses within both lhe fil1 and arch ring, means that in practise few, if any, of
the smalJ scale models have identified masonry material failure (crushing) or
70 STRUCTURAL ANALYSIS OF HI$TOR1CAL CONSTRUCTIONS

attcmplcd to propcrly model lhe intcraction betwten lhe fi l! and arch ring which is
strcss dependent.

Ooe of lhe known advantages of small scale tcsting is lhe accurate definition of
lhe boundary conditions and lhe accu rate repcatability of lhe tesls . Since much tcsting
work is unde rtaken lo prov ide benchmarks against whicb numerical analysis can be
comparcd and because both of these goals have becn difficult to ach ieve wit h large
modcl scale work this is secn as a major advantage. The large physical size of lhe
large laboratory slruc lurcs together wilh lhe appl ication of large loads has causcd
sevcre problcms of rcpcatabi l ity and lhe poor dcfinition of boundary conditions .

Prcv ious unil gravity work on masonry has becn undCrlaken thal proves lhe
viabilily of scalc modclling of l~lason ry and Ihis is being cxtended in lhis ncw work to
include cllrvcd masonry slruclurcs and the soillstructure intcraction. This work is
cllrrenllyal 1/6111 scale bul wi ll cvcntually inclllde 1I18th scale models.

2.4 Pailu re modes idenlificd in leSIS

Thcrc are Ihrce main

failurc modes Ihal have becn
idcnlificd for masonry archcs
!-linge
wh ich are shown, llingc
diagramll1atic,1l1y, in Figure 4.

i) Thc formalion of a
mcchanislll, bere joinl
I'ossiblc S11ap Throllgh
opcning gradllally
progresses throllgh lhe Figure 4 Simp[ificd failure modes of masonry
ring ai, aI lcast fOllr
arch bridges
locations, lhe joinls Ihcn
rolate and lhe Slructure gradual1y collapscs.

ii) Elaslic snap-Ihrough prior 10 lhe f ull forlllation of hinges. Hcre lhe joint
opening can cause significan l secliolls of vcry tlcxiblc ring carrying large
Ihrusls. The progress of mechanism failure is pre-emplcd by a rapid snap
Ihrougll.

iii) Material failllrc whcre significanl parts of lhe ring crush.

Clcar[y cOlllbinalions, of lhe above, are also possiblc. Tablc 2 shows lhe basic
rcsu[ls of nine of lhe TRL IcSIS anel givcs a briel" descriplion of lhe actual failure
modcs observed in cach casc.

Whcn a [oad <lpplicd, aI or near Ihc quartcr span of an arch, is increased four
cracks or hinge points will gradually formo T hesc hinges nOflnally occur one at cilhcr
abuIIIK'lll. ÚIlC under lhe load poinl <lI onc approximaleJy half way bClwccn lhe load
poinl anel lhe f:)f abutment. This failun; mode bccomcs more complicalcd by lhe
 T H_HUGHES I Masonry areh bridges 71

Name Sp"n Failure Failure Mode

(m) Load (kN)
Prestwood 6.55 228 FOLlr Pins
Bridgemill 18.30 3 100 Snap Ihrough
Bolton 6.00 1170 Four Pins
Shinafoot 6.16 2500 Four Pins complicated by random brick
conslruclion
T orksev 4.90 I080 Snao Ihrough
Baroower 10.36 5600 Crushing bclow load
Preston 5.18 2100 Crushing bcJow load
Strathmashie 9.43 1325 Not we ll defined with material falling out of
existing cracks
Barlae 9.87 2900 Heavily skewed. Snap through followed by
shcar failure in spandrel wall

1'able 2 Failure modes of ful1 sizc test bridges

introduet ion of spandrcl walls and fill material. and also becomcs Icss clear whcn the
arch ring is constructcd Df wL'aker and lcss homogcncolls materiaIs, for cxample
random rubble or 510nc.

Material failurc of lhe arch ring undcr lhe cxccssive loading is causcd by lh e
strcss resulting from joint opening. Tlle collapse cOIHlllcncing without the l'lIll
l'ormation of ali lhe hinges. Thi5 local l'ailure can, in thcory, occur aI any of lhe
in cip icnt hingcs but whcn it initiatcs fail urc it usually occurs in the extrados under lhe
load point. Thc loading conditions at this locat ion are extremely complcx and there is
a clear lack of knowl edge cOll cerning the mec hani sm of failure by crushing. Th c
curvcd nature of the ring , togcl her with morlar-block intcracl ion , lhe presence of th e
large laleral applicd load. ali occurring ai lhe extreme edge oI' the Illason ry ullil, will
producc lII1LJsual conditio115 beyond currcnt maso nry modcl s. Back analysis, 011 arches
that showed no signs of such a failurc. has 11O\Vevcr indicatcd lhat the material
su rvives I11l1cll longer Ihan that delermined from traditiona! masonry wall eri[('ria.
1'11is typc of crushing fai l ure is mosl COlllmon in archcs constructcd oI' lhe wcakc r
materiaI s, for exalllple s!one rubbl e archcs.

Thcrc is also lhe possi bilily oI' lhe local ised failure oI' 1l1:tsonry bridges, in
cases whl're high in[ cnsily 10:lds and poor or inadcquale ll10rtar cause sec li ons of lhe
arch ring lo fali oul by punching sllear.

2.5 Sitc data for bricl!!c mo(]elling

The COSI of collecting good site data, es pccia lly for bridges wh ich carry
vchiclIlar traffie, can frcqucntly excced the cost of lhe analy sis . For oruinary brid ge
load aSSCSSlllent s it is lI s11:111y lI ecessary to sta rt wilh the sil1lple basic geollletric data.
Th is includcs span (s), centre span ri se (r). ring lhickn css (d) anel soil dCplh (I)
togc Lh cr with lhe bridge width. lhe mosl criti cai dilllcnsion is lhe rin,!; lhi ckncss and
sincc Ihi s CdnnOI bc I1lcasu rcd propcrl y wilhou l a Irial pit il C:ln, slIbjcc l la a sil1lil:l r
72 STRUCT URAL AN ALYSlS OF HI STORICAL CONSTR UCT10N S

experience with simil ar bridges in lhe locality, be based on lhe visible edge af ring
beneath the spandrel walls. In lhe authors experience the actual thickness is rarely
significantly less than tha! visible. In addition the initial site visit should identify the arch
material, the position af any cracks ar distortion af the masonry ar other defects, and
other faetar likely to be relevant, for example rutting af the road etc. A preliminary
assessment can then be undertaken.

Ir lhe assessment is in any way marginal it will be necessary to gather more site
data . Ir trial pits are possible then these can i) confirm lhe assumed ring thickness, ii)
allow the type ofbackfill material to be assessed and if possible iii) determine the extend
of any haunching above the abutments. It is difficuh to determine this infarmation from
cored bo reholes since there is frequently a merging af masonry into the fil! particularly in
stone ralher that brick arch bridges. Additional information on waterproofing, previous
attempts to graut the til! or saddle the arch, and services can also be ascertained .

It is unusual to abtain samples from lhe masonry for laboratory tesling. This is
because of j) the inherent difficulty in obtaining decen! samples, jj) the damage to lhe
structure jji) the large variabilily in results requiring a number of samples and iv) lhe
value of lhe data in lhe analysis is ofien less important than frequently imagined This
las! poinl resuhs from the important distinction bclwcen masonry units, ie bricks, blacks
stones, etc. and masonry, and because lensile strength is more important in analysis than
compressive strength.

The development of flatjacks for lhe determinati on af lhe insitu slress SI ale,
modulus of elasticity and even material strength values, to be discussed later, has an
importan! role in lhe future development af masanry arch bridge material testing.

3. ARCH ANAL YSIS

The analysis of masonry arches has been used to perform a range of fllnctions. ·
]n the UK bridge assessment has beeo the major concern, associated with the need lo
assess load capacity of bridges for the 40 tonne and eventually 44 lonne vehicles. The
current analyses can be conveniently divided into a number of groups.

1) Empírical equations and ideali sed c10sed form sollltions used solely for the
purpose af bridge assessment.

2) Simple numerical analysi s, using rcadily availablc engineering general elasti c

analysis software, again used solely for the purpose of bridge assessmcnt.

3) Simple mechanism analysis tailored for application to arch bridges , developed

to directly include fill passivc rcstraint and simple arch ring crusl1ing .
Developed both as a researcl1 tool and for exploitation as commercial masonry
arch assessment software.

4) More sopl1isticated rigid block kinematic models, including sliding, cru shing
and multileaf analysis. Only used, to date, as research too ls.

5) Discretised cracking elastic analysis.

 T. H. HUGHES / Masonry :m:h bridges 73

3.1 Advanced analysis methods

The analysis covered by group 5) is considered in more detail in the eurrent

paper. It ean be eonveniently divided into l-D models, which are now regularly used
in the UK as analysis and assessment tools, and the more eomplex 2-D and 3-D
models whieh are still under development and are generally only used for research
purposes.

The earliest work into the use of non-linear models in the analysis of masonry
arches was carried out in the early 1980's. The initial applications showed that not
only was there potential in using ~hese methods in predicting tbe ultimate load
carrying eapacity, but they would also be useful in determining lhe areas that would
crack under abnormal loads. These non-linear methods have some obvious
advantages over lhe other analytical solutions, as they can provide information on the
extent of cracking, lhe stress and strain leveis wilhin lhe arch and lhe deOections at ali
positions under any load situation.

In both the l-D and 2-D rnodcls it is usual to assume a I melre strip load. The
limited experimental data suggests that the load is rapidly laterally distributed once il
hits the arch ring beeause of membrane action, this would suggest that the bridge load
should be assessed rather than the loaded lane but it is still common in lhe UK to use
a lane based analysis.

J.1.1 Dne Dimensional Models

The tirst work was carried out on the development a tinite element program
used beam elements, with a no tension formulation and a parabolic stress strain
reJationship in compression [17]. The unorthodox formulation was limited by its lack
of inclusion of a soil model and the possible effeets of moving abutments. However ,
this inilial application showed the advantages to be gaincd by lhe use of the tinite
elcment method when considering the serviceability of masonry arches.

A more conventional program was developed using curved ring elements and
using a smeared continuum to establish the extent of lhe cracked zone [18,19]. Eight
Gaussian integration points were used through the depth of the element to establish the
extent of the tension zone and thus the craeked area. AS with earlier work use was
made of a simple tension cut off model, where lhe masonry was not allowed to
develop any tensile forces, prior to cracking. However, the later models developed
allowed for a tension softening zone on unloading, but still most of the analyses
undertaken limited the tension to zero. Tbe eompression models used were general!y
elasto-plastic rclationships with a limiting stress levei and a strain cut off, after which
it was assumed that the material was cornpletely crushed. As the software was
developed it was realised that lhe soil models were becoming increasingly important;
initially sai 1 was taken as dead Joad only. Later work showed that these effects were
very signifieant, 50 horizontal one dimensional spring elements were incorporated into
the models to include for the efreels of lhe tills passive resistance. The models
74 STRUCTU RA L ANALY$I$ QF !-I 1$TOR1 CAL CONST RUCTION$

included geometric non linearity and, by using deflection control, were able to
simulate post peak load behaviour.

A cracking elastic analysis was developed. based on the early elastic work cf
Castigliano, which incorporated joint opening by reduction Df ri ng thickness [20].
The basic simple model was extended to include geometric non-linearity and the 50il
model detailed Ja ler. As with lhe finite element analysis lhe incípicnt hinges emerge
frem the ana1ysis and do nOl always conform to lhe usual assumed positions. Figure
5 shows lhe application cf this method to Presto" arch. This is a semi elliptical arch
which produced
l N: '.0.0.0
crushing under lhe
load, lhe l-D
analysis show lhe
position af hinge
formatian with
hinges forming at
the springings,
this method
assumes no shear
failure at the tight
curvature near the Figure 5 Predictcd hinge positions for Prestoo arch.
sp ringings. It is
likely that such a geometry would in fact have haunching and as a flat arch would
then likely fail with an elemen t of crushing. This l-O numerical model has becn used
to investigate the influence of the various material parameters on the predicted (ailure
load [21] and a parametric study , in probability space, has been undertaken that
suggests that these l-O type moclels provide more robust load assessments than the
simple elastic analysis that fo rms the basis of lhe current UK masonry arch assessment
standard [22]. In addition the use of these advanccd anaIyscs have identitied the need
to determine lhe initial stress state cf these highly redundant structures if normal
serviceability parameters are to be considered.

Furthcr work on 1-0 tinite elements used tapered straight bcam elements
[23,24]. The mcthod allowed for the alteration of the effective depth if cracking,
crushing or both occllrred during the itcrative processo This method of tapering
allowcd for the inclusion of the sections of the arch ring which were in compression,
excl udin g zones which had developed tensile stresscs. The formulation allowed (oor a
compressive yidd stress levei after which lhe material was assumed to be crushed.
The soil model used was inc1uded as a number of horizontal one dimensional spring
elements conncctcd to the arch ring. However, these were only activated when the
arch ring movcd in to the till, providing on ly passive resistance. Tbe tapers were
based on a linear variation of section inertia along the element thus considerabl y
simplifying lhe ele ment matrix formulationo

The l -O elastice cracking models have providcd a valuable insight ioto lhe
beh aviour of masonry arch bridges. AI lhe limit if lhe soil model and the geometric
non lioearity is removed then the results should tcnd towards lhe mcchanism sohJlions
and this provides a good test on the numcrical procedures. Since the analysis wiJl fai!
 T. H. HUG HES I Masonry arch bridges 75

and this provides a good test on the numerical procedures. Since the analysis will fail
if lhe elements become too thin the mechanism represents and upper bound.

The inc1usion
",~(P:':',,':"~'_ _ _______ _______---,
of the geometric non
linearity a1ways
reduces lhe predicted
collapse load. Some
o.•
argue that masonry
arches are too thick o.•
lO suffer geomelric
effecls but as lhe 0 .4

arches crack and

0 .2
lhus 'thin' lhe
effective sections o L---~~--~----~----~----~
reduce and lhe line o 0.5 1.5 2.5
tE I Em
af thrust straightens,
which ali contributes Figure 6 Variation of predicled test bridge failure loads with
to lhe effecl. elastic modulus [25]
Parametric stlldies have been undertaken, using l-O cracking models, on lhe effect of
variations in lhe governing material properties [25]. Figure 6 shows lhe variation in
the predicted failure load, for the bridges detailed in Table 2, with variations in the
assumed elasticity. The results indicate a factor af 2, increase or decrease, in the
assumed value generales only a 10-15% change in lhe predicted load.

More
important is the '" r" -"-".::"'-'- - - - - - - - - -- - - - - --,
varialion resu lting
fram changes in lhe
~
principie sai l o.•
para meter , lhe angle
of internai fríclion, o.•
this variation is
o.•
shown in Fi gu re 7.
The soil parameler 0 .2
produces a more
o L-----~ ______ ~ ______ ~ ____ ~
sensitive response o.• o.• , ,.,
with a 10% change I~ I ~n)
producing a 0- 10 %
change in the Figure 7 Varialian of predicted lest bridge failure loads with
resulting failure soil friclion angle (25)
load.

The effect af variations in the compressive stress, shown in Figure 8, produces

a much less sensitive response, in fact in same ardes there is no visible effecl. This
early work l-O has been confirmed by lhe results of recen! 2-D analysis and poses
sevcre questions abou t assessmen! methods based solely on compressive strength.
 76 STRUCT URA L ANALYSIS OF HI STOR ICAL CONSTRUCTIONS

Ali the above U ~tP:é/-'P~"),---_ _ _ _ _ __ _ _ _ _ _ _ _-----,

work suffers from
lhe consideration of
masonry as a single • •
material and lhe o.•
therefore required
average or smcarcd o.•
material properties,
o.•
and their
simplification of the o.,
compressive state of
the ring. For o L-----~----~----~----~--~
u u u u U
example mosl (d I a'ç)
modern mason ry
Figure 8 Variation of predicted tesl bridge failure loads with
theory suggests lhat
compressive strength [25]
lhe failure in
compression invol ves the tensile splitting of lhe units caused by lhe poisson ratio
effect generated in lhe softer mortar, thi s is often attributed to a different poisson ratio
when in fact it is a result of lhe difference in the unit and mortar stiffness. The
concept of "crushing" of masonry as a global material failure is lhus suspect
especiall y when this is occurring in the extreme fib res of an eccentrically loaded arch
ring. Similarly the models require lhe masonry elastic modulus al lhough lhis
smearing is nol considered to cause significant errors. In a l-D analysis the lack of
any tensile strength in lhe mortar joint is relatively easy to consider since lhe joint is
always at righl angles lO lhe ring. Some methods update lhe geometry and effective
ring thickness whereas others incorporate lhe effect via smearing in the numerical
integration to determine the beam stiffness.

The are a number of drawbacks in considering arches using these I-D models
{hese are listed in comparison wi th 2-D modelling.

1) The soi! can only be

SoU Prusure O .
treated as a vertical .r---r-
load plus, at best, a
c . ~ Uu l u • • • S"b~,.d.
non-linear hori zontal A• •• " ••
spring. The besl soi! Kpo,
model assumes initially
an , at rest, lateral earth - - - -,---"
pressure at dead load I
Kaa,

and allows deflection -6 o 6-

dependant increases in H or izontalOeUection

lateral pressure as lhe

Figure 9 Soil spring model used in l-D analysis
arch moved into the soi l
mass and similar decreases as the arch moved away from lhe soi!. These
changes in the soi! pressu res are lhen limiled by the relevam aclive and passive
coefficient cut-offs, this is shown diagrammatically in Figure 9 . There can be
no allempt to properly model either the lrue restraining effec t of lhe soi! or lhe
possibility of lhe soil actualJy arching in its own right. Similarly lhe loading
 T. H. HUG HES I Masonry arch bridges 77

has to applied to lhe arch ring, invariably using some distribution law, ralher
than the true distribution obtained as lhe soi! deforms under load.

2) The treatment of the abutments is simplistic, either assumed as rigid or at best

acting as some son of spring.

3) The proper modelling of cracking and phenomena such as ring separation is

impossible.

4) Many arches are sufticiently thick and curved such lhat some l-D
simplifications are reaJly invalid.

3 I 2 Two Dimensional Madels

To date ali the models developed treat masonry as a single material and are
undertaken at 'macro' levei, the treatment of an arch as a 'micro' moclel with the
individual masonry units and the mortar joint separately modelled by eJements with
different material properties is realistically outside lhe scope of present hardware.

Al! 2-D models have been based on a plane strain consideration of a section
through lhe arch. The first 2-D simulations [19] maintained the use of 1-0 ring
elemenls but incorporaled 2-D elements to moclel soi l behaviour. These models nol
onJy improved lhe soil modelling but also overcame lhe problem, in the I-D model,
of distributing the live load through lhe sail to the ring. This had previously been
based on a specified load di striburion angle.

lt became apparent that lhe simple I-D models would be of limited use in lhe
modeIling of arch ring behaviour. This led to lhe development of two dimensional
models in which the fill cou ld be modelled directly using standard FEM soil s models
and lhe arch ring could be moclelled using more Ihan one element through ils
thickness.

The 2-D model developed by Choo et ai [2 6] used traditional eight noded

isoparametric elements to medel lhe arch and, ralher surprisingly, horizontal elements
l-O elements as the fil!. These fill elements only provided resistance when the arch
ring moved into lhe fil!. To enable cracks to form when tcnsion forces develop the
rather cumbersome method of the discrete disconnecting and reconnecting of nodes
was incorporated. This method could mede I cracking in both the radial and tangential
directions, as would occur in the case of ring separat ion. A1though lhe disconnecling
of nodes is not ideal i( is more useful in masonry modelling lhan concrele modelling
because the failure planes do coincide w'i lh either lhe bed course or the mortar layer
between adjacent rings. Since the easiest way of generating finile elemenl I1lcshes is
to use equal radial and circumfrential element boundaries these naturally coincide with
the directions of these weakjoints.

The two dimensional finite element· program developed by Loo and Yang [27]
used a linear elastic tension-tension zone with a tension cut-off, lhis was followed by
a signi ficant linear slrai n softening region , between 5 and 10 times lhe strai n at peak
78 STRUCTURAL ANALYSI S OF HISTORICAL CONSTRUCTIONS

stress, and a quadratic compressive stress strain relatio nship with an equally large
linear strain softening zone. Failure in lhe compression fcompression zone was via a
Von Mises plasticity mode!. The soil model developed by Loo ct ai was in lhe form
Df a separate tinite element program with vertical fill elerncnts, each held in position
at the base by hinges in line with lhe arch extrados. This enabled lhe vertical and
horizontal forces to be detcrmined at these hinge points due to the soilloading.

Neither cf Choo's Cf Loo's analyses included lhe soil as a 2· D media. Ali Df

lhe above models also surfer from their consideration Df masonry as a homogenous
isotropic material, indeed the governing failure properties in Loo 's modcl are the
masonry compressive st rength and the masonry tensi le strength neither of which has
any real meaning.

Recem work has been undertaken using a modified concrete model at micro
scale to model the behaviour of brickwork loaded in the manner of arch brick rings
[28]. The results of a series of analyses with differing compressive and tensile
stresses applied normal to the main material axis has been used to develop an
anisoptropic masonry model where the failu re surface is governed by the mortar
compressive and tensile st rength and the masonry unit compressive and tensile
strength as shown in Figure 10. Whereas in traditional masonry modelling it has been
necessary to develop a 3-D failure surface, by rotating the direction of the principie
stresses to the material axis, to properly model the brick mortar interaction, this was
not considered necessary as lhe principie stress directions in the arch ring of a
masonry arch bridge coincide c10sely with lhe material axes, That is, the principie
stresses are essentially ci rcumfrential and radial. In addition when the angle between
these stresses and the material axis are aI their largest these generally correspond to
regions of lhe arch whcre lhe stresses are low and still wilhin the elastic region.

,-
Itll
,- ~1tJ- o o o o o o o: Íl ' -
I
., ,;
_._-_ -
. ...
." "
1.... 01 \ · 1).. ",, _
."
SI ....... O•
.!IO

" o

.\
- - -.-- _.
Figure 10 Failure surface for curved masonr model [28]
 T. H. HUGHES I M:lSonry arch bridges 79

3 1 3 Three Dimensional Models

To date little work has been undertaken 00 masonry arches in three

dimensions.

Some initial work has been undertaken using plate elements of variable
lhickness coupled to soil springs in ao auempt to model the effeets of spandrel walls
and to provide some indication of lhe effeet of diagonal eracks [29]. This later work
was necessary to assist in the determination of so called condition factors for load
assessment. These studies use the same basie approach as the l-D analysis where the
region of lhe arch where teosioo is initially delermined are effectively removed from
consideration by reassigning lhe plale thickness ai lhat location.

More advanced work on lhe 3-D analysis of masonry arehes has recently been
published using lhe so-ealled homogenisation technique [30]. Homogenisation is a
process whereby a staeked series of bricks and the mortar between are initially
considered and it is assumed that under a uniform load there is eOlllpatibility of lateral
and transverse st rain between the units and mortar. This together with consideration
of force equilibrium allows lhe overall response of lhe staek lO be modelled by an
anisolropic layered analysis where lhe elastic components are funetions af the brick
and mortar elastic properties and of the geometry of the brick and mortar joinl. The
seeand stage homogenisation involves assuming that a series of these staeks are
formed with mortar layers between. The homogenisation process is lhen repeated.
Finallya the fuH set of elastic constants for a three dimensional analysis are obtained
again dependant on lhe separate unit and mortar properties and lhe unit and mortar
dimensions. Following an anisotropie homogenous analysis the average masonry
stresses and strains are detennined and this can then be di saggregated into scparate
briek aod mortar stresses. In the receot work eraeking was included aod lhe
homogenisalion of the eracked elements undertaken and applied lo masonry arch
.b ridges (31]. The work has yet lo be exlended lO include lhe soil modelling. This
form of analysis is considered appropriale for modelling in plane loads in walls and
has becn used in laterally loaded walls but lhe laek of through wall bond in arehes
coupled to lhe limited number of rings suggests that a modifieation to this process is
required for masonry arehes.

3 2 Modellin o lhe Initia! Stale

Ali lhe above sophisticated analysis require an assumplion of lhe initial state of
lhe strueture. This is importanl becallse lhe masoory areh is a high!y redundant
structure and its response to loading depends critieally on its inilial statc. This
important inOuence has, in fact, becn ignored in almost ali lhe above analysts.

The initial state of an areh, thal is the eurrent dead load state, depends 00 a
number of faetors, the constrllction teehniqlle specifieaHy including when the
formwork was st ruek, the consolidation of lhe fiH material and specifieally the al resl
pressu re it provides to the arch ring, lhe hi storieal movements of lhe abulments dnd
lhe creep wilhin lhe masonry. It has been shown Ihat arehes ean remain stable under
80 STRUCTURA L A NALYSIS OF !-IISTORI CAL CONSTRUCT10NS

a wide range of initial states and that at working loads the deflections, stresses and
craeking are dominated by the assumed initial state condition [32].

The importance of the initial state has lead to recent rescarch into the use of
flatjacks in lhe areh ring masonry to determine both the insitu elastic properties of the
masonry and lhe initial stress. These flatjacks operate using lhe gauged stress relief of
a slot within lhe mortar which is then relurned to its original stressed state using lhe
flatjacks. The work to date suggesls lhat masonry tends to be a very forgiving
material and that the extremes of initial stable states do not prevail with the measured
dead load stresses being associated with more uniform stress distributions within the
ring [33,34]. The work has recently been extended to determine lhe live load stress
within arches by mimieking lhe tive load ring strain behaviour using a computer
eontrolled pressure within the natjaek.

3 3 Future model1ing developments

The use of l-O and 2-0 non linear elastics analyses has produeed a large
amount of useful information on lhe load earrying characteristics of masonry arehes
at, and prior to, collapse. There are however a number of unresolved matters that
need lO be addressed in future work and eonsideration of lhe above progress to date
ean perhaps provide some pointers.

The l-O models appear to have a Iimited future in providing any further
insight into single span masonry arch behaviour. If lhe effeet of lhe spandrel wall
stiffening of multispan piers can be neglecled they eould yel be expanded to model
multispan arches.

To date geometric non-linearity has been included in a number of 1· 0 models,

it has, however, nol yet been used in 2· 0. This needs to be addressed if the 2·0
models are intended to be used for a more advanced consideration of ultimate load ,
this is particularly important if work on lhe developmenl of new shall arches
progresses.

The relative merits of using 2-0 soils models with 1-0 arch ring models or }.
D soils models with 2·D areh ring models needs to be addressed. There is general
aeeeptance of the need and value of soi l models but 10 date no overall acccpted
direction for 2·D work exists. If lhe approach adopted for lhe 2-D work is a plane
strain analysis whcrc more sophisticated arch e1ements are develupcd lhen there seems
little rcason to use l -D soi ls models. The general plastieity models that will form part
of the arch ring model could be used, with appropriate parameter values, to better
model the soi! behav iouT.

To lhe aulhors knowledge ali the areh bridge models cu rrent1y used both in 1·
D and 2-D modelling, with one exception [31], use a composite model of masonry
behaviour, these use either a smeared or tapered mason ry model. Many of the
problems associated with modelling masonry arches are exceptionally severe when
compared to those found in normal masonry walled slructures. For example lhe
crushing under the point load, previously discussed, also the interaetion between the
 T. H. HUGHES I Masonry arch bridgl!s 81

arch ring and both internal and externaI spandrel walls. In addition to moving to
general 2-D models it may also be necessary to use some of the more advanced
masonry models.

In the tanger term there is also the possibility of developing fuH 3-D models so
that the fuI! stiffening contributions of the spandrel walls as well as proper
consideration lhe fill could be incJuded jnto the analysis. This type of 3-D modelJing
will Iikely prove too expensive to be used as anything other than as a researc h tool.
The models could however provide valuable information for use in determining more
appropriale 2-D models for general use.

4.0 ARCH ASSESSMENT

The assessmenl of lhe load capacity of existing structures creates significantly

more difficulties than the design of new structures. The strength parameters and even
lhe geometry are often unknown. There is however evidence of the condition of the
structure and even its existence, in the current loading regime, says something about
its load capacity. In the UK the MEXE [li] melhod has been used since the 50's and
is based on an empirical equation of unknown origino The equation provides a
provisional axle load based, PAL, which is then modified by a large number of
factors lO take into consideration the joint material and width , the arch material,
backfill material. arch shape, arch condition etc.

PAL
= 740 (d + I) ,
s"

Despite lhe considerable investment into the research of masonry arch bridges,
alluded to in this paper, over lhe last 10 years the new standard published in 1993
[IIJ retains the MEXE equation as the primary method.

As an alternative
700 i~
'I=
M ='~='="=" =L'=.=,~,,~,"~"~
..~)____________________-,
an elastic two pinned 0- ... _ 1

anaIysis of lhe arch ring .00

+ '''cll. "
is permitted with lhe '00
lP-p l n ... "
o ""' cll. 2
soil aCling as dead
.00
, c. ... ~ 2 o
weight only and lhe
compressive strength on 300 •
the arch material being
,"O
lhe load determining
+
parameter. The '00 5
development of ten sile
stresses is conveniently • .00 200 300 .00 500
Experimental load (Tonnes)
'00 '00
ignored. The results of
the application of this
Figure 11 Variation of the theoretical loads for five
method wcre compared
assessment techniques compared to the experimental loads
by its originators with
[I I]
lhe other mechani sm
and l-O crack.ing elastic methods discussed above, the results suggested indicated a
82 STRUCTUIlAL ANALYS IS OF HISTORICAL CONSTRUCTlONS

wide scaUer. These resulls are presented in F igure 11 which shows a comparison
between lhe theorctical toad, determine using lWQ mechanism, tWQ 1-0 elastic
cracking and the recommcnded 2-pinncd analysis, and lhe experimental load for lhe
10 full scale bridge tests to failu re. Although lhe figure does 001 identify a clear
"winner" it is argued by many that there are too many additional effects presen! in lhe
fuH scale teS1S that are nOl in any way included in lhe analyses for 5uch an exercise to
be uscful. Thc rcsults are 3150 prone to large variations dcpending 00 lhe choice of
materi al parameters
adopled by the methods ..,F:allur. LO.d R.tlo

x
• ,
proponcnts. Previous Plnn." EI.Ulc

.-x"""
--
'h chul ...
work [22) suggests that
• C.ac; k lng EI.. uc
lhe result obtained from , xxxX
x
5uch a tWQ pinned xxxxx

analysis is dominantly ,
dependent on the X XX XX
o.• 'gP» XX XX X
assumed
strength
eompressive
which is x
,..-
o.•
notoriously diffieult to
aceess. Figure 12 shows
the variation of a the
o.•
o """"" o., o.• o.• o.•
assessment load , non
dimensionised with Figure 12 Variation of lhe predieted tes! bridge failure
respeel to lhe median loads wilh ellmulalive probabilit [22]
value, against the variation in lhe material properties for the two pinncd elastie,
meehanism and l-D elastie craeking analyses . The olher methods are more reliant on
lhe geomelry and weight which are betler determined and conform more to lraditional
views on lhe origin of the strength of tltese Slructurcs.

In the authors view lhe mechanism methods, and empirical equations

developed from lhem, should form lhe basis for lhe simplesl assessment tools. They
have a fundamenlal base in the actual fa ilure modes of the struclUres and with the
addilion af passive fill pressure and, much less importantly, malerial c rushing Ihey
provide very useful assessment lool s. A number of these models are available
commercially and a number of engineering consu!tancies have developed similar
software in-house.

The l -D elastic c racking models are much more complicated and only lwo are
currently available comme rcially. They provide lhe assessing engineer Wilh
significant additional infonnalion which must be treated with an appropriate degree of
scepticism. They provide useful information al serviceabilily and can mode! slow
moving loads. There is a significan! increased in vestment in lhe Iraining associated
with these pieces of software.

The "is no published work on lhe use of 2-D analysis for regular bridge
assessmellt, most of their work has been either for research purposes o r for bridges
lhat posed particular problems. The economic realities of lhe number of bridges
requiring assessment suggests that these types of models with remain relatively
 T. H. HUGHES I Masonry arch bridges 83

uncommon in general use. The investment associated with the growth of new
masonry arches may change this.

5.0 CONCLUSIONS

There are a number of general conc1usions that can be drawn from lhe work detailed
in this paper.

i) Ma,onry arche' continue to play an important role in the tran 'port

infrastructure.

ii) There has been a very significant investment into the research of these
structures. The experimental work has been significant and has been used to aid the
development of numerical tools and to validate assessment methods.

iii) The inveslment into the development of numerical analyses has been less well
co-ordinatcd and has resultcd in a number of differenl approaches. Whil sl lhe
analyses have provided additional insight into lhe structures behaviour there has
unfortunately been little take up of the more advanced nurnerieal tools by industry.

iv) The experimental and some analytical researeh work is moving more towards
serviceabilily.

v) Development of better masonry models for use in areh struetures is

continuing.

ACKNOWLEDGEMENT

This paper eontain s referenees lO the work of many mueh other researehers. The
work of my researeh student Martin Baker and researeh assistants Robyn Pritchard
and Paul Taunton under contracts [rom lhe Engineering and Physical Sciences
Research Council contraclS GR/ K04262 and GR/K76296 i, gratefully acknowledged.

REFERENCES

1. BRIDLE, RJ. - The prcservation of the bridge stock, Cardiff University

Centenary Conference, 1983.
2. HEYMAN,l -"The masotlry arch" 10hn Wiley, Chicheste r 1982.
3. PAGE, 1. - "Masollry Are" Bridges· TRL State of the Art Review, HMSO, 1993.
4. C. MELBOURNE (EdilOr). Proceedings o[ lhe Firsl JnlemOliona/ Co,,[erence 01/
Are" Bridges, Bolton, UK, Sept, 1995.
5. ROYLES, R. and HENDRY, A.W. - Model teSl' on ma,onry arche" Proc. In s!.
Civ. Engnrs, Part 2, June 299-32 1, 1990.
6. MELBOURNE, C. and GILBERT, M. - A Sludy of the effeelS of ring ,eparation
on the load carrying capacity of masonry arch bridges, 2nd Inl. Conf on Bridge
Management, University of Surrey, UK, 1993
7. PIPPARD, A.l.S and CHITTY,L. Repeated load tests on a voussior arch, } In st.
Civ. Engrs. Pape r 5268,79-86, 1942
84 STRUCTURAL ANALYSIS OF HISTORICAL CONSTR UCTIONS

8. CHOO, 8.5. and HOGG, V. -Determination af the Serviceability Limit State for
Masonry Arch Bridges, Proc. 1st Int. Conf. on Arch Bridges, Boltoo, UK, 529-
536, 1995.
9. BROWN, G. , PRETLOVE, AJ. , ELLlCK, J.C.A. HOGG, V. and CHOO, B.S.
- Changes in lhe dynamic charactcristics af a masonry arch bridge subjected to
monotonic loading to failure, Proc. 1st Int. Conf. on Arch Bridges, 8olton, UK,
575-383, 1995.
10. MELBOURNE, C. BEGIMIGIL, M. and WEEKES , L. - The load yleSl lo
collapse af a 5 m span brickwork fiat arch barrei, Prac. 1st InL Conf. on Arch
Bridges, Bollon, UK, 397-406, 1995.
11. DEPARTMENT OF TRANSPORT. The assessmellt of higlllvay bridges alld
struClUres . Dcpart ment af Transport, London, Departmen t Standard 8D21 /93
1993.
12. SUMON, S.K. and RICKETTS, N. - Repair and strenglh of masonry areh
bridges, Proc. ISllnl. Conf. on Areh Bridges, Bolton, UK, 501-508,1995.
13. CHOO, B.S. PEASON, C.H. and GONG, N.G. - Relative strength of repaired
areh bridges, Proe. ISllnl. Conf. on Areh Bridges, Bolton, UK, 579-588, 1995.
14. GARRITY, S.W. - Retro - reinforcement - A propose~ repair system for masonry
arch bridges, Proc. 1st Int. Conf. on Arch Bridges, Bolton, UK, 557-566,1995.
15. MELBOURNE, C. AND WAGSTAFF, M. - Load tests to eollapse of three large
scale multi-span brickwork arch bridges. 2nd lnt. Conf. on Bridge Managemen t,
Surrey 1993.
16. DAVI ES, M.C .R. HUGHES, T.G. and TAUNTON, P. - Considerations in lhe
small scale modelling of masonry arch bridges - Proc. 1st Tnl. Conf. 011 Arch
Bridges, Bol ton , UK, 365-374, 1995.
17. TOWLER, K.D.S. - Applications of non-linear finite element codes to masonry
arches, Proceedings of the 2nd international conference on civil and structural
engineering eomputing. VaI. 2, pp 197-202 , 1985.
18. CRISFIELD, M.A. - Finite element and mechanism methods for the analysis of
masonry and brickwork arches, TRRL Research Repor! 19, Transport and Road
Research Laboratory, Crowlhorne, Berks, 1985.
19. CRISFIELD, M.A. and WILLS, J. - Non- linear analysis of conerete and
masonry structurcs., in, Finite element methods fro non-linear problems, ed.
P.G. Bergan et ai, Springer-Verlag, 639-652, 1986.
20. BRIDLE, R.J. and HUGHES, T.G. ·- An energy method for areh bridge analysis,
Proe. Inst. Civ. Engrs., Part 2,89,375-385,1990
21. HUGHES, r .G. and BRIDLE. R.J ., "The innuence of certain paramclers on the
strength ofmasonry arches", British Masonry Society Prac. 4, 58-6 1, 1990.
22. HUGHES, T.G. and BLACKLER, M.J . , "Sensitivity Analysis of current assessment
techniques" , Proc Int Sym. on Computer Methods in Structural Masonry, BU Int. ,
Swansea, 207-215, 1991.
23. CHOO, B.S. , COUTIE, M.G. and GONG, N.G. - The application of the flnite
element method to the study of cracking in masonry arch bridges. Proc . Int.
Conf. on applied stress analysis, Nottin gham, UK. 476-485, 1990.
24. CHOO, B.S. , COUTIE, M.G. and GONG, N.G. - Finite element analysis of
masonry arch bridges using tapered elemcnts. Proceedings of lhe in stitution of
civil engineers, Part 2, pp 755-770, 1991.
 T H. HUG HES I Masonry arch bridges 85

25. CARDIFF TRRL ASSESSMENT PACKAGE (CTAP) Assessment Manual,

University of Wales, Cardiff, 1991.
26. LOO, Y .• Collapse load analysis of masonry arch bridges, Proc. 1st rnL Conf.
on Arch Bridges, Bolton, UK, 167- 174, 1995 .
27. LOO, Y. and YANG, Y. . Cracking and failure analysi~ of masonry arch
bridges. Journal of Structural engineering, ASCE. Vol. 117, No 6, 1991.
28. BAKER, M. - The finite element analysis of masonry with application to masonry
arch bridges and tunnels. PhD thesis, Cardiff University of Wales, 1995.
29. CHOO, B.S. and GONG, N.G. - Effects af diagonal cracks in masonry arch
bridges. Proc. of the Centenary Year Bridge Conference, Cardiff, 205-210,
1994.
30. PANDE,G.N., L1ANG, LX and MIDDLETON, l .. Equivalent Elastic Moduli
for Brick Masonry. Computers and Geotechnics, 8, 243-265, 1989.
31. PANDE,G.N., L1ANG, LX and MIDDLETON , l .. - Homogenisation techniques
for masonry structures and its application to rch bridge analysis. Proc. 4th Int.
Masonry Conf., Londoon 1995.
32. HUGHES, TG., "Modelling the effect cf initial slress state on the serviceability of
masonry arches", Proc Int Sym. on Computer Methods in Structural Masonry,
Swansea, 50-59, 1993 .
33. HUGHES, T.G., PRITCHARD,R. and BAKER, M.G., "The determination of in-situ
stress in masonry rings" , Ccntenary Bridge Conference, Cardiff, 199-204, 1994
34. ABDUNUR,C. - Direct assessment and monitoring of stresses and mechanical
properties in masonry arch bridges, Prec. 1st Inr. Conf. on Arch Bridges, Bolton,
UK, 327-335 , 1995.
 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS
P. Roca, j.L González, AR. Marí and E. Onate (Eds.)
© CIMNE, Barcelona 1996

CHARACIERIZATION OF THE MECHANICAL BEHAVIOUR

OFMASONRY
C. Molins Borrell
E.T.S. d'Engillyers de Camills, Cmw/s i Ports de Barcelolla
Ulliversitat Politecllica de Cata/rmya
Spain

1.- INT R ODUC TIQ N

For ccnhu'ies, the two main const1'uction materiaIs \Vere masonry a.nel timbc1'.
However, beeause of masonry 's dU1"ability, Hlost historieal eonstructions still
standing are huilt of this mat.erial.

Tlw stl'uctural analysis of thesc const1'uctious 5ho\Vs some specific difficuItics

derived from its own internaI Il;:Lture as well (lS the divel'sity of loads excrted
O\le1" the yea rs. Both aspects ha ve to be considcred in a detailed analysis of thcse
stI·ucturcs. The prcsent wo1'k deaIs with the mechanical behavioul' oi" masonry
anel tlw con5titl1tive equations for 1'epresenting it.

Masonry is a materiüI composed of units sucb as bricks, blocks: ashla1's ,

or adobes , and characterized by being mallufacturecl according to a svecific
geonwtrical order. In ancien t ma.sonry, I1n its are usua.lly joined by lime mort.ar.
The huge numbpr of possib le cOll1binations generated by the gcomet.ry, nature
auu ilnijllgl'lIl\'lIt uf 1II1it.~ <.uH] t.he cha.l"a.ct.(~IÜLic~ of the 1ll0rL<lr ca~t~ doubt
on the suitabilit.y of t.he generic tenn "masonry" (Figure 1). Nevertheless, its
mechaIlical bf'haviour has one common feature: vcry low tensile strength. This
propert.y is so important that it. determines the behaviour Df masonry structmes.

Most resem'ch carried Qut in recent decades has been concerned with brick anel
concret.e block ma sonry dllf' to their relevance for building designo This grcater
interest 5ho\\'n by research is reflected iu the followin g sections. Nevertheless,
most of wlwt \ViII hc de5cribed hereafter can be applied to auy lUasonry
composed of regular hexahcdral units joined with mortal'.
 C. MOLlNS BORRELL I Mcchanical bchaviour of Illa$onry 87

RU88LE MASONRY ASHLAR MASONRY

8RICK IAASONRY

Figure 1 - Diffcl'cnt kinds of masonI'y classified by PIET-70 (1970 ).

2 .- ST R ESS-S T RA I N B E H AVI OUn.

2.1 Facto r s a ffecti ng st r ess-s train b e h aviour

Gcnerally, it is ass umed that the most importallt factors affectiug stress-stl'ain
anel strengt lt behaviour of ma sonry are (Hendry. 1990):
From the uni ts: st reugth , type anel geometry (solid. p erforated , etc.) anel
absorpt ion.
From the mortal': strengt h , rela tive deforma tion charac teristi cs anel relative
t hi ckncss .
• From the mas onry: bund between uuits and morta r , directian of stressing
(anisotropy) and lo cal stress raisers.

It is well known tha t mor tar bed joints act as planes af weakness inside the
masonry. Moreover, biaxial anisotropi c behaviour af \Vall s ill their own plane is
determined by the exi stcnce of t hese plan es of weakness . In other word s, the
stress-strain characteristi cs of masonry chaIlge with t he anglc between the bed
joints and the main loading clirections.
 88 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCT IONS

-----------------------·Oõ
--_- ....--_

Figure 2 - Uniaxial stress-strain curve in compressioll, perpendicular to

bed joilltS.

2.2 U niax ial b e haviour

Ustlally, the component mateI"i als of masonry exhibit very di stinct mechanicill
p roper ties . Nonnall y, units are stiffer t han mortal'. This differcn ce is more
pronounced in an cient m asonry, built with lime mort al" . \·Vhen masonry is
loaded in uniaxial compressiol1 , the mortal' of the bed joints tends, due to
its lesser stiflness, to expaml laterally more t han u ni ts . The hi gher stiffncss
af th e units does noi allow thi s lateral extension of the mortar , confining it .
As a COllscquence, the units rea ch a uniaxial OI' biaxial tensile sta t.c of stress
in a plane parallel to the bed joints . According to t hi s, uniaxial campressivc
failur e af masonry ini tiates through vertical cra cking of t he units . Most uniaxial
compression tests find that t he compressive strengtli af mas onry is higher than
the st rength of mortar , due to its confined sta tc.

It is a.ccepted t hat the stress-strain rc1a tionship of rnas onry und e!" unia.xi al
compression resembles <ln elastoplas ti c Olle (Figure 2). EC-6 ( 1992 ) p rovicles a
parabolic rectangular st ress-strain curve (Figure 3), for ultima te li m it stat e
design of masonry sections . The m a in parameter s of t hi s relationshi p are
campressive strength (17m ), penk stra.in (c m = 0.002) ancl l.l ltima tc straiIl (c" =
0.0035) . These parameters are used as for con crete designo

EC -6 also a llows, like ot he r European design st. and m·ds, to app roach t he secant
modulu s af cleformatio n for masonry (E) by 1000! ", where h is the cba racteri stic
com press ive st rength af ma sonry. FOI" the shear moclulu s (G), EC -6 provides
 C. MOLl NS BORRELL I Mechanical behaviour of rnasonry 89

(J

I, ____________________. ___~-~
- _-----~

0 .002 0 .0035 €

Figure 3 - Icleali zecl st ress-st rain relatiol1ship in comprcssion acco rcling to

EC-6 (1 992 ).

OAE , which is equiva lent to ass uming a Poisson ratio (11) of 0.25. Th ese values,
resulting frum com press ion test ing of sa mples in the direc ti on normal to the bed
joints. are appropriate for structural elements such as vertically loaded waUs 01'
arches, in ",hich the loacl is actua lly normal to the becl joints.

There are two differcnt ways to estima te the deformation l'wdulus of ma sonry.
First.. the empírica I relationships betwcen st rength and defonniltion , ,vhich have
been adopted in most des igll codes (e.g., FL-90 , EC-6 ). Second , by models baseei
on t he strain characteristics of t he com ponents (units anel mortar ). Aelvantages
of composite material mo deis are their ",ide!' range of appli cab ili ty, due to
considerat ion of the actual dimensions of uni ts anel joinrs, anel thei!' elastic
characteristics.

An attempt to estimate the eleforma tion moelulus on t he basis of the properties

of the components was made by Brooks (1986 ). This autho!' proYides the
expression
1 byCA w 1fI y(C+ 1)
(I)
E H (E~y A ~+E71\ A 71\)+ HEm
wltere by is the dep th of th.e unit , C is the number of courses . J-/ is the height. A" .
ih anel A m are the cross-sectional areas of ma SOl1ry, ul1its, anel perp encl mor tal'
joints respectively, Eby and Em are the effectivc mocluli of the unit between bed
fa ces and of t he mortar , and m y is the thickness of the mortal'.

The above expression results in the addition of the axial ficxi bi lity of the two
component materiaIs. Brooks (1986) verifieel the relationship (1) for maSOlH'~'
piers made of elay bricks anel oth er s made of conCl'ete blo cks. Nen' rtlw less.
as Anthoine (1992 ) points out , t he prediction of t he mechanical prop f'n i ('~ of
90 STRUCTURAL ANALYSIS OF HISTOR ICAL CONSTRUCT IONS

mcl sOllry hom t. he geome try ônd elast ic proprr t ies of ulli b illl.d 1l1Ort il l" nwy ))('
iu appropriate in the case of insufficient knowledge of tIl(> COIll!H)UPllt prop (·rti f's .

S<ll11<lr<l sing lte ct aI. (1982) ohtainf'cl very inte rC'stiug rc.s ults by t ('s ting elay
bri ck ma SOlll'y square pônels wi t h 150 mm sides undf'r uniaxial com p ressioll
varying hed .i oint. ori cn tation to the applied load. Fi?;ur e 4 illust rates the grea t
inHucnce of b ecI joint ori enta t ioll 011 st rength as well as stiffness , showing a
highly n onlinear b cha.viour . Thc pcak strain values are st wn gly depC'lldcll t OH
t he bed joiut orientation , from (li = 0.0003 for O = 67.5" to Cu = 0.00'1 for 0 = n".

20

'~-----r'O'-----,"r-----'3mO------'"

Figure 4 - Uniaxial compress ive st.ress-s train curves af paneIs varying bed
joint orient a.tion (Samarasinghe et aI. , 1982).

Regarding the cyclic compressive behaviour of ma sonry, Naraine & Sinha. (1989,
19913 , 1991 b ) observed tha.t the envelope of peak strcngt h points approxima.tely
coincides wi th the monotonic stress-strain curve for clay brick masonry. These
a.uthors formulat.ed a geomctric model for the unia xial cyclic behaviour of
brick masonry (Naraine & Sinha , 1991b) , consist ing of a. group of focal points
from whi ch it is possible to completely reprodu ce the behaviour of ma.sonry
under cyclic loading parallel and perpen dicular to b ed joint.s. The test results
necded to determine t his model \Vere carricd oui. 011 low-strength brick masonry
speClmens.
 C. MOLlNS BORRELL I Mcchanical bchaviour ar masonry 91

2.3 U niax ia l co mpressive stre ngth

The e\"aluation of the uniaxial compressive strength of ma sonr)' ha s been an

important goal for many resea rcher s.

As ri con.seq\lPIH'P of t.he' comprehensive experimental stuclics llndertaken over

recent decad es to characterize the ultimüte strcllgth of ma sonry unclcr uniaxial

compression , nowadays there are some empirical rdat.ionships for it. Some
of these empi rica l relationships are compi lcd in the design cudes (EC-6).
Generally, thcsc formu las provi de the compress ion strengtlt of masonry 011 the
basis of the mechanical properties of ullits anti mortal" , ta king iuto account
their geomctric arrangement (bed joint thickness). For exa mplc. to ca1culate
the characterist ic compressive strength of masollry (/I,) in lhe c"ent of lacking
experimental results. and for bed joiot thickness cOInprised bctween 10 and 15
mm , EC-6 ( 1992) provides the following expression
(2)

where /b and 'm are Lhe average compressl\"(' strength of units and mor tal"
respect ively (in MPa) and K, o and {J are constants whi ch al"e not yet dcfinitively
fixed ( EC-6 suggests (l = 0.75. 8 = 0.25 and [\' = 0.41.:, whcl"e 1. ' = (,;)0:1:1:s I.:)
for un its af h < 15M Pa }}!ld mortal" strength lIot st ronger t hall ),110, othcrwisf'
!./J :;; I). Othel" des ign codes, such as FL-90 , PIET-70 and the It,tlian co de (1987 ),
provide tables to calcu la te the strength on the basis of the pmamet el"s in\"ol"f'd.
An extens ive account of empirical expressions for the compressive strcngth of
masonry can bc found in the hooks by Tassios (1988 ) anel Hcndry (1990).

2.4 Uniaxia l co mpr ess ion failur e crit e rion

Two main groups af approximations ha\"e been de"eloped for compressive failure
criteria fo r m<lsonry, 011f' based 011 elastic stress distributioll s and t hr other 011
il!'''umed biaxial Ol" t riaxial st ress distributions usi ng limit a nalysis.

Hendry (1990 ) pro\"ides an cxteosive accollnt of the main developments ma de

O!1the basis of elastic stress distribu tions in the componcnts. In thi s section ,
beca use of their rele\"ClI1ce. onl~' limit analysis based criteria are treated.

Hil sdorf (1969) estab li shed perhaps the first thcoret. ical failure criteria for
masonry under uniaxial compressioll on the basis of strength proper ties of
componen ts. He assumed that the mortHr (lf the bed joints is kept under a
s ti'lte af trii.lxial compressioll while the unit.s are under "ertical compress ion and
horizontal tCl1siOI1. i.lnd the stress di stributions are unifol"m (Figure 5). Th ese
uniform s trf>SS dist ributiolls do Ilot verify the equilibrium at t he edges.
 92 STRUCTURAL ANALYSIS OF H ISTORICAL CONSTRUCTIONS

a,
unit

bed mortor joint

Fi gure 5 - Masonry prism under uniaxial compression (a) anel the stress
distributions in units anel hed joints (b).

The cquilibrium conditions betwcen mortar anel units are (Figure 5):

(3)

(4)

where ;'0 is t he height af the unit anel 11 m

is the thickness af the mortar joints.
If O' is the ratio betwecn li'" anel h b , then (1"'" = _(To / a:. As a. gencralization, the
geometry of brickwork auel blockwork fulfills < 1. (t

If the failure niterion for units anel mortal' is af the Mohr-Coulomb type, the
following expressions can be writteu,

(5)

(6)

where 0"1 anel 0'3 a re the principal stresscs anel 17T < (f"c are the tensile anel
compressive strengths af the matcl"ials. UsuaUy, the compressive strength af
the units is higher than that af the mortal' (ct > "'2).

According to thc lower-bound theorem of limit analysis 1 ".: stress verifying

eqllilibrillm conditions (4) and failure cri teria (5) and (6) is a lower bound
of the ultimate load. The two last expressions (5) and (6) can be rewritten for
t.he limit couditioIl as
(7)
 C. MOLlNS BORRELL I Mechanical bchaviour of rnasQnry 93

(8)

Finally, by isolating o., the failure stress is obtained:

where (9)

Although the resulting modellooks quite sim pIe, it shows certain features such
as an increase in the ma son ry compressive strength with: increasing unit and
morta r compressive strength and unit tensile strength , and with decreasing
relative t hi ck ness of the bed joints. However , t he predi ct ions provided by (9)
are in disagreement with test results. Hilsdorf suggests th.'It the Mohr-Coulomb
criterion is inappropriate for reproducing failure of units and mortaL In fact ,
Hendry (1990) rcfers to some works that point out the dev iation introduced by
Mohr-Coulomb cri teria in the case of elay bri ck.

Following the sa mc direction of Hilsdorf, Biolzi (1988) proposes a more complex

developrnent that takes into account t he true cond itions at the (rce edges and
the nonunifonn s tress di st ributions. Nevertheless, numerica l implementation is
rcqu ired to evaluate the st rengt h from the res ulting expressions.

AlI the above approaches 'lre ha sed 00 limit analysis. assuming a perfect plastic
behaviour of materiais. Another kind of development employed in order to
determine the uniaxial compressive strength of masonry assumes a perfectly
brittle behaviour of both materiais. unit s and mortar , and imposes hori zonl <l l
strain compatibility at the unit-mortar interfa ce. Au thors sllch as Tassios ( 1988)
and Atkin son & Nol and (1983) uscd the sa me llniform di srribu t ion of triaxial
stre:;scs as Hilsdorf (1969) to eva luate the failure 10ad as df'scribcd below,

lf units and morta l' have an e!astic aod isotropic behaviour , the horizontal
strains from ex pressions (3) and (4) are) respectively.

qb Vb
O .. 1JbOb
, . =O'-
Eh
-+E6
-+ -
O'E
(11)
b

",here Em, V m , Eh and Vb are lhe Young modulus and the Poisson's ratio of mortal'
and unit s respecti'-ely_ and (m and (6 are lateral strains in the two materiais.
The compatibility of the hori zontal straios gives an expression relating YCrtiCiJI
compression (o.) and horizontal tension in the units (Ob):

( 12)
94 STRucrURAL ANALY$I$ OF HI$TOR ICAL CONSTRUCTIONS

Tassi os ( 988), <-LIso u:,; illg the Mohr-Coulomb fnihlH' criteria (6) for the units ,
f01lnd t he following ex pression fo r uniaxi,J\ co mpressive strcngt h:

11~.
( 13)
",- ol,,~ f:. v .E ... )

lu equa t ion (13). the compressivc strcllgth does not depenei 011 t h e compressive
st.rcngth of tlw rnor tar. In fact , Atkin sOll (1983) observed that, according to
t1li5 modd , failure iui t.iates in the com prcssf'd mortaL Because of this , he
p roposcd ca1cula ting lateral tensile stresscs iu lhe units (O'b ) taking into a ccount
t he actual nonlinear behaviour a f mortal' iu com pressioll and an e1astic perfect
brittk behavioul' for the unils (Figure fi). Thc reslllting increment ai expression
IS

( 11 )

a,
Unear mortor
Nontineo r mortor

E
. I
\1
,
'-I
-
,
~

a' Ten sion Compression

Figure 6 . Evolution of lateral str<'ss in units anel mortal" considcrin,e; linear

and nonlinear behaviour of mortal' (McN ary, 1985),

McN ary (1985) vCl'ified experimenta lly cxprcssion (14) 011 brick\Vork sa mples
madc of different kinds of mortal" and cla.y bricks, To dete. .rminc the bchaviour
af the componcnts, tria..óal comprcssion tcs ts on mortal' anel biaxial tension-
compression tests on bricks \Vere carried out, In general, calc\llated strengths
and strains \Vere 30-40% lawer thau experimental anes. The authar suggests
lhe inap propriateness af the uniform strain distribution employed in the model
 C. MOLlNS BORHELL I Mcchanicul bchuviour of Illusonry 95

to ex plain t hi s system atic difference observed. 111 tlU' same way, lI(' points OUI
that lIonlincar beha viour of ma sonry in compress ion depend s aImost compl f' t f' I~'
DO the softcning behaviour of mortar.

2.5 U niax ia l t e ns ile str e n g th

In ma sonry shuctural elemems. tensile stresses are usually dHe to loading

ecccnt ricities and / or acting shear forces. It is well knowll that ma sonry exhibi1.s
lo\\" res is tance to tension du e to the joints. whi ch act as planes of wea kness. In
fa t.:!. the tensile strength of ma sonry is d etennined by the bond strength bet\\"een
morta r and units. Some researchers have studi eJ thf' origin autl thc strcllgtli
of thi s bond. Hendry ( 1990 ) refers to some works lhat indi cate Lhe effects of
the moi sture content of bricks, the fin es contem of sand. the pore sizc of brick
material and the suction of the clay bri ck, and the type of mortar. In the same
\\"ay. this author points out that. <l ccording to recent research. the boml between
elay bricks and cemellt morti.ll"S is mechani cal in nature , although with other
mat eriais there may be chemical bonding as wcll.

2.u Bia xia l b e h avi o ur

The detailed biaxial behaviour af masanry presents many challenging difficulti es

due to its highly nonlinear and ani sotropic behaviour. This anisotropic
beha\'iour ari ses for high stress \"i.llues. The mos t thocough tes ting program
ou hiaxial behaviour wa s carried out by Puge (1981). This <lUthor tes ted 180
on e sixth scale brickwork pancls. The main rcsult of thi s testing prognnl1
\\"a s hi s well· known bii.lxial failurc envelopes (see 2.8.1 ). Panel strains \Vere
measured simultaneously until values af load closed to failure when m.easuring
instruments wel'e remo\"cd from their place. The stress·s train relationships
obtained from these tests ( Figure 7) show the highly nonlinear hehaviour of
nHl sonr~· ullder biaxial compress i\"e stresses. while in tension· compression the
stress·strain curves are almost linear until failure.

Homogenization techniques de\"cloped for layered composite materiais have also

been used for the evaluation of the mechanical properties of masonry fcom the
known mechanical and geometri c characteri sti cs of the components (units and
mortar ) . These techniques have become a very useful tool for evaluating elasti c
orthotropic properties of ma sonry. Some works on this subject are described in
the numerical macromodclling section.

Narainc &:: Sinha (1991a ) also studied the biaxial compression behaviour of
96 STRUCTURAL ANALYSIS OF HI STOR ICAL CONSTRUCTlONS

• ""_1 .U ...·... ,..I 'tnl,

• ..... U.I .t'....·'.r.II....... ,..
... a.. .•u ... ·S. . . , Sual,
..
•
..
..
(a) - .. - - - - Stratn X 10....

~r-----------------------------,
• N.r. . 1 Su ••• · No ... 1I Su.l"
PUloU •• 5tr. . . · ' ••• II •• Sual"
'''lo •• Su ••• · S ..... Ih r.t.

· " . ..;,--------._;,--------...~--------...~
..r<-~.------,
(6) Stratn. X 10-&

Figure 7 - q - ( curves measured in Illa sonry pancls under biaxial

compression and tension-compression loading (Dhanasckar et
aI. , 1985a).

masonry undeI' cyclic loading. Although t heir work was restri d ed to the load
C<l.ses fnll'allel and norma l to the bed joints (O :;::: O", O= 90°), for several ratios of
principal stresses (O, 0.2 , 0.6 , LO, 1.67, 5, 00) their result s are very usefu l for
dctermining Lhe maximu m stress leveI for brick masonry under cyclic biaxial
compression loadiug.
 C. MOLlNS BORRELL I Mechanical behaviour of rnasonry 97

2 .7 Biaxial strengt h

The treatment of design codes of the strength under bia.xial states of stresses is
usually simplified by treating separately the normal and shear strengths at the
joints. In other word s, on the bases of the shear and axial acting for ces . For this
purpose EC·6 only takes into account characteristic shear strength (j"J.) parallf'1
to bed joints and provides two ways for evaluating it : by testing masoury panels
(macroelements ) or, analytically, using the express ion
( 15)

where 1~J.o is t he shear stress strength in the absence of normal st resses (Ud = O),
Ud is the lowest vertical comp ress ioll act ing on the wall and h is the llniaxial
verti cal com pressive strengtb of the ullÍt si I".. is also limited by certain tabulated
\"alues.

Equation (15), corresponding to a Mohr· Colllomb failure cri terion , ha s been

llsed by m<l.ny researchers and adopted by most European design codes.
Hendry's book (1990 ) provides a comprehens ive list of the values used by
different authors for t he cohesion (J~ o) and the internaI fri ction term (11 )
depending 011 the components uscd. T he cohes ion term (1"0) is strongly
dependent 011 the shear strength of the interface between unit and mortar.

2,8 Biaxial failur e cri teria

The abo\'e mentioned treatmcllt of the biaxial strength of masonry is adequate

for design purposes but it does fiOt form a complete failure criteri on for the
analysis of stru ctural elements under biaxial stress states.

\Vhile the biaxial behaviour of masonry before cracking is almost isotropi c.

biaxial strength strongly depend s on the orientation bctween princi pa.l stresses
and bed joints, This highly anisotropi c behaviour is due mainly to the geometric
distribution of the components. accentua ted by the low strength of mortar in
comparison with units. In addition, this anisotropy wiU be more pronounced if
the units are also anisot ropic due to , for example, perforations.

Generall y, biaxial failure cri teria are defined as 3D surfa ces where the
dimensions are form ed by the following variables UI , U 2, 9 or U ", u p , r according
to t he reference system used , where (TI and U2 are principal stresses and O is the
orientation of the bed joint related to the principal st rcss direction , and 0'" and
(1p are the normal and parallel st resses alld r is the tangential stress to the bed

joints (Figure 8).

 98 STRUCTURAL ANALYSIS or HISTORICAL CONSTRUCTI ONS

a
T b

Figure S - R efercnce system s used fo r defi ning biaxial failurc cri teria for
masonry.

Magenes (1992) and Anthoine (1992) distinguish four difl"erent types af biaxial
failure criteria according to the methodology used in its formu lat ion . These are:
• Expcl'imentCll. Cri tcl'ia obtained as a result af a. com prchensive testing progr am
of maso nry panels.
• Phenomenological. Cri teria based 01'1 a pnori mechanical interpretation of the
failurc, derivcd from testing observations.
• Empi ri ca l. Cri teria obtained by fitt ing fai lure surfaces fmm the avai lable test
results .
• Theoretical. Failure cri teria. deduced fram the geornctric and rnechanica l
properties of the compoucnts (ullits anel morta r and unit,-mmtar interface ).
The interest of this cri teI'ion resides in its imm ediate numerical implemcntation
in finite elem.ent analysis pl'Ograms,

2.8.1 Experimenta l c rite ria

Few com prehensive bi axial failure cri teria have been obtai llcd expcrimentally
due to the large amount of tcsts rcquired . Tests 0 11 m<ls onry panels under a
macl'Oscopic uniform stl'CSS statc are the most widely llscd in defining failurc
cri teria , beca use the maximum measurcd loads coincide with failurc strength in
terms of principal strcsses.

Page (1981) carricd Ollt aIl extensive testing program on square , half scale
brickwork panels under biaxial com press ion and tension-compression SLrCSSCS,
va rying loading directions, In biaxüd compI'cssion two difl'erent kinds of failurc
 C. MOLlNS BORRELL I Mechanical beh:lviour of Ill:lsonry 99

\\'t're observed depending on the ratio between principal stresses. Thus ) in most
tests failure occurred by splitting af a plane parallel to the paneI. Neverthcless.
when aue principal stress was preyailing. failure was reached in a plane normal
to the panel by cracking and sliding of the joints 01' in a combined mechanism
im'olying bricks and morta r depending OH the bed joint orientation (O in Figure
8). From Lhis fad h e deduced that Lhe inftuence of angle O was significant
unly when one of the principal 5Lresses predominated. He al so pointed ouL
that uniaxial compress iye strength is a low bound of the biaxial compress ive
strength. Figures 9 and 10 sho\\' the biaxial compressiye failure envelope that
was obtained by Page.

Figure 9 - Biaxial compressive failure envelope in terms of {T].(T:t.O ( Page.

1981 ).

In the tension-compression tes ts, Page ( 1983) found that the inftuellce af the
b ed jOillt orientation on the strength was decisive. In fact , there i5 a pronoun ccd
decrcase in strength when the principal st ress direction approaches lhe normal
direction of the bed joints. In addition, he observed that the failure was in
fhe interface between or in combined mechani sm s im'oh-ing mortal' and hricks.
Brittle failure of paneIs was observed in alI tests.

Dhanasekal' et a!. (1985b) developed an empiri cal criterion con5istillg of an

analytical surface fitting Page's failure envelope in the space {O'".O'p.r}.

Samarasinghe &: Hendry ( 1980 ) also tested oue sixth scale brick ma sonry p(lllf'l~
 100 STR UCTURAL ANA LYS IS OF HISTORICAL CONSTRUCTIONS

Figure 10 - Biaxial compressive fadare snrface proj ect.ed Oll to plallc {(Tl , (1 2}

(Page, 1981 ).
under a biaxial tension-compression st ress state obtaining a failure envelope
t hat , like the one by Page, emphasized the significa nt influcnce af the bed joint
orientation and the tencieney af the envelope to adopt a COIlCHYC shape.

It 1S interest ing to note t hat , in t hese tension-compression te5t5, ten sile strength
ha s a maximum for average va lues of the compression acting in the other
principal direction. fi gure 11 , showing failure mechanisms for difl"crent stl"CSS
sta tes and bed joint orienta ti on, di sp lays the differcnce bet\Vcen tensi le failure
in parallel dircction to bed joints with ar without normal compress ive stresses.
\~l ithout normal compression , failure oecurs by splitting af vert ical joints in
tension and shear in thc beci joints. Ir the lattcr are compressed, shear strength
increases and fail ure occurs by spli tting in vertical joints and cracki ng of unit s.

Samarasinghe & Hendry (1980) fitted a surface equation for thc failure envelope,
expressed by
UI :;: 0. 7e-o.I~.,., - 1.:14 E..
, - 0.02 C" MP. (16)

where 0"1 ,0"2 are the principal tensi le and com pressive stresses, respectively, and
eis the angle between 0"1 and the bed join ts (in radians). This su rface considers
all failure mechan isms in tension-eompression stress states.

Page (1980) simulated biaxial tension t.esting by means of a numerical micro-

lllodel based on the finite element met.hod, since this kind of test is especially
complexo The model assumes an indefin ite elastic behaviour for bri cks while bed
 C. MOLl NS BO RRELL I Mech::mical behaviour ar masanry 10 1

,,
"
.-
-11-='
~• -~-
22· S·

.,-

6 1· S·
-~-
90-
-11-
BI AXIAL
COMPRESSION
J.'

Figure 11 - Different fai lu re mechan isms for brick masonry panels under
biaxial stress statcs ( Page, 1983).

joints have li mited tensile and shea r strengt h. Rcsul ts showed the considerab lc
inft uence af () and the ratio between shear and tensi le strengt h af t he unit- morta r
interface.
102 STRUCTURAL ANALYS IS OF HISTORICAL CONSTR UCTIONS

2.8.2 Phe nomenological cri t e ria

Most allalytical formulations fOI" defining failure of masonry are bascd on

mechanical interpretations of the observed phenomena. in the experimental
tests. These fo rmu lations become diffcrent dependiug on the fa ilure hypotheses
cunsi d ered. In this sectio n, so me phenomenological cri teria of considerablc
interest are prcsentcd.

T he criterion of Turnsek & Cacovic (1970) \Vere developed in arder to evaluate

the strength of a waU panel under a xial and shea r Joads. This criterion assumes
that failurc is associated with a biaxial tension-compl'cssion state of stresses in
the centeI' of the \Va lI.

GenerallYl the failure of a panel under shea r loacis consists of the develapment
af diagonal cracks, parallel to the dircction of principa l compressions, and it
initiatcs dose to the centre of the pane!. Turnsck & Cacovic (1970) fonnulated
the hypothesis that failure initiates when t he tensile st.rength is reachcd in the
centre af the pane!.

In thi::; criterion, norm,.1I and tangential stresscs in the central section of thc
panel are calculated by means of beam theory. In this calculation, the behav iou r
of masoury is assumccl to be linear, elastic and isotropic.

According to it ,
O"m = N/A y T"'dr = 1.5 V/A (I7)

where 0"", is thc aver age compression in thc panel, Tm~r is the maXllllum
tangential stress in the \ValI, N and Vare the axial and shear forces, and A
is the surface of the crass scct ion of the \ValI . In the case in which the stress
parallel to the bed joints is null, thc principal stress in the centre af the panel
can be evaluated as
(18)

Solving T maz in (18) results in

(19)

Anel the horizontallo<ld tha.t initiates crack ing in the panel is

(20)
 C. MOLlNS BORRELL I Mcchanical behaviour of masonry 103

where f I" is the tensile s trength of the masonry. ~ote tha t the anisotropy of
the biaxial tensile strength of the masonry is not considered .

It was verified. by means of the finite element method l that stresses calculat~d
from beam theory are reliable for ratios between the height and the width (V
greater than 1.5. For lower ratios of tI stresses parallel to the bed joints have to
be taken iuto account and some errors are also iutroduced iuto the calculation
of Tma~' Turnsek & Shepp<nd (1980) propose approximtlting Tma r. for (t) ratios
lower than 1.5 in the following wa)': for (t) :s; 1 T ma.r = r, and for I < (t) < 1.5
by using a linea r in tcrpolation.

Yokel &: Fatta! ( 1!J76 ) st udied the different failurc mechanisms associated
with biaxi al tension-compression st ress states . Their criterion is based on lhe
definition af the failure for units and bed joints. In order to verify their proposal,
they carried out an experimcntal program on square masonry panels under
uniform distributed vertical load and an in crcasi ng diagonal point loa d.

They noticed th at, in order to predict splitting af bricks. a linea r combination af

principal for ces fittcd t he test results better than a preselected value for tensile
strengt h. For bed joint failure. they cmploycd a 1'\llohl'-Coulomb cri tcl'ion. The
stresses rcquired for this criterion fl.re macroscopic) provided by Cl linear elastic
cCllculation assumiug isotropic behaviour of th e masonr)'. B cca u ~e this model
works with ma croscopic 01' nominal strcsscs, panels ratheI' than components
must be tested to calibrnte the constants requircd.

The main cOlltrih1\tion of thi~ model was to take into account IIH' mlÍ sot ropy
induced by the bed joints.

::vIann & Mül!er l s (1982) criteri on incorporates some concf'pts of the beha\'iour
of masonry not included until then. Sta l'ting fmm the fact th,H mortal' is
uSllally much more deformable tha n bri cks. they consider that the latter are
like rigid bod ies in a mortal' ma trix t hat allows its relati\"e dis placement. In
this way, bri cks sho\\" rigid body mo\"ements and rotations inside lhe mortal'
matrix. If thc strcss s t llte is homogencolls (O"r , O"r'O"r,) and oU the bricks htl\"c
lhe same geomet ri cal chm'a ctel'istics, thesc all h,n-c the sfllue rot<ltion and t he
bed mor ta l' joints are st ressed (Figure 12). Then. an additionnl nOIluniform
stress di stribution appears in t he bed joints.

lvlann &:: Müller (1982) consider that verti cal jaints are not C<lpnble af
transmitt ing tangentia l st resses for three reasons: first. ycrtic,ll .ioints are
usuatly not cOll1pletely filled ; second. the existcIlce of adhcsion in tlH' wrtical
joints is very ulllikel~' beca use the)" are no t comprcssed : ando finally. I hat in the
 104 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTlONS

Figure 12 - Rigid bricks in a mortal' matl'ix, relative rotation of units and

deformation of mortar (M ann & Müller 1 1982 ).

_ _ 'ty _ _

+---~. ----J.

Figure 13 - Stress state af a brick accorditig to Mann & Müller (1982).

case of shear failure these joints have no residual strength .

With the hypothesis made (Figure 13), and assuming that. tangentia! stI"ess
distribution in the unit-mortar interface is uniform, in equilibriutn conditions,
results
(21 )

where 'XII is the taugential stress. In addition , the shear force at the centre of
the brick QII is
QlI = 2TC::.Y (22)

The authors defined three possible failure mechanisms:

a) Failure af beci joints. This occurs when failure shear strength is reached in
the less compressed (0-,,) part af the bed jaint (Figure 13). Mann &: Müller
 C. MOLlNS BORRELL I Mcchanic:l1 bchaviour or rnasonry 10,

associ<ltf> this failurf' with a steppC'eI crHcking pattl"rIl (l(TOS!' h('J Hml q:'rrical
joinh (wilhollr brick cracking).

" ) Sheflr-tensioll failure in the ullits. This O("("IlrS \\'hcu tlw 1Il1lXilDtllil prillóp,d
st res!-' at the ("<'lItre of the unit rCíH.:hes tellsih" stn'ngth.

c) Compression failure of ma so nry. Th is occurs \\"hen the Pilll(·j is unde1" high

compression st resscs anel the total vertical stress (Ily + .1t7 y ) f'xceeds th('
compressi\'e strf'ngth of ma so nry.

T
\ 11' "

" , .,
, -!"~,,+_,_o__"--'.~.~""''"''''-'_'_'__"__'_0-l~'"o'!!'" ~.. 'JJ G. \ 11'"
r,.... '..".

Figure 1-1 - railurf' em'dope for brick masomy_ from ~danll!.:. )' Iüll er (1 982 ).

T he em-dope defined by these tlm'c diffcrenr failme mcchanisms is ShOWll

iH Figure 14. Al so. this figure displays tlJ(' good agref'mf' nt achic\"cd with
experi mental stre-ngths obtail1cd in pallels IIlld(,1" lIlacroscopic stress ::;:tates. In
<tdd ition. til(> criterion cort"f'ct Iy predicts the t\\"o shear failurc mechanisms «) alld
b) associatf'd with joint and brick failure-_ rf' speet in'> I~·. Final1y. it is importallt to
emphasize that the ratio of vcrticêll and horizolltal dimensions of units ( .1J' . ó.y)
is a fundamental parameter of the shear behil\'iour of thc wall.

L<lter. Dialer (1991) experinwnt<llly \"erified that ).l;u1I1 &: )'!iiller·s model
underestimated the shear st rcngth of panels \\·hen \"ertical joints \Vere wf'1I
constructed OI" t,here \\'ere compressive stresscs parallel to the bed joint. Because
Qf this. Dialer ildded the shear strength of \"crtical joint.s to the criterion.

:\"owadays_ there is an open rcsea rch field in lhe application of homogenization

tcchniques to pcriodic materiais with nonlinea r beha\"iour such as masoIlry.
in order to obtain their constitutive rehüionships and biax ial strength.
Homogenization enables thc dctermination of the bl11k properties of a
compositc material from it:-; cOlllponent propertics. given that their geomctrical
dist ribution is kllo\\"n. Lourellço (1995a ) citf's some of the main studies in this
106 STRUCTURAL ANALYS IS or HISTORICAL CONSTRUCTlONS

subject.

3.- L ONG-TERM B EHAvrOU R

There are relatively fp\\, works on t,hp long-term behaviour a f masollry. It is

well known that this material clisplays movemenls throllgh time l1IHler sustained
loads. This bchaviour h as bccn tlH' subject. af experimental research programs
(Lenczner, 1981). Nevcrthelcss, most rcsearcllcrs , as EC -6 points out, coincide
in emphasizing the huge variability of sllch featurc a r the bchaviour af rnasonry.

Gcnerally, it is acceptcd that ela)' <\ml stonc units do not show significant Jong-
term dcformatioll, which is conccutrated in the mortal'. Thu s, the parameters
affecting long-term behaviour are: the geometry of thejoiuts, the chamdcristics
of the mortal' (e.g., watcrfcement fatio and typ e of eemcnt), the moisture
content of the components during constrnction anel the absorptinn of the units.

EC-6 provides a table of coefficients in meleI' to e:;timate the fin<'ll creep anel
shrillkage ueformations (Joo, anel (h) depellding on the uuit materi<ll. The final
creep co('fficient is elefineel as
(23)

where { ~ ... is the Rua l crcep strain , lfl = ulE is the short~term strain of the
masollry, anel ih.., is the long~te rm v<llue of the shrinkage st raÍn. Somet imes,
moi sturc deformatioIls céln be expansive for elay bri ckwork masonry.

Although FL~90 does Ilot provi de explicit. vallles of ereep eoefficients, the long~
term deformation modlllus (E) of bri ckwork including alI deformation (ioo),
s hort ~t ('}"m alld rheological, CHn be eval nated in the following WHy:

(24)

wherc /d is the clesign :;trcngth of masonry ancl (00 is the long-tcrm tota l stl"ain.
which depends OIl the natme of the brick anel the mortar used.

lt call be pasily proved that long-term strain values provieled by FL-90 are ver)'
similar t.o those provicled b.y EC-6 ( 1992).

4.- NUMERI C AL SI MUL AT I ON DF M ASON RY ST R UCTU RAL ELEM E NT S

The <}ctu<rl mechanical properties of masonry hindcr its assimi lation as an

elast.ic isotrapi c material. Becausc af this. aver the last two decades, sincc
 C. MOUNS BORRELL I Mcch::lI1ical bch.1viour of 1ll.lS0nry 107

the development of modern computing techniques based on the Finite Element

Method , specifically deyeloped models for the analysis of masol1l'y structures
have been the subj ect of research. These models takc into account some features
of their behavioul" su ch as actual tensile aud compress ive strength s, stiffness
dcgradation in tcn sion or shcar cracked zones, etc.

First , somf' nnidlll1cn!-;ioll<l1 models, develop ed for the analysis of structurcs

cOl1sisting af linear clemeuts, are presented. Thell , the I!"lost impo rtant biaxial
models developcd for ma sonry are di scussed in Section 4.2.

4. 1 U nidim e ns io n a l m o d e is

It is well kn own that the nonlincar beh;l\·iour exhibit f'd by ma sonr)"

constructions is due to its very low tensile strength. Because of thi s. ma sour)"
easil)" develops intense rracking under severa I act ions. Starting frol11 thi s
fa cto the eas iest wa)" to reprodu ce most of the no nlinf'ar bcha viour of this
material is to consider its limited ar null ten sile strengt h. Formulations for
limit analysis have a higher level of simplificati oll wheu consid ering, in additioll.
rigid behaviour iu compress iol1 (Heymau. 1982). Adoptcd compress ion strf'SS
di stributions cOlTespond to an clast.jc brittle behaviour or a perfect plastic on('
( Figure 15 ).

Figure 15 - Common adop ted stress dis i.ributiol1s for lirnit analysis
of unidimen s ional s truclurcs cOlls tru ctcd with 110 t cn s ion
materiais: elas ti c bri t tle and pcrfect plasti c.

PIET-70 and FL-90 consider a perfect plast ic beh<'lviour of ma ~onry for the
ultima te limit state desigll of sections under normal strcsses. although forces
have been obtained fwm a strl.lctural analysis that does 1I0t take inio account
thi s behaviour of ma sonr)".
 108 STRUCTURAL ANALYSIS QF HISTOR1CAL CON$TRUCTIONS

4.2 Two-dime nsional and three-dime ns ional mod e ls

There are two ma in groups of two-dimensionaI models for st.ructural anaIysis

of masonry: micromodels anel ma.cromodels. An1.hoine (1992) distinguishes
auother group caJled micro-macl"O-models for 1.hose developecl by means of
hOIllogeni:.-.;a 1.ioll tedmiqucs of perio uically layer·cu lIlilLeI"i ab.

Its main a.pp lica1.ion is, obviously, the analysis of two-dimensional structural
clements such as walls and vaul1.s.

4.2.1 Micromodelling

Micl"Omodelling consis1.s of simulating 1.he globa l behaviour of a composite

material by meaus of a deta iled geomet.rical discretization , based on t he fi n ite
element mcthod, and providing nonlinea r constitutivc laws for each componen1. .
Generally, 1.he actual application of micromodels to composite ma.teriaIs such
as concrete or masonry requires a huge am ount of computing effort elue to
the fine meshes anel iterative solving algorithms needed. Because of this , thc
application of micromodelling in masonry structures is generally restricted to
the study of isolated panels, walls wit h a limitcd amount of openings and local
effects on walls due to concentrated loads. Never1.heless, micromodelling is,
wi1.h experimental tes1.illg, very useful for unders1.anding 1.he actua l beha\·iol.l l"
of rnasoIl ry.

In micromodels: uni1.s and bed and vertical mur1.ar juints are simu latcd
separately. For units , plane s1.ress elements aTe lIsed while for mortar joints
two d ifi"erent teclmiques have been used clepending on the compromise be1.weeu
precision anu simplicity uf modelling. R ots (1991) n.neI LOlll"cnço (1994)
distinguish between:
Deta iled micromodelling. In this 1.echnique joints are represented by continUlll11
clements .
• Simplified m icromodelliug. BeJ anel vertical joints are represented by
eliscontinllous elements (F igure 16).

4.2.2 Examples of microl1lod e ls

P age (1978) WilS the first rcscarcher t.o obt<lin some result fro111 the npplication
of micromodcll ing to brick masonry. Tbe ma in features of his model anel 1.he
testing program llscd t.o ca libra1.e and ver ify the ll10del ar(' b riefly described i1l
 C. MOLl NS BO RRELL I Mechanical bchaviour of masonry 109

Urut ( bric>:, block. .cc ) Unil

Headjoint

Bed
joint Inlerface
Urutlmortar
~:;:z:;-;=::::E::::r-=

a) Masonry sample b) Detailed micro-modelling

"Unit" "loint" Composite

\ I /
________ __ ___ ___jf ~ _ ____ _ __ .. L _______ _

--- ---
---~I:-
__ _
-- - ---
- - ___ _ ___
~I;,.-
~ I :-
~I;"
---... ,..- ------- ... ,..---
_ __ .J L _ _ _ _ _ _ _ _ .J L __ _

------- -l r- ----- -- ---- - -- - ... ,.--- - ----

c) Simplified micro-modelling d) Macr~modeUing

Figure 16 - Bid imensional mi croUlodelling for ma sonry (Louren ç.o, 1994 ).

t he followin g pal'agl'aphs.

111 thi s modeI. bri cks are ass um ed to be isotroJlic anel indefinitely el<l sti c. Failul'e
occurs whell the shear tensioll failul'c critel'ioll is I"cached in the bed or verticn!
joints.

80

•
~
60
~

•• 4.0
~ bnck
masonry
~o monar

4()() 800 l::!OO 1600

slraln (e·OS)

Fignr(' l"j - f1 - ( rela t ionship in uniaxial compress ion for m(l S ()ur~', bricks
and mort al' (Page, 1978).
110 STRUCTU RAL AN ALYSIS OF H1STOR ICA L CONSTR UCT1 0N$

20

·
~

"
15

•
~ 1.0
"••
~

•
bnck
05
~ masoory
mortar

1()()() 2000 3000

shear stram (e..()5)

Figure 18 - T - r shear rclationship for masonry, bl'i cks and mortal' (Page.

1978).

Thc sl rcss-strain relat iolIships under normal anel shear stn'sses of joints are
considered nonlinca r anel uncoupled. Figures 17 alIei 18 show brick an d
masonry stress-st raill l'elationships rr - ( and T - i", obtaincd by te:-;t. ing. Mortal"
elcfonnability, in its actual confined sta tc in th<: bed joints, wa s indirectly
obtained B S the ctifkrence betwcen the results from masonry anel f!'Om bricks.
The fai lltl"f' crit('rioll of joints was experimentHJ1y elcrived from tbe tcst result.s
(Figure 19).

30

·
~

"•
• •
~
~ 20
iR
~
~
\;
. Mean 01 1Z lests

'3 10

Region Z ! Reglon 3
I
20 40 60 BO
Region 1 compreSSlon - _
ulllmate normal stress MPa

Figure 10 - Failurc envelope derived for joints (Pa gc, 1978) .

 C. MOLlN$ BORRELL I Mcchanical behaviour ar masanry 111

If joints a re subj ected to tension when the shea r f<lilnre cri teria i5 reached,
no residual strength is considered. OIl the contrary. if joints are under high
compression 1 r esidual shear st iffness is estimated from the T - r relationship of
morta r (Figure 18) while the axial stiffness is kep t constant.

Figure 20 shows how bricks arc simulated by rccta ngular plane strcss elements
anel l beca use of the reduced thickness of the Joint sl nodes 1) 2 ) 3 and 4 al·e
assimilated to nodes A and B. The real thickness of the joint (tm) wa s taken
into account in the evaluation of its uniaxial and shear stiff·ness .

I
L 'I: ri'" ~I'
I
JOlnt element

"
• •
.mJIt Bnck element

Figure 20 . Finite element micromodel by Puge (1078) .

Page (1978) sho\\"cd that t hi s model wa s abl e to reproduce the st rf'ss

disLl"ibutions in a bending tcst on a deep mas onry bea m unde!" verti ca l load.
It mll st be tn ken into account that complete two·dimensionnl faiIure cri teria
for ma sollry \\"cre not yei developed when Pa ge presented hi s modeI. Note that
this modeI could not accurately reproduce the behaviour of mnsoluy under high
normal compression stresses beca use it did not include tension fa il ure of units.

Chiostrini oS.: \ -ignoli (1989) de\"eloped a simplified mi cromodel to l·eproduce

the behaviour of a brick masonry pane! under ver tical and shea r st resses. As
Page did, they used plane stress tinite elements for bricks while jomts where
represented by gap elements . .--\.11 nonlinear materi'JI beluwiour was achieved by
the loss of stifl"ness of t he ga p e1emcnts in their nonmd and tangen t ial direction s.
This approach wa~ capab!e of sill1ulating shear-wall behaviour whcn only sl iding
Illccllêlnisms iu the joillts wcrf' obscn·ed. As in P<lgcls model, fai lure bricks wcre
not included .

.-\ Ii &: Page (1988) stud if'd the behaviour of ma sonry panels under hi gh stress
gradi ents causcd by concentrated loads by mcans of a complete micromodel.
 112 STRUCTURAL ANALYS IS OF HISTORI CAL CONSTRUCTJONS

T hey dcveloped a p lane tensioll smeared cracking model in whi ch uni ts and
mor t al' where simulated separatcly by four no de finite eIements. Uni t and
mort al' behaviour was considcred elastic-brittIe in tension and nonl inear in
com preSSlOn,

Igna,t a ki s, St avra-kaki s &.:: P enelis (1989) dcvcloped a more sophisticatcd

complete mi cromodel that incorpor ated t riaxial mo deIs with smeared crackillg
for the behaviour of uni ts ano mor tal' (FiguI'C' 21 ), T hey assimilatf' the tria.xia l
behaviour of units and mortar to that of concrete, with th e sarne lluiaxial
compressivc stl'ength . In additiol1 , th e model in cluded a shear failure criterion
for mor tal' joints (Figure 22 ), as a vcrificati on of th e mort<lr elements. Uni ts
and mortal' are represent ed by mea ns of triangular plane stress elemen ts , The
model ill corporates a m ethod to evalu ate the tnl.l1 SVCl'Se prin cipa.l stress from
the geometri cal aud t he mechani cal p ropcrt ies and the stress sta te in both
componcnts,

(b)

(.,
,. '
,/ ".,
/
/
,
/

. "

Fi gure 21 - Tr iaxial failu re surfa cc, (a ) Hydrostatic sect ioll , (b ) Deviatory

sections. (c) Biaxial scction (IgIlataki s ef. al. , 1989).

This modcl was vCl'ified from the results obtained by Page (1981) dming hi s
tes ting progr am ou m asonry pallels in or der to define the b iaxial failure envelope
(Sect ion 2.8 .1) . A recen t papel' by Stavrakakis ct. a.l. (1995 ) prcsents stress-
 C. MOLlN$ BORRELL I Mechanical b~haviour of masonry II J

,
I"

I"
o, o••

Figure 22 - Failure cl"iteria for joints ( lgnatakis et 31. , 1989).

strain relationships, complete failure envelopes anel thei r mechanisms for a

Roman brick ma sonry. the most notable cha racteri stic of which is the relative
thickness of the beel joints (/",) over the units (16), normally t",/tb = 0.8 to 1.0.
Figure 23 shows complete biaxial failure envelopes, their mechani sms anel st ress-
strain relationships obtaineel analytically by Stanakakis et a1. (1995 ) for Roman
brickwork. Results f!"Om some of t he 130 F.E. analyses made in order to define
the failure envelop e ar(> graphi cally presenteel in Figure 24.

Ridclingf.on &: Ghazali (1990) develop eel an experimental and numerical study Oll
the shear behaviour of ma sonry panels. For the numel'ical analysis they worked
with a detailed micromodelling in which units and mortal' are represcnted
by means of plane stress finite elements anel the brick-mortar interfaces are
represented by links as gap clemcnts. Linear elastic behayiour i5 a5sumed for
the units while a nonlinear one with smeared cracking is used for the mortar.
To verify the interfaces a lv[ohr-Coulomb shear type criterion or failure is used.
Post-peak b eha,-iour at the interfaces is tI"eatcd approximately. as it wa s by
Page (1978 ). :.Jote that these models do not consider bri ck strength.

More recent micromodels, by Lotfi et aI. (1994 ) and Lourenço (1 99-4 ). have gone
more thorol.lghly into tlle numerical simulation of the plast ic with dilat<U1 c ~·
behaviour for joint elements. The use of this joiut clement in simplified
mi cromodelling is "ery interes ting in order to reduce computing time and
storage space.

Lourenço (1994) d c\"eloped iI plane stress sirnplified mi cromodel for bri ck

ma sonry in which bricks are rep resented by rectangular element s. anel joints
are represented b:v interface clcments_ The g eo m et r~- of thi s modd is quite
dose to Page-s (1978 ), ShO\Vll in Figure 20. Ne,-ertheless, it incorporates te nsile
114 ST RUCTURAL ANALYS IS OF HISTOR ICAL CONSTRUCTIONS

t ~n ,, ~c ~ ,~ .' ~." .. ~_ lO"" . . .",,'" i<>I.to

~ u.. .. ue. ,,,o I! ... ' ico! oO\d doei iolnh
111 """""".;"11 01 ... 01_ ;01 .. " ·5"";"; 01 kO 1011"'-
·v. , tlcol 'f1>d' ''I1 01 "' "' ~.
Iy SI..,;"'I/tJn l t.... t~ 0/ . . .. "'~ "'"10-
_V•• ~ c ol c 'O , ~ II> , . , ..,;. ••
v c ...... ,... • , ... 4 1.",.. - Ç ._ ;" , -Sol1 '~"9 o' .... 'oh
VI C",",,;,,~ 01 • .,Nc:o' 0<1' ...
~ jooInl._ SpI!Hi<>q 01 bt l < ~ •
.... s_ , oi ..... ~n "

o.
....::
:::r"
!

-~-r-'-
1.0 1) 5

Figure 23 - Complete biaxial failure envelope anel its mechani sm s for O =

() O (90") , 0= 22.5" (67.5") anel 0 = 45° . q - ( rclations hips for rOlllélU
masonry in uniaxial and biaxial comprcssioll (Stavrakakis et .11.
1995).
 C. MOLlNS BO RRELL I Mcchanical bchaviour or masonry 11 5

 Crac.ked brlck OI' mortal' elemenl A Tlansversely splitted brlck Cf mortar elemenl
4 Slid joint (mor1ar 019rn6rlt) ... Crast'tod bfick OI' ITIOftQ' elemoot
~ UnsI!.Jc:k joint (morta' elOO'Ient) Â DoubIe aackeo brick OI' mor!&' elemen1

Figure 24 - Darnage patterns ar failure: (a) I) :;; O",u,,/up :;; -1.0/0.0, (b)
O ~ O°.'" / 'P ~ -0.7/ - 1.0. (c) O ~ 0° .• "/., ~ -0.4 / + 1.0 . (d )
I) :;; 45" ,uJ/(J'1 ;: +1.0/ - 0.6. (e) I) :;; 22 .5", udu2 :::: +0.<1/ - 1.0, and

(r) O ~ 22.5°.• ,/., ~ -1.0/- 0.7 (St av .. akakis el aI.. 1995).

failure of units by means of an interfa ce elemelli located <'It t he centre of t he

unit that makes it possible to simulate the scpa rati an af the two hah-es. This
tcchnique permits the a\'oidance of some nurncrit;al problems dctected when a
116 STRUCTURAL ANALYSIS OF HISTOR ICAL CONSTRUCTlONS

smeared cracking model is used for bricks (Lourenço ct a1. , 1993, Lotfi et aI. ,
1994).

Lourenço (1994) carricd ouL a detailed vcrifi c.:"l.tion. in severalload cases, of the
values of the dcformat.ion modulus for joints, whi ch have no thi ckllcSS bccausc
the units are rcpresented as being larger th an actual size.

Thc results obtaillcd by Lonrcnço 's mi cromo dclJing , using mechani cal properties
derived from COmpOI1Cllt anel micro-elcIncut tcsting , \Vere quite gaoeI , especially
for the prediction af the complete behaviour of the deep ma sonry bcam under
vertical lo.d te.ted by Page (1978).

4.2.3 Mac ro m o d elling

Macromodell ing works fram the hypothcsis that masonry is a continuum with
a determinate constitutive equation includi ng the strcss-strai n l'elationship and
failure cri teria . There are two basie groups of macromodels: firsL those based
OH existing macromodels for rei nforced or plaill con crete ,md, second. those

specifically developed for m asonry.

Exi sting con crete models have an immediate appli cation to ma sonry of col1 crete
blocks, especially if it is completely grout.ed, i. c ., its bchaviour is quite isotropic.
Anthoine (1992) pl'CBcn ts an exposition of some macl'Omodcls for bl'i ck masonry
formerly developecl for conerete. This approaeh wa s also followed by Rots
(1991 ) for the global analysis of masonry structures by mcans of an extensiol1
of smeared cracking models b ased 011 the deeomposition of defonna tions.

Lot.fi k Shing (1991 ) rai se some obj ection s to t.he use of smeal'ed cracking
m acromo dcls. In their work t hey fouud that , while flexul'e response is accmate
using these models. the brit.tle failurc mechani sm by shear in t.he diagona l of the
panf'ls 1S actually uot well represented. This is due to the inadequn cy of these
smeal'cd cracking models to reproduce accurately the kinemati c conditions in
the crack s opencd . aud the resu lts depend on thc mcsh size due to the softening
behayiour in tcnsion and in compression.

?\'louels specifi caJly dcvelo ped for masonry takc into account . in their conception.
tbe ani sotl'opy cau seu b,\' mortal' joints. Samarasinghe et aI. (1982) proposed
a sim pIe macromodd for nonlinear analysis of brick masonry undel' biaxial
tellsion-comprcs!-i iou states consisting of an I"lastic-brittle material behaviour
alieI the empíri cal fail\lre cri teria proposC'd by Samarasinghc &: Hcnd ry (1 980 )
(Equiltion 16). The proccss of the iterati vc analysis consists of the c1imination
 C. MOLlNS BORRELL I Mech:mical bchavi our a f masanry li?

I t I I

sn... ",'I' DI.... 43'

I IW I I

~ ...... Jl C(.. - 50,

,
W/P - t, ,, «
-- - hperio'netll,)/ l'!IIX\e aI'a il....
~ =~=:.~iIor'''II~'bla:ci.1

Figure 25 - Comp<-ll'isoll of analyt ical <mel cxpcrin1f'ntaJ cr;lCking pattf'rns

(Samarnsinghe et aI., 1982 ).

of strength and stiffness in the elements where the failurc critl"'rion \\tas reached,
The comparison of lhe numerical results with shear compress ion tests on panels.
inuucing - according to fhe hypothesis ma de - tcns ion-compression s tates of
strcsses. was positivc in the prediction of collapse load and cracking pat tern
( Figure 25 ).

Dhanasekar & Page (1986) developed a more sophisti cated ma cromodel that
included both the nonlinear stress-st rain relationships of Dhanasekar et aL
(1985a ) (Scction 2.6 ) and the empirical failure cri teria by Dhana ~ ekar et aI.
(1985b), presell tcd in Sectioll 2.8.3, In the iterative resolution procedure , lhe
t \\'o sources Df nonlinear beha\'iou r. plast.icity in com p ression and cr acking
in tension , \Vere treated sepa rately. Numer ica l inst<ibilities \Vere solved by
kecping artifici<d residual stifFness in yield('d 01' crackcd arca s. This model was
successfully ilPplicd to the st lldy of the infiuem:e of masollry in -fi lled frames.
for which therc were a\"<.lilable experimental results.
 IIS STRUCTURAL i\Ni\LYSIS OF HISTORICAL CONSTRUCTIONS

b) Unit ceU

First homogeoeization FU'St bOUlOgeneization

(along x axis) (aloog yaxis)

Secood homogenisation Second bomogenisation

(alODg y axis) (aloog x axis)

Figure 26 - Homogenizat,ion pattcrns for masonry (Lourenço , 1995).

To conclude this section on rnasonry ma.cromodelling , some recent resca rch

on the applicf\.tion af homogenizatiol1 tecbniql1cs in ma somy is prescnted.
Due to the promising res ult s obtained by tlwse homoge-niZ<ltioll r,echni ques
in the <lIlalysis of periodic layered materiaIs, some resea rchers proposed to
exiend their use to masonry eVf'n tbough it is not strictly a periodi c material.
Homogeuizatioll of the unit cell is clone in t wo phases. one for eilch dircction
(Figure 26 ). Pa ndc ct ai. (1989) successfull y appli cd this tf'chni que lo df'tf'l'mine
the eh!'stic m thoth ropic prop erties of mil son ry.

Lourenço (1995a) developcd a.n dastoplastic formulati on of these

homogenizat.ion techniques to reproduce t he nonlinear cr<l cked lwhaviour of
masonry. Neverthelcss, thi s researcher verified that t he application af t hese
techniques in the nanlinea r range presents some difficul ties due to. first, verti cal
joints not being actually con tinuolls 311d, second, that st rain distributions
generated in t,he unit cell \\'hen t,he mas onry is cra cking are extremely
nonunifonn.
 C. MOLl NS BORRELL I Mechnnicnl behaviOllr ar rnasonry 119

4.- CONC L U5 IO NS

},/fasonry is a material which exhibits a complex mechanical beha\·iour. Due to

its internai constitution. it sho\\"s au anisotropic behaviour both in the ela stic
and plastic ranges. being more pronouneed in the hltter. Its beha\"iour is affected
by fadors such as the geometl"Y of tlw componcnts and their <llTangf'tncnt , the
nwchanical beha\"iour of each compolwnt and the specific response of the unit-
mortal" joint.

Despitc its complexity, the belu.l\-iour of ll1<lSom~' is nowêidays better understood

thanks to the analytical and experimeutal rcsearch c<llTied out O\'cr the last
decades. Among the results of this rf'scilrch <lXC experimental biaxial failure
em-elopes. phenomcnological critcria "nd numerical micro- and macromodelling.

In i.lncient structurcs. masonry presents a \Vide rêlllge of geometrical

arrangcmcnts and different materiais. In facto c8ch ancient structurc could bc
regarded as being built of its o\Vn particular lllflSOOly. Despite some commoll
cha ra cteristics such as low tensile strength. the calibration of a constitutive
modeI requires a comprehensive testing programo Howe\"er. a first ::itep . used
by sc\"enll authors. can be found in the combination of simple models as an
elastic-brittle relationship ill tension togeth("l \\'ith <in dastic-pbstic 011(-' in
compresslon .

The combined use of micro- anel macromodels allo\\'s the accurate simlllation
of the response of masonry elements on the basis of a Iimited étInount of
experimental data. rVl icromodelling is used to ohtain the required input
data for the macmmodel. such as the failure surfaces and the stress-str ain
relationships, fram known geometrical and meclwnica! properties of the
components. J'.'licromodelling can also be used for the detailed anal~'sis af ZOl1es
of special interest.

REFERENCES
Ali. 5.5. &. Page, A_W. (1988), "Fini te element mudei for masolll'~' subjf'cted to
concentrated loads", }ournol of 5frucluI'ol Ellymee/"(/)!I ASCE. 114. (8). 1'61 - 178-l .

Allthoillc. A. (1992) , III-plane beha/}/our of masollry: a lileralure ren e!!". Commisiol1 of

the Europe<lll Communities . In dustrial Processes, Luxemburg.
AI.kiUi'Oll. R..H. 8..: No land. J.L. , (1983) . "A praposed Í<lilurc t!lC'ory for hrick
lllélsollry in comprf'ssion", Proc. 3rd Ca l,adlal' Maso"r!} SYlI!po.m Jl".
 120 STRUCT URAL ANALYSIS DF HISTO RICAL CONSTRUCT ION S

Dio l'll i. L. (1 988), "Evaluatioll of Cornpr('ssiv(' Stn'ugth of ::V[asom . .· \:\,T,d ls hy

Lilllir Allill ....·sis'·. )ourllo( o[ S trud ura ( EII 9w eerm9 ASCE, 114. ( 10 ) , 2170 21 89.

Drooks . .1..1. (1 986 ), "Compositc modcls for predi cting f' b st ir ,md I011g- tnm
1l10Vpmf'llts i11 brickwork walls'· . ['ror . fJ n l. M a ~ . Sor .. 1 .

C hios t.r iui. S . , FOl'ah osch i. P . & SO l"aCI!, S . (198D L "Problems COIlIl('d f' U with ti\('
Arrangclllcll t. of a Nem -linear Finite Element Method to the Ana lysis of Masomy
St.nICtuf('s", S /n ldura l RCJlal1· alld :\I aill lclI (/ I!N' o[ fl lS lo1"1 cal !1l1 ddlll!J.~, Compllt ational
M('dlllni cs Pll blications, 525- 534.

C hios tl·i ni. S . & Viglloli. A. (1989). " Appli cat ioll of a Nl.lmeri ca l Method to
th e St u dy of Masonl'y Panels wit h Vm·ioll s Geomd ry lllHl er Seismi c Loa ds" .
S trllcl llral Rel,air (/n d Mrli nt enallce o[ fli stonral /Jlllld owç. Compll tational lvlechani cs
Puhlications.309- 317.

D ecre t.o M iuis t e riale (1 987), "Norme techniquc per la progf' tta zionc, csccuz ione
e coll audo elegli edifici in mtl1"il tura e per il 10m collsolid;ll nellto" , D.M.
20 / 1I / 198i , !tal)'.
Dhanase kal' , M. , K leeman , P.W. & Page, A.W. (1985a ), "Biaxial stress-strai n
rela tions fo r bri ck ma ~ onr y" , ) ol11'nal o[ S lrucf ura l Engin ee rill .q ASCE. 111 . (5) ,
228-
244 .

Dhanasekal'. M. , Page, A.W. & K lec llHUl , P . W. (1985b ), "The failure of bri ck
ma sonry unu er biaxial stresses" , Proc. Ins t/!. Civ. Engr5., Part 2, 79, 295- 313 .
Dhanasekal'. M. & Pagc , A.W. (1986), "The inftll eO(:c of hri ck ma sonry infill
propcr ties Oll the behaviour of jnfilled frmn es" , I~/Y)r. Irlstn CIV. ElI gl's. , Part 2,
81 , 593- 605.
Diale l'. C. (1991 ), "Some rcma rks on the stt·cngth and defor ma tion beha vior of
shear stressed masonry panels under sta tic monotoni c loading· ', Proc. oi Ihc 9th
IIJM AC, 276- 283.

E mocod e N o 6 (1 992 ), "Common unificd ndes for masonry stru ct.ures", Report
EUR 9888 EN.
FL-90 (1990), "Muros resistentes de fá b ri ca de ladr illo ll , Norma Básica de la
Edificación , Ministerio de Obr as Públicas y Transportes.
H ellcll'y, A.W. (1 990 ), S lruclu ral Ma sol!ry, Ma cMillan Educa tion Lt d , London.

H e yrnau , J. (1982 ), TII e Mas01l r·y A rc}" EUi s Honvood , Chichester.

Hil sd o l'f, H.K. (1969), ulnves tigation jnto the failure mechani sm of br ick
m asonry load ed in axial compression·' . Proc. h ,t Syml!. on Mason ry S lr1J cfu ral S ystcms,
Texas, 34- 41.
 C. MOLlNS BORRELL I Mec hanical bchavio ur of rnasonry 12 1

Igllat,akis. C. , St.avn.\kakis. E. & PCllelis. G. (1 989 ). "Analyrica l !\ Iodel for

iVl asollr~' using tlH' Fillit(' Ele nwnt :\Iethod··. Slru clura! Repalr aná M(lI1l/ ena nce of
!ltslor/cal /JuddwY5. Com puta tional :\kchanicR Publicatiolls. 511 - 523.

Le lll.:z u í'l" . D . (1 981). '· Brickwork: a guidf' 10 CI'f'Cp .. IlIl r rllallOlI(I! JOIl1"/wl of .lI aso/l l"!}
ConslI"urlIOIl. L ( 4 ). 127 133,
Lutfi . H.R. &.. Shiu g. P. B. (1991). '· An appraisal of smeared crack models fo r
ma sollry slwar wall <ll1al ~· s i ~··, Compulcrs aliá SII"uc!uI"n. 41. (3) . -U3- -l25.

Lot.fi . H.R. õ:. 5hiug, P, B. ( 1994 ), "Interface model applied to fr acture of ma sonry
struCLur('s", f ourllU! of Sl rI/rI'lm! /:;ngweerfll!l .'1.';CI::. 120. ( 1 ). 63- 80.

LOlll"CIl c;tl. P,D. oS.: Ro ts ..L G. (1993). "011 til(' tis!"' of micro-mod!"'lling for the
<lllalysis of ma sonry slH'ar-wa ll s·'. I 'rur. 2 11d 111/ . Syll! /l. 0 11 COIllf!u/a/lollOl .lI etbod~ li!
S/nlc/ural Ma sonl"y (Eds. G.:\". Pallde &; .1 . :\ liddletoll). Swansea . "C. h-: .

P.D. ( 1994 ), AnalYi>ls of masonry slrudurei> u·/lh III/elface cfellle llk· Theory aud
LOlll'c u C;O.
Dplft Ullin"l"sity ofTccl ll101ogy. Rqlort ILO. 03 ~ 21-22 - 0 · 0 1. DeJft. fh e
UpplWl l /OIIs,
f\:et herhlud s.

L O lll" C Il t,:O. P.D. ( 1995»), Tlle cla~toplasllr elllplr/lll'n l(lflon of homogell/:.ul,oll Irchll/quflj
w, lh alI eT(e llSlQn to ma sonry S / I"IIr1UN'~.
Delf! Lilliye r s i t~· of Tcchnology. Report no.
03-21 - 1-31-02, Delft . the Netherl<lllds.
P.D. ( 1995b). ;I" or/ho /I"opl e fonlmlllllll /II odef fOI" lhe allaly5li> of m llso nl"9
LOllre ll ço.
sl,. u clllre~.
Dclft Cn iwrsity of T('c hnology. fl <'pon llO. 03-21-1-31 -27. Delfi. lh e
Nctherlnnd s .

M a gclics. G. (1992). Compol"/all/ol/o HHl/l fO d, ma/"/II/1re d, lI/(ltlOII/ /"e,/:./"":;U r

m"CCfUI t-· III I di rot/uI"a dI lIIa:.cll/ 111111·"1"'. Doctoral Thesis. Depa rtll1<:'llt of SIl'udural
:\Iechani c.-;. Unin:rsity of Payia.
M allll . W . &.: Miillc l' . H. ( 1982 ), ""Failure of Shear-Strcssed )' Ia sonry - Ali
Enlarged Theory, Tes ts and Application lo Shea r \Yalls·· Pror. of /hf' lJnl"h
Ceramlc .';of/l' ly. 30 , 223- 235.

McNII l"Y. W ,5. &.: Ab r UlIl s. D .P. (1985 ). "?\ I('d Llu lics of :\lasonry iu COlllpressioll··
.}O/II"n,,! of Sl rll ctural EI/g/llltr/ll!l ..tSCE. 111 . (4). 837 8 70.

Nal'aill C. K . &.: Si Ilh a. s. ( 1989). ··Bch<l\·ior of Bri ck :\bsolll"~' Hlldf'r C~'clic

ComprC'ssi\"(" Loadillg'\. JouI'lIa/ Df COlIs / rurlwll 1:.·llgH!('~ nllg /Jlld .\/11 11 {/!Jflll CIl I ASCE.
115. (2) . 1432- 1-1-15.

Nar a illt ,. K . &. s illha. S. (1 991 ). "Cyclic Bpha\'ior of Brick ~ I asonry undr-l" Biaxia l
Comprcssion··, Joul'1I1/1 of Slrudurul EII!J/II(,(,1"III!1 A:5CE. 117. (5) , 1336- 1305.
Narai u í'. K . .s.:: s iuh a, S. (1 991 ) .. L:\ Iodel for Cydic CompreRs i\"{' BelH.wior of Briek
:\ lasonry · . ..t el S'lrll('/lIm! JOlo·""I. 88. (5).603 609.
 122 STRUCTURAL ANALYS IS or HI STORICAL CONST RUCTlONS

Palllll ', G.N ., Liau j.!; . .I .X . ~ Mitltlle toll . .l. (1989), "ECjuinl lt'nt ('last ie JIl ()r\ll li for
brick ma sollry" COllllllllrf"~ UI/li (,'rolrrl/llUIj. 8, 243 260.

Pa).!;L'. A W. ( 1978 ). "Fiuik Ek'lll('nt J\ Iod t·1 for r\'b ~()nl'y"'. JUIlr'ual ui IIH' ,'i/ r I/duraI
f)I/II~/01I .I SCE. 104. (S T 8) . 1267 12 8~.

Pag(·. A.W. ( 1980 )......\. binxial failun' nil('rion fo r hr ick lIlaSOlIl'y in ti\(' kusion-
t(,ll~iOIl l';lIlge"
lI/f. jflllnllll of M/I~QII"y lOIl~·I'·lId/Ql! . 1. ( 1),26 29.

Pa).!;I', A. W. (la81). "TII\' Li<lxi;l\ COmlll'('ssi\'(' st r (,lI~ t li oi' bric k Jn<l s o1l1'i'. I'roc.
11I~/1I ('til. E1/91's. 71. (2), 893 gOG .

Paw·. A. w . (1 983 ), "TIl(' st rf'ngth of brick ll1ilsomy lllldcl' biaxial tellsio n

26 31.
COIllIH'('ss io !l", 1/11. )0/1,.,,/,1 of Ma sonr.'1 Con~("l/rI/(J/I. 3. ( 1).

P"gt -. .T. ( Edit.or ) (1993 ), .I{ asonry Art" IJndg('s. $ /rl'" oJ Ihc Ar/ R rtr l ftll, T1"I,n~port
n escm·ch Lab . H MSO Publi calions Cellti"(,. Londoll.

P.I.E .T. 70 (1970 ). " Obras dp r.:ihr ic;,··. I'r~scnJi("wn(', def 11I~111!llo Ed'lllrdo Tm-ro}u,
1 Indrid.
RidJiugt.ou . .l.R. &, G llazali. M.Z. (190 0 ), ·· Hypo lhf'sis for slwar félilUl"(, in
ma!,(jlll"~' joint s", Prac. If/ $III. COI. EUf!7".'., Parto 2. 89 , SO 102.

Rot.s. J .G. (199 1). ":\u l1lcri ca l s imnlati oll of cra ck illg il l structtlra l lllll:-;onry
Urrou, aG (2), 49-G3.

SaUlIU'<ISiughp, W. Y Ht"'lldry. A.W. ( 1980 ), "Strcagth of hri ckwork undf'r biaxial

st rC"ss". P1"OC. 7,h 1711. SYW/I. Loadflfa'·I1I!l /Jnrkw/Jl"k . L01ldoH , Dr itish C"UlIllic Soc iety,
Stokc'-Oll- Trcnt.

SlHIHU·" s iu g h e. W ., Pagc·. A.W. oS.:: HClldl"y. A.W. ( 19S2). "A fin it<, ckmellt modcl
for the iu-plane behaviour of brickwork". /'roc. IIISIII CII·. ElIgrl> . 71. (2) , 171 - 178.

Stavl"akaki s, E. , Iguill.aki s. C . oS.:: Pl'up lis . G. (10a5). " Bchavi011i" of Roma ll

masoul"y under biaxi<ll loadi ng: au:dytic<.Il ;I SS" SSllwut of complett"' failllre
niter ion". Slrucfural SllIdlt't> oJ Hls l oncal /Jluldmgs /l '. 2, Complltatiol1tll ~"I (,l"ha ni cs
Publi catioll s, 339 347.

Tassio:". T.P. (1 98S ), Ri\{ t'fT(Wu:a d"l/,.. IIIIII·/Ifurc·· LiglloJ"(' Ed itori, Napoli.

TUl"lI1w k. V. Y Cacovic. F. ( 1970 L ·· SOIlW c'xIwrimellLlll rt"" ~ml t B on tlH" sf. n 'll gt h of
bri ck masollry walb", PrOf. 2"d {"I. /Jnrk Ma ,'·"'II·Y ConfrrY"IIN". 14D- l['G,

1'lu·w;('k. V. Y SIH"ppal"fl. F. ( laSO ), "TIl(' :-;hea r aud H.. x mal resistallc(' of llWsollr y
walls", Prol'. of fhe 1111 . I(r~f(11"fh ConJat'llff 1111 f;/lrlhqua/.:, /·; U.lfluf,..rwg.

Vohl. F.V . oS.:: F'It.t.:11. S.G. (1 076 ), "Failu]'(" hypoth('sis of maBOlll'~' s!l eHI" \\,(llls ",
jourl/al of 111 r .'ilrudllrul f)m'lOlI, ASCE , l() 2. (ST3 ). [,13 ,")32,
 STRUCTURAL ANALYSIS OF HISTORlCAL CONSTRUCTIONS
P. Roca, J.L. González, A.R. Ma rí and E. Dilate (Eds.)
e C IMNE, Barcelona 1996

STRUCTURAL ASPECTS DF THE REHABILITATIDN DF THE

MEXICD CITY CATHEDRAL

R. Meli and A.R. Sánchez Ramírez

Ill stitllte of Engilleering
National Uni versify of Mexico
Mexico

SUMMARY

The M exico Cily Cathedral is undergoing a major rehabilit,lIion process aiming

aI correcling a significant portion of its differential scttlemcnts, that reached
2.4111. Man y slructural investigaliolls are been performed to support lhe decisions
taken during the processo The most relevant 01' them are brielly described in Ihis
paper, as those related to assess the s<lfety undcr vertical IO<lds, the effects of lhe
differenti<ll seulelllents anel their correc tions, and lhe seislllic aClions. The cxtense
Illonitoring systelll illlplelllenled lO control lhe struclllral respo nse is descri bed.
Laboratory and in situ tests 01' lhe mechanical propcrti es of lhe materiais. as wcll
as in si lu delerlllinalion s of the dynanlic response and of the stalc of slresses \Vere
carried ou!. Tlle corrections achicved after I\VO years of \Vorks are c011ll1lented.

I. INTRODUCTION

The Mexico City Cathedral , probabl y lhe most important colonial Illonument in
Ameri ca , has been severely <lITccteel by diffcrcnliaJ setllcmenls since lhe beginllitlg
Df its construelion in lhe XVI eenlury. The 1ll0lllLlnenl is extremely hcavy
(127.000 tOI1) anel is locateel on vcry soft elay deposils, which in some pari s had
been previollsly con solidated by AZlce temples and pyramids over whose remains
lhe Cmhedral W<lS ereelcd. 111 this century the intense pllmping of the undcrground
waler Ilas severely agg ravmed the sllbsi den cc.

Dllring lhe prolracted period of its constru etion (240 years) lhe great differenlial
settlements forced lhe builders to make significanl acljuslll1enls in lhe geornclry of
lhe rnonulllent. After its completion . lhe Cathedral has been subjeeted 10 an
almost uninlerruptcd activilY of repairillg, especially in order 10 seallhe crads in
lhe roof in order 10 avoid lcaking.

In recem ycars. lhe concli lion 01 dislOrlion of lhe building \\as such Ihat se riotlsly
unclermined ils struc lural safely. The maxinwl11 differenlial selllcment reached 2.4
tll <mel was increasing aI a rale of 12 mm /y r. Some of lhe main COIUIllIlS
supporling the roof showed an Olll-of-plumb near 10 3%. Severe cracks in lhe
rool". t100r and walls evidenlialecl lhe elfects of lhe clifferenlial sett lcment s.

Con si dering 1hm the regional subsiclence would not be elilllin<lled in lhe near
124 STR UCT URAL ANALYSIS 0 1' HI STORICAL CONSTRUCT IONS

fulure because underground water needs to be extracted to satisfy lhe demand of

lhe city, measures were taken to restitute lhe building to a stable and safe
condition. A major rehabilitation project was started in late 1991. The main
structural aspects of the project will be described heTe, focussing on lhe
experimental and analytical studies performed. A description of lhe geotechnical
prob lems and of lhe underexcavation technique can be found in Ref. 1.

2. DESCRIPTION OF THE BUILDING AND ITS PAST PERFORMANCE

The temple is constituted by five naves. The roof of lhe central nave is formed by
a ci lindral vault supported by arches and by 16 stone columns. The lateral naves
have hemispherical vaults. A dose array of robusl masonry walls divides lhe
extreme naves in smaJl chapeIs. These walls along with lhe facades and some
buttresses constit ute a peripheral belt providing great lateral strength and stiffness
to lhe monumen l. A large dome aI lhe intersection of the cen tral nave and lhe
main transverse nave, constitutes the heaviest and mOSI criticai part of the roof.
The main features of the construction can be appreciated in Figs. 1 and 2. The
struclll re is supported by a grid of foundation beams (3.5 m deep) and by a
foundalion mal with a thickness of aboul 2m . Timber piles (with a diameter of
0 .201 and a Jength of 2-3 m) are spaced every 0.6 m undemeath lhe foundation
mal.

The primary construction material is a kind of poor concrete constiluted by

volcan ic stones, of different size and unit weighl according 10 the structural
member, agglutinated by a lime-sand mortar. The properties of this material will
be discussed later. Andesitic stone sills were used in arches and columns.

The most clear evidence of the many adj ustments made to lhe dimensions and
shape of the construction members in order to cope with lhe large differential
seulemenls originated since the early stages of construction are lhe following: lhe
lenglh of lhe colurnn varies according to lhe settlements experienced of their bases
at lhe time when lhe arches and vaults were built. The maximum differences
being 0.85 m. Several rows of sills with variable height were placed at lhe facade
to correct lhe inclination aI lhe lime of construction. The span and rise of lhe
arches and and vaults were varied in order to obtain an uniform levei of lhe roof.
After lhe completion of the roof, lhe structure acquired a grealer stiffness, Ihus the
differential settlemenls increased aI a lower rale. Nevertheless lhe large cracks
and the inclination of lhe upper parts of the columns reflecl the very large
distortions suffered by the struclure once it was completed.

Fig. 3 shows lhe pattern of differential settlemenls measured in lhe floor of lhe
Cathedral some time before lhe beginning of the rehabilitation work. Another
lIseful representation of the subsidence is given by lhe curves of equal rale of
differentia1 settlements in lhe year before the commencerncnt of lhe works (Fig.
4). Two major mechanisms of dcformation can be appreciated. One is the
sinking toward lhe Southwesl comer, anothcr is lhe "emergence" of lhe central
nave in the Northem part o
R. MEU and A. R. SÁNCHEZ RAMíR EZ I Th!! Mcxico Cily C.:tthedral 125

~~~ ~q:>C?
-9./4-11.24-15. 76-11.24-9./4-
I I I I I I

ffi::::~=r1-~~
0-
~ f"1
-i ~ ~ 1 [~ ~
1

~:~II~:-rl[~
0- -I..!
@
@
'"
'"
~
@--- 10(0 _ 'i.;USfUM (() '"
/2 50 /5 90
0- - 1

@---'rO-
@-'yo_
=-,°
&- '° - 1
1070
0-) -
/360
@- \ -
DimerlSoon
mm - - - 6 3 6 3 -- -
CA fH[ORAL

Fig I Plan view of C(J(hedral

Fig 2 Front view

126 STR UCTURAL ANALYS IS OF HISTORICAL CONSTRUCTlONS

Fig 3 Differential selllemenls, in meters (Oec 1989)

~ 'F<'_'=
0-
0- ~I.';" /.II
0-
0-
0-
0-
0-
0-
0-

Fig 4 Annual rate ofincrease in differential seltlements, in cmlyear (1991)

 R. MEU and A. R. SÁNCHEZ RAMíREZ I Thc Mexico Cily Calhedral 127

The tirst mechanism has produced a pattem of transverse cracks in the roof and
walls, especially near the central dome, and some separation of the Southem
facade, wilh its very heavy towers, from the rest of the church.

The second mechanism has produced lhe outward rotation of lhe columns and
lateral walls, and the opening of lhe vaults and arches in lhe roof, originating a
pattem of longitudinal cracks in lhe roof, floor and foundation. This second
mechanism is lhe criticai one from the structural point of view, because of lhe
inc1ination of lhe columns receiving lhe greatest vertical loads, especially those
supporting the central dome. As can be seen in Fig. 5, the shape of lhe column
shaft shows changes of direction due to corrections made during lhe conslruction.
The total eccentricity between lhe upper and lower part of lhe column is, for Ihis
case, O.6m representing 25% of lhe size of lhe column . Some vertical cracks aI
lhe upper pari of the colum n are auributecl to lhe compressive stresses generated
by the eccentric compression.

3. STRESS ANAL YSIS AND ASSESSMENT DF STRUCTURAL SAFETY

The weight of the temple represents a very severe action on lhe structure, the
foundation and lhe soi!. The differential settlements have significantly a((ected the
stresses induced by lhe gravity loads. The effects of these actions have been
studied mainly by finite element analyses of different structural models. First. tri-
dimensional models of the central part of lhe tem pie including its foundation were
studied (Fig 6) to determine the overall slate of stresses in lhe structure. A
schematic representation of lhe flow of forces derived from the resuIts of sueh
model of the undeformed structure is shown in Fig 7. It can be appreciated thal
the weight is transmitted towards the foundation essentially by axial forces in the
vaults, arches, columns and walls. The levei of stresses in lhe different strllctural
members is well within the range of capacity of lhe materiais.

More detailed models of critical parts of the structure were studied, as the
Southwest tower, lhe adjoined Sagrarium, and the central dome. The effecI of lhe
deformed shape of lhe structure were sllldied mainly by bi-dimensional models.

The assumption of a continuous. linear-elastic, isotropic structure. involved in lhe

finite element analysis is (ar (TOm being representative of the condilions of a
struclure of this kind , cons idering its low tensile strenglh , its large cracking and
the many sources of discontinuity. Ncvertheless, such an analysis can still provide
a useful picture of the overall state of stresses and identify critical zones deserving
more accu rate stud ies. Quantitative results of the finite elemenl analysis cannot
be directly used for assessmenls of the structural safely.

An evaluation of lhe accuracy of differenl types of analysis can be made by

examining lhe state of stresses in the central columns due 10 the gravity loads.
lhe finite element analysis indicated an almost constant levei of stresses Ihrough
the cross section of the column. Table I shows the results for column C ~ 7. On
lhe contrary, lhe state of stresses determined in the columns by the nat jack
technique described laler in thi s paper indicated a linear variation through lhe
128 STRUCTURAL ANALYS IS OF HISTORICAL CONSTR UCTIONS

Columo Elevallon

Fig 5 Deformed shape of a central column and applied forces due {O selj-weight

Fig 6 Fin;te eiemem madel of lhe central ponion of lhe Cathedral

 R. l\I1ELI and A. R. SÁNCHEZ RAMíREZ I Thc Mexico City Cathedral 129

Table I Compariton of compresive stresses on one of lhe central columns

(C-7), in MPa

Isolilted Flilt
Cross Section Point FEM Column jack
13.5 19.3 measur
'm' 'm' ement

---
C
C-7 1 2,31 0.88 0 .28
, No.-th 2 1.75 3.19 3 .80 ---
S
)( : Eost ___ o

3 1.76 3.48 4 .26

8

~~
3
, 4 2.30 0.60 -0 . 19 ----
, 5 2 .35 0 .62 -0.12 2.06

5
, 6 1 .71 3 .43 4 .17 5.50

ffirEI1 (0,0 47.00 48) 7 1.72 3 .61 4 .46 7 .65

:gI H,13 (~0,2 4 9. -0.199)
8 2.34 0.49 -0,3 4 1.22
+H19 (-0,38 4.-030 4)
2.03 2 .04 2 .04 4 .11
(!;r J , <-0588.-0422) "-
130 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

section evidentiating lhe presence of large bending moments . The c!ifference is

attributed lO lhe fac! Ihal in lhe analysis cf lhe whole structure lhe eccentricities
cf lhe loads generated by lhe roofs and by lhe inclination of lhe structure, are
transmitted direclly to lhe transverse walls, whose buttressing effect has been
panially 10sI in lhe aelual structure due to lhe olltward rotalion of lhe lateral
naves. The analysis of an isolatcd column sublllittcd to lhe vertical load acting
with the total eccentricity prodllced hy its out-of-plurn h , gives a Tnllch better
picture of lhe state Df stresses. In addition, lhe study of lhe internai arfay of lhe
stones in lhe cross section of lhe colum n indicates lha! only part of the stones !lave
proper contact across the horizontal joints. Therefore , the effective cross seclion
is significantly less than the olle considered in lhe analysis.

To reproduce the non -linearities of lhe struclure, different modificalions were

introduced inlo lhe Slrw.:lural mode!. " Weak elements" with very low moduli of
elasticity were used 10 represent major cracks or arcas wherc tensilc stresses
appeared in lhe analysis. The analysis of models of parts of lhe Slructure limited
by the lines of major crack ing anel ignoring Iheir interaclion witll lhe rest of the
structure gave a good insight on lhe stales of stresses.

In addition, plastic analyses of the Illain struClural Illembers werc perfonncd 10

identify their mecl1anislll of co llapse and the safely faetor involvect. The main
concern at Ihis regarei was lhe stability of vaulls anel domes and, especially, lha!
01' lhe columll oI' lhe central nave.

The outward motion 01' the SllppOrlS in lhe Northern part of lhe central vault !las
produced a lllcehanism of non-linear deformation of the roof as the one
schematically shown in Fig . 8a . T he presen t deformed shape 01' lhe valllt is still
well wilhin a slab le configuration . On lhe cOlltrary. the 3na lysis of lhe slability
of the cenlral col\lmns shows lhal bcncling mOlllents inclllcect by lhe eccelltricity of
the force produced by lhe wcight 01' lhe rooL increase the maximlllTl eompressive
stress abolll 2.3 times above tha! computed for the undeflected shape 01' the
colllmn. In this fashion stresses are now very near to lhe maximllm capac ity 01'
the stone, which in lhe upper part of lhe column is ralher low, as it wil1 be
explaincd later in th is papeL

The seismic safety 01' monumental SlrllClures like lhe Cathedral cannot be assessed
by lhe procedures prescribed by building codes for modern slruclures. It IllUst be
eonsi dcred Ihat the aClual shaking incluced in Ihese very heavy and stiff Slrllcturcs
founded 011 a very soft soil is mucll smaller tllan for ord11lary builclings. because
a significam part 01' lhe energy that lhe grolllld tries to impose lo the structure is
aClually retllrned lo lhe soil throllgh radiatioll. Addiliona lly, pari ot' lhe seismic
energy can be dissipated throllgh opcning and c!osing 01' cracks, Illrough relalive
motions bctween parts 01' the slrllctllre and through upl i t'ting. For {!lese rcasons
monumental bui!dings have showll all oUlstanding eapac ily to withsland severe
earthqllakes in Mexico City. even when modem and ap parently stronger strlletllres
have been badly damaged. Earthqllake damage of mOllumenls has been generally
associaled lo cases 01' ext reme degradation 01' lhe materiais or to severe previolls
damage due lo differcntial sculcmcnls, especially associaled lo out-of-plulllb 01'
columns and walls.
 R. MEU and A. R. SÁNCHEZ RAMíREZ I The Mcxico City Cathedral 131

Load in ton

Fig 7 Flow of self-weighf loads in the zOTle under fhe cemal dome

Pressure IIne 7

t ~~ t
aI Shorteni ng of the span

Pressure IIne ...

b) Open ing of the span

Fig 8 Mechnisms of inelasfic defonnatiom due to displacements of fhe suppons

of lhe central vault
132 STRUCTURAL ANALYS IS OF HISTORICAL CONSTRUCTIONS

T he previotlsl y described li near elastic tr i·dimcnsional l1lodel was used for a

dynamic analysis to compute lhe modal shapes anel pcriods oI' vibration (Fig. 9).
It was fou nd lha! in plan deformations of lhe roof favored torsional modes Df
vibrat ion anel 1hal lhe large concen trated mass Df lhe central dome produced
significant vertical vibration. Computed periods were compareci with thase
derived by measuring lhe víbration of lhe building in ambient conditions.
Co nsistently . computed periods are slllaJler than 11105e measured . This is
attributed to lhe lack of fixily of lhe base of lhe strllcture where lhe founda tion
all ows significan! rotations of lhe walls and colulllns . Adclitionally, lhe exlensc
cracking of l he structllre sig nifican tly reduccs lhe stiffness, I!WS incrcasi ng the
vi bralion period.

For the asseSSll1enl of lhe seismic safel y, a constan t spectral ord inale ofO.2 g was
assumed. 11 was considered Ihat, becallse of lhe existing cracking anel of lhe small
tensile slrength of lhe materiais. parts of lhe structure cou ld vibralc independently
from the rest. Therefore, the seismic safcty of the mOSI criticaI portions was
assessed independently, alld primarily by silllplified mClhods. [n general terms,
it was concluded Ihat lhe slrong peripheral bel t consliluted by lhe fac<lde walls and
by lhe walls surroun ding the chapels providcd a salisfaclory ovc rall safety.
Nevertheless, lhe addilional lateral displaccll1cnls of lhe colulllns during their
seismic vib ralion could lend 10 their lateral inslability and to a local collapse.

4. CORRECTION OF DIFFERENTIAL SE1TL EMENTS AND ITS EFFECTS

ON T H E STRUCTURE

As sh own in the prcvious section, in ils present conditioll IllC Catlíedral i s

slructu rally unsafc d\le mainly to lhe grea! inclinatioll 01' lhe central C01UIllIlS, and
the operation of lhe tclllple is arfeclee! by lhe excessive slope 01' lhe t100r and the
great cracking.

After evalualing severa l alternalives, a Icchnique called lIndercxcavation was

adopted to correel parts of lhe settlements. Briel1y, a conlrol led subsidenee of lhe
most elevaled parts of lhe ground is produced by lhe extraetion of soil frol11 lhe
deep strata of soft clay. As show n in Fig. 10, sl1lal1 diameter radial boreholes are
excavaled frol11 2:\ shafls opened to a deplh of abollt 25 111 . The closing of the
holes due to lhe weight of the soil anel lhe strucIlIre prod\lccs a scttlelllcnt at lhe
surface. Through a careflll progralllming of tllC alllO\lnt anel position of exeavaled
soil , a preselecled configuration of grOlIl1(] sctllcmellts can be achieved with great
precision.

The preparatory works Slarted inlale 1991, wilh lhe excavalion ot"lhe sharts. The
undercxcav<llion process beg<lll in AugusI 1993 with some trial work in the
Northern pari of lhe tClllple. The nlllllber af lhe shafts being atlacked and the
volume of soil eXlr<lcted in each of them was progressively adjusted unl i! a
satisfaclOry pattern of differential sett]emenls was oblained. Frolll the very
beginn ing, lhe roor was thoroughly shored with a stee! pipe stnlcture provided
with sc rew jacks in order 10 adjust ils height following lhe geometrical changes of
lhe bui lding eauscd by lhe induced seulemcills.
 R. MEL! and A. R. SÁNCHEZ RAMíREZ I The Mcx ico Cüy Calhedral 133

,
.J
, -,
T '" 0.323
T '= 0.303
Tronsverse

MODE MODE 2

, "'
.J
, -,
T _ 0.223
Vertical

MODE

T .. 0.184

Ta rsian

MODE 5

Fig 9 Moda! shapes of vibralion from a jinite element model of lhe central
ponion of rhe Carhedral
134 STRUCTU RAL ANALYSIS OF HI STOR ICAL CONSTR UCTIONS

Fig 10 Position Df lhe 11Ulin shafts and radial peiforations for lhe
correcrion of lhe differential selllemenrs

o la 20 30 m
I , I , I , I

escala grdfica

0-
®-
0-
@--
@-
@-
Fig 11 Differential settlemenrs berween Ocr 1991 anti Sep 1995, in milimeters
 R. MEU and A. R. SÁNCHEZ RAMíREZ / The Mcxico City Cathedral [35

The objective was lo produce a general rotation of the monument, as a rigid body,
toward the North~East, plus an inward rotation of lhe lateral naves in lhe northerrn
par! of the chllrch.

The reslllts of lhe process, can be appreciateel from Fig 11 where lbe elifferential
seltlemenls originaled in lhe tloor of lhe monulllents from OClOber 91 to
Seplember 95 are showll . By comparison witb lhe previolls sett lem en ts
rcpr~senled in Fig 3, il can be cOllcluded Ihat lhe Irend of lhe correclion is
favorable . Of lhe 2.4 f1l of maximum differential scttlclllcnt, 0.42 m !lave been
reduceel so far. A target rate of correction of 20 111111 per 1110nlh has been
established. In general terrns, lhe correction has been carrieel oul wilh minimum
damage 10 lhe structure. Most ex isting cracks have reduced their width, some
prcviously repai red cracks reopened, anel some local crushing or fali of ornarnents
appeared.

The displacemenls of lhe colul11lls anel walls have becn, in 1ll0St cases, in a
direction correcting the existing inclinalion. Nevertheless, as it call be seen in Fig
12, the al110unt of the correction has been ralher small, less than that
corresponding to the rotation of lhe fioor in lhe same position, lhus indicating Ihal
Ille vertical struclural l11embers <Ire nol completely fixed at Ihcir bases.

5. STRUCTURAL MONITOR I NG AND TESTING

Several parai lei IllcaSllrClllcnl sySIClllS have becn implemcnted lo monitor the
displacelllcnts of lhe struClure . !3imonthly level!ing of lhe position of several
hllndrecl poinls of the struclllre is perfol"lned through hjgh~accuracy surveying.
Additionally, lhe slope oI' lhe columns aI differenl heights is being Illonilorcd
tllrough eleclronic inclinomelers. The posilion anel width of cracks in the lllain
slruclural elemenlS is checked ever)' monlh. Ali lhe inforlllalion is processed in
a cOlllputer SyslCIll providing Illaps <lnd graphs of lhe deformalions.

Additional1y, some more sophisticaled. special pllrpose measurelllenl syslems have

been implemcntecl. The loads taken by the shoring towers afe delerlllined by
measuring the unit Slrain in the vertical pipes throllgh 312 slrain gages. An
autolllated contillllOllS lllonitoring systelll !las been providcd by lhe ISMES Inslilllte
of Bergall1o, Ilaly. It is composed by 4 telecoordinollleters Ihat Illcasure the
change in lhe inclinalion of lhe colulllns: lell exlcnsollleters measure lhc changc
in lhe slxlIl of arc!les; four temperalure transducers are located inside and oulsidc
the lemple. Ali lhe instrurnenls are cOllllecled lo a celllral data acquisilioll sy~[C'm
which takes aUlOmatic readings every four hours.

Fig II shows an exalllple oI" such Illcasuremenls (rem July 1994 lo Scptelllber
1995 . The a!lllost conslanl rale of correetiol1 of COIUlllll slope can be apprecialed,
cxcepl for lhe pcriod whcn the undcrexcavation works \Vere Intcrruplcd duc to
adlllinislrative problellls . The cOlllinuolls Illoniloring systelll will be Cunher
cllhanccd aI lhe end of 1995.

Another illlponant ser of meaSllrements pcrforllled hy ISMES i.') lhe lIleasurement

136 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

-....

Initiof out-of-pfumb in percent

Correc lion of lhe slope in percenl

Fig 12 Comparison Df initial out~ of-p/lImb wirh lhe correclion of lhe slope
altamed by lhe underexcavalion
 R. MEU nnd A. R. SÁNC HEZ RAMÍREZ I Thc Mcx.ico Cily Cathedral 137

TELECOORDINOMETER (COLUMN C7)

(JUL 94/ SEI' 95)

",.
/'
V ~
".
/ ,.... V ~
E
S ... / /'
V~

1 ---
t~
~ "....-/ t
E
•
E I

~.
! U/OO/9
~
, """ """ "
50 - 24{(J7JQ
;5

00
'}/ ---

' 00 "O
"'"
Time (days) "'" "'" .'"

EXTENSOMETER AXIS 7
PUl 94!S[1' 95)

"
o. M
~
--.
·0.5
t, '1A/U71'1
Ê
'" ---
~ f--
É.

I " ~ I
!
14 "~11"

.""
!• '0 1
"\r--+-
~
IAIJfJ!>

ls.
I I
" '--f - '
lO
I~
'-v'"
"• '00
"'" "'" ""
Tifl'lC(doys) "'" '" '00
"'"
Fi~ /J Graphic t!f tlw c!UlIIgl' in Ilu' sfopC' 01 a ('oluml/ aJul til(' .\'l!aru' l/ing
(1 01/(' Cl'lIfral areI!, lrom 111(' f{'cOI'(lillg (1 ,lu' contimwlIs
mOf/ilOriflg -")'.\'ll'IlI
138 STRUCTURAL ANALYSIS OF HISTORICAL CONSTIWCTIONS

oI' l he state of st resses i n COIUIllIl S and walls. by l he nat-jack techn iqu c dcvcloped
by tha! institute for ils application to masonry struc tures ( ReI'. 2).

Bricfly, l he lcchniquc consists of insert ing vcry thin na! jac::ks i n g roves previullsly
open in lhe joints between stones, an el in mcasuring lhe pressure I1ccdcd to
eli minate lhe shortcning oI' lhe joint duc to lhe cutting procedurc.

As previously cOl1l11lcnted l he leveis of Slress detcl"lnincd by this Icchnique were

(ollnd to bc significantl y greatcr Ihall thosc com::spondin g to a linear el<tsti c
analysis oI' lhe whoJ e Sl fllclure. I\nolhcr , more detailed, 1"0\111<1 oI' IllcasurcnWllls
\Vi II be ca rrit.!d oul in l<l !e 1995 .

Th!.! I1lcchan ica l propertics of lhe main st rllc[ural materiaIs have bl!l!n clirectly
delermincd in lhe building. Salllple!l have been l!xtractecl rrolll colulllns and walls
10 determine unil weighls, moduli ar elasticily , an el cOlllpressive anel lensilc
s lr~n gl h s .

Frorn Ihese lesls, lhe arrangemcnt of ~tone!l \\ ilhin lhe cnlumns cross seclion cou ld
bc determ ined (Fig 14) , along wilh Illcir crkl.'livt' rôi\ling arca. The mosl
signi lican t linding was th,H the qua l il)' 01 \\one'\ \Va ... changcd at lhe IOp 01' lhe
co lullln s afler one of lhe man)' inlerruplionl, 01' Ihc COIlSlfuclion \\'orks. \Vhilc in
lhe tower pari a high qualily SIOlle \\ilS llsed with " comp rcv.,ive !llrenglh of
approximatcly lO Mpa, in lhe upper pari oI' Illl: COh11111l\ lhe ... In.:nglh i!l belwecn
8 ;\!lcl 10 Mp;\. This finding 5ho\\'cd Ihal Ih!.! sa fdy f<lcLnr ag,lill\\ lhe collapsc oI'
some colulll ns due to eccenlric compre..,\ion is nl'ar 10 Ol1ç.

6. CONCLUSIONS

A very favorable pallern ar dirrerenlial .~t~lllelllcnts has hecn producé'd by lhe

underex cavalion work. Thc maximulll differenLial senlcml..'nl of thl..' tempk ha\
becn reduced by 420 1111n durin g three years oI' \~ork (one of prcparalion and 1\\'0
o f ul1c1erexcavalion). represellting almoSI 18% of Lhe lotaI.

A vic\\' 01' the correc l ioll in lhe shape 01' s~lIlement~ 01' lhe maill facade i~ given
in Fig 15 .

The span 01' lhe cen tral nave has experiçnced a clo!'ing oI' 32 111111. AhnoSI ali
col untn have reduced tlleir slopc . For lhe 1110S1 critica I column bcncath lhe centr<ll
clame, a reduc l ioll of abouI 10% in lhe oul oI' plulllb \\íl!l achit:ved.

Avi e\\' 01' lh e rela tive posilion 01' tlle upper pari 01' lhe COIUlllllS wilh n::gard to
Ihei r base i s givcn in Fig 16. A s il cal1 be <Ippreciat~d Ill<lny of lhe COIUlllll!l are
still in a very c ri ticaI condilions in Icrl))S uI' Iheir out -o f plumb.

II is not clcar aI Ihis timc how far t!ll..' co rrcclion proceS!I could go. Tlle rolalion
oI' lhe 1ll0nU Illenl as a ri gi d body can hl! perforllled wilh no major 11<In11 to the
SlruClure. Neverth elc~s , lhe inward rOlalion oI' Ih!.! 1'Il!.!r,\1 l1avt:~, aJollg with olher
local correctio!ls . face Ihc i!lcrea~illg OPp(1.~jtion oI' lhe \truclilral sliffncl,s . II is
 R, MEL! ano A. R. SÁNCHEZ RAMíREZ I Thc Mcxico City Cathcdra! ! 39

~ Poor mosonry

Dimensions, mm

Fig 14 Cross secriOll and sfOne arrangemefUs for Iypical columns

-025
]: -0.3o
-.... r--- ~~.~~~
~ -0.35
,.. 33./0.7
~ -0..4O V "" "- >< 3270.&
~ -0. 45
i;j
/'" """ --- 'C" f"'" ~
32.30~
g
~b/
-O 5 Jl.90~
-05 JI.SOe::
A B c o [ r G H J •
AJ"S

Fig /5 Fronl view of lhe remple wilh ils original profile and lhe correcrion
achieved
140 STRUCTURAL ANALYSIS OF HISTOR ICAL CONSTRUCTIONS

anticipated lha! aI leasl twice lhe amounl a f l he presen t correction cou ld be given
witholll major prob lc llls. 1'l1is will Icave lhe slrUClUre in a more stable position .
NcvertheJess. il is likely 111a1 some struclllra l rctrofitting will be needed aI lhe end
af lhe process, espccially in lhe COltllllllS oI' lhe central nave whose out-of-plumb
would be co rrected on ly in ti minar parI.

The rehabili tation lechnique adopted htl'\ never heen used before for a bui lding of
thi s size, complexity anel materiais. Thercfore, lhe dctai led analysis af lhe effeets
and lhe close monitoring af lhe slfllcturaJ response are csscntial in arder to select
lhe mos! convenient and safe progralll oI' work. Up to Ih is mOlllcnt, lhe results
have been complctely salisfaclory, nevertheless lhe moSI criticai part of lhe
process is sl ill to come.

REF ERENCES

1. Talll cz E., E. Santoyoand A. Cuevas. "The Metropoli tan Cathedral in Mexico

Cily. Correct ion of lhe Bchavior of il s Foundation". Mcxiean Socicly Df Soi!
M cchanics, M exico. aug 1992.

2. Modena c., P. Rossi. " In silll invesligations. slructural analysis and

slreng th cning of a slone masonry be l llowers". ISM ES. l la ly. july 1992.

ACKNOWLEDGMENTS

The rehabilita tion project for lhe M exico City Cathedral is direcled by Sergio
Zaldivar, head of lhe Office 01' Hi stori e Monlllllcnls. Enrique Talllez and En rique
Sí\l1loyo are responsible of lhe underex cava tion projecI and of i15 i mplel1lentation.
 STRUCTURAL ANALYSlS OF HlSTORICAL CONSTRUCTIONS
P. Roca, J.L. González, A.R. Mari and E. Oftale (Eds.)
© CIMNE, Barcelona 1996

THE ANGKOR'S TEMPLE IN CAMBODIA: PRELIMINARY

STRUCTURAL STUDIES ANO STRATEGY OF
INTERVENTION
G.Croci
Ulliversi/y of Rome "La SapiCllza"
Viu Eudu::.::.ill/!fI 18
I~00184 Roma, Italy

SUMMARY
Angkor represenls a comp lex of temples built belwecn lhe II Ih and 13th centuries in lhe
aneienl capitais of lhe Khmer Empire Angkor's Temples, very important examples Df
brickwork masonry and dty stone bloek Slructure, was built in lhe middle of the
Carnbodian rain forest and they are neglected till 19th cent ury
Luxuriant vegetation and seaso nal rnonsoon flooding are lhe mos! important causes 01'
structures decay and partial or total eollapse.
In order to avoid the use of reinforced concrele, and to propose new strategies for
safeguard inlerven lion , which are more sllitable and revcrsiblc, if possible, some studies
are carried out
In Ihi s pape r the preliminary strllctu ral bchaviour studies are presenled and related
possiblc stnltegies orinlervenlion are proposed .
I. HI STORI CAL SURVEY
Angkor represents a com plcx of lemples mainly bu ilt between lhe IX and XI II
centuries; Angkor Thom and Angkor VaI, the aneien! capitais, are a witness of lhe
splendou r ofthc Khmer Empire (fig. I) .
A peculiar charactcrislic of lhe main Icmples is lhe pyramidal shape that the complex of
lowers, galleries and Icrraccs assumcd as a whole, so thal usually they are called
"lllo11ntain temples" (fig 2): lhe Bakong, IX century, is Orle 01' lhe first cxamples of this
co ncepli on .
In lhe X ccnlllly lhe "tllollnlain tem pie" 01' Pre Rup was built and is one o f lhe lasl
irnpor1anl exatllples oI' brickwork Illasonry belore il was substitutcd by dry sto ne block
struclures. The ternples were progrcssivcly surrounded by a system of barray and
tlloals. Following lhe Iradition of lhe Khrner ruler vcry frequently each king abandoned
the previo us capital lo build a new one always in lhe some area; lhe sp lendid "Baphon
lemple" was buil! in thc X I centu lY; lhe houses and Royal Palaces, differenlJy from the
temp les, were built on timber so Ihal no trace oflhe buildings remains today
In lhe X II cenlllly "Angkor VaI", undoubtcdly lhe masterpiece of Khmer al1 was buil!;
shor11y aHerwards lhe eapila lmoved lO Angkor Thorn, a large arca in the centre ofwhich
ri ses lhe IlluulLlailL lcrnplc "Bayon" . The cnd 01' lhe XII ccntury represem lhe end of lhe
construct ion of these exceplional lemples, although the splendour of Angkor continued
for anolher centUly. Angkor was lina1Jy abandoned in lhe XV cenlury and il has been
only in lhe halfofthe XIX century lha! Angkor began lo be rediscovered to show to lhe
world lhe exeeptional, the rigorous geometrical organisat ion, of ils architecture and lhe
incredib le unily in a strucl ural conccpti on Ibat has developed Ihrough four centuries.
2. THE MATERIALS AND THE DECA Y
2. I Soil and t'oundation
The soil of lhe area \Vere Angkor rises is substanlially madc of sand and clay; its
bchaviour is conditioned by lhe prcsence of waler bOlh as deep and superficial leveis
relaled to lhe seasonal rains
T he temp lcs generally are based 011 a layer or compaeled sand afier having removed the
superficial parto The foundations, that anyway remain superficial, are realised usually
142 STRUCTURAL ANALYS IS OF HISTORICAL CONST RUCTIONS

Fig I : AII$!,kuI" Vaf

i'ig 2: Tlle AIIKkor's "mO/llllail1 temple.\""

G" eROO! The Angkor's Tcmpk in C:lInbodia 143

Fig 3: Dolt/ages dl/e 10

lhe !)"oi! .\"eUlemellls

Fig..J: Sleel slirrlljJs 10

co1llraSl sfone flake oi!
144 STRUCTURAL ANALYS IS DF H1STOR1 CAL CONSTR UCTlON S

Fig 5: lJefl'noral i ol/ dl/e

lo lhe r o()!s
 G. CROCJ I Th!.; AngJ..:.or's Tl.:mpJc in Cambodia 145

with lat c rit e blocks and oOen !lave sullered relevant selllelllenls producing inc1inations
and dalllages to lhe strllClures (lig 3), lhe cause nf lhe seltlements is likcly due lo lhe
cyelical variatioll of lhe w,Her lable rclated lo lhe seasonallllonsoon rainfall , il is possible
however Ihal the varialioll of lhe hydraulic condilions especially dllring the cen luries
when Angkor ha s bcen abandoned have playcd a rcleva nt role as wel!

22 StnJclural materiais
Three are lhe main kind oflllilleliab Ihal h,l VC becn used
- brickwork slructllrcs made by sun cooked brick5, connecled with a monar Iha l in
rea lity is a kind ofvegclable g lue (pre Rup) .
- lalerile it is a reddish very porous material used as blocks for lhe foundaliol15: il1 some
cases il is uscd for wa ll s and lhe lower pari 01' slruc lures The slrenglh is good especially
10 the surti.cc where an hard cruSI (aboul 20 em) is prodll ced, benealh Ihis CnJ st lhe
material is yclJower and so l h.:r~ lIslIally lhe deeay is limited
- sandstollc (grcs) it is a more compact material uSlIally g rcyi5h used as blocks 101' lhe
conslruclion oflhe bllildings Tbe strcnglb is good blll it sufTers a decay subsla nlially due
to the wa ler (rising waler fór capi llarily. e lc ) espccially visible aI the basis of Ine piers.
so mc picrs bcsidcs Ilake ofl lo ngi tll dinally as in the qllarries Ihey were cut horizontally,
lo llowing the natural st ratiti calion of lhe sandstone which represent lines of
discontinuity. vely freq uent sleel sti rrups to con lrast Ihis phenomenon arc visible (fig 4) .
A peculiar kind of coarsed-gained sa ndslone, Is represented by Ine arkose Ihat Is also
frcquently used ror lhe linlcls and the small COIUlllllS in lhe brick buildings

23 Thc decay
Thc causes 01' dccay, besides lhe natural processes, can be related lo
physioehemical-biogeOl:helllical aspccIs. 10 lhe abnormolls devclopmenl o f lhe
vegclalion especially in tne cenluries when Allgkor was aballdoned, so Ihat even Irecs
over lhe ICllIples (fig 5) ha ve dcvclop alld Ihc roots could enlargc thcjoinls and displace
lhe block s 11 i5 lhe \Va lcr ho\Vcvcr Ihal ha5 playcd lhe main role penetrating inlú lhe
slones dlle lo Ihc rain. Ihc percolation and lhe risillg walers, thesc wa lers, rich of
minerais brought fi'o lll lhe 50il, caused dissolulion and oxidation oflhe slone il self. Little
by little lhe cohesion was brokcl1 dO\VIl, and di si nlegralion progressed, phenomena of
e ro sion and l1aking oflhc sandslo ne are frcqucll[ (fig 4)

3 TlI E STRUCTU RAL BEIIA VIOUR

3. The sl ru ctura l ty pologies

We ca n consid er thrcc main Iypc 01' slruclures
- lhe galleries and Iheir cOllncclions (Iigs 5.6)
- lhe lowers (fig 7)
• lhe slructure5 lllassive5 which provides lhe Icrra ces of the "Illounlain lemples" (fig 8);
these st ruc lures are partieularly subjcclcd 10 the cOcei 01' lhe rainwatcr wl1ich penetrales
through lhe joinls crea ling slIpplcmcll lary prcssures 0/1 the retaining waUs and possible
selllemenls on the foundalions. lhe now oul of lhe \Vater, besides, creales possible
seepage and loosing of lhe sandfill w hich represcnls a cause of funher deformalions.
The slructures reali sed wil h dry blocks presel1t vertical joints little e flccli ve as lhe
overlapping is insuflicienl OI' IOlally absenl 50 Ihal lhe longitudinal slnJclura l con linuil y i5
mi ssed o r very \Veak, Ihis sl ruc tural imperfcclion could bc juslified by thc need to leI the
5tructllres p1ll11y lO lollow lhe settlements oflhe soil. probably we1l known to the Khmer
builders, due 10 lh e seasonal varialion of lhe \Valer, during the dry season in fact the
Ircnd of lhe soil is lO contracl, whilst in the 1ll0llS00n seasol1lhe soi l expands .
 146 STRUCTU RA L ANALYSIS OF I-II STOR ICAL CONSTRUCTI ONS

Fig 7: A IVlVer ~f
AIIgkor VaI

Fig 8: lhe ma:;sil'f! Il!rrCICe slrllc ture

 G. CROC I I Thc Angkor's Tcrnplc 111 C:unbod ia 14 7

3.2 The vaults

The main slruclural characleristic of the Khmer's architecture is lhe conception of
lhe vaulls Ihal are realised wilh corbelled blocks wilh horizontal joinls instead than with
lhe struclurally more efficienl wedge shaped blocks with radial joints as currently used in
lhe European architecture; lhe former in fact presents a lower bearing capacity and a
larger sensitivily to any rotalion of lhe foundalions (fig. 9).
A malhematical analysis of Ihi s kind of vaults shows anyway Ihal in relation with lhe
limiled spans, lhe Slresses are cornpatible wilh Iht;: rnalt.!rials (fig . 14) and in particular lhe
horizontal shear stresses in the joints do not exceed the friction strengl h; in other words
lhe inclination of lhe resultant force remains inside lhe friclion cone (figs. 10, 11). The
vauhs Ihen, in lhe o riginal situation, were stab le. The cho ice of the Khmer buildcrs does
nol appear Iherefore related neither to lack of kn owledge nor to "ritual beli ves" as
somelime il has been said ("lhe wedged arches does nol rest , only lhe corbelled arches
repose") ; it is likely instead thal Ihis Solulion came from a precise evaluation of lhe
following advanlages:
- simpler cutting being parallel the two main Opposile faces;
- simpler and quicker construction as il was possible lhe firsl blocks lO be placed without
centering; lhe upper ones bcsid es cou ld take advanlage of a certain longitudinal co-
operalion finding support on lhe blocks of lhe arches already buill , 50 lha! ali lhe vaull
cou ld have been realised without centering;
- efficient water proof; in facI whilst in the radial joints lhe waler can vel)' easi ly
penetrate, in the horizontal joints il becames more difficult. In some vaults lhe
walerproofing is further improved realising lhe joint s with a slight outward inc\ination
although this however worsens lhe friction slrenglh; this lechnique is probably relaled
with the large tradition o f lhe Khmer builders in the co nslruction of wood roofs, where
the tiles are just superposed wilh an outward incJination.
The slide between lhe blocks (fig . 12), Ihat easierly occurs ai lhe upper leveis were the
load is lower, can be related lO a reduction of lhe friction due to lhe wealhering of lhe
surfaces (vegetation, lichens, bacleria, elc.). The main more dangerous cause however is
due lo lhe increase of the horizontal forces produced by settlements and rOlations o f lhe
foundati ons; when Ihere is an oulward rotalion more Ihan lhe sliding it is a geometrically
instability oflhe blocks to be easily reached (figs 13 , 14).
In gallcries just one side stands on a continuous wall and the olher on disconlinuous
pillars; in Ihese cases lhe gallery on this side is compl eted by half vaults associated with
some horizontal struts, 10 ensure lhe contrast; oRen these slruts have been broken (and
subsequenlly rebuilt) due to rei ative vertical settlemenls so lha! a slide belween lhe
blocks has been produced .

3.3 The towers

The towers follow lhe same general concept ion of lhe vaults with two main
adva nlages:
- the const ruclion is easier as il is possible to grow with horizontal rings withoul any
support;
- the shear forces are lower so that lhe ri sk o f sl iding bet ween lhe block s is usually
excJuded.
On lhe other hand lhe highness of lhe towers associated with lhe lack of efficient
connections on lhe vertical planes makes these Slru Clures very weak to even little
differential soil settlements which create supplementary horizontal Slresses and wide lhe
joints; till unstable silualions and collapses of some blocks can be crealed. The
progressive developmenl of detachmenls is clearly highlighted by a non linear analysis
 148 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCT10NS

Fi).:.Y: (·ro.\,\' ,\'1.:(."//(111 \ '1('11' 'l li ('(whdkd 1'(1/(1,

hg lO: JJ{ffercl/I kmd,\' 'lore:/,,:.\' he!J(l\'/(Jffr

 G. CROOI The Angkor's Temple in C~mbodia 149

limi t friction
cone

true orch corbel orch

Flg. !! : !·"'c llo!Jal he/UII'IUIII" (/11(/ III/II! ("(Jlle

Fi;..:. 12: "lhe -'"lide ht'llt 'l'L' /I lhe IIppel" h!ock.,

150 STRUCTURAL ANALYSIS OF HISTORI CA L CONSTRUCTIONS

Sai I settlements: inwa rd rotation

t hrust
Ihear
.tflngt~

AB C O N

increcsl'! the thrust

- possible sliding of the bJocks @
- ne w equilibrium with deformed shope (J)

Ng 13

Sai l settlements: outward rotation

the thfust does not Bubstontiolly chonge

- rocking af the blocks
- possibJe instobility of the shope
- reduction of the sheor strength as the contoct between the blocks
is reduced

Fig/4
 G. CROC I I Thc A ngkor's Ternplc in Carnbodi.:l 15 1

which shows how lensile slresses are produced in the corners and in correspondence of
the lintels (figs. 15, 16).
In lhe dry block slructures lhe crack and possible fai lure af lhe lintel can bring to a
progressive phenomenon ofloosing and then collapse oflhe overhangi ng blocks.

4. STRATEGY OF INTERVENTION
The Angkor temples presenl a so large variety of situations, decay and damages
Ihal each one requires a differenl solution; besides lhe studies are nOI presently
compleled 50 that JUSI some general criteria will be presented taking in mind thal the
solution muSI be found balancing lhe minimum alteration with the stability requirements .
We can distinguish substantially two situations (fig . 17):
a) Interventions on lhe foundations. these interventians are necessary in lhe case of non
slabilised soil settlements; it must be realised anyway that some cyelical deformations
due to the cyelical variation of the waler table cannot be eliminated.
When lhe deformalions have became incompatible with lhe structure stabilily it would be
necessary to recover part af lhe settlemenls and inclinations with special eq uipments
Uacks, etc.).
b) Interventions on lhe structures; these interventians musl mainly achieve one or bolh
lhe tasks :
• to improve lhe friction in the horizontal joints; this can be obtained, insertillg some
pieces oft he same original stone (laterite or sandstone), or of other suitable material , in
some holes drilled in lhe horizontal plans of the joints, alternatively incJined in two
different directions, these holes can be realised fram lhe interior wilhout reaching the
externai surfaces;
- 10 improve or to create global horizontal conneclions; solutions of this kind can be
obtained inserting transversal tie rods, (treated against lhe corrosion), slightly
preslressed .
 152 STRUCTURAL ANALYSlS OF HI$TORICAL CONSTRUCTIONS

Fi)!.. 15: Maflll:J/Il1liclIllllodel (!! I 'I"!' /(111'\ !OIl'CI": dcat! /o(/(I m}(l~I 'sis

J'ig lfí: (ú/Ck's /11 lhe

10m'/' Imlel
 G. CROCII The Angkor's Tcmple in Cambodia 153

Interven tion crit erio

al Carrection af lhe ratations a t the foundol ion levei (big d is plocements)

- Inlerver'llionB on lhe
foundolionB ..a t- oetive
forc .. (jock, •... )
- monitorln; to contrai
th. d.lormot ion

b) Improvement of the structurol behoviour (little displocements)

- intervenHon on the joinlS lo inerees e lhe st-eer resistonce

- red uetion of lhe thrusl wilh tie bor~
thrust

H· lo H

~ BC D N

Fig / 7
 STRUCTURAL ANA LYSIS OF HISTORICA L CONSTRUCTIONS
P. Roca, l.L González, A.R. Marí and E. Oilate (Eds.)
© CIMNE, Barcelona 1996

THE COLOSSEUM: SAFETY EV ALUATION AND

PRELIMINARY CRITERIA OF INTERVENTION

G. Croci
Uni versity of Rome "lA Sapienza"
Via Elldossiana 18
/-00184 Roma, Ita/y

SUMMARY
For many years the Colosseum has been the object of interdisciplinary research aimed
at assessing its structural behaviour and safety leveI. Studies of the historical
documentation, in situ investigation and analytical modelling have allowed us to
understand lhe partial collapse causes.
In this paper we are too proposed some mathemalical madels, which are carried out to
allow lhe safety assessment in present situalian after the restoratian works in the 19th
century.
The results are discussed and the weakness of the monument with regard to
earthquakes is outlined. The zones requiring strengthening are studied in delail and
some preliminary criteria ofintervenli on are praposed.

1. THE HISTORY AND THE STRUCTURAL BEHAVIOUR

I. From 72 a.O. to 443 a.O. - The first centuri es

I. I The hi story
The Flavius Amphilheatre, whose construction lasted about ten years. was
dedicated and opcned to public in 80 a.O ., during lhe reign of Titus (tig. I) . The
works presumably started in 72 a.O. with lhe simu ltaneous co nstruction offour similar
portions separated on plan by lhe two principal axes of symmetry; Ihis explains the
speed of the reali sation, some structural imperfections, such as lhe in sufficient bearing
lenglh of lhe blocks in some zones, and lhe prcsumably weak connection o f the four
porti ons in correspondence to lhe two principal axcs .
The Colosseum was covered with a velariu m whose documentation has been
completely lost; a possible configuration designed on the basis of st ructural analysis,
lha! we have carried out taking into account the characteristic of the hemp ropcs that
probab ly have been used
Th e second peculiarily of the construction, is lhe site Ihat has probably played an
important role in the damages and col1apses as we shall see later. The site was
originally the lake of lhe Domus Aurea and was then partially filled to create the basis
for lhe foundations; the result was hetcrogeneous soil and foundations (partially
represented by a concrete bed) and an alteration of the hydrographi c settlemenl.
In the first centuries ofthe Colosseum's life, il was affected by fire s, small earthquakes
and flo ods. In the 217 a. D. a tire, caused by lightening, bumt the timber structures Df
the top roof; importam fires also occurred in 250 a.D. and 321 a.O. It is important
however 10 observe that . although lhe restorati on works following lhe 2 17 tire were
reco rded as having taken place over tive years, it is unlikely that they substantially
affected the stonc element s calcined by lhe tire or lhe masonry damaged by lhe fl ood.
However, in spil e of these events, the Colosseum maintained lhe original strength and
soundness.
 G. CROCI I Thc Colosscurn 155

Fig. I: Viell' ofColosseum

L \ n r \, r! II 1 r \ ,~ ! .. ,

Fig. 2: lhe Colosselll11

after lhe earlhqllake of
}3~9
156 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

1. 1.2 Mathematical models related to lhe dead load

In arder to verify the original safety leveI s a first finite element mathematical
medel considered the Colosseum in it s original form o
The static analysis has demonstrated tha! lhe original structure performed well ,
showing lhe sk ill and fo relhought Df lhe ancient designers and builders. The structure
appears to be aver dimensioned, reaching a maximum compressive stress of only 24
KN/m 2 at lhe base af the travertine picrs. Whilsl the strength is of many hundred s
KN/m2
The medel has a1so shown lha! lhe arches inserted in lhe elliptical wall s induce a
horizontal thrust with a radial componen l corresponding to each pier, lhal is added to
lhe thrust produced by the ambulatory vaults; the corresponding tensile hooping
st resses are anyway lesscr Ihan I KN/m2.

1.2 From 443 to 1703 - The Four Slrong Earthquakes

1.2. 1 The history

The first important event thal effectively caused some destructions of lhe
Tllonument was the earthquake in 443 a.O., estimalcd of VIII-IX grade Merca11i Scale,
with lh e epicentre in lhe roman region. Paol us Oiaconus in hi s "Historia Ro mana " said
tha! "tam terribili terremolu Roma concussa est, uI plurimae aedes eius et aedificia
co rruerinl ".
The Colosseum suffered damages on lhe sitting grades, in lhe arena and podiulll areas
and at the attic leveI. 11 look three consulales to be reslored and some of the works
ca n sti11 be secn a! the 10p of lhe externaI wall, where a chaotic cyclopic masonry was
seI.
After this, news about the building seems to fade: lhe las! ludo mceting is recorded in
the year 523 a.D.; lhen, in the year 663 a.O., Constantinus 111 ordered lhe tearing
down of lhe bronze plalcs of the atti c. 1I is believed thal in this period lhe locking
slirrups amo llg the blocks were also taken Oll" Ihis spoiling callscd lhe ruplu re of
many blocks, Ihus 10ca11y modifying the stress stale dist ribulio n. After lhe 5th century
the Co losseum was nOI only disllsed and abandoned but also sma11 earthqu akes
occurred and expanding plant rOOlS opened exist ing c racks and created ncw ones.
There was Ihen a IX grade carthquake in May 801. with probable epicentre in the
Abbruzzese Appennino . The cyclopic order columns a! lhe attic in ner levei fe 11 into the
cavea, causi ng huge damages in the inclined barrei vaults and arena; other less striking
structural cracks also accurred . We believe that thi s is lhe dat e a f the loss of continuity
at the top of lhe e11iptical wa11s. The next information about the monument is dated at
lhe beginning o f lhe 12th century. lt was Ihen inhabited by patrician families
(Frangipane and Annibaldi) who fartified the Iwo lower leveis af the sout h·eastcm
side.
Unfortunately there is no iconography available of this period, with lhe exception of a
commemarative medal ofLudwig the Bavarian (year 1328) lhat shaws the Calosseu m
behind o lher buildings wilh the upper row sti ll intact; the occasion of lhe
representation, lhe cOIllIllcmoration of lhe new emperor, and the small scale of the
piclure expIa in the lack af reliabil ity. Minor earthquakes occurred in 1255. 1287,
1300, 1321,1348
Even if lhe pope Innocenzo IV ordered lhe destruclio l1 of lhe fortificat ions, lhis must
have been more or less ho w the Colosseum was in 1349 when anather destructive
earthquake occurred in Rome \Vith VIII · IX grade. The most famous witness, Petrarca,
said "magna clIm partem collapsat" . This carthquakc. the most destructive in central
 G. CROClI Thc Colosscunl 157

Italy, with epicentrum in the Umbro-Abbruzzese, produced the complete failure of the
two externai cylindrical walls on the Celio si de of lhe Colosseum.
Afier lhe earthquake of 1349 the Colosseum, reduced to a ruin, was abandoned and in
1362 the heap of rubble, especially from lhe IWO externai walls, constituted a small
hill, named "Coxa Colisei" (fig . 2). The use of lhe materiais (marble and brick) was a
bane of contenlion belween the reman people, the Frangipane famil y and the
represenlatives of Pope Urbano VIII , as il was required for lhe conslruction of some
roman palaces and St. Peter's Basilica.
It seems, anY'-vay, that the removais were usually from lhe fallen portions; there is no
evidence of demolition except for Iimited ones authorised by the popes.
Some "spontaneous collapses" are recorded in 1646 and 1689; lhe cause of these
phenomena Ihat cannOI be direclly related to particularly events, nor to seismic
aCl ions, will be explained later.
In 1703 the fourth and last slrong earthquake occurred; lhe origin was once again in
the Appennino region and crealed wide destrucli ons in lhe city af L'Aquila. Although
lhe energy characteristics were similar to thase of 1349 earthquake, the damages were
much less extensíve, as just one arch is recorded as havíng coll apsed in the outer wall
and three arches in lhe second wall ; ít appears therefore to be a reverse in the trend of
the increase af damages and failures in the succession of lhe earthquake of 443-80 1-
1349. We have nol aI present complete data to explain this differenl behaviour; one
possible explanation however can lie in lhe different characterislics of the soil beneath
the foundalions: Iherefore, once lhe Slructures on lhe weaker bed have collapsed the
remaining have a greater resistance . Deeper analysis and invesligations are presently in
progress o

1.2.2 Mathematica l models relat ed to seismic aClions

The theoricaJ behaviour ofthe Colosseum under lhe effecl of seismic acti o n ha s
been analysed slarting with a global elastic linite element model laking into account
lhe differenl characteristics oflhe soil
To quanlity lhe seismic aClion lo be used in lhe malhematical ana lysis, two ditTcrent
melhods. which gave similar resulls, have been considered :
a) lhe hislorical data aboul seismic events recorded in the area : il can be seen Ihat an
earthquake of VIII 10 IX grade. in M.S. scale, has a return pcriod of abou! 500
hundred years;
b) sludies 00 seismic characteristics of lhe area ; some studies assessed Ihal a Richler
inlensity of 6.68 can be attached lo a period of 400 to 500 yea rs and therefore a
ground acceleralion of 0.05 to 0.06g can be deduced . Evaluating lhe amplificalion of
the masonry structure up lo 25 to 3 limes, a design value of 0. 15g, is obtained. very
elose to the value 0. 16 slated in the Italian Seismic Code for lhe 3rd calegory area. this
Is the valuc that we have adopled .
The models indicate that lhe zone of higher stresses is in the allic wall. and in
pan icular in lhe zone of the minor axis toward s lhe Colle Celio; thi s is the area where
lhe firsl breach in lhe wall may have occurred. In fac! il is here Ihal the maximum
horizontal stress (up to 8 KglClll 2, figs. 3, 4), due 10 seismic action in lhe x-direclion.
is reached . This levei of tension cannot be sustained by the friclion present in lhe
horizontal dry joints ofthe blocks because oflhe low vertical Joad .
Similar results have been reached with lhe seislllic action in y-direction . A delailed
examination of the resulls of the seismie analysis allows us to consider Ihal lhe allic
wall could reach a criticai condition near the minor axis (on lhe Celio side) with a large
range of possibJe eanhquake direclions as opposed lo lhe small range of earthquake
direclions which could afreel lhe wall around lhe major axis.
158 ST RUCTURAL ANALYSIS OF HI STO RI CAL CONSTR UCTIONS

Fig. 3: Tensile horizOJJfal slresses due lo lhe seismic aclioll in lhe X-dire clioll
correspolldúlg lo lhe 3nd mude

"\ :'.',~ .....

~
.,-

/ , \

".

", .

-.0 4 , .' !· l . ... ;

Fig. 4: Tensile horizonlal slresses due lo lhe seismic aclioll in lhe Y-direcliolJ
 G. CROCI I Thc Colosseum 159

Another zone where the model shows a criticai situation is in the first levei of piers in
lhe inner elliplical wall, near lhe major axis. Particularly wilh seismic aClion in lhe x-
direction some piers' sections are partialised by lhe presence of tensile stresses and
reach compression of about 50 KN/m 2.
The tensile stresses anyway are not high enough to cause lhe visible damages of lhe
first levei piers and it is evidenl thal a single eanhquake could nol have caused lhe ruin
of a large portion of lhe monument. It is necessary lO inlerprel lhe elastic and post-
elastic behaviour of lhe Co losseum in its original configuration to understand tht: leal
phenomena. Referring to fig o 5, during an earthquake, while lhe vertical axial force
"N" remains constant, there is a transfer of effort, from lhe stresses, "Til, which
characlerise lhe circurnferenlial behaviour, to lhe bending behaviour "M", in the pillars,
This behaviour is schemalically represented in fig o 6: in the elastic phase il is lhe
circumferenlial tensile st resses in lhe elliplical walls which give the main contribulion;
il is only after they reach Iheir peak value linked 10 lhe friclional limil ofthe horizontal
joint, lhat larger deformations (qualitatively represenled by õ in the diagram), and
relevant bending moments are generated in lhe pillars. In conclusion, we pass
progressively fram an "N-T" behaviour to an "N-(T)-M" behaviour, 11 is important to
note Ihat, after lhe firsl earthquake, permanenl deformations (and thus permanent
bending moments) remain so lhat lhe global behaviour, even for low levei seismic
actions, becomes weaker and weaker; fig o 7, show lhe sinusoidal deformations. signs
ofthe earthquakes on lhe externai wall oflhe Coli seum.
This interpretation of lhe phenomena is in accordance wilh the hislorical document s
Ihat indicate Ihat lhe two earthquakes of 443 and 801 caused only a loosening of lhe
masonry in lhe allic levei , as well as in lhe horizonlal bands of lhe lower levei s, with
consequenl sliding between the blocks, particularly in lhe zçme oft he minar axis on lhe
Celio side. These sliding movemenlS caused incremcnls in lhe annular length and Ihus
oul of plumb Moreover this iniliated lhe vert ical cracks which began in lhe attic wall
and which were leR. unrepaired after the earthquake of 80 1. Despile lhe small ent il Y of
Ihese damages. Ihey comprimised the efficiency of the Iwo-dimensional behaviour of
'lhe attic wall , so Ihal il was more easily damaged during successive minor earthquakes
( 1255, 1287, 1300, 1348), This weakness of lhe attic wall, which extended to lhe
lower leveis, explains lhe ruin of lhe large portion of lhe IwO ou ler elliptical walls near
the minor axis on the Celio side during lhe earthquake of 1349.
An interesting queslion is whether lhe failure of the Celio si de is mostly due to the
weaker foundalions. In effect lhe remaining side shows tensions thal surpass lhe limit
for lhe iniliation of frictional movement (wilh lhe seismic aClion in y-direclion in lhe
malhemalical model) . Thu s il is possible Ihat lhe foundation factor was important nOI
onl y during seismic aclions (amplifying deformalions and lensions o n lhe Celio side as
visible in lhe model for aClion in lhe x-direclion), bUI al50 after and belween Ihose
cvcnts, wilh diAcrcntial settlement due lo lhe effects of the oUI-of-plumb, especially on
lhe Celio sid e; these aspecls are the subject of deeper invesligations.

I3 From 1703 10 1979· From the Ruin lo lhe Restoration Works

AR.er the 1703 earthquake lhe Colosseum rema ined abandoned and consequently
suffered progressive deterioration. The 19th cenlury is lhe century of lhe biggest work
evcr carried oul on the Colosseulll. Attenlion was inilially focused on lhe laleral
borders of lhe surviving façade, where lhe lack of continuity and Ihus lhe
disappearance of circumferenlial stres5es, created lhe mos! unfavourab le situation,
\Vorsened by lhe dynamic actions Ihat affected lhe existing boundary of the collapsed
zones
160 STRUCT URAL ANALYSIS OF HI STORI CA L CONSTR UCTIONS

Fig. 5: Maill slresses aClillg 011 lhe Colossf!um

Fig. 6: EVO/Ufio" of lhe main -"tresses afta b/ock !ric/ioll overlake

 G. CRQCI I Thc Colosscurn 161

Fig. 7: SiKIIS of earlhq/lakes (sill/lsmdal deformarioll) visihles 011 lhe exlemal wall
162 ST RUCTURAL ANALYSIS OF HISTORICAL CONSTR UCT IONS

The abutments of Slern (1805-1807) and Valadier (1822-1826) stopped lhe

deterioratian pracess and generated a new, adequate static situation (fig. 8); the same
cannal be said for lhe dynamic behaviaur as we shall di scuss later. From 183 1 to 1846
important work s were carried out by Sal vi with lhe reconstruction of the interior
circunferencial wall on lhe Celio side and lhe installali on of a system of radial tie-bars
in the alignment s around the imperial entrance. Ahhough Ihese works were reali sed,
and a1though no slrong earthquake has occurred since 1703 , a dangerous situation,
very elose to collapse, occurred in 1979 when we were asked by the Archaeological
Superintendent of Rome to check some piers in lhe inncr circumferential wall , ncar the
Stern abutmenl. A decp proccss of eracking (fig. 9), was immediate1y evident, so that
urgent shorage and intcrvention were requ ired.
This event, which iniliated lhe study summari sed here, gave us lhe o pportunity to
highlighl the facl Ihal lhe earthquake produccd not only direcl co ll apses, but also
started a process of deterioration related to lhe hig h stress leveis rcached, which
created cracks and micro cracks sensit ive 10 thermo-hygronometric conditio ns, and
particular1y to lhe effects of frost provoking an íncremenl of internai st resses in lhe
outer layer ofthe blocks, Ihus facilitating spall ing.
This phenomenon justifies lhe numerou s "spo ntancous collapses" that have contributed
to the large quantity of material that has becn losl during lhe ccnturies; lhe facl tha!
only lhe collapscs corresponding lo lhe earthquakes and few olher situations are
recorded by historians is nOI a surpri se: for centuries the Colosseum has been an
abandoned ruin, no one being interested in whal happened to it . Fig. lO shows fram a
qualitative point af view the evol ution af the collapses and the weakening of the
structure.

2. THE PRESENT SITUATlON AND THE SAFETY EVALUATION

On the basis of what has been shown ir is possible now to give a judgement of
the prescnl safely levei by a criticai interconncctio n of the information by means of
three different processes: the direcl observation, lhe mathemalical analysis, the
hislorical survey.
I - The direcl observation - A carefully observatio n of the Colosseum highli ght s the
main alterations Ihal can be summari sed as follows :
- the materiai s: dcteriorations are present in many parts ofthe Monument , linked to the
fire, weathering and to the effects of almospheric attack, frost, etc .... : high stresses
and cracks have accelerated the related phenomena ;
- lhe geometry : alterations are particularly cvident in the "earthquake signs", such as
lhe frequent separations between the elliptical and radial walls (fig. 11), lhe sliding of
lhe blocks in the walls in lhe vaults and lhe arches, the oul-of-plumb of walls and
pillars, lhe deformations of the facade (fig. 7).
Thi s last aspect is of particular interest as the circumfcrential si nuso idal deformation,
nol always in phase from a levei to another has also praduce relative rotations abo ut
their own vertical axes and twisting effects in some pil1ars.
- the previous works: the collapses have deeply changed lhe o riginal constmctio n
whose most evident aspects are the Stern and Valadier abutmenls, the works of Salvi,
the st rengthening of some elements (arches, pillars, ... ) and lhe in sertion of metal
chains, (fig. 12) related to lhe out-of-plumb which pragressively increased over time;
11 - The Malhematica! Models - Different modcls were made to take into account
differenl hypothesis on the structural behaviour of the monument , due to different
hypotheti cal efficiency of lhe slruct ural elemcnt s.
The malhemat ical analysis indicates Ihal lhe surviving façad e, belween lhe Stern and
Valadier abutmenls as wcll as lhe abutments lhemselves, reach tensil e stresses, initially
 G. CROC! / The Colosseulll 163

Fig. 9: Cracks and

crushillg discol'(!red ill
1979 in some pillars

lHE EVOLUTlON OF OAMAGES DUE TO EARTHOUAKES

ANO OTHER REDUCT10NS IN STRUCTURAL RES1STANCE OVER TIME.

S, F ,
r- =. I
~--;-,1-- ..,--1--_-___ _
,
I

. '~:1:·.:.~r,,="'t.:r":,r:r.~
- ... ......u .. "

... - ; ..... _1 . .. "11 .. · " • • ,·

0. • • _ " f.. ' ,

--
~ .""í. ...... -
. ....,.._
...., ...

r ...... ... -
- ..... 01 """..... ,
~
• c""", . ..
_lo -""I.....

Fig. 10: Qualita/h1f! evolllfloll offailureé1· al1d loss of material

164 STR UCT URA L ANALYSIS OF HISTORI CAL CONSTRUCT IONS

Fig. 11 .' Separa/íon

helweell circlIIII!erellfial
anti radial walls

Fig. / 2: ('hoin.\' plm:t'd

in lhe 191h c.: enlllly
 G. CROCl I Thc C010SSCIIIll !65

in the eireumferential direetion, whieh overtake the frietion between the bloeks; this
means thal sliding between the travertine blocks ean be produeed, and as a
eonsequence ofthe increased bending moment in lhe pillars, there is a reduction in lhe
effeetive section of the pillars Ihemselves that can lead to criticaI situations. The
mathematical models also show that lhe current situation is weaker than the original
situation due to the loss of cireumferential eontinuity.
111 . The histarieal survey strong earthquakes (VIII and IX degrees on the Merealli
scale) dacumented in Rome every )·5 eenturies, have always produccd damagc and
collapses; besides they have initiated deterioration processes which have brought about
collapses ar dangerous situations decades ar eenturies after the event itself.
To interpret the lesson from history regarding the present situation we have to take
into account lhree aspects :
· the increasing weakness of the structure aftcr each earthquake in relation 10 lhe
sliding belween the blocks, the deformations, lhe out·of.plumb and lhe increasing
stresses in the columns; from this poinl of view lhe present situation is worse than the
situation in lhe pasl.
· the role of lhe soil and foundations; lhe zone \Vhere lhe characteristics are \Vorse
correspo nds to lhe most damaged part of the monument; from this point of view the
capacity of the surviving structures, especia11y lhe tàçade, appear better as they
probably bear onto efl'icient foundations;
• the different geometry and the configuratian afthe aetual struClure, and in particular
the façade including lhe Stern and Valadier abutments; this situalion has never been
tested by strong earthquakes so no specific information can be extracted from hislory

3. PRELlMINARY CRITERIA OF INTERVENTION

The interaction bet\Veen lhe different information mentioned above and Iheir
critica! analysis, makes it possible to have a reliable piclure of lhe situation and allows
us to express some concern, especially \Vith the prospect of the nexl slrong
earthquake.
The exceptional value ofthe monument requires however prudence in lhe judgements
deeper analysis and more sophistieated mathematical madels are going to be
developed, and we expecl in few years to !lave acquired elements la reduce lhe
uncertainties: il will be possible therefore 10 ensure an adequate margin of safety and
durability, minimizing any alteration corresponding to lhe interventions.
The preliminary indications on lhe possible strategies for interventions on the façade
can be synthesi sed as fo11o\'ls :
• to conneCI the externaI \Vall to the ambulatory vaults and lhe radial walls (after
cheeking lhe effectiveness ofthe chains placed in lhe Ilineteenth century)
• to improve the circumferential resistance which is currentl y only provided by lhe
frietian between the blocks:
· to improve lhe transverse flexural resistance af the abutments where the
circumferenlial collaboration is missing .
Some cables (steel or synthetic fibres), passibly prestressed should be useful for these
purposes; as previaus mentioned, howcver, these are JUS! preliminaT)' ideas la be
verified and detaiJed on lhe basis of the new data that will be acquired , taking inta
account thal. due to the dynamic nature af the problem, any possibilily lo dissipate
part ofthe energy must be favoured .
 STRUCTURAL ANALYSIS DF HISTORICAL CONSTRUcrlONS
P. Roca, J.L. González, A. R. M~rí and E. On~te (Ed s.)
© C IMNE, Ba rcelo na 1996

ANAL YSIS, DIAGNOSIS AND PRESERVATION OF ANCIENT

MONUMENTS: THE ST. MARK'S BASILICA IN VENICE

F. Mola R. Vitaliani
[JoliteCllico di Mi frm o 15 ti/ l/f a di Sciwza e TeCllica de/lc Costruziolli
P.zza L. Da Villci 32 Facoltii di hlgeglleria-Ulliversifii di Padava
20133 Milano Via Marzolo 9
Ita/y 35131 Padova
It a/y

SUMMARY
An operadonal procedure for the analysis. lhe prognosis and lhe monitoring of lhe
structural behaviour of ancient monuments is presenled. After a delailed discussion of lhe
basic aspects of the lechniques necessary to collect significanl data about lhe state of
stress and strain of lhe monumental complex. lhe results derived from a wide survey_of
experimental tests performed on lhe S1. Mark's Basilica in Venice are presented.
Panicular emphasis is devoted to lhe flat-jack tests and to the ones connectcd with lhe
evaluation of the main propenies of the stIUctural materiaIs. The criteria followed in order
to fonnulate a fmite element modelo the results obtained from some panicularly significant
case studies and the basic aspects of lhe monitoring equipment are lhen exposed. Finally
the general assumptions which have govemed lhe assessment of lhe statical reliability of
the monument are discussed, wilh the basic aspects of lhe restoration works which have
been prescribed for some structural parts exhibiting an unreliable statical behaviour.
Some concluding remarks and the characten stics of lhe future works necessary to assure
an adequate staticaI behaviour of the monumenl elose lhe paper.

I. INTRODUCnON
The problem connected with the preservation of lhe statical efficiency of andent
monuments has assumed great importance in lhe last two decades as a consequenee of the
development of operational techniques whieh have allowed to scientifically proceed for
lhe analysis and lhe diagnosis of these particular structures. The procedure can be
subdi vided into three main slages, namely: diagnosis, conceming lhe assessment of the
structural reliability prognosis, defining possible restoration works and prediction of the
time evolution of lhe strllctural behaviour based on the results derived by means of a
proper monitoring cquipmcnt. Owing to the complexity of the statical behaviour of
ancient monumcnts, lhe three preceding stagcs require to be exhaustively investigated to
take into account many aspects whieh are nOl strict1y connected with lhe typical ones
pertain ing to structuraI mechanics, so that other complementaI)' concepts have 10 be
in troduced. 'l'he first is lhe historical investigation about lhe major events which took
place during the strUctural life-time, in order to deteet degrading phenomena and lhe
works done for gening rid of them. The following step , concerning lhe geometrical
description of lhe monument, represents a quite difficult task since lhe architectural
shapes are alw ays complex and extremely variable. The third step is re lated to the
investigations necessary to achieve an adequate knowledge about the mechanical
properties of the various suuctural materiaI s. T his can be perfonned by means of a
conveniem sei of experimental tests carried out on specimens eXtr"J.cted from particularly
sil,,'T1ificant struc tural pans.
It is Ihen necessary to achieve reliable informations about lhe 51ate of stress and strain
of the structure in order to obtain a reaIistic picture of the strUctural behaviour for the
calibration of the theoretical methods of strllctural analysis. The collection of geometrical
 F. MOLA and R. VITALl AN I I SI. Mark's Basil ica in Vcnicc 167

and mechanical data allows to formulate a lheoretical model, based 00 lhe fmite elements
technique, in order to correctly describe the complexity of the various structural
arrangements.
The approach now discussed in its basic steps has been codified by means of specific
documents published by the Associations dealing with lhe statical preservation of ancient
monuments. In particular, lhe Proceedings of the Symposia and of the Conferences
recently held on this arguments, [e.g. I, 2, 3] represent a valuable contribution in
enhancing lhe knowledge for lhe theoretical and practical solution of such many complex
and critical problems.

2. BASIC ASPECTS RELA TED TO lHE ANAL YSIS OF ANClENT MONUMENTS

2.1 Historjcal sUlyey

The maio goal of historical surveys is lhe achievement of data able to give a reliable
and detailed picture of lhe struclural behaviour. The complete lack of documents
regarding the hislory of a monument is a quite extraordinary event, since during its life·
time lhe monument has generally been observed and deeply s[Udied, 50 lhal a number of
documents and reports is available. From these documents and from local inspections we
ean derive usefuJ informations about events occurred in lhe pasl, which have produced a
marked structural degrading, i.e. frres , partial collapses, seisms.
Moreover we have to observe that many monuments have been continuously
eontrolled during their life·time so that it is generally possible, especially for lhe
monuments built in lhe second millennium, 10 have ai our disposal historical documents
describing lhe events whieh have driven lhe structures to lheir present configuration. This
represents the initial condition for the structural analysis and for a correel formulation of
lhe basic hYPolhcses governing lhe suuetural response under the applied actions.

2 2 Gcometrieal Descriotion
The geometrical identificarion of lhe various structural elements is a fundamental step
for a correet modelling of the structural behaviour. This operarion requires high precision
and it is quite complex, as the structuraI members present a marked variety of shapes.
The most reliable technique for lhe geometrical description of structural members is
the photograrnmetrical analysis. This technique, even if some negative criticisms have
been addressed to it in lhe past, has shown good precision and general applicability, so
that at present time it represents lhe main operational procedure for lhe geometrieal
description, [4] . The results of the photogrammetrical analysis, togelher wilh the ones
conneeled with lhe direet survey of cracks and dislocations, for which it is possible 10
proceed by means of a careful visual check, allow to reliably define lhe struc tural
configurarion and to proceed for the modelling Df the structural behaviour.

2 3 Experimental Tests
The structural members of ancient monuments are generally constructed by masonry
or stones, whieh should be experimentally investigated to define their physieal, chemical
and mechanieal properties by using appropriale tests. In order to perform the
experimental tests it is necessary to extract some specimens, pnxiucing local damages in
the structural members. This is not always possible and in any case lhe local damages
have to be redueed at the maximum extent. Therefore it is nOI generally possible to fully
perform ali the necessary experimental tests. In fac! it is well known thal a realistic
evaluation of lhe bearing capaeity of a masonry structural member ean be derived only
extraeting a quite large specimen, bUI this represents an intolerable damage for a
monumental complex, so lhat a different approach has to be followed.
168 STRUCTURA L ANALYS 1S OF 1-II STOR1 CAL CONSTR UCTIONS

For this reason the tests perfonned 00 cylindrical specimens of small diameter (Le.
$ ~100 mm), extracted by a drilling machine, assume great imponance. These tests
introouce only small damages, and therefore they can be perfonned with convenient care
also in monuments of particular value. The drilling operarion, which can be driven 10 a
significant depth, allows to obtain a conrinuous boring able to give reliable data about the
mechanical propenies of the materials up to the inner parts of the members. In the bores
produced by the drilling machines, it is possible to execute tests for evaluaring the
soundness of the internaI material by using micro video-camera investigations, or
ultrasonic tomography. In particular the last operation does not produce local damages,
so Ihat it can be also used in the zones where decoralions do not allow to perfonn
destructive tests. Finally, the data regarding the state of stress and strain parameters can
be achieved by using lhe flat -jack technique. This methoo, introducing a small local
damage, is of greal imponance and it allows to correctly define the structural properties to
be introouced in the mathemarical model.
The basic stages of lhe experimental analysis here recalled, are exhausrively discussed
in [5], and the theorerical and practical requirements of the flat-jack tests have been stated
by Rossi in [6, 7]. The aforementioned operations have been applied for the ana1ysis aod
the diagnosis of ooe of lhe most important historical monuments in Italy: lhe SI. Mark's
Basilica in Venice. During lhe years 1990- 1994 lhe ISMES Co., Bergamo - ltaly, under
the supervision of the authors, has perfonned a detailed survey of e:ICperimental tesls on
the Basilica, together wilh the implementation of a mathemarical mooel for the structural
analysis which has allowed 10 reliably assess lhe structur-ll safely and to derive some
interesling infonnations about the time evolution of Lhe structural behaviour of the
monwnent.

3. ANAL YSIS DF THE STATICAL BEHA VIDUR DF THE ST. MARK'S BASILICA
SI. Mark's Basilica in Venice (ltaly), Figure I, is one of lhe mosl important ltalian
monuments and its historical heritage is well known world-wide. In the year 1994, nine
hundred years elapsed since lhe Basilica dedicarion and to celebrate this exceplional event
the ltalian Ministry of Public Works (MPW) charged ISMES Co. to perfonn a wide
program of experimental tests in order to check lhe statical reliability of lhe structural
complex and to develop a malhematical model for the theoretical analysis. The authors
were charged to supervise the procedure, to define the various operational stages and to
state reliable criteria for the formulation of lhe assessment of structural safety.
The suggested procedure has been developed in five stages, namely: historical
analysis; geometrical description; experimental tests; fonnulation of a mathematical
model; placing of a monitoring equipment. The procedure, which is in agreement with lhe
general discussion expressed in the preceding points, has suggested to prescribe
restoration works for a structural member exhibiting ao unreliable statical behaviour. The
most significant results derived fTOm the various steps of lhe diagnostic procedure are
discussed in the following, pointing oul the aspecls connected with lhe correct
formularion of the theoretical model for the structural analysis.

3.1 Experimental Tests

The main features of the Se Mark's Basilica are shown in Figure 2. The construction
phases have been developed during three periods, e:ICtending from the early IX cenlury
until the dedication year 1094. In Figure 2 we can observe the primary kernel of lhe
Basilica, with five domes and a planimetric arrangement having the fonn of Greek cross.
The bearing structures can be subdivided in elementary sub-systems fonned by four
arrangements, each one composed by four pillars, disposed at the corners of a rectangle.
These sub-systems represent lhe bearing structures of the women's gallery and of four
arches bearing the domes, connected with them by a cylindrical tambour and by four
spherical triangles.
 F. MO LA and R. V ITALlA NI I SI. Mark's Basilica in Vcnict: 169

Fig. I The SL Mar"'". B,I . . d ll.:a 111 lI ... pn.~ . . el1l t:o nfi gll lilllOIl

5 1 Dome 01 San Glovannl

2 Dome of Penlecosle
or 01 lhe SPlrlIO Sanlo
:3 Cenl ral Dome 01lhe A:;CUlI'~ IUIIU
4 Dome 01 lhe Chorus or 01 Enullanucle
5 Dome of San Leonardo

FI !! J Ge neral vicw o f l he plan 0 1" Ih~ H,N Ilt:a

170 ST RUCTURAL ANALYSIS OF HI STORICA L CONST RUCn ONS

The elementary sub-systems, shown in Figure 3a, are connecled in the volumetric
di sposition of Figure 3b, where the central sub-system represents lhe basic structural
arrangement. It is notewonhy to observe lhat the horizontal reacrions for lhe equilibrium
of the domes are almost fully supplied by the sub-systems as the externai walls are ralher
thin and consequent1y unable to counteract significant lateral forces. The masonry domes
have a spherical configurarion and are overhung by external w<X>den domes covered by
lead scales. The woodeo domes do not represent structural bearing parts, they have only
an architectural charaCler.
The slatical configuratioo existing at present time is lhe result of valious restoration
operations, done in lhe pasl centuries and described by Saccardo [8J, Forlati [9] and the
Restoration Division llO]. The first experimental tests planned were lhe vertical and
horizontal drilling operations to be perfonned on the foundation structures, in lhe bearing
soi! and in the masonry structural members. The drill-holes were aIso used for lhe
ultrasonie tomography tests and the video-camera inspeetion, while funher holes were
drilled to proceed to complementary sonic tests eonsisting in measuring the propagation
velocity of a some wave between two parallel holes. The results related lO lhe foundation
structures and to the foundarion soil are ilIustrated in Figures 4 and 5.
The correlation between the results of lhe bore-hole lests and the sonic ones has
allowed to define, even if in an approximale way due to the small number of data, Lhe
prerequisites of statieal efficieney of the fouodation arrangement. Regarding the sonic
teslS the following eathegories eonnected with Lhe measured propagarion velocity of the
sonic waves in me various parts were stated:
v < 2500 m/s is associated to a discontinuous or highly degraded material;
2500 rn/s < v < (4000 + 5000) m/s is eonneeled with a moderately discontinuous
material, exhibiting a suffieienl preservation state;
v > 5<XXl rn/s belongs to a well preserved material with negligible discontinuities.
The cathegories here introdueed have allowed to subdivide lhe foundation sttucture in
four zones, namely:
poor zones where coarse macro-cavities are present together with v < 2500 m/s. In
these zones the structures present high degrading;
uncertain zones, where moderate cavities are associated to a velocity v < 2500 m/s.
The material even if sufficient1y dose, presents sigmficant degrading;
rather reliable zones, where macro-cavities are associated to good values of the
propagation velocity. The material, even if discontinuous, exhibits a good
preservarion state;
reliable wnes, where small discontinuities are present together with high velocities.
The di slocation of the results has JX>inted out thal lhe poor zones are prevailing. They
are extending over 50% of the tested zones while the uncertain zones and lhe reliable
zones respectively represent 30% and 20% of the tested zones.
Even if the negative results are more frequent, we have to observe that the small
number of lests is oot sufficient to exhaustively assess lhe statical efficiency of lhe
foundation structures and to detect the need of prescribing restoration works in short
time. Refening to lhe zones exhibiting high uncertainties it appears necessary to place a
monitoring equiprnent in order to collect data about lhe stnJctural behaviour. The teslS
performed on the foundation have also been prescribed for the vertical bearing structures,
in the zone of the women's gallery, which was lhe only one where moderately destructive
tests could be performed since in the inferior parts lhe structures are covered by marble
plates and in the upper parts golden mosaics are presenL
Ultrasonic lomography has pointed out a marked scauering of the results as the
measured velocities belorig to lhe interval 600 m/s < v < 2000 m/s, with a prevalence of
the lower values connected with materials exhibiting poor mechanical properties.
 F. MOLA and R. VJTALlAN I I SI. Mark's Basilica in Venice 171

a)

Fig. 3 a) Basic struclural sub-assembly: b) general view or lhe spatial structural arrangement
 172 STRUCTURAL ANALYSIS OF HISTOR1CAL CONSTRUCT10NS

Bllck m3:wvy

050

0.00

mOO
0.50

10_0 ' .00

1~0

2.50

'"
3.00

CJ S,lty clay ~ Filhng material

r;3 S,lty sand ~ Slalic cone penellalion lest . cone res,stance

Fig. 4 Experimental tests on foundations: soil stratigraphy and investigation on masonry walls

Vp (m/s) Vp (m/s) Vp (m/s)

2000 JOOO 4000 sooo 6000 2000 JOOO 4000 50006000 2000 JOOO 4000 5000 6000
0.0 0.0 0 .0
H (m )
H (m )
1.0
H (m )
1.0
" 1.0
li
) Ii
15
f )
1.5
\ 1.5 1/

2.0 2.0 2 .0
I<
1/ l)
25
I 2.5
/
I 25
J "
17
I

,
[)
30 [ili' 30 3.0

I I I
35 3.5 3 .5

Fig. 5 Resu lls of the ullrasonic lesls performed on (he foundation

 F. MOL A :md R. VITALl AN I I SI. iIo b rk's Basilica in Vcnicc 173

These results have been confinned by the ones derived from lhe bore-holes tests,
poinring out the poor quality of the masonry structures from which, in many cases, it was
rather impossible to exlract solid specimens as only desegregated material was found.
The video-camera investigations have also pointed out frequent and significan t cavities
associated to a material with reliable strength disposed only on the external surface of lhe
pillars, exhibiting a ralher small thickness. At the in terior of this external shell, a ralher
homogeneous material exhibiting poor mechanical prerequisites is present, so that from
lhe statical point of view, lhe pillars can be assumed as hollow-core members.
The results derived from a bore-hole test are shown in Figure 6, while two
tomography tests, related to a not-restored and a restored pillar are shown in Figures 7a
and 7b respectively. It is noteworthy to observe that lhe tests based on lhe measure of lhe
propagarion velocity of lhe ultrasonic waves is an efficient 1001 to define lhe soundness of
lhe material at the interior of the structural members. As it appears from Figure 7a, lhe
most frequent situation, confumed by lhe bore-hole tests, is lhe one correlaled to the
presence of poor material, exhibiting a low propagation velocity. In lhe pillm reslored by
means of mortar injeclions, lh e velocity, as shown in Figure 7b, exhibits high values in
some zones of significant extension, but not everywhere. The results derived from the
uhrasonic tornography pointed out that the restoration technique based on mortar injection
has nOl given reliable results as the physical and mechanical properties of the material
present at lhe interior of the pillars are not sufficienl for reaching a reliable degree of
soundness and homogeneity.
The single flat-jac k teSlS perfonned to derive infonnation aboul the slatt:: of stress in
the various parts. are also affected by a marked scauering, as we easily can deduce from
Figure 8. This scattering takes place not only as a consequence of lhe different
planimetrical position and geometry of the pillars, bUI it is also connected with lhe
restaration works performed on these s(fUctural rnembers which have modified the
structural stiffness producing stress redistribution.
The stresses higher lhan 1.2 MPa represem significantly high state of stresses in
pillars exhibiting poor material. Analogous considerations can be rnade regarding the
elastic moduli which, evaluated by double flat-jack tests, took place at the interior of lhe
inlerval 700 MPa < E < 5000 MPa, or in a more detailed way we can write 700 MPa < E
< 2000 MPa for nOl restored masonry, 1()(X) MPa < E < 4000 MPa for restored masonry
with monar injection, 4000 MPa < E < 5000 MPa for pillars wholly rebuilt aboul 30
years ago. The correlalion between the results have allowed to define four situations
associated to four intervals for the elastic moduli, fo r lhe propagation velocilY of
uloasonic waves and for the state of stress as spedfied in lhe following:
E > 5000 MP" 3200 m/s < v < 4200 m/s, O MP. < (J < 0.4 MP.;
3000 MP. < E < 5000 MP., 2200 m/s < v < 3200 m/s, 0.4 MP. < (J < 0.6 MP.;
1500 MP. < E < 3000 MP., 1200 m/s < v < 2200 m/s, 0.6 MP. < (J < 0.8 MPa;
700 MPa < E < 1500 MPa, 400 m/s < v < 1200 m/s, 0.8 MP. < (J < 1.2 MP• .
The preceding situalions are reported in Figures 9, 10, 11, wherc the lcttcrs A, B, C,
D painl out the direction of increasing risk, associaled to low velocity and small elastic
moduli. Figure 12 points out lhe highly degraded or lhe mechanically poor elements, for
which only 70% of lhe material extracted by the bore-holes lesls was sufficien lly
compact. The contemporary presence of the letter O for lhe whole set of lhe perfonned
tests is connected with situations of high statical ri sk.
Another interesting informa.tion deriving from single ar double flat-jack tests, is the
indication af lhe state of stress present in lhe masonry on the stress-strai n diagrams. As
illustraled in the Figure 13, the stress-strain relationships can be assumed as three-linear
diagrams, where the achieving of stresses belonging to lhe second or lO the third branch
of lhe diagram points oul a stalical behaviour exhibiting significantly non linear aspects.
Thi s fact has been found in one pillar, where the contemporary presence of nega tive
results in ali the other te sts has induced to proceed to restorarion works.
174 STRUCT URA L ANALYSIS or HISTORICAL CONSTRUCTIONS

BORE-HOLE TEST C3 1V
O56mm

\-< ------ I~
~to

Fig. 6 Bore-hole test performed on masonry structures

 F. MOLA nnd R. VITALIAN I I SI. Mnrk's Basilica in Vcnicc 175

"
"
I ~

r·:~
I"~) ,· 1
r:
6,.. ir , -
,1;I': . -."}'

~I "
. . ;<
~

_.
--_.-.
- ...
.~
~

y.

.~~
... .,
----. .-,--
= "">
~

-- ,-
o'. a)

---- -.' ....

.
.
... .
~

-
• ."
.", ,
, "'

=
1.1" " 11

-
"1.1 " ",,,

-
"" I
_ J~M

- j ~? ''''
j ~'~I
, lo/I'I
"' -
,.M
~"~
"'~.
b)
Fig . 7 Tomography lesl execuled a) on a nOI-restored pillar: b) on a reslOred pillar
 176 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

y '" J200 - .(200 m/~

Slress ( t.l p,,) 2200 - J200 m/s
1200 - 2200 m/s
y :; .(00 - 1200 m/.

Fig. 8 Results of single flat-jack lesls Fig. 9 Results of lomography tests

A [ > 5000 t.l Po 0= 0.0 - 0 . .( l.I f'o

B 5000 I.IPo
E '" .!IOOO -
C ('" 1500 - JOOO )'!Po
O E = 700 - 1500 wPo

Fig. 10 Mean exper. values of the elastic moduli Fig. 11 Mean exper. values of lhe stresses
 F. MOLA and R_ VITALl AN I / SI. M:1Tk's Basilica in Vcnice 177

percentage
01 voids

Fi g. 12 Masonry in a poor preservation state

,
I
i I I
" I

%
i

~
AVERAGE STATE
OF STRESS 0.85 MP.

-',,._'. -'t'I ::_JLLi._

,
, .-
,, , I j I I ,' 1
, i
O'
, ,
00
·250 200
TR ANSV STAAIN
ISO 100 50
Vi I 50 100 ISO
AXIAL STR AIN
,
200 xla-

Fi g. 13 Stress-strain diagram for masoory stmctures. with the iodicatioo 00 the actual
measured statc of stress
 178 ST RUCTU RAL ANALYS IS OF HI STOR ICAL CONSTRUCrI ONS

Thc elaboration Df lhe resu lts has allowed to fonnulatc a j udgemenl abQul lhe
cfficiency and lhe preservation af lhe slruclural alTangcmen t, based on lhe fo llowing
situations:
conlemporary presence of negalive results in ali lhe perfonned te ... I.... The structure is
statically inadcqualc and even ir further invesligalions can be prcscribed in order to
bettcr define lhe repairi ng lechniq ues, restoratian wo rks have to bc pcrfonned within
shon time;
presence Df dcgrading phenomena togelher w ith signilicant discontinuities in
homogeneous material subjecled to low stresses. In Ihese cases il is recommendable
to imposc lhe s ingle nat-j ack teslS in arder to check lha! lhe slresses are sufficien tly
low. This condition is essent ial to consider statically efficicnt a structural member
even if affcc tcd by sign ifica0! mechanical inadequacies:
presencc of slalically effeclive configural ion which does nOI rcquire restoration
works.
As we can observe in Figure 14, the zones cxhibiting a not negligible risk are quile
sma l!. They have 10 bc subjected to restoration works as the degradi ng and their intrinsic
inadequacy have produced situations which are not consistent with lhe prescribed safely
leveis.

A P1IIOrt c lo.slll. " o. typ. I

P~ l oro c!o n iU. d o. "rl>. 2

5 0'" P'-Iro Trlt>uno 50 ... Clom onl. Tri buno

Fig. 14 Zones wilh significant slatical !'i sk

4. THE MATHEMATICAL MODEL AND T I-IE NUMER ICAL ANA LYS IS

The results of the experimental lesls !lave allowed to get ready a Iheoretical model for
the structllral analysis based on lhe sub-assemblies FEM techniqlle. The model has been
employed fo r lhe cvalllation of lhe slale of stress in lhe various slmclural members and
lhe rclatcd displacement s produced by lhe following actions:
 F. 1\ 10 LA ,:md R. VITALlA NI I SL Mark's Basi lica in Vcnicc 179

local load. evaIuated for the structural pans as volume force proportional to the
density of lhe material. Regarding the pans which have nOl slruclural tasks, the
permanent load , connected with their presence has been introduced as e:ltternal nodaI
force applied to the bearing rnembers;
two lemperalure fields applied at the e:ltternal surface of the scructuraI members. The
first related to a temperature variation of 1 CO. with respect lhe isothermal slate
unifonnly applied at the externaI surface of all elements; the second connected with a
temperature variation of 1
lhe top surlace of the roof;
C" with respect the isothermaI state unifonnly applied at

imposed displacements to foundation structures according to the resuhs obtained by

means of lhe monitoring equipment. The values are normalised assuming as a
referenee unit displaeement the maximum measured one having value 1.39 mm.
The presence of cracks and dislocations, particularly frequent in lhe roof has been
taken into aceount in two different ways. The firsl way provides for lhe adoption of a
reduced elastic modulus in the zones where there is a diffuse microcracking. We can so
take imo account, by an approximate but substantialJy correct approaeh, lhe local
reduction of StruClural stiffness produced by cracking. The second way can be used when
lhe cracks are wider and more localized and it consisls in intrc:xlucing an opportune
splitting of two adjacent elements crossed by the crack. It is so possible to introduce in
the structural mode! lhe degrees of freedom corresponding lO lhe disconnection operated
by lhe cracks. making il more realistic and dose to the aCluaI struclural behaviour.
The geometricaI discretization of the struclural comp!ex has been performed following
the results derived from the photogrammeoical analysis. while the mechanical properties
of lhe malerial have been assumed, as flTst approximation, consistent Wilh a linearly
elastie homogeneous and isotropic behaviour. The numerical values have been assumed
according to the experimental teslS perfonned on the masonry members, in particular the
double nat-jack ones. The numerical analysis has been carried out by means of the
MSC/NASTRAN Code [11], working in the linearly elastic domain while the geomeoical
discretization has been operated by AUTOCAD [12]. FinaUy lhe graphical elaborarion of
the obtained results has been processed by the I-DEAS Code 1"13). The complexity of the
model and lhe dramatical1y high number of the degrees of freedom (about 2.5 x lOS) has
suggesled to operate by means of the sub-assembly techniques. In this way besides
making the numerical calculations less cumbersome we can take into account without
particular difficulty the differenl mechanical behaviour which can take place in parts
subjected to high stresses or exhibiting marked degrading phenomena. In fact we can
vary the mechani cal behaviour only with reference to the sub-assembly affected by
degrading. assuming lhe remaining surrounding parts as an elastic srructure interacting
with lhe sub-assembly in the linking nodes. assuming as unknowns lhe related
displacements which can be processed by means of the condensed stiffness matrices.
The srructural complex has been subdivided into the nine sub-assemblies represenled
in Figure 15. namely lhe foundalion mal together with a convenient stratum of the bearing
sai!, the six pillars, lhe externai walls and lhe roof. The model, in its complete
confi guratian is sha wn in Fi gure 16a, while Fi gures 16b and 16c show respectively the
soil-foundation discretization and the venicaI section of the complete finite element
mode!. The sub-assemblies have been chosen in order to correctly evaluale the stale of
stress in the strUclural pans panicularly sensitive to stress variation s. This assumption is
quite useful for the analysi s of lhe pillars, for which the experimental tests have induced
10 improve lhe analysis assuming a non linear behaviour. Finally the sub-assemblies have
been identified assuming c1early defined interfaces requiring for their kinematic
description a moderate number of linking nodes.
The elemems used to describe the various sub-assemblies are quite different. In
particular six -nodes exaedrical elements have been assumed for lhe soil stratum,
represenled by a pri sm having squared plane with base of 180 m and 30 rn high. The
pillars have been described by solid tetraedric elements and the wall s have been
associated to plate triangular elemenls.
 180 ST RUCTURAL ANALYS IS OF HI STOR ICAL CONSTRUCT IONS

a) b)

c) d)

Fig. 15 Sub-assemblies assumed for the mathematical modcl: a) foundations: b) bcaring wall s:
c) pillars; d) roof
 F. MOLA and R. VITALlANII 51. Mark's Basilica in V\!nicl! 181

,,)

Fig. 16 Thc finitc clemcnt rnodel: a) the global di!<.crctization: b) lhe soi l roundation
dl . . <:rctization
182 STR UCTURAL ANALYSIS OF HISTOR 1CA L CONST RUCT IONS

Fig. 16c The vertical section of the complete tinite element model
Finally quadrangular and ttiangular pIate elements have been used for the domes, the
arches and lhe spherical ttiangles, whi1e eight-nodes brick: elements have been assumed
for lhe tambours. This discretization required some interconnection operations between
lhe generalized displacements of rwo different elements. This is the case of lhe interfaces
zones between lhe domes and lhe related tambours where the pIate elements of lhe
domes, having six degrees of freedom are connected with the solid exaedrical elements of
lhe tambours having onIy three degrees of freedom. In order to make compatible these
different elements lhe rotations of lhe pIate eIements have beeo related to lhe nodal
displacements of lhe exaedrical elements by means of convenient connecrion functions.
This technique has beeo applied according to the procedure outlined in [14].
The mechanical description of the various strucrural pans has beeo perfonned as
follows: for the e)(ternal walls and lhe pillars lhe high number of experimental tests has
allowed to formulate a Slatistical treatment of lhe results, by defining the cumulative
distribution function of lhe elastic modulus, showing three calhegories corresponding to
poor, sufficient and good masonry. These cathegories are connected wilh values lesser
lhan lhe 33% fractile of lhe result distribution, or lying wilhin lhe interval 33% - 67%
fractile ar greater than lhe 67% fractile, respecrively. The corresponding numerical values
are: El = 850 MPa, E2 = 2000 MPa, E3 = 6000 MPa. For lhe roof, where lhe domes, lhe
arches and lhe spherical triangles have beeo subjected to restorarion worlG in lhe past, lhe
value E = 5000 MPa has. heco assumed, while for lhe tambours, where a diffuse state of
cracking is present, lhe reduced value E = 3000 MPa was chosen. For all lhe structural
members lhe subsequent values for the density "t. lhe Poisson's ratio v and lhe linear
thermal expansion coefficient a. have been selected: y=18kN/m 3• v=O.15. a. =4.SxIo-6 °C· l .
Regarding lhe stratum of foundation soil, the results of the standard penetrarion tests have
suggested. to subdivide it into three partial strata, having lhiclrness 3.5 m; 10 m; 16.5 m
and elastic moduli Et \ = 1500 MPa, Ea = 500 MPa. Et3 = 250 MPa, respectively. The
Poisson's ratio has been assumed. v = 0.3.
In the Figures 17a, 17b and 17c some results of lhe numerical investigations are
shown in a graphical fonn. In particular Figures 17a and 17b show the displacements and
the state of stress in the roof due to pennanent loads, while Figure 17c shows lhe
displacements due to pennanent loads in the vertical section of the Basilica.
 F. MOLA .1nd R. VITALlANII SI. Mark's Basilica In Vcnicc 183

Fig. 17a Displacements due to permanem loads in the transverse section of

lhe roof

1 1 f ""f

D , (lP

Fig. 17b State of stress in the roof, due to permanem loads

184 STR UCTUR AL ANALYSIS OF HI STO Rt CAL CONST RUCTtONS

Fig. 17c Displacements due to permanent 10ads in the vertical section of the Basilica
These results al low to formulate some considerations about the reliability of the
mathematical modelo Regarding the pillars the calculated. stresses are generally in a
sarisfactory agreement with the experimental ones. The results of the theoretical model are
underestimated of about 20% and in quite restricted. zones the theoretical values of the
stresses are one half or one third of the corresponding experimental values. On the
contrary, in tbe roof we observe a stress distribution which, even if globally in agreement
witb the experimental stresses, assumes in some restricred zones values which are too
high when eompared with the strength of the material. In the connective zones be[Ween
the tambours and the arches the theoretieal tensile stresses, even if in good agreement
with the presenee of cracks. have too high values. which, if effectively present, would
represent a serious situarion of starieal risk for the roof. However we have to take into
aecount that the inferior surface of the domes is covered by precious golden mosaics. so
that the execution of experimental tests for the evaluation of the stresses was not possible.
For this reason we are lacking of the direet eomparison between theoretical and
experimental results, and the only cheek is related to the contemporary presenee of
theoretieal tensile stresses and of a diffuse eracking state. As the fust are evaluated by
means of a theoretical model considering lhe structural members in the uncracked stale,
high values of the tensile stresses numerically computed ean justify the presence of
cracks. The obtained results point out a substantial correlarion between computed stresses
and observed cracks, so that the theoretical medel seems able to correctly predict the
structural behaviour even if some local results cannot be considered sufficiently reliable.
GeneralJy speaking we can state thar the theoretical result.'\ show a sufficient1y reliable
distriburion of stiffoess with regard the pillars while tOO high stiffness is present in the
connecrion zones between domes, tambours and arches. We have also to consider mar the
peak-stresses experimentally evaluated in some pillars and not emerging in the theoretical
analysis suggest to operate a convenient refinement of the mechanical descriprion of the
material in the sub-assemblies representing the pilIars, while for the edge zones of the
domes this refinement should be accomparued. by a more detailed geometrical descriprion.
Even if the theoretical model points out too high local stresses, the numerical results,
together with the experimental ones, have allowed tO reliably assesss the structural safety
of the Basilica and to state an oprimal srrategy for its sttuctural preservarioo which wiIl be
briefly illustrated in the following.
 F. iVIO L;\ anel R V ITALl AN I I SI. Mark's Basilica in Venict! 185

5. OPERATIONAL STRATEGY FOR TIIE PRESERVATION OFTIIE STATICAL

EFFICIENCY
The results of lhe theorerical and experimental anaIyses have pointed out quite high
stresses in two pans of the structural complex, namely in lhe south-wesl pillar of lhe San
Giovanni Dome, Figure 14 and in the connective zones between the domes and the
arches. The high compressive stresses in lhe pillar and the marked tensile stresses in lhe
edge zones of lhe domes have been obtained experimentally in the pillar and theoretical1y
in the domes. These different sources for the knowledge of lhe state of stress. have
suggested to consider statically inadequate the pillar prescribing for it immediate
restorarion works, while the lack of experimental data has induced to prescribe lhe
placement of a monitoring equipment for lhe domes in order to control lhe time evolution
of their structural behaviour.
Sefore proceeding to lhe restoration of lhe pillar, an additional set of experimentaI
tests has been perfonned at lhe pillar base, after removing the marble plates. The resuIts,
substantially similar to lhe previous ones, have confumed the statical unreliability of lhe
pillar justifying lhe immediate starting of lhe restoration works. The repairing operation
consisled flrst1y in lhe whole substitution of the mortar existing in lhe externaI surface of
the pillar, extended along the totaI lhickness of the bearing parts. The second operarion
consisted in injecting, under low pressure, the internaI pan of lhe pillar with
cementitious, moderatel)' expansive mortar. The chernical-physical and mechanical
properties of the mortar have beeo established by assuming lhe compatibility with lhe pre-
existing materiaIs in terms of linear thennal expansion coefficienl and permeability,
avoiding mortars able to produce chemicaI composites containing sulphates or to generate
ao increase of lhe internai humidity of the pillar. Another aspect concems the value of the
elastic modulus and of the strength of the repaired pillars which has been calibrated on the
basis of the corresponding value of a well prese rved masonry. In this way no
concentrations of stiffness have been introduced.
In Figure 18, a detail of lhe restoration works is show n. The possi ble slatical
inadequacies connected with lhe high stresses theoretically obtained in the edge zones of
the domes have beeo controlled by means of a monitoring equipment opponunely placed.
The quantities to be measured are: the relative aod absolute generalized di splacemenls of
significant points of the structure. the opening of cracks and their width. lhe thennal
variation, lhe interstilial pressure of the soil, the relative di splacement between lhe
foundation mat and lhe olher peinlS of the bearing soil.
The monitoring equ ipmenl consisls of wire extensomelers, pendulum with
telecoordinometer, inclinometers. extensometers, thermo-couples. electrical piezometers,
assestometers. These equipments, extensively illusttated in the technical lilerature, e.g.
[151. have been placed according to the scheme of Figure 19 and the related data are
contin uously recorded by means of electronic stations which are able to represent in a
numerical or graphical form lhe collecled results. Particularl)' signiflcant for lhe
evaluation of lhe possible negative statical effects connecled with the high state of stress
in the domes, are the wire eXlensometers placed along lhe chords of the bearing arches,
the inclinometers placed on the pillars, the extensometers and the lhermo-couples. These
instruments allow to point oul the time evolution of lhe defonnations induced by
temperature and by the associate state of Stress.
In Figure 20 lhe diagrams performed in lhe year 1992 are reported. We can see thal
lhe di splacement measured by the wire extensometers are in good agreement with the
lemperature variarion
Funhennore aI lhe end of lhe flrst year of recording the residual displacements are
practically negligible so Ihat we can state that the presem statical situarion of lhe complex
fonned by lhe domes and by the arches is substanliall y stationary. This fact has
suggested 10 nOl prescribe immediate restoralion works of lhese parts, poslponing this
deci sion when the data derived from lhe monilOring equipment shall indicale tha! lhe
structure is oriented to ao unreliable staticaI behaviour.
 18 6 STR UCTU RAL ANALYSIS DF l-ll STDRICAL CDNSTR UCTI DNS

Fig. 18 The rCSloralion works of lhe

pillar in lhe San Gi ovanni
Dome

PZ prezome' .r Cf Wire ( !de nsom ete r

AS Anesl om e t er El ( . 'ansam" te r o n Ih. ini ury u '"
, 2Or:--i-)--U-i-:::j~lm~--:1
.; ~--: : ChannÍtI016-T5 : :-r--:
~ ol_ --':__ :
~ 10 l ____ :- __ ~ __ ·_:- ~· 7 - Air TemperalUre --~--- ~---:- ---;-
! : , ., , , : :
.J_.J----l.--<-J.--"---'--'----'-
: :
Gen Mar Mag lUO Seu No v a I

Ê
É.

"~
20.:----:----:---"f---~-
I ,' ,, ,, ,, Wire - e ~rensome
- -· EFl
CharmeIOO1 ' , , -! BJ'
,
t.r --:----:----:---1-
,,
1 0:::::H~~: ~--t-i_b_-_,L n_:- _-+

i_. J--+- ··j:j=t--i- --j ~j~~r~--r---~--+--+

(Ll
o G," ,.'"
Ê
R6 É. 20,--,-õ._-,,-,,------_--.------;;--,-,-,
, , I 'ChanoeI006 . EF6

,~ ~~~~~:~:::r:+-- w:re e~tens~~_e~~:r:!::+

10
t

"
, , , , I , , , , , I
---:----:-- · -: ---~---t--- t- --t--- t- --:_- --:_--_t- - _t
-[ O I I , , , , , , I , , I

li "', cl
'" 1 20'r-.,,,-.,-,,-., -., -.,-.,---..,-,r-,-..,

~
I +::f::i:::i:::i: ~;~:~~~;f~~l~::f:::f::+::+
I
' I ; ,
O_~ ~~'
', ------:-,
t :.~-_ ·_--· _-- ·
, ,
,~,
: j :"-

Gen M3' t.!ag Lug Sen Nov d)

Fig. 20 Time variation :a) temperature:

Fig. 19 Sche me o I' lhe mon ilorin g cquipment b). c) and d) relative horizontal
displacements mcasured by lhe
wi re exlensometers I, 6 e 7
 F. MOLA a lld R. VITA Ll Ai'\I l I SI. Mark's B:tSilica in Vcnicc 187

6. CONCLUSI0NS AND FUTURE WORKS

The experimental tests perfonned on the St. Mark's Basilica and lhe subsequent
fonnulation of a theoretical model for the numerical analysis, have allowed to achieve a
fLfst consistent set of data for the assessment of structural safety and for formulating
correct decisions about the need of restoration works. The process which has govemed
the diagnostic approach is based on a sequence of operations which, with reference to
experimental phases, can be considered well defined and accepled world-wide.
The obtained resuIts are good and they allow to define the most imponant factors
goveming lhe slructural behaviour. In particular the slale of stress and the strain
parameters evaluated by means of single or double flat-jack tests stand as basic reference
points to which compare any olher result deriving from complementary approaches, both
experimental and theoretical. The evaluation of the mechanical properties by means of
bore-holes, video-camera and sonic tomography tests has allowed to compare the actual
stresses existing in lhe structure with the srrength of the material and 10 derive a reliable
judgement about the statical adequacy of the monument. The theoretical model has
provided results in good agreement with lhe experimental ones. However, the presence
of local anomalies has suggested to refine the model and this task will represent one of
lhe mOSI importanl items of the future works. On this subject the improvement of the
geometrical description and the refinement of the mechanical properues have to be ta.k:en
imo account. About the fLfSt point, a more detailed kinematic description of the interface
zones is necessary, while lhe second pcint requires a more realistic approach taking into
account the non linearities which are unavoidably related to cracking phenomena or to
marked compressive stresses. The models able to descnbe lhe non linearity of lhe
materiais, associated lO the development of the degrading process, like the model
described in [161, seem to be profitable for a generalized improvement of the theoretical
structural model.
In the authors' opinion it is however possible to state that lhe operations perfonned
for the experimental and theoretical analysis of lhe statical state of the S1. Mark's Basilica,
illustrated and descnbed in [17], represent a basic reference point. When the procedure
will be funher refined by means of the improvemenls previously discussed and of lhe
additional data denved from the monitoring equipment, it wiU give reliable results for a
correct and realistic fonnulation of a judgement about the statical efficiency of this
monumenl which is lhe living expression of a unique historical heritage.

REFERENCES

1. Balasubramaniam A.S .• Ed., Proc. of lhe Symp . on Geolechnical AspecIs of

Restoration Works on Infraslruclures and Monumenrs , Bangkok, 1988, A.A.
Balk.ema, Rouerdam/Brookfield, 1990.
2. lABSE (IntemationaJ Association for Bridge and Structural Engineering), Proc. of
lhe Intern . Symp . on Structural Preservation of lhe Architecrural Herieage,
lABSE Repon. printed with the financial s uppon of Italian National Research
Council, Rome, 1993.
3 . Brebbia C.A ., Leftheris B., Ed. Proc. of ehe Symp . on SlrucluraJ Srudies of
Historical Buildings . Chania - Greece, 1995, Comp. Mech, Publ. , Southampton,
UK, 1995.
4. Vio E., Galeazzo G., Vitaliani R., Rilievi, controlli dimensionali e analisi strutturale
nella Basilica di S. Marco in Venezia (in Italian), Bo/lettino SIFET, Società Italiana
di Topografia e FOlogrammeJria, N. 3/4,1988.
5. Riccioni R., Rossi P.P., Restauro Ediliúo e Monumenrale (in Italian), i1 Cigno
Galileo Galilei. Edizioni di Ane e Scienza, Roma, 1990.
188 STRUCTURAL ANALYSIS OF H ISTORICA L CONSTRUCTIONS

6. Rossi P.P., Peano A., Carabelli E., Determinazione sperimentale delle caratteristiche
meccaniche delle murature (in ltalian), Comportamento Slatico e sismico delle
strutture murarie, Sacchi G., Riccioni R. Ed. eLUP, Milano, 1982.
7. Rossi P.P., Recent development of the fla t-jack test on masonry structures,
USA- Italy Workshop on Evaiuation and Retrofit of Masonry Structures, 1987.
8. Saccardo P., 1 restauri deUa Basilica di S. Marco dai 1878 in poi (in Italian),
Venezia, 1905.
9. Forlati F., La Basilica di S. Marco attraverso i suoi restauri (in Italian), Trieste,
1975.
10. Direzione dei restauri della Basilica di S. Marco, Rapporti sulio stato dei lavori di
restauro dell'Angolo di S. Alipio daI 1909 ai 1914 (in Italian), Procuratoria di S.
Marco, Sezione Documenti, Venezia.
11. MSC/NASTRAN version 67, The MacNeal-Schwendler Corporation, Los Angeles,
1992.
12. AUTOCAD version 12, Autodesk inc., 1992.
13. I-DEAS, Structural Dynamics Research Corporation, 1990.
14. Bathe KJ., Finite Element Procedures in Engineering Analysis, Prentice Hall,
Inc., Englewood Cliff, New Jersey, 1982.
15. Malerba P. G. ed., Monitoraggio delle strutture dell'ingegneria civile (in Italian),
CISM International Centre for Mechanical Sciences, Collana di Ingegneria
Strutturale, Vol.N.9, Udine, 1995.
16. Creazza G., Saetta A., Scotta R., Vitaliani R., Oõate E., Mathematical Simulation of
Structural Damage in Historical Buildings, Proc. oflhe Symp . on Structurai Studies
of Historical Buildings, Chania - Greece, 1995, Comp. Mech. Publ.,
Southampton, UK, 1995.
17. Mola F., Rossi P.P., Vilaliani R., Melhodological Approach for Sttuctural Analysis
and Inspection in the S1. Mark's Basilica, Venice, Proceedings of the Symposium
on StructuraJ Studies of Historicai Buildings, Chania, Greece 1995,
Computational Mechanics Publications. Southampton, UK, 1995.
 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTlONS
P. Roca, J.L. González, AR Mar! and E. Onate (Eds.)
© CIMNE, Barcelona 1996

STRUCTURAL ANALYSIS AND DURABILITY ASSESSMENT OF

HISTORICAL CONSTRUCTIONS USING A FINITE ELEMENT
DAMAGE MODEL
E. Dilatenl, A. Hanganum, A. BarbattU, S. Oller(1),
R. Vitaliani(2), A. Saetta (3) and R. Scotta(2)

(1) E.T.S. Ing. de Camillos, Callales y Puertos (2)/s/. di SciCllza e Téwcia del/e COllstmziOlli
Ulliversidad Po/itéClZica de Cala/mia Unillersitii di Padova
08034 Barcelona Padova
Spain lta/y

(3)Dipartnmellto di COlltmziOlli del1'Arc1ziletfura

IUAV
Venezin
/ta/y

1. INTRODUCTION

Nowadays tbe rc is a gcne ral COIH':CI"II for prcscrvation of historical COll-

strurtions [1-3]. This will illvariably require to pay grcatcr att,cllt. ioll to I ht'
rcliablc asscssmCllt of tbc structural conditions of l he tllOllUlllcnls alld lO
t il(' corrcct dcsigll of illlcrvcntion Illcasurcs. It is ob\'iowi that. as tilllt' gtWS
011. l.he lIumbcr of historical COllsLructioll:; incrcases. In particular. 111;\11.\'
exist ing bridges. to\\"c rs anc! dams as v..'dl as sonw relevam bHildin~!'i (';111
1I0W U(' ("ollsidcred in cvc!"y rcspcct hist,orical cOllsl rurtion s (3].
Thc philosophy of consrrvatioll has the goal to pres<'l"n" t lu.' orig-illill
arcl!itcc I.ural Illcssagc of <lny 1Il0llUlllent. The first. st.ep to l"t'<1ch sueI! oh-
jt'l"I ive is tl!e accuratc predietion of Lhe actual sa feLy levei of t.IH' sI ruC"llln'.
This illdlld('s tbe kllowlcdgc of bot.1! the lI1echanislll of detcrioratioll 01" t ht'
lua! rrial anel of t he structural cOlllponellts and tlle evolutioll of degrada! i011
\\"ilh t ime. The causrs of dCLcrioratioll usually can be subd iv id l·tI in 1\\'0
11Iaill g roups physical-chcllli cal-biological aggressivc agellts alld 111('('11<\11-
ical probkllls, Thc lattc r can bc lIowadays accuratcl.\' stlldi('c! b.\' ]]](';111:--
01" rOllpled IIIl111Nical- ex p erillle nt.al pro cedures. where the slruclllrill lH'-
ha\'iour is analyzed by ;,dvalH.:ecl nOll-lillear finit.e elr lllcllt lI1od('15 whirh are
;'lc!t'fjllat.c!.v calibratéd lI si ng cxperilllcll tal data (5-28] . On the Mher halld.
IIH' illlalysis of Lllc pllysical-chclllical-biological dcleriorarioll rC([I,ir<'s 111('
dt>fitlit ion of" ad hoc" pl!ysical anel lllllllcriral llIodc\s for sol\"i ng t IH' pqHa-
I ions of diffusioll anel 1ra lls port of the aggrcssivc s pcc ics in meleI" to pl"l'(lin
IIU' l'\"olution of dcgradatioll phcllomena duc LO differcllt externai <l1!.t'uts
[.1.2~.JI [.
190 STRUCTURAL ANALYSIS OF HISTOR ICAL CONSTRUCTIONS

The objcctivc of this papel' is to dcscribe a mct hoclology whicb C<lll

be effectively used for assessing the structural conditions alieI durab ility
of historical masonry anel concretc constructions. T his incJudc~ L1H~ prc-
diction of local anel globa l behaviouf up to structural fai lurc untlcr static
anel dynamic conditions. The app roach combines the use of l he conccpt
of damage to reprcsent the non-linear cletcrioration process of the material
with advanced finite element mo deis (induding physical-chelllical-biological
degradation) anel cxtcnsive experimenta.l tcsting to calibratc and val idate
the overall nurncrical mo dol.
The contcnt of the papel' is the following. I n the next sccLioll a brief
historical background of damage modc ls is presented. Thcn the bas ic in-
gredients of the damage model used are dcscribed togethcr with il,s illl pk'-
mentation for finitc elemcnt analysis of histori<.:al cOllstrudions. Fill all y.
examples of prelill1inary applications of thc 1lI0d.el to the analysis of t lt e
ccntral dome Df St. Marks Basilica. in Venicc are presented.

2. BRIEF REVIEW OF CONSTITU T IVE MODELS FOR NON

LINEAR ANALYSIS OF C ON C RETE AND MASO N RY
STRUCTU RES

Extcllsive experimental stuelies have becll undertakell to clmracteriz(' 111<'

response anel 1Iltilllatc strcngth of masonry and plain COllcrete lInder Illulti-
axial stress statcs [6-8]. Considcrable SCi.\lter of results has bccn observctl
and collaborative studies havc bccn undertaken to identify the principal
facto rs influencing t his varia.tion [7, 11 ]. Several approadlcs. lmsed 011 ('X-
perimental data , have bccn uscd to rcprescllt the coustitutiV<' relal iOllShip
lInder multi-axial strcsses and thcse Cêln bc catcgorized illto the fivc fol-
lowing groups: (a ) linear anel 110B linear clasticity theories. (I» pcrrcc t anti
wo rk-hardcning pla.'iticity theories, (c) endochronic thcory of plasticit y. (d)
plastic fratturing theo ry and damagc t.hcory.
A sil1lplc and popular mode l for nOIl linea r finite elelllcnt analysis of ("Ol!-
<.:ret(~ and lllasonry struttures assurncs clasto-plastic (01' viscoplastic) C()JJ-
sti t utive eq uatiolls for co mpression bchaviour, whcreas a cOllce ptually IIIOJ'('
simple smcared ehl.sto-brittlc rnodel is uscd for defin ing Ollsct alld progrf'S-
sion ar cracks ato points in tensiol1 . Diffcrent versions 01' t.hi:; 1110dl"'I hiLVI'
becll successfully used for llon-lincar analysis of masonry as wcll as plaill
and reinforced concrcte structurcs [3, 10,11 ,24]. A summary of SOIII(' 01' t.il"
more recent contrihut.ions in cach of t.hosc theori es call he fOlllld in [6.1O.11J.
The elasto- plast.ic-brittle smearcd llIodcl, in spite of it.s poplllarit.y. PI"('-
seuts variolls controvc rsial featurcs s ueh a.s the Ilced for dcfinillg II l1co\lpled
bchaviour along cadl principal stress (01' strain) directions: the usc 01' a :-;hcar
rctentiol1 factor to cnsurc some shcar rcsist.ancc along t,}H~ crack; the lal'k
 E, ONATE CI aI. I Finitc clcrncnt damngc modcl 191

of equ ilibrium at the cracking point when more than onc crack is fornlf'd:
the difficulties in defining stress paths following t he opening and c10sing of
cracks under cycling loading conditions and the difficulty for dealiug \\'iLh
the combined effect of cracking and plasticity at t ile dam aged points [2~ 1 .
1t is well known that micro-cracking in COBcrete and masomy takcs
place at low load leveIs due to physica! debonding betwceu aggrcgatc alld
mor Lar particles. 01' to simple micro-cracking in the mortal' luca, Cra<..:king
progresses following a nOB homogeneous path which combines t he two nwn-
t ioned mechanisms with growth and linking between micro-cracbi along
differeIlt dircctiolls, Experiments carried out on mortal' SPCcilllC ll S silo\\'
that the distribu tion of micro-cracking is fairly discontinuotls \\'ith arbitrar,""
oricntations [6], This fact is supported by many ex perimellts which !jho\\'
that crncking can be considered at microscopic levei aS a 1l01l dirl..'cLiollal
phenomenon and that the propagation of micro-cracks follows an (-,ITatic
path which depends on the size of the aggregate partidcs. Thus, t he dOllli-
nant cracking di~'ect ions can be interpreted at macroscopic le\'el as the loclis
of t.rajectories of Lhe damage points (Figure 1).
A- A
~
~
. \ ;, ,, . .,

;jF' , .•' ~,>

:' .'. "\ ' . ' , .. ~
OR'IGINAl ' ' INCREASE IN
DE,8DNDING DE80Nb.ING ' .

..CRACKS I~
MATRIX

Figurc 1. t-.lechauics of dalllage anel propagaI iOlJ 01' a !ll<j(T ()~('lIpic

crack in plaill couC!'ctl'

The é\bove co ncept.s support the idea that. thF 1101/ lill ('o,. bc hu/'/Olll ' oI
COI/.Cl'ete a.'/Id 1I1.asotl.1'y ca'!/. be m,odeled usillg cOl/ cepts of damO,l}f' t!t('o/'y 01/111
192 STRUCTURAL ANALYS IS OF HISTOR ICAL CONSTRUCTIONS

[10 , 12 , 13 , 17-21] provided an adequate damage fUHction is dcfincd for

taking into account the different respanse af couerete under tensioll ane!
compression statcs. Cracking can, thereforc, be illt erpretcd ,,1."; a local d(ulI-
a.Qe eJJect, defined by the evolutioll Df known material paralllctors aHe! by
ono OI' several functiolls which contral the Ollset alld cvolut,ioll of dalllag:I'.
One of the advantages of such a model is the illdepcndence of the allal-
ysis with respcct to cracking dircctions wh ich C<l11 be simply idcntificcl (f
posterioTi once the non-linear solution is obtailled [10,12,13]. T bis al\ows
to overcorne the problelTls associated t.o most elastic-plastic-brittlc SIlICcUNI
cracking !nadeIs. In this papel' a model dcvcloped in recellt years by l.he
authors group [10-16,22-28] for non-linear analysis of concrete based 011 til('
conccpts of dCLmage mcntioncd above is extcnded for stnH..: tllntl analysis of
historical cOllstructions. The model takes iuto aCl.: ount, ali Lhe importa!!!. a:-;-
pects which should be cOllsidered in the 1I0B-linear analysb af cOlltTd(' and
rnasonry structures such as the diffcrent responsc 11lIder l.ellsion and ('0111-
pressioll , the effect of stiffness dcgradation due to mcchanical and ph.ysic;d-
chcmical-biological cffcCLS anel t he problem of object,ivi ty of the l"rsults wit,h
respect to the finite eIemcnt IIlcsh .

3. THE CONCEPT OF DAMAGE VARIABLE

Ia order to clarify the conccpt of dalllage col!sider a smface C!eIlH'u1.

in a darnaged material volume. This surfacc has /;til arca large ClloUglt l.o
cOlltaill a representativc number of dcfects , but still enabling t.o 1)(' n'!á]"('d
ao; pertaining to a parti cu lar material point. Thus, ir 5" deBol.es 1.111' /lVl'rall
scctioll anti 8" the cffective resisting arca. (5" - S'" is Lbl~ CII"(,C\ olTupil'd In·
the voids ), the damnge w!·rúLble d" associated to th i5 surfacc is (F igurl' 2 )

d = Su- 8u= 1 - oS" ( I)

u S" 5"

Clearly, d u represents the surface dCllSity oI' lIIaterial dcfects and il \Vill
ha.vc a ""ero value whel! the material is in Lhe uwlalBaged virgin sl.all'. ('OH -
versely. t,he reduction of the cfTcct.ive resisl.illg arca \ViII I(, .. d 1.1) ,\lI ill("l"(';(S( '
01' dalHagc ullt il rupture defincd by SOlllC critica I vallu' of d " (IHlllll(]('(] I).'"
L11C Illlrcadla.ulc valuc of d" = 1). Not,e that t,bis is a ([in'l·tiolull 1II'fil lit il)l) ()I"
dalllage. In I llany cases a siug!e scalar ITIH"Csclltal,ioll of (I;Ultaj.!,l ' i~ ;111()!)I!"]
(i.I' . /1. " = li ) which snfficcs to ensurc rca lisl.ic Itl<ll.('riill !llolkl [I :~.]7- 22 1 . I1
is wort,h lloting 1 t,!tal. in t, his case cracks at a micrusl'opi c poillt ]H'('tI !lul lo
havc not particular direction and a Illacroscopic crack is Lh(,tI ddil!('d ;\."; t IH'
locus of damage points as previously men t ioned.
An \lsdul concept for 1ll1dcrstanding t he effect 01' dillll<lgl' i~ [!til] of
 E. ONATE et ai. I Fínite ekment damagc model 193

(b)

a (f (f

,L Ló k
~s
k Ld
~
K ,. r;
"
(c)
a

Unloodingj reloodi ng

Figure 2. (a) D;:utlaged surface; (b) Cauchy s tl'CSS a anel cf[('I't ive
strcss ã ; (c) Evolution af uuiaxial strcss-st.raill curve

effectiue stress. The cquilibriUlII relationsh ip bCL\. . ccn Lhe slallda rd Cauchy
strcss a and the :'effecLivc" :;trcss. ã. in thc dalllagcd bar speci llle n 01' Figl\l'(,
2 is

(JS = ãS (2)

and from (J ) and (2)

(J = (1 - d)ã = ( I - d)E€ (O)

Clcarly whcn a damaging process is occurrillg. the exte rnai loaclill)!,' i~

rcsisted by the cffective strcss arca anel. Lherefore. ã' is a morc ph,vsically
rcprcsentativc para meter thall a (Figure 2) .
194 STRUCTURAL ANA LYS IS OF HISTORICAL CONSTRUCTIONS

4. CONTINUUM OAMAGE MOOEL FOR MASONRY ANO

CONCRETE UNOER PHYSICAL, CHEMICAL ANO BIO-
LOGICAL EFFECTS

4.1 Basic assumptions

As mentioned above, the detcrioration mcchan isllls that ocem in lll<ln'y
concretc ma.sonry st ruct nrcs can be divided in physical-chcmical-biological
causes alld mcchanical ones. T he lattcr are usually d HC to an CX(;(':-;s 01' SLI"l'SS
act ing on the material anel can be ~tll di ed by menns of a st.andard dalllag('
mode l as cx plai ned prcviously. T he physical-chcmical-biological Cil.IISPS Df
degradations are duc LO Lhe interaction bet,wccn t,he cllviron Il H.'1l1 alld til('
bu ild illg mater ial. Hcre the prese nce 01' \Vater plays a primary ro le. T Il<'
main causes of degradations are listcd in Tablc I [29-32 .34}.

Table 1 C lassificat ioll of phYl:!ical, chcllI ical a nti biologit:al

degradation causes [:1-1J.

Ty pe Description of the sourccs oI' degradaI iOIl

1. 1 freezing-thawillg
l. phY::iical J. 2 leach ing out
causes ] .3 crystalisatioll of solublc salts

2. 1 rcactioll bctweell sulpha t() alld hydratr ('i\kill111

silicatcs wi l,h productioll of ettringite
2.2 reactioll betwccll slIlphatc and hydratc alt u lI illa-
2. dlcmical tes with production of thaulllasite
causes 2.3 re..tction bctwcclI calciulll chloride anel calciu llI
hydroxide with product.ion of hydratcd calcilllll
oxychloridc
2.4 alkali-aggrcgate reac ti oll

J. biologil"a l 3.1 lormation of a lgac. lichclls alui fungi

causes
 E. ONATE c{ aI. I Finite clement damage modcl 195

4 .1.1 FÍeld equations for environmental variables

The elifferential equations governing moisture, heat and aggressivc ~pc­

eies flo\Vs in a porous material (such as masonry anel concrete) in the h,v-
pothcsis of existence of chemical reactions anel by consielering both diffusion
and transport mechanisms can be written as

oh ~T ~ I
- = v Cv ' + - +
oh, K-
aT+oh,
-
at at at at
,aT _ '1Tú'1T aQ" aT,
pc,at - + at + at

ae
- = '1
T aw+ac,
D~v c+-c - -
at ' " at at
wherc !t , T anel c are respectivcly the rclative bumidity contenl,. Lhe telllper-
ature anel the diffusivc !:ipecies cOllcentratioll (e.g. chloride. sulph atc. CO:!,
etc.) anel w is the waLer content. Qu is the outflow of heat per unit volulIl(,
of solid, bis the thennal conductivity anel âC)c!ât denotc~ changes dllr to
the carbonation per time unit. A more precise definition of these :;Ylli!)ols
can bc fOllnd in [29,30].
The eliffusivities of rclative hUll1idity h and the aggrcssivc sp('('ic;-> D,.
are asslImed to be strongly dependent on the pore humidit.v. t(~1I1 p eratur('.
degree 01' hydration of cemellt (that is the equiva1cnt curing Lime ) alld the
precipitate concentration cf (if t he product of the chemical reaction is a prt'-
cipitate, like the calciulll carbonate in the carbonation process ). Tllcs{' lat.I {'I'
t\Vo facts slow the diffusion phenOlnenon for long tenll rcs u!t,s due LO UH'
reduction 01' the porosity. The (noll linear) expressiow, 01' the diffusivit.ic:; as
a function 01' h , T anel c call bc found in [34]. Note that thc lllathclIlal ical
form of the problern is similar to lhat of t,he corrosion problelll 01' UH' sl.<'(-'l
in conerete anel the frec(';e-thaw problem [29:30.33].

4.1 .2 Numerica/ so/utio/] of en vÍro nmentt:d varÍabJes

Due to the complexity resulting from the !lOtl linearities occurrillg ill
the definiUoll of the diffusivities and the coupling of the differclltial eqll~)­
tions (4), ànalytical solutiotls are very difficult to obtain allCl ;-l 1I1l1llCrica J
approach should be preferred. There the finitc elemcnt Illcthod S('('lllS t.o 1)(,
the more adequate procedure to solve lhe space-timc cquatiolls Hlld it, lias
been successfully used by thc authors in this cOlltext [29 , 30,3~ J .
Application of the well-known Galerkill procedurc arter space di scl't,ti-
196 STR UCTURAL ANALYSIS OF HI STORICAL CONSTR UCTIONS

sation lcads to Lhe follow ing systc lIls of cqu ations of coupled ol'dillal'y dif-
ferential equations in time

D" + C h _ K Dl' _ Dh, = 0

Dt ' Dt àt.

(5)

DC + DC _ C. Dw _ DC, = O
Dt 3 Dt Dt
wherc h , T and C rcprcscnt t he discrete re lat ivc humidi ty contc nt, t.1l(' d i::;-
crete temperaturc and the discrctc diffusive s pecies conccll t. rat ioll ved o !":;.
For the detailed cxpressiolls of the matri ces in (5) see [29.30.33].
The systclll (5) <.:al1 be writtcn in t.hc fo llowing concisc fonll

A:i: + Ux = q (6)

where the mcanillg Df the tcrms in the cq. (6) follows inullcd iatcly fro lll cq.
(5). The solut ioll in time Df eq. (6) ca ll bc casily a t tc lIlptcd fo ll ow ing a
s tandard onc stcp algori t hrn giviug

[~
CJ.t
A + a u ] x"+ ' = q" - [( I- a) U - ~ A]
CJ.t.
x" ( 7)

where (.)" d ellotes va lucs aI. time t". ,ó,1, is the t.ime incrcllIcn t anel n is a sI (t-
bili ty parametcr [35]. Tlle algcbraic 11011 lillcar cquat iolls (6 ) i:; t hlls sol wd
by a d irect a pproach. Dcta ils of t he sLabili ty a nd con vergcltt'C' nm d it io ns ill
the solut.ion of eq. (7) can be round in [33.3G].

4.2 Coupling environmental and m e chanical damage

T he coupling or pllysical-chcmical-biolog ical crfccb; wiUI st n ' ll g"l li ('lmr-

acteristics cun bc t'lkcn iu to aCCoull t, iH t he da lllage 1II0(\(>[ hy IIlOdi !\" illg
cq. (3) "'
IY = {3(1 - rl )&

where {3 is a physical-chcmica l-biolog ica l (P C B) para IlH 'I1 '1" SI WI! 1hitl
L The valuc of (3 approachcs Lhe cri t. ica l ]O\wr va il H' It as 1lI!' PCI3
fie< (3 :S
dcgradatioll process rcachcs it.s ma,xillltll'll encet . Eq. (8) (';\,11 1)(' n '\v r it,j (,1l as
 E. ONATE CI aI. I Fillite e1emenl do.rnagc model 197

with d' =1- j3(l - ri) ( la )

Clearly the effcct of (3 is that of reducing the meehanical strcllgt h. ~ot('

t hat as (3 ----7 /3.-: thcn d' ----7 d. The evolur.ion of /3 depends Ol! t hat o r
moisture. temperaturc and aggressive species wit hin t he (porous) IlIat('rial.
The physical parametcrs in th is diffusion process depenei aIso 011 the OVC'I';:1i1
degradation levei th rough t he damage vari ablc d and the problclll is t hlls
fuHy coupled [33.34]. A fuH coupled aualysis illvalves the finitc cle l Jl('lIl
so lution of the differential eq uations governing Lhc transport af cliffusi ll g
species together with t hose governing the structural behaviour. The P CI3-
mc('hanical eoupling can bc taeklcd by means of a staggcred sehelll\? [J3.3C>].
Very fcw applieations of the ful! eoup led so lutions can be found in til('
li tcratu rc. Preliminary results for simplc test problellls ha\·c bcell I'ccc lllly
reported in [33.3.J]. Thc ana lys is can bc obviollsly s implificd by ('l!oo!'iiug
a fixcd value of (31 which is cquivalent to accepting a certain leveI 01' P CB
elcgradation at the timc of analysis. OI' by defining all appropri at (· tÍlIlC'
cvolution of lhe pe8 parametcrs.

4.3. Three-dimensional m o deI and evolution laws

Equation (8) can bc cxtendcd to define thc conslitllti,·c cqllation or ...
three dimensional s pccimen in \"cctor f01'1l! as

II = (I( l - rl)õ- = j3( 1 - d) D< ( 11 )

where D is the elastic co nstitutivc matl'ix U and é are t lH~ s talHlanl SI ]'('SS
and strain vectors.
Thc Illodel defined by eq. (11 ) requires the knowleoge 01' lhe d;Utlilg:4'
and peB variables d and (3 .rcspectively at every stage of the history 01' 111('
construction. For this purposc one must define:

FoI' the da1l1age variaIJ/c d

a) A s uitablc sealar Bonll T of lhe strain tellsor (O I' alternaI in'I.\· oI' 1111'
ullclall!agecl st rcss tC lIsor). Here. seve ral possib ilitics exist and a sltil ;tlll('
option for ('OIH'l"f'tp anel lll1:\somy is [10.23]

T= " l-O)[ II-TD'-j"

(0+-- II ( 12 )
n

\\'herc n = f" /.!'I is t.hc ratio bet\veen Lhe éOll1prcssion and u'll:-iio1! lilllil
st rC'llgths.
198 STRUCTURAL ANALYSIS OF HISTOR ICAL CONSmUCTIONS

,
L <17,>
g = '-', with <±â,>= ~ (l a,l± oJ ( I:J)
L lâ,l
;=\

Expression (13) accounts for the different limit bcha.viour of the nta\'0rial
under compression anel tension states.
b) A damage criterion formulated in the strain of the undarnagcd stn'ss
spaces. The simplest form of this cal1 be written as

F (T, 1' ) = T - r SO ( 14 )

where T is the norm defined in (8) anel 7' is the darnage t hreshold value.
Damage grows when t he TIorm T excceds the current t;hreshold valtlP.
In particular, damage is initiated whcn T cxcecds for lhe first t ime Lhe
value r' (typically r' = f;/VE is taken [10,13]) .
Figure 3 shows the form of the limit surface TO _ 1,0 defining t he ollscL
af damage for the exprcssion of T given by cq.(12).

,
(JJ

f't

f',

----- f~=nf~

Figure 3. Limit damage surface and uniaxial stress-strain C\Uvü

for the model of eq. (9)
C) Evolution laws for the damage variable d and the damage threshold
value 7" . These can be written as [28). [311.

d = G(1') l' = lTIaX {To) T} (15)

where G is a suitable monotonic scalar function taken as

G(7") = 1- ';~ exp {A (1 -,'~) } (16)

 E. ONATE ct 31. I Finitc clement dnlll:lgc model 199

Note that G(TO) = O and G(oo) = 1 as expected. The parametcr A

is determined from the ene rgy dissipated in an uniaxial tension tcst as
[10,23]

( 17)

wh ere 9, = C,/ L". C, bcing the spec ific fra cL ure cncrgy pcr unit arca
(take n as a material property). r is the characteristic lClIgth 01' the
fractured domain. As always 9, 2: U:)1/2E (the material must clissipat.c
at least the energy stored when the elastic limit is reached). para,Hlelcr
A must be positivo [28]. Defining 9~ = 9,- (f~)1/2E. an altcruaLivl'
exp ression for (17) is

( 170.)

For tlie PCB variable fJ:

a) A suitable scalar Ilorm of the residual strcngth. Here the simplest opt ion
is to choose the same Ilorm r proposed for lho danuige \"ariablc.
b) An evolution law for the PCB variable (J. This call be writtell followillg
[33] as

fJ = 1'(7. C) (1Sa)

\. . ith

( 18b)

whcre '"Y = .;.-

I,cf
is the ratio between the actual st,rcngth 1I 0rJIl T and
t hc residual strcngth norm. COlTcsponding to c = Cj.ef' i.c. lhe referellt('
cOllccmration oI' the eliffusing species (ch loricle, sulphate, CO~. t'tt:.) ror
which the PCB dcgradation process reaches its maximum effcet.
Th e experimen tal cha mclerization of the model requires the follo\\'illg
llIaterial paramcters : Young modulus and Poisson ratio. tension anel COlll-
prcssioll limit strcngths alld specific frac ture energy obtained rrOIl1 ulli"lxial
teSls as well as the time cvolution of the conccntration of tllP cliffusing
specics obtained from separa te solution of eliffuse transpo rt equations 01' by
experimental measurements [29-34J.
200 STRUCTU RA L ANALYSIS OF HI$TOR1CAL CONSTRUCTIONS

4.4. Finite element structural analysis

The damage model presented above is extremely simple in cOlllparison
with more sophisticated modeIs for frictionaI materiaIs. The finite element
implementation follows the stand ard process [35]:
a) Finite eIement discretisation of the st ructure and interpolat.ion 01' t.he
dispIacement fieId within each element as :

u = Na (19)

where u is the dispIacement vector) N is the shape function rnat.rix alld

a is the vector containing the displacements of the cIemcnt nodes.
b) Discretisation af the strain and stress field!) as

c = Lu = LNa = Ba (20 )

a = 1l(1 - d)D€ = 1l(1 - d) DBa (21 )

where L is the appropriate strain operator and B is t.he stl'ain l1latl'ix.

c) Derivation of the (non-linear) discretised equations. SlIbstituting cqua-
tions (19)-(21) into the principie of virtual work [351

Jbc" adV = JbuTbdV + J JuTt dS

" v S
(22)

gives after standard algebra

w= p- f (23 )

p =J B Ta dV (24 )
r

(25 )

In the abave equat.ions b and t are hody forr.(~ anel rli~t.rihll t.p.d roJ'r.~ V~t'I'. Ol'S
respectively, 6é and 6u denote the virtual strains and virtual displacements.
respectively and W is the so-caIJeel residnal force vector which expresses t he
equilibrium between the ext.ernal forces vector, f anel the intel'llal forces
vector, p .
The system of equations (23) is non-linear' due to the dcpclldencc of the
stresses 00 the damage anel PCB parameters d anel f3 through eq. (21). A
summary of the main steps of the non-linear solution is shown in Box 1.
 E. ONATE CI :11./ Finite eremcnl d:1m:1gc Illodcl 201

nt h load increment, ith iteration

Compute displacement increment

~a '.,'= - [H "]" -'w"

'11 :': residual force veetor (= f, . B TudV - f)

H ;': iteratioll matrix (i.e. tangent stiffness matrix)

Update displacements and strains

Evaluate stresses
(1) Compu te t1lldamaged stresses: U:'+L = De ;'_L
(2) Evaluate ,.:'., using eq. ( 12):
(3) Update r. d and (J

(4) Update stresscs: a:'+1 = ( I - d ;'"!",) a:'"!",

Evaluate residual force vector: \jI ;'... ,
Chcck co nvergenee: 11 '11 :',,11 :S <II ! II .,
No: Continue iterations: i = i + 1

Next load increment: 11, = 11 + 1

Box 1. Quasi-static nOB linear [inite element structural solut ioll \lsillg
the envirOIllllcntal-mechanical damagc ll10del
202 STRUCTURAL ANALYSIS DF IiISTDR ICAL CDNSTRUCTIDNS

Note thal the process is relativdy sim pie as no special al go l'it,hm for
integration 01' the constitutive equatiolls is llcedcd as t he stresscs are explir-
itly given by eq.(21). Further information including details af the consistem
com putat ion Df the tangent constitutive matrix can be found in [28].
The concentration Df diffusing PCB species, c, a t cach stagr oI' lhe
structural lire should bc determ incd by a sepal'ate transient sol utioll of Lhe
PCB t ranspor t equations which in turn depend on the damage levei as ahoV<'
explaimid. The solution of the fuU coupled problem in t ime is ~dICllli.lt ically
shown in Box 2 .

o. New time incrcmcnt t'1+1 = t il + t1t

1. Solve evo lution cquation for C:l+l

3. Solve structural problem at time t 'I +1

• Solve p (a JH1 ) - JII+I = O for a " +I , e 'l+ l anel a " +1 usi ng
• a 'I +1 = {3"+\ (1 - d 11+1) e ,,+l

4. Repeat steps 1. to 3. until convergence of env irolllllcnt.al

anel mechan ical solutiolls at time t ll + 1

5. Compute global damage index D'I+ ' at t.ime t " +1

6. Next t ime increment - go to s t.ep O.

Box 2. Full coupled transient solut.ion linking envirOlllllC'llt.al

anel rnechanical dalllage.

A simpler alternative is to assume an ullcoup led lJf'haviour 1)('1\\'('1'11

the mechanical allel cnvironmental dalllage parametcrs alld Lo Ohl aill ali
analytical express ion of C by solvi ng indcpcndcntly the t.rans port. ('qll al iOll ....
for the d iffusing s pecics under differe nt degraelatioll asSlIIllpl iOlls PU]. T il"
lIlechan ical problclll can thus be separately solved for a fixed vrlhl(' oI' ri 10
obtain the structuraJ respOllSC and failurc loaci Df the cOIISLrll ct ioll ar <llly
de~ired time Df its his tory. This process is sket.ched in Box 0.
 E. ONATE c( aI. I Finite element damage model 203

1. Uncoupled transient environmcntal da ll1age analysis

1. Solve for Lhe time

cvolutioll of (',

.................... C,

C,

Time

1.2 COlll pute {3( I)

w
'"
E
1'l
g
c
w
E
e
;;
c
UJ
{3" f-------"'-.

(:Jc ------ --------------------

I" Time

2. Pel'fol'lll quasi-static structural analysis aI tillle t = I"

• {3" = {3(1")
• Sol\'(' p(a") - f I! = O for a" e" alld a " usillg:

• (7 " = {3"( 1 - d ") e"

• CO lll putc global dalllagc index D "

Box 3. Quas i-static structural analysis lISillg illfol'lllatioll frolll 1111-

co\\plcd t ran sicnt cll\'irollllll'lltal dalllagt' ('OIII\>llt <11 iOIIS.
204 STRUCTURAL ANALYS IS OF HI STO RICAL CONSTRUCTIONS

The dynamic solulion under t ime dependent forces follows a similar

pattern. Inc1usion Df dynamic effects in t he virtual work equation (2 2)
!eads after discretisation to the well known t ransient cquation system [35].

Mii. + Cil + Ka = J (t) (26)

wherc â anel a are nodal velocity anel acceleration vectors, respcctive!y. M

anel C are t he mass anel damping matrices a nel J (t) is the time dcpendcnt
nodal force vector.
A fuH non-linear t ransient solu tion af eq.( 26) is now possible Ilsing stall-
dard t ime integration algorithms giving the displacements, velociLies, acccl-
cratíons, strains a nel stresses as well as the local anel globa.l damagc indexes
at each time of the deformation processo A detailed descript ion of t he dy-
namic finite element solution falls outside the scope of this papel'. Thc
interested reader can find fuI! detai ls in [28 ,351.

5. THE CONCEPT OF GLOBAL DAMAGE

The global streugth of the structure can be assessed by means of a global

damagc indcx D. The simplest definitioll for D is the following

D = 1- -
o (27)
U

wherc f) and U are the internai encrgies corresponding to the dnmaged f1.1111
undarnaged sta~es , i.e.

U
- = aTI, \f
T
B a dV = aTI,\ . J3 ,.(1 - d)ã dV

U =a
T r T
lv J3 ã dV (28)

In (28) the total energy of the strllcture is obtaincd uy S lIlI1 of t he

~ulltriUul.iuJ ::S iu I..he sl..alH.lard JIIallucr.
c1CllICllt
Note that global st ructural failurc corrcsponds to a value of D a ppl'oach-
ing unity. Thus, t he computation of the local and global damage indiccs
provides a useful tool for monitol'ing in dctai l the evolution of the lIon linear
responsc of the st ructure up to fai lure.
 E. ONATE ct aI. I Finitc element damage model 205

z
x
Figure 4. Views ar the 3D finite element mesh of St. Marks Basil-
ica Dome.
206 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCT IONS

6. ANALYSIS DF ST. MARKS BASILICA IN VENICE

Thc rncthodology prese nted is currc ntly being applicd by t.hc ali! 1101'S
to analyze struct ural elcments of St. Marks Basilica in Vcnice for which
cracking patterlls have bccn det.ected . Thc first exam ple prcscntcd iti t hc
preli minary damage analysis of the central dome under various stal..ic ;:md
dynarnic condi tiolls using a fin ite element code developed by t he au! hors
[281
c:
«
ro Global Damage -
~" 0.9
Ui
Q; 0.8
E
o
O 0.7
"""'ro 0.6
::;;
ui 0.5

0.4
Oi
>
Cll
...J 0.3
Cll

'"
'"
E
0.2
O'" 0.1
ro
.c
o O
a O 0.5 1.5 2 2.5 3 3.5 4 4.5 5
Sei! Weight Factor

Figure 5. Global damage cvolution for the stat ic al1alysis.

Figure 4 shows differp.nt views of the finite e lClncnt meshcs oI' t he Si ruc-
ture using standard 20 Bode isoparametric hexahedral elements. Ali points
at the base (y = O) have the vertical displacemcnt restrict ed. In acldit.ioll
the displaccment iH the plane y = O of two bl:\Sc points havc bcen also r('-
stricted to avoid rigid body motions. The materia l propertics asslIllu'd iH
the analysis are the following :
 E. ONATE CI :11. I Finile elcmcnl damagc model 207

Guse 1:
Stone
• E = 600.000 kg/c m2 v= 0.15
• Tension limit strength = 40 kgjcm 2
• Compression limit strengt h = 400 kg jcm 2
• Density = 2.7 g/c m 3
Maso nry (vertical cylindrical wall elerncnts)
• E = 200.000 kg/ cm 2, v=0.15
• Tension limit st rcngt h = 10 kg/cm 2
• Compression limit st rength = 100 kgjcm 2
• Density = 2.4 g/c m3
The first solution attcmpted is the analysis of the limit static ~tr u c lural
strength under self weight loading. The st udy has bcen undcrtakell for a
co ns tant value of the PCB paramc ter (3 = 0. 9 for simplici ty. Figure 5 siJo\\'s
Lhe evolution Df the g lobal damagc index D foI' illcr~asin g valtH's of II\(~
self weight up to fa ilure characterizecl by a value of D rapidly approadting
lInity. Convergence of Lhe 110n-linear soluti on was lost for D = 0.8 auel
a safety coefficient of == 4.6 was found. Thc cvollltion Df t he local daltlage
indcx d contours is plotted in Figures 6a-c. Notc t!lat the m<l.ximuln vah w Df
d reaches 0.9513 at t he dome base where existing cracks have beell oh:;ervcd
in practice. The contours of equal vertical displacc ment anel a picture of
Lhe deformed s hape Df t he strllct.ure at the failurc load arc ploLtf'C1 in Fig.7.
Thc same analysis was repeated for reduced values of til(' III<lSO Il1".V
strcllgth Df
CASE 2
• Tension limit st.rengt h = 5 kg/<':1l12
• CompressioIl lirnit s trength = 50 kg/cm 2
CASE 3
• Tcnsion limi t. strengt h = 1 kgj cm 2
• Compression limi t. ~trcngt h = 10 kg/cm 2
while keeping the rest of Lhe material propcrties cons tant and pqual to ! lto~·J('
chosen in Case l.
Figure 8 shows the evolution of Lhe safety coefficient for ! 1](' ,.hn>('
differcnt cases stlldied. lt is interesting to note Lhat t he safe t")' codTic-i t'!l1
does not substa nti a lly increase for vaJucs of t hc tension limit S t.ITlIp';1 h 01"
the maso nry g reater t han 5 kgjc m 2. This is due to the chauge of Lhe d pfor-
mation mode of t he St. rllct urc as the st rengt h of the masonry wall i nCTI" L':.('S
which leads to accelerated failure Df t he st.o ne arches. as ex pc<.:t.('d.
Thc next study is the dynamic analysis of the structure UluleI" H lI.\"-
pot hetic earthquake. The material parameters anel boundary cOllditiollS
coincide with those of Case 1 described above.
The analysis was pcrformed in tWQ steps. First 1 the static solutioll linde!"
208 STRUCTURAL ANALYS IS OI' HI STORICAL CONSTRUCTlONS

Pigure 6.3 Contours Df the damaged zones for load factors af 1.0,
1.2. lA and 1.6. rp.spective ly.
 E. ONATE el :lI. I Finilc elernenl darnagc modc1 209

Figure 6.b Contours af the damaged zanes for load factors af 1.8.
2.0.2.2 and 2.6, respectively.
210 STRUCTURAL ANALYS IS or III STORICAL CONSTRUCTIONS

Figure (i,c Contours of t he damaged zones for \oad factors of 3.0.

3.ó, 4.2 ano 4.6 , rcspcctively.
 E. ONATE el rll. I Fjnítc clcmenl damagc modcl 211

Figure 7. Contours of equal di splacement and deformed shape aI

,ollapSt'.
212 STRUCTURAL ANALYS IS OF HI STO RI CAL CONSTRUCTIONS

E 4
.""
'""o'
u 3
~

'*
<n 2

o
Tensile strength (MPa)

Figure 8. Safcty coefficicnt evolution against the tensilc strcngth

of masonry.
sei f weight was obtained and next the sy nthetic acceleration shown in Figure
9 was applied to the base cf t he initially self-loaded structure. The evolution
af the global damage index D with time is plottcd in Figure 10. Note that,
global failure is obtained after 3.02 seconds. Failure occurs due to fracture
af the masonry elements at the rnid-hight af the cyli ndrical wall as clcarly
shown in Figures lla-b. The deformed shapes af t he structure at differcnt
times during the earthquake can be seen in Figure 12.
Thc last example corresponds to t he fulI t hrec-dimensional analysis af
the five domes of St. Marks Basilica. Figure 13 shows the mesh of 7676 20-
noded hexahedra involving 48505 nodes used fo r the analysis. 2265 15-noded
triangular prisms were also used as transition clements in some lIoues, The
material properties for the stone were taken the same as for the analysis of
the single dome in the previous example. The properties for the eylindrical
walls masonry wherc taken as:
• wall 1 E= 30000kg( crn2, v =O.15 , a,=40kg( cm 2, a, = lOkg( cIl1 2
• wall 2 E=60700kg/cm 2, v-O.15, O'c= 40kg/cw 2, a; = 10kg/cm 2
• wall3 E=250000kg( cm 2, v=O.15, a ,= 300kg( cIl1 2, a,=30kg( cm 2
• wall4 E=90QOOkg( cm 2, v=O. 15, a ,=40kg( cm 2, a, = 10kg( cIl1 2
The density was taken equal to 2.4g/ cm3 in ali cases. A quasi-staLic study
was performed for a constant value of t he PCB parameter {3 = 0,9.
Figure 14 shows lhe evolution of the global damage index for incrcasing
values of the seif weight. Convergence of the solution was lost for 0 = 0.84
and a safety coefficient of 5.75 was found. T he contours of equal displace-
rnent at that load levei are displaycd in Figure 15 whereas lhe clall1agc
 E. ONATE el a!. I Finite c1emcnt damagc modc1 21 3

150

100

50
N
E"'
"'-
c
O
.9
~ -50
"
Q;
<.>
<.>
« -100

-150

-200
O 2 3 4 5 6 7 8 9 10
Time (s)

Figure 9. Synt hetic accelerogralll.

«"
<.> , ,
E Global Damage -
'"c
>-
0.9
O
;,; 0.8
E
o
O 0 .7

'"''""
::;
0.6

êií 0.5

DA
a;
> 0 .3
"
...J

"'" 0 .2 ~
E
O

a
'"
'"
<li
.o
o
0 .1 ~

O
O 0.5 1.5
f
Time (s)
2 2.5 3 3.5

Figure 10. Global damage evolution for the dynamic analysis.

2 14 ST RUCTURAL ANALY$ I$ OF IIISTORI CAL CONSTR UCTIONS

Figure 11.a Contours of the damaged zones during t he dynam ic

loading history, corresponding to the triggering of the
damage process, for the time instants of 1.6,1. 7.1.84
and 2.0 seconds.
 E. ONATE C( .:11. I Fini!c clcll1cnt d.:llmgl! modd

Figure ll.b Contours of t he damaged ZQnes during the dynamic

loading historYI correspond ing to the final stages be-
fore structura,1 collapse, for the time inst.ants of 2.2,
2.9 , 3.0 and 3.02 seconds.
2 16 STR UCTURAL ANALYSJS OF HI STOR 1CA L CONSTR UCT10N$

Figure 12. Deformed shapes corresponding to the vibratioll af the

slruclure immediate1:v befare callapse.
 E. ONAlE c( a!. I Finite element damagc modcl 217

Fig ure 13. Fini te Element mes h for the full 3D analysis o f l he 5
do mes .

0.9
0.8
x 0.7
Q)

"
E
Q)
0.6
Ol 0.5
'"
E 0.4
'"
O
0.3
0.2
0.1
O
2.5 3 3.5 4 4.5 5 5.5 6
Weight Factor
Figure 14. Evolu tio n of global damage funct ion of l he self-wcight.
facl.or.
218 STRUCTURAL ANALYS1S OF HISTORICAL CON$TRUCT10NS

Figure 15. Contollrs of eq llal displacement for a 5.75 self-weighl.

factor.
 E. ONATE cl al. I Finite elernenl darnage rnodcl 219

F'ip:un~ 16. Damag~ contours for a 5.75 self-weight factor.

220 $TRUCTURAL ANALY$I$ OF HI$TORICAL CQNSTRUCTIONS

,lf"" , -~

• •
f

l'

,
•
-~
lê'

l'
T .•

--=--:-.~~ =--
~
• -==:
?''' ~'
. .,' ..
~ ?~ .c ,

" ,
,,.
~

... 4"·

"',., ~ . ~.

\
\

Figure 17. Crack patterns for a 5.75 self-weight factor.

 E. ONATE el aI. I Finile demenl darn.3gc rnodcl 221

F I ~HI"c 18. Abo"{': The cenl,ral dome or SI.. ~Iark'!" na.<;ilica.

RdO\.... : PcrspccI.ive o r SI.. Mark's sq1l8r(' rrmll , IH" 1"00f.
222 ST RUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTlONS

contours are shown in Figure 16.

Figure 17 shows finally t he crack pattern at failure. Cracks have been
assumed to appear at cach integration point in the orthogonal direction to
the maximum principal strain. The size af cach crack has becn defined as
proportional to the damagc levei at that point. Figure 17 shows that failure
Dccur due to accu mu lated damage at the base of the masonry walls and at
the central stone arehes.
Figure 18 shows finally a close-up picture af the central dome taken
during a recent visit ofthe first author to St. Marks Basilica in the company
af the architect rcsponsi ble for its maintenance Mr. E. Vio. A perspectivc 01'
St. Marks squ are taken from t he Basi lica's roor is also show n in the seconel
pho tograph.

7. CONCLUDING REMARKS

The environmental-mechanical damage methodology presenteei can be

applied succcssfully to assess the structural conditions and cstimatc t he
safety levei anel durability of historical constructions uncler static anel dy-
namic loading. A full coupled solu tion taking into account physical-chemi-
cal-biological degradation anel mechanical effects is nowadays possible anel
this will allow to t race the history of structural pathologies and to design the
correct intervention measures. A particularly intcrcsting imlllcdiate app li-
cation of the damage model here proposed is the evaluation of the structuml
strength incr"ease fo r different intervention measures, thus allowing to opti-
mize any restorat ion investment .

REFERENCES

[I] ENCO (cel ). Materiali 1Iegli Edifi ei Stod ei: Degmdo e Restaur·o. by
Collcpard i, M. , Coppola, L. , Spresiano 1990
[2] lABSE (ed ). Struetural Prese"",Ú01l D/ the A,~hitec tnral Heritalje.
lABSE Sy lllposiurn , Rome 1993
[3] Brebbia. C.A. and Leftheris, B. (Eds.) " Strnctuml Stnr/ies o/ Hist01'Í.cal
lJuildiuüs" Com puto Mech. Publication 199fi
[..J ] Collcpardi, M. Degradation and restoration of masonry walls of histOl"·
ical buildillgS. Material and Structnres, 1990, 23 , 8 1-102.
[5] Oüate, E. Reli"ability analysis of coricrete structures. Numcl'ical anel cx-
pcrilllclltal st uelics, Evoluzione nella speT-imentazione lJer te coslnLzion1..
pp. 125-14li, Seminal' CISM, Merano , April1994.
[li] BmlH.nt, Z. , " Mechanicsofdistributcd cracking" , Appl. Mech. Rev., Vol.
39 . 1'1'. 676-705, 1986
 E. ONATE ct a!. I Finitc elcrncnt d:l.Iuage rnodeJ 22 3

[7] ASCE Committee on Concrete and Masonry Structures, Task COIll-

mittee ou Finite Element Analysis of Reínforeed Conerete Strueturcs:
A State-of-The-Art Report on Finite Element Analysis of Reinfol'ccd
Concrete Structures, ASCE Spec. Pub. , 1981
[8] Wastiels, J., " Behavíour of eouerete under multi-axial stresses a review"
Cement and Concrete R esearch, VoI. 9 , pp. 35-44 , 1979
[9] Kupfer, H.S. and Gerstle, K.K. , "Sehaviour of conerptp. lIncier hiaxial
stresses" , ASCE Jouma/ of the Eng. Mech. Div. , VoI. 99. N° EM4 .
pp. 853-866, 1973
[10] Oller, S., "Modelí zaeióu numérica de materiales friecionales·· . .MallO-
graph CIM NE, 1991
[lI ] Mang, H., Biéanié, N. and de Borst, R. (Eds.) . " Com!",ter mode/in" of
concrete structures" , Proe. EURO-C, Innsbruck. Austria. 1994
[12] Ooate, E., ~Uer , S., Oliver , J. and Lubliner, J., " A constitutive model
for r.raeking of eonerete based on the incrementai theary of plasticiI,Y" .
Engng. Comput., 5 , 309-20 , 1988
[13] Lubliner, L. , OlIer, S. , Oliver, J. and Oll ate, E. , " A plastic dalllagc
model for Ilonlinear analysis of eoneretc", lnt. J. SolúL StnLct .. Vol. 25.
3, pp. 299-326, 1989
[14] Barbat, A.H., OlIer, S. , Oilate, E. and Hanganu 1 A.. " Simulatioll af
damage phenomena in required eone rete buildings subjeeted Lo seis lIli('
aetions", Numerical Methods in Engn. and Applied Scie nces. H. Aldcr
et a!. (Eds. ), CIM NE, Barcelona 1992
[15] Hanganu, A. , OlIer, S. , Onate, E. and Barbat. A. H.. ,. A fin ite elclllcnt
model foI' damage analysis of nuclear reaetor eontainment shclls··. f!d
National Conference on Boundar'lj and Finite Element. Sibiu. ROlllauia.
May, 13- 15, 1993
[16] Cervera, M., Oliver, J .. Hcrrcro, E. and O ilate, E .... A computatiollal
model for progressive craek ing in large dams due to s\Vclling of concr('f.('·.
Engng. Ftucture Mechanics, 35, No. 1,2,3, pp. 575-85.1990 .
[17] Lemaitrc, J., " A Continuous Damage Mcehanics Moclel for Duetilc Pl'H C-
ture" , Jouma/ Engng. Mater. Tech. , VoI. 107 , pp. 83-89 . 1985.
118] Lemaitre, J. , "How to use Damage Mechanics·' . Nuclem' ElIgil/ eerll/ !1
and Design, No. 80, pp. 233-245, 1984
[19] Simó, J.C. and Ju, J.W. , "Strain and Stress Based Co ntillUllll1 Darnagc
Models-I. Formulation" , lntenwtionaJ Journal Solids f:J St7'uGlUI'e.s, Vul.
23, pp. 821-840, 1987.
(20) Simó, J.C. and Ju, J .W .. "Strain anel Stress Based Cont inulllll O;UIl-
age Moelels-Il. Computational Aspects··. Inlenwtional Jou7'1wl Solids e~
Structures. VoI. 23. pp. 841-869, 1987.
[2 1] Kachanov, L.M. , "Continuum lVlodel of Medium with Craeks··. Jou.rnal
of the Engineering Mechanics Division. ASCE. VoI. 106. No. EM5. PI'.
224 STRUCTU RAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

1039-1051 , 1980.
[22] Faria, R. and Oliver, J. ) " A rate dependent plastic-damage constitutive
model for large scale computat ion in concrete structurcs" MOnO.llHlph
CIMNE No. 17, Barcelona, J anuary 1993.
[23] Oliver, J. Cervcra, M. } OUer, S. and Lubliner, J. ) " Isot ropic dalllagc
models and smeared crack analysis of conerete" . Proe. 2d int. Conf.
on Comp. Aided Analysis of Concrcte St ruct. ) Zell am Sec, Austria. pp.
445-57, N. Bicanic et aI. (Eds.), Balkema 1990.
[24] Oilate, E., Oliver, J. and Bugeda, G" "Finite elcrnent analysis of no 11-
linear response of couerete dams subject to internalloads., " Europe-US
Symposium on Finite Element Methods for Nonlinear Problems, (Editcd
by Bergan, Bathe and Wunderlich) Springer Verl ag, 1986.
[25] Arrea, M. and l ngraffea, A.R.} I' Mixed rnode crack propagatioll in lllOr-
tar and concrete", Comell Univ., De]Jt. St11LCt . Engng. Repo1't 81 -1 .1,
IUraka, New York, 198!.
[26] Cervera, M., Oliver, J . and Galindo, M. ," Simulación Numérica rie Pu-
tologías en Presas de Hormigónll , Monografía ClM NE no. 4, Barcelona,
June 199!.
[27] Cervera, M. , Oliver, J, and Galindo, M.," Numerical Analysis of Dams
with Extensive Cl'acking Resulting from Concrete Hydrat ion: simula-
t ion of a real case", Dam Engineering, VoI. 3 , Jssue 1, 1992 .
[28] Hanganu, A. , IIAnálisis Dinám ico No Lineal de Estructu ras de Horllligón
Armado con Modelos de Dano", Ph.D. Thesis, in developmellt.
[29] Saetta, A. , Schrefler, 8., Vitaliani, R. " The carbonation of COllcrete
and the mechanism of moist ure, heat and carbon dioxide flow through
porous materiais, Cem. and Cone. Res. , 1993, 23, 761-772.
[30] Saetta, A., Scotta, R., Vialiani , R. ll The numerical analysis of chloridt:
penetration in co ncrete, AGI Mat. lour., 1993, 90,441-451.
[31J C reazza, G. Structural Behaviour of San Marco Basil ica, Venice IABSE
SEI Volume, 3 Nu mber 1, February 1993.
[32J Collepardi, M. , Coppola L. , Monosi, S. Chemical attack of calciulll c:ltlo-
ride on the Port land cement paste, il Cemento, 1989, 2, 97-104.
[33] Schefler, B.D. , ':Finite elements in environmental engineerillg: COlllpll!le
t hermo- hydro-mechanical processes in porous media includiug poll11tanl,
transport" ,Archives Df Computational Methods in Engineer'ing, 2, J, 1-
55, 1995
[34] Creazza, G. ) Saetta, A. ) Scotta, R" Vitaliani. R. anel Oüate. E.. "j\llat.!J-
ematical sirnulations ofstructural damage in historical bui ldings". Stnu;-
tural Studies 0/ Historieal Buildings IV, C. A. Brebbia and B. Lcft.l"'ris
(Eds. ), Vol. 1 , pp. 111 ,1995
[35J Ziekiewicz, O.C. and Taylor, R.L., "The Finite Element Me t,had .. Mc:-
Graw Hill, Vol. I , 1989, Vol. 2, 1991.
 STRUCTURAL ANALYSIS DF HISTDRlCAL CONSTRUCTIONS
P. Roca, I.L González, A.R. Mari and E. Oi\ate (Eds.)
e ClMNE, Barcelona 1996

CHARLES BRIDGE IN PRAGUE

y
D. Novák and J. Zák
Technical University of Bmo
Veveri 95, 662 37 Brno
Czech Republic

SUMMARY

The last controversial reconstruction of the famous Charles bridge caused that
the bridge has changed its structural behaviour. This resulted in excessive
defleclions and cracks. The aim 01 lhe presenl sludy was lo undersland lhe
structural behaviour of lhe bridge using the finite element modelling in
combination with a statistical simulation respecting different materiais and their
interactions. The variability of seleeted input parameters is taking into aeeount -
they are eonsidered to be random variables. The statistical characteristics of a
struetural response were calculated based on the advaneed Monte Carla type
simulation. lhe distribution functions of selected stresses and deflections are
estimated providing a basis for confidence limits. The stochastic sensitivity
analysis was also performed , which gives information about the influence of
random. variables on the structural response . Temperature load cases were
analysed in arder lo assess lhe presenl slage and two suggesled varianls 01
the reeonstruction . The results should be used for the evaluation of the
proposed reconstruction and the selection of the most suitable way of repair
works.

1. INTRODUCTlON

The aim 01 Ihis paper is lo describe FEM modelling and probabilistic

assessment proeedures used for the analysis of the present stage and
proposed reconstruction . lhe statistical approach is described in a way which
can be generally useful for lhe statistical analysis of historical structures. lhe
description of techniques used , lhe assumptions made and the interpretation of
resulls are included . The lechniques presenled should conlribule lo lhe effort 01
co rrect assessment of safety of structures.

lhe safety can be considered to be a degree of assurance to maintain

its required function and lhe capacity of resistance against deteriorating
ca uses. Generally. salety leveis should be dislinguished and lhe salely should
be evalualed aI lour basic slages: The planning slage. lhe design slage. lhe
construction stage and the maintenance stage.
226 ST RUCT URAL ANALYSIS OF H ISTORI CA L CONST RUCT IONS

Considering historieal bridges the last item - the maintenance 5tage, is

the only rema ining cne for us to save our historieal heritage for future
generations. Our daily experience shows what would happen if either
inspection ar maintenance works were ignored . For a required functioning Df a
bridge the inspection and feas ible maintenance treatment are inevitably
necessary. Nowadays il is generally accepled Ihal lhe so-called mainlenance-
free bridges do nol exist Th is applies, of course, for medieval masonry bridges,
100.

Generally, lhe Iraditional melhods of analysis Ireal lhe physical

uncertainties Df various parameters in a sim pie and efficient way. Minimum
quaranteed values cf these parameters, usually following the relevant standard
ar code of practice (e.9 . strength, modulus, etc.) are used for the analysis.
This is not applicable for some more complicated structures where lhe minimum
(or maximum) quaranteed values may not produce the worse structural
behaviour. Charles bridge is a typical example of such a structure. Due to the
composition of different materiaIs one cannot predict the structural behaviour
under temperature loads.

A statistical approach provides us with a sim pie option how to deal with
the above problem . If input parameters are significantly uncertain and structural
behaviour unpredictable let the selected parameters be considered random
variables. lhen the treatment of structural system becomes feas ible at higher
quantitative levei by efficient structural analysis as well as advanced stochastic
procedures such as Monte Carlo type simulations. An acceptable determination
of input parameters when analysing historical structures is ofien very limited.
lhe statistical approach is, in some cases, th e only way how to get a realistic
prediction of structural behaviour.

2_ THE HISTORY OF CHARLES BRIDGE

Charles bridge, Fig. 1, belongs lo the mosl valuab le hislorical

monumenls in lhe Czech Republic. II is probably lhe biggesl lourisl attraction in
Prague and the Czechs are very proud of it.

The bridge was designed and buill during lhe reign of King Charles IV
by famous architect Petr Parléf and was finished in 1402. lhe bridge is 515.76
m long and lhe average width is 9.40 m. II has 16 arches wilh lhe span from
16,62 lo 23.38 m supported by 17 piers. Charles bridge was originally called
"Slone Bridge" and was buill as lhe second slone bridge in Prague. The
bridge was always the most important and busiest traffic route in the city and
still creates an inseparable part of Prague's panorama with the most
impressive view of lhe Prague Caslle. The Old Town Bridge Tower and many
historical sculptures contribute to the outstanding general impression of the
bridge.
 D. NOVÁK and J. ZÁ K I Charles Bridge in Prague 227

The bridge was damaged many times during wars and by fioods. The
fioods that partly damaged the bridge were in 1432, 1496, 1784 and the worse
one, in 1890, completely destroyed two piers and three arches. lhe damage
was considered a national disaster and the bridge was repaired in 1892. lhe
damaged piers were founded on caissons and other piers were also secured by
caissons. lhe damaged arches and piers were repaired using in most cases
different building stones instead of the original ones. lhese repairs resulted ,
together with different technologies used , in a great diversity of the bridge.

lhe filling materiais were removed and replaced by expanded clay

concrete and a reinforced concrete slab without expansion joints during the last
controversial reconstruction (1966-1975) . lhe slab was anchored to the
parapet walls and granite paving was put in cement mortar. lhese changes
caused the bridge to become toa rigid and resulted in several cracks mainly
paraJlel to the lengthwise axis of the bridge. The cracks did not occur in joints
only, but some of them appeared in the ashlars. Excessive deflections of the
parapet waJls were detected on severa I arches.

Figure 1: The views 01 Cha rles bridge.

228 STRUCTURA L ANA LYSIS OF HI STOR ICA L CONSTRUCTI ONS

lhe above rnentioned damage together with the surface erosion af

building stones initiated a new investigation supervised by the Charles Bridge
Committee in 1994. Many bridge and structural engineers, architects, chemists,
geo109i5t5, mineralogists and state administrators were involved in a
comprehensive preliminary investigation under several sub-committees lhat
resulted in a propasal af new reconstruction af the bridge. lhe second author af
this paper had a chance to participate as a member af the sub-committee
Structural Analysis .

3. DETERMINISTIC COMPUTATIONAL MODEL

Finite element package ANSYS [1 ] was used as a basic tool for the
present analysis. This commercial package offers a good selection af suitable
2D. 3D and special elements. It is still practicaliy impossible to analyse the
whole structure like Charles bridge using 3D elements. 2D solid model of
selected cross-sections (without modelling joints between different materiais)
was developed in the first step. Then the resu lts were compared to the resu lts
of corresponding 3D model of a quarter of the arch. It was confirmed that the
differences were quite significant.

20 models of selected cross-sections respecting joints between different

materiais were developed for the present analysis in three different va riants
corresponding to A) the current state and the state after the proposed
reconstruction in the two variants - B) without expansion joints and C} with
expansion joints. Arch number IX was selected for lhe analysis. Two cross-
sections (on the top of the arch and on the foot of the arch) were analysed
using 2D plane strain finite elements (Fig. 2).

Figure 2: FEM mesh of the bridge cross-section.

 D. NOVÁ K and J. ZÁ K I Ch:lrlt:s Bridge in Pr:lgue 229

Three different load cases were considered in lhe analysis as a· result

of the study of meteorological data: I1 warm top surface and cold bottom
surface, 111 warm surface and cold core, 1111 cold surface and warm core .
Respecting the collected meteorological data and previous analyses a non-
stationary temperature field analysis was carried out to obtain possible
temperature gradients in the structure under different conditions. The results
were used as boundary conditions lor the lollowing stationary analyses . 2D
temperature elements (PLANE55) were used to obtain the distribution of
temperature field in the cross-section in lhe next step of the solution. The
calculated temperatures (see Fig. 3 for the load case 11) were converted into
initial strains and then transformed into equivalent externai loads for the
structural analysis. This is the last part of the analysis Ihat was carried out
once, and was not a part of the stochastic repetitive analysis.

Figure 3: Temperature distribution for the load case 11.

paving
erpanded clay concrete
concrete slab
arenaceotLS marl
(3 sandstone

Figure 4: Materiais 01 the bridge.

230 ST RUCT URAL ANALYSIS OF HI STOR ICAL CONST RUCTIONS

lhe temperature elements were changed to the corresponding

structural plane strain elements (PLANE42), and then a standard structural
analysis was performed cne hundred times for various values af slochastic
parameters. lhe bridge was built af sandstone, arenaceous marl , unknown
materiais and granite paving in sand . lhe current materiais are sandstone,
arenaceous marl, concrete slab anchored to the walls, expanded clay
concrete and granite paving in morta r, Fig . 4 . lhe proposed repa ir works
comprise the removal of the concrete slab, expansion joints Df expanded clay
concrete andgranite paving in sand . Different materiais were connected
using special contact elements (CONTAC12) which enable us to model friction ,
normal and tangential inilial stiffness and possible gaps. These elements
generally require Newton-Raphson iterations to find the correct status of
contacts (opened or closed). Correct input of initial status avoided time
consu ming iterations to be dane for ali runs of stochastic analyses .

It was quite difficult to input correct coefficients of thermal expansion in

the longitudinal direction and for expanded clay concrete with expansion
joints. The long itudinal strains are zero for the plane strain analysis but the
coefficient of the thermal expansion would produce unrealistic stresses in this
direction. On the other hand. zero va lue of the coefficient of thermal
expansion would not be correct for this analysis. As a compromise the
coefficient of the thermal expansion in the longitudinal direction was
decreased . It would also be difficult to model expansion joints. We assumed
that the strains perpendicular to the expansion joints did not almost produce
any stresses. So we again assumed the value of the coefficient of the thermal
expansion to be decreased in this directio n.

The deterministic procedure can be summarised in the following steps:

• Oetermination of material characteristics . Sources: tests, literature, previous

projects, engineering judgement.
• Collection of meteorological data , analysis of non-stationary temperature
field . Result: boundary condition for stationary temperature field analysis.
• Stationary temperature field analysis. Result: temperature distribution .
• Temperatures converted into initial strains. Dane automaticaJly in ANSXS.
• Structural analysis. Newton-Raphson iteration if necessary. Resu lts:
deflection. strain, stresses.

4. STATlSTICAL APPROACH

4 1 General remarks

The rel iability methods used for the probabilistic assessment of Charles
bridge are described in this section . They have a general use and can easily be
applied to a different computational model, to a different bridge , to a different
historical structure. The authors believe that the so-called probabilistic risk
assessment procedures will be applied in the field af histarical structures more
 D. NOVÁ K and J. UK I Charles Bridge in Prague 231

often in the future , and the techniques described here represent one possible
alternative how to deal with uncertainties of historical structures.

A statistical approach was utilised to model important basic material

properties of Charles bridge that were considered as random variables.
Their statistical characteristics were estimated and then used for statistical,
sensitivity and probabilistic analyses. A powerful numerical simulation
technique Latln Hypercube Sampllng (LHS) [2], [31 in combination of FEM
package ANSYS with different realisations of input random variables were
used a hundred times for each load case and each variant, together nine
hundred single structural analyses were performed . The aim was to obtain
the scatter of stresses and deflections - to estimate their mean values and
variances in selected nodes or elements and to use these results to evaluate
the current state of the bridge and the proposed repa ir works.

The statistical analysis of Charles bridge has been performed in three

steps: Firstly, Latin Hypercube samples were generated for random variables
separately for ali simulations without calculating structural response by FEM
immediately. Then structural analysis of the bridge by ANSYS code was done a
hundred times utilising once generated data according to LHS scheme.
Selected response variables were saved in files . Finally, statistical sets are
statistically evaluated, basic statistical characteristics of selected response
variables are obtained . The triangle of lhe solution is sketched in Fig . 5.

Latin Hypercube
Sampling

100 time simulated

randam variables

Reliability assessment:
ANSYS - FEM res anse = Statistics
modelling variables = Sensitivity analysis
= Distribution functions

Figure 5: Three basic steps of the analysis.

232 STRUCTURAL ANALYS IS OF HI STORICAL CONSTR UCTlONS

4.2 loput random yarjables

In probabilistic calculations the definition cf the set Df random variables is

lhe primary lask. li consisls of two sleps: Firslly, lhe choice of lhe paramelers
which will be considered stochastic and will be simulated as random variables.
In our case cf modelling thermal effect we consider only the most influencing
material properties Df lhe bridge - Young modulus of elasticity and the
coefficient of therma l expansion for ali materiais used in the bridge structure to
be random .

Secondly, every random variable X should be described wilh lhe aid of

the probability distribution function . Such a function is in many cases (two-
paramelric dislribulions) characlerised by a particular type of dislribulion (e. g. a
normal distribution ar a uniform distribution), mean value J.1 (X) and standard
devialion ,,(X). Very oflen lhe coefficienl of varialion COV(X) ~ ,,(X)/ p( X) is
used as a relative measure cf variability.

If slalislical dala is available for ali lhe variables, lhe slalislical

characteristics can be estimated and a theoretical model of probability
distribution function can be assigned by well known procedures of mathematical
statistics. In most cases there is scanty data available. The statistical properties
will then have to be entirely estimated based on different assumptions .

The following sources for estimations (also for deterministic parameters

involved) were used :

• in situ measurements
• information published in literature
• experience and opinion of experts
• engineering judgement and intuition

A synthesis of ali information resulted in the estimation of limit values, X'''):h 'X ,,,.... '

above or below which the relevant variab le will not be in ali probability situated .
For the normal distribution it is well known there is 95 % probability that a
variab le will have a value situated between J.l - 2a and J.l + 2a. Based on this
assumption it is th us possible to estimate the mean value and the standard
deviation :

JJ(X)~~lx"" + x,,,.} ( 1)

,,(X) ~ ±lx,,,,,, - x,,,.} (2)

For a log ~ norma l distribution the corresponding formulae are:
Il( X) ~ Jx"" .x,,,. (3)

cov(x) ~ ±Inlx""/x,,. } (4)

 D. NOVÁ K and J. ZÁK I Char[es Bridge in Praguc 233

lhe choice between using the normal and log-normal distribution will
depend on the physical nature of the random variable . Log-normal distribution
is to be preferred for variables which are unable by nature to take on negative
values. Bul il should be noled Ihal for small coefficienl of varialion (COV s 0.10)
the difference between normal and log-normal distribution is negligible for
practical purposes.

lhe above assumptions resulted in a choice of the type of distribution ,

the mean value and the standard deviation of the probfem random variables.
Random variables XI' X l " ' " X I O considered here are listed in labia 1. lhe
mean values and standard deviations adopted are those values which were
rated as providing the best estimates of the type indicated above. The normal
probability distribution was considered to model Young modulus of elasticity
(E) and the log-normal distribution for the coefficients of thermal expansion
(,1.). The statistical correlation among random variables was not considered
here due lo lhe lack of sufficienl informalion. For varia nls B) and C) lhe lable of
random variables without concrete slab was used.

Material x, .t "," x/o"~ .u(X) a(X) cov(X)

sandstone E - X, 5.0 15. 10.0 2.50 0 .25
). = x : 0.7 1.0 0.84 0 .09 0 .09
arenaceous marl E = X, 3.0 8.0 5.5 0 .23 0.23
). = x . 0.7 1.0 0 .84 0 .09 0 .09
concrete slab E = X, 18 . 28. 23 .0 0 .11 0.11
À = .\'0 1.0 1.2 1.10 0.05 0 .05
expanded clay E = X, 8 .0 12 . 10.0 1.00 0 .10
concrete À = XI 0.4 0 .9 0.60 0.12 0 .20
granite paving E = x~ 4 .0 8 .0 6 .00 1.00 0 .17
À = X,O 1.0 1.2 1.10 0.05 0 .05

Table 1: Random variables (E in Gpa , A in ' C ' x 10-' ).

4 .3 Statistical analys js

Response funclion g( X) (FEM compulalional model) of random

variables X =X I .X 2 •.••.•• X IO gives us response variables (e. g. deflections,
stresses) . The response variables are arranged into vector Z :

Z = g(X) (5)

lhe aim of the statislical analysis is the estimation of basic statistical

parameters of response variables , e.g. mean values and variances. It can
2:14 STRUCTURAL ANALYS1S DF H1STOR1CAL CONST RUCT10NS

easily be done by Monte Carlo simulation, by repetitive ca lcu latians of th e

structure. In case of simple Monte Carla simulation such a process is time-
consuming .

A special type of numerical simulation Latin Hypercube Sampling

enables us to use a small number of simulation (a hundred) for satisfactory
good estimates of basic statistical parameters [2 ], [31. The probability
distribution functions for ali random variables are divided into N equivalent
intervals (N is a number of simulations), the centroids of intervals are then used
in simulation process o This means that the range af the probability distribution
function F( X J ) of each random variable X I is divided into N interva ls of equal
probability 1/N, see Fig . 6 . l he representative parameters of variables are
selected random ly based on randam permutations of integers 1, 2, ... i, ... ,N.
Every interval of each variab le must be used only once during the simulation . A
tab le of random permutations can be used conven iently, each row af such a
table belongs to the specific simulation and the co lumn corresponds to one of
the input random variab les [31 .

F(X)

,",'
3
2
, , X

Figure 6: l he divisian of the distri bution function into intervals.

4.4 Sensitiv jty analysjs

An impartant task in the structural reliability analysis is to determine the

significance of randam va riables - haw they influence a response fun ction of a
specific problem. There are many different approaches of sensitivity analysis , a
summary of present methods is gi ven in [4].

LHS simulation can be effieiently used to obtain sueh an information. lhe

sensitivity analys is is obtained as an additional result. lhis provides us with the
information about the importanee of basie random va riables. Dominating and
nondominating variables can be disting uished using certa in sensitivity
 D. NOVÁK and J. ZÁK I Charles Bridge in Praguc 235

measures. Here two different types of sensitivity were used : Nonparametric

rank·order correlation and sensitivity in the form of variation coefficients.

lhe relative effect of each basic variable on the structural response can
be measured using the partial correlation coefficient between each basic input
variable and the response variable. lhe method is based on lhe assumption
that the random variable which influences the response variable most
considerably (either in a positive or negative sense) will have a higher
correlalion coefficient lhan other variables. In case af a very weak influence lhe
carrelatian caefficient will be quite close to zero. In case af Latin Hypercube
Sampling this kind of sensitivity analysis can be peliormed almast directly
without any particular additional computational effort. Because the model for
structural response is generally nonlinear, a nonparametric rank·order
correlalion is utilised . lhe key concept of nonparametric correlation is this:
Inslead of the aclual numerical values we consider lhe values af ils rank among
ali olher values in lhe sample. lhal is 1. 2•.... N. lhen lhe resulling lisl of
numbers will be drawn frem a peliectly known probability distribution function .
lhe integers are uniformly distribuled . In case of Latin Hypercube Sampling lhe
representative values of random variables cannot be idenlical . Iherefore there is
no necessity to consider mid·ranking . lhe nonparametric correlation is more
robusl Ihan lhe linear correlation, more resistanl lo defects in data and also
distribution independent. lherefore il is particularly suitable for lhe sensitivity
analysis based on Latin Hypercube Sampling .

As a measure of nonparametric correlation we use the statistic called

Kendall's tau . II uses only the relative ordering af ranks : higher in rank , lower in
rank , ar the same in rank . Since it uses a weak property of data , Kendall's tau
can be considered as a very robust . For a detailed description of ca lculation ,
see [41, here we presenl anly a symbolic farmulae . Kendall's tau is lhe function
of ranks q I' (lhe rank of a representative value of the random variable X, in an
ordered sample of N simulated values used in the j·th simulation which is
equivalent to the integers in lhe table of random permutations in lhe LHS
method) and 17 , (lhe rank in ordered sample of lhe respanse variable abtained
by lhe i-Ih run of lhe simulalion process):

r . = r(q " .p, ). }=1 .2... , N (6)

In this way the correlation coefficient r, E(- Ll) can easily be abtained
for an arbitrary randam variable and we can compare Ihem . lhe greater
absolule value of r , for a variable X, ' lhe greater influence has this variable
on the structural respanse . An advantage of Ihis approach is the fact Ihat a
sensitivity measure for ali randam va riables can be obtained directly within one
simulalion analysis.

Anolher method for lhe sensitivity analysis can be applied also based on
LHS method . lhe melhod is based on the comparison of variation coefficients
of random variables and a response variable . Let us designate the partial
236 ST RUCT URAL ANALYSIS OF I-II STO RI CA L CONSTRUCT IONS

coefficient af variatian COv(z,) for the case in which the randam variable X; is
the anly one treated as randam in the simulation process o COv( X,) is the
coefficient af variation of this variable. Other basic variables are kept at their
mean va lues. The partia I sensitivity factor a ; for lhe random variab le X, may
be defined as:

cov(z, )
, =cov( X, )
a (7)

Such factors express the relative influence of individual variables on the

variability cf structural response . If ali n random variables are considered to be
random , the fol lowing approximate formulae may be written :

,
COV(Z) =[t (a,cov(x,))']' (8)

It can be seen that the absolute influence is represented by the va lue af

COV(Z,)' . Such sensitivity may easily be depicted using pie charts. The
absolute influence of variable X, on the variability af response variable
(COv(Z ) ) is best presented in percentage terms.

In case of sensitivity analysis in the terms af COV the simulation process

has to be repeated n-times. This fact represents a great disadvantage for
reliability problems with large number of random variables and time-consuming
FEM calculation of structural response . For such cases, as modelling Charles
bridge, Latin Hypercube response approximation can be efficiently used {5).

The aim is to approximate response function g(X) in order to avoid a

time-cansuming solution of the system of linear algebraic equation s. This can
be done by Taylor"s series around the expansion poinl. Neglecting higher order
terms a secand-order polynomial can be considered in standard space of
ra ndom variables ); = (x, - Ji( X, ))1o-(X,) , where mean va lues are expanslon
points:

" " "

g(Y) =ao+ L) ,Y, + LLc, ):t; (9)
/. 1 , a i ./ . 1

bpc" are caefficie nts related to the value of partial derivatives and ao is the
va lue of the response function at expansion poin!, ao = g(,lI(X)).

Based on this con cept a multiple regression model can be used to

determine coefficients. The most decisive task is the seleclion of interpolation
 D. NOY ÁK and J. ZÁ K I Charles B ridg~ in Prague 237

points and the way of perturbing them around the expansion poinl. The Latin
Hypercube Sampling ensures that multidimensional space of basic variables is
covered by generated points quite regularly with a small number of simulations
in comparison with the simple Monte Carlo random sampling to reach a
comparable accuracy, the procedure can be called the Latin Hypercube
Response Approximation Method (LHRAM) [5). This advantage can be used to
obtain interpolation points for an efficient response approximation . Because
tensor C'I is symmetric, at least N = 2n + 11(11 - 1) / 2 paints must be generated to
obtain polynomial coefficients. Generaliy higher N can be used, which leads to
the multiple regression mode!. In our case of Charles bridge analysis we use
N;;100 which is greater than the minimum necessary N;; 65.

4.5 Probabilistic analvsis

Comparing structural response variable 2 j (X) with required values

2 /. ",'1 in a certain chasen range , one can estimate the complementary
cumu lative distribution function of response variable Z I

<D(Z,)= p(Z, (X) ~ Zi"" ) (10)

In th is way the probability that the structura J respanse variable (stress ar

deflection) will be greater than the prescribed value is defined . Confidence
limits for response variables can be estimated. A concept of the probability
distribution functian in the complementary farm is used in order to express
"subjective failure".

Here the numerical calculat ion of fa ilure probabilities for the estimation of
the comp lementary cumulative distributian function was based on a curve
fitting approach [3) , [6) . The selection of the most suitable theoretical model for
the probability distribution function of a response variable was fi rst done
comparing theoretical and empirica l probability distributions obtained from the
simulation . The most suitable theoretical model among competing ones was
selected and then used for the calculation of probabilities.

5. NUMERICAL RESUL TS

5.1 Oeterministic results

Sorting the results we had to solve the usual problem of any FEM
analysis: there were too many available output data so it was difticult to select
the most interesting and important ones. We considered node deflections,
stress components and principal stresses in the selected elements.
2~8 STR UCTURAL ANALYS IS OF HISTOR ICAL CONST RUCTIONS

The lollowing results should be understood as a sample 01 possibilities

given by ANSYS package and are related to the top 01 the arch, load case li ,
variant A Contours of principal tension stresses on deformed shape a re shown
in Fig . 7 . Contours of principal compress ion stresses are displayed in Fig. 8.
Gaps between different materiais due to the temperature load appeared in both
figures . Oirections and magnitudes of principal stresses are illustrated in Fig . 9.

Figure 7: Principal tension stresses.

Figure 8 : Principal compression stresses .

 D. NOVÁK :lIld J. ZÁK I Charles Bridge in Prague 239

Figure 9: Principal stresses.

5.2 Statistical results

From ali results obtained by the statistical simulation the results for the
principal tension stress at position a (see Fig. 10), for the horizontal deflection
ai lhe lop edge of lhe parapel wall (posilion fJ) and for lhe principal lension and
compression stress at the position r is shown here.

(a) The principal fensio!1 sfress Clf jJosifion a

lhe complementary probability distribution functions for variants A , B

and C in ali load cases I, 11 and 111 are presenled in Fig . 10. One can easily
observe mean values and fuI! probabilistic information showing confidence
limits (5 and 95% percentiles) . It can be seen that the principal tension stress
significantly decreased in lhe case of variant C comparing to variant A In case
of variant B ;t happened only in the load case I.

The results of nonparametric sensitivity analysis in the form of Kendall's

tau for load case 11 are presented in Fig . 11 . lhe plus sign represents a positive
influence and the minus sign a negative influence of a random variable on the
principal slress, lhe higher lhe more importan\. The dominaling variable of lhe
current state is the coefficient of the thermal expansion of expanded clay
concrete (.\'8) - Fig. 11 A lhis dominancy diminishes in the state after the
proposed reconstruction in variant C where Kendall's tau for variables of
sandstone and expanded clay concrete increased.
240 STRUCTURAL ANALYSIS OF J-[ISTOR 1CA L CONSTRUCT1 0NS

0 .8
~: i -- O B
.-::. 0 .6
;e: 0 .4 --+. ~, -- - C .
,
-i ~ ;:::I
0 .2
~\; , I Z, [kpa]
O
O 10 20 30 40 50 60

I ~;;:-iI -- OR
~
1'1 ~
li

--:.- 0 .6 -.: ~I ---.e

N
e 0.4 3...
0. 2
O ~ Z, [kl'a]
O 10 20 30 40 50 60
" _A- li '
0 .8
~ 06
N' . +---:r--
e OA r,- I ! ! I ~ I

0 .2 I -l ~I""': jg'J~,=_
~ ,- =---. Z, [kPa ]
O
150 200 250 300 350 400 450 500

Figure 10: The principal tension stress (position a)-

the complementary distribution function, load cases I, 11 and 111 .

OA ,,-----,-,-,-,-,-,-,-,-,---,
0.2-l1--++l-++l-++f---I

-'ro" -o.2-l1--H++-Hr--
'O
cQ) _OA.jj--H+ +-H!-
" -0.6 -lI--+++-O-+-+-
.0.8 -ll-"';"'-':'''':'''''I-r--r-r'--r--I
A 1 2 3 4 5 6 7 8 910 C 1 2 3 4 7 8 9 10
Random variable X Random variable X

Figu re 11 : The principal tension stress (position a) - nonparametric sensitivity ,

load case 11 in variants A and C.
 D. NOVÃK and J. ZÃK I Cha.rles Bridge in Prague 24 1

The influence of variability of random variables on the coefficient of

variation (eOV) of the stress (0.53 - variant A, 0.17 - variant e) is expressed by
pie charts in Fig . 12. Here the same trend observed in Fig. 11 can be
interpreted as follows : The contribution of X. is 80 % for variant A but only 14
% for variant C .

,% 0%
11 Xl O X6 C
== X2 ~ X7
2%
• X3 UIII X8 16%

• X, I!!I X9
O X5 11m Xl0 41%

Figure 12: The principal tension stress (position a) - sensitivity in percentage of

cav, load case 11 in variants A and e.

(b) 711e hori=onral defleclion ai lhe top edge oflhe paropelll'oll (posilion p;
The proposed reconstruction modelled by variants B, C resulted in a
decrease of the horizontal defiection at the edge of the parapet wa ll. For load
case I the situation is shown in Fig. 13 . 90 % confidence intervals are [mm]: A
(0.449 . 0.586) , B (0.286, 0.490) and e (0 .114, 0.175). It is obvious that variant
C leads to better confidence inlervals also in this case of the horizontal
deflection . On the contrary to the principal stress in load case 11 the importance
of the coefficient of the thermal expansion of expanded clay concrete did not
decrease , Fig . 14 (CaV 0.08 - variant A, 0.13 - variant e).

0.8
0.6
N'" 0 .4 '----'
e 0 .2 - - i-
1. I
0 - - ' ,.--,
0.1 0.15 0.2 0 .25 0 .3 0.35 0,4 0,45 0 .5 0.55 0 .6

Figure 13: The horizontal deflection at the top edge of the parapet wall (position
fJ) - the complementary distribution function , load case I.
242 STR UCTURAL ANALYSIS OF HISTORI CA L CONSTRUCTIONS

3%
C
lII\l X1 O X6

:J X2 ~ X7

11 X3 11111 X8

• X4 ~ X9

O X5 lI!ii. X10
94%
Figure 14: The horizontal deflection at the top edge of the parapet wa ll (position
(3) - sensitivity in percentage of cav, load case I.

(c) The principalwnsion and compression Slress aI position r

The complementary probability distribution functions of the principal
tension and compression stresses for variants A, B and C in ali load cases I, 11
and 111 are presented in Fig . 15 and Fig. 16.

0.8 o
~ 0 .6
90A
•
~

0.2
O
_I Z, [kPn]
O 10 20 30 40 50 60 70 80 90

.. II
O
.-... 0.6
N' •
a OA
0.2
-.. Z, [kl'a]
O
O 50 100 150 200 250 300

0.8
.. 111
O
___ 0.6
N'
a 0.4
I I
0.2
Z, [kl'n]
O
20 40 ~
I
~ 100 1~ I~ 1~ 100 ==
Fig ure 15: The prin cipal tension stress (position r) - the complementary
distribution function , load cases I, 11 and 111.
 D. NOVÂK and J. ZÂK I Charlcs Bridge In Prague 243

A significant decrease of the principal tension stress for va riant C can be

observed for load cases I and 11. Concerning principal compression stresses ,
variant C resulted in significantly lower values with small scatter in load cases I
and 111. For the warm surfaee and the eold core, load case 11 , it is opposite . But
the shape and position of functions is similar, i.e. the coefficient of variation has
approximately lhe same value for ali varianls (0.3).

0.8
-;:0.6 ..
'"
,&0.4
0 .2
• Z, [kl'a ]
-50 -45 -40 -35 -30 -25 -20 -15 -10 -5 O

11
0 .8
-:: 0 .6
N .. I

'90.4
0 .2
O
• - Z, [kl'a]
-220 -200 -180 -160 -140 -120 -100 -80 -60 -40
111
0.8
---0.6
",'
..
'9 0 .4
0.2
O
• , Z, [kl'a]
-400 -350 -300 -250 -200 -ISO -100 -50 O

Figure 16: lhe principal compression stress (position r ) . the complementary

distribution function , load cases I, 11 and 111.

6. CONCLUSIONS

It should be noted that the present statistical simulation was based on

the 20 model that cannot produce accurate comprehensive results for th is
type of structures. lhe presented analysis was done for comparative purposes
of thermal effect in different variants of the structure . Significant differences
were found eomparing struetural deterministie results on 20 and 3D models [7] .
3D model should be used for more reliable results , but we realized that.
bearing in mind contact elements , a total number of elements and a number of
244 ST RUcrU RAL ANA LYSIS D F I-II STDR ICAL CONSTR UCTIONS

analyses , requirements on hardware and the time for the development of the
model would exceed our facilities . Although 3D model would produce more
reliable results the 20 analysis was performed as an initial step and 3D analysis
was left for future investigations.

It should also be mentioned that we did not work with random fields , with
spatial variations of material properties. There are two main reasons to be stuck
with the concept of random variables for the time being . Firstly, random fields
req uire parameters difficult to obtain as the type of correlation fu nction and
correlation length . Secondly, the applicability of presented methods is easier
and straightforward via an interface to standard computational FEM software.
Random fields modelling requires a specialised software based on lhe
stochastic finite element method and the separation of modelling process into
two parts (deterministic computational model and statist ical simulation) cannot
be made so easily.

It was shown that more reliab le results were obtained here by using
the statistical approach rathe r than th e classical deterministic analysis. The
aim was to acquire more detailed knowledge of the behaviour in order to
identify and prevent hazardous situations. These results might be beneficiai for
a better understanding of the structural behaviour and for the proposals of
future reconstructions of Charles bridge .

Our deduction can be summarized into the following items:

• Stochastic results clea rly indicated that significa ntly lower levei of stresses
(both magnitudes and scatters) appeared in variant C of proposed
reconstruction (removed concrete slab, expanded clay co ncrete with
expansion joints) .
• Special contact elements are necessa ry for a realistic prediction of structural
behaviour.
• FEM modelling in a combi nation with statistical simulatl0n Is a suitab le tool
for analyses of historical masonry structures .
• The variability of stresses and deflections appeared to be quite high (COV
reached 0.5 in some cases).
• Distribution functions give us better insight and a basis for confidence limits.

7. ACKNOWLEDGEMENT

Th is research has been co nducted at the Faculty of Civil Eng ineering as a

part of the research project No. B18/94 "Sensitivity and Probabilistic Analysis
of Charles Bridge" and has been supported by the Technical University of
Brno, Czech Republic. The partiaI fi nanciai support from the grant No.
103/93/2037 (Grant Agency of Czech Repub lic) is gratefully acknowledged.
 D. NOVÁ K and J. ZÁK I Charlcs Bridge in Prague 245

8. REFERENCES

1. "ANSYS User's Manual, Revision 5.0" - Volumes I, 11 , III and IV,

Swanson Analysis Systems , Inc., Houston , 1994 .

2. McKAY, M.D., CONOVER , WJ. and BECKMAN , R.J . - A Comparison of

Three Methods for Selecting Values of Input Variables in the Analysis of
Output from a Computer Code, Technomelrics , Zi, 239-248, (1979).

3. NOVAK, D. and KIJAWAlWORAWET. W - "A Comparison of Accurale

Advanced Simulation Melhods and Lalin Hypercube Sampling Melhod
wilh Approximale Curve Fitting lo Solve Reliabilily Problems ", Rep .
No. 34/90 , Institute of Engineering Mechanics, University of Innsbruck,
Austria , 1990.

4. NOVAK , D., TEPL Y, B. and SHIRAISHI , N. - 'Sensitivity Analysis of

Structures: A Review ', Proc. Int. Conf. on Civil & Structural Engineering
Computlng CIVIL- COMP'93 , Edinburgh, Scotland , 1993, p. 201 .

5. NOVAK, D. and SHIRAISHI , N. - "Lalin Hypercube Response

Approximalion and Sensilivily Analysis for Reliabifily Problems ", Rep.
No. 93-ST-01 , School of Civil Engineering , Kyoto University, Japan,
1993.

6. FLORIAN , A. and NOVAK, D. - The Statistical Model Selection for

Random Variables, Software for Engineering Workstations Journal, ,4 ,
No. 3, 158-162 , (1988).

7. KRATOCHVILOVA. D. - "The Analysis 01 lhe Inffuence 01 Temperalure

Changes in Charles Bridge ", diploma thesis, Technical Un iversity of
Brno, Faculty of Civil Engineering , Czech Republic, 1994, (in Czech)
 STRUCTURAL A NALYSIS OF HISTORICAL CO NSTRUCTIONS
P. Roca, J.L. González, A.R. Mari and E. Onate (Eds.)
C CIM NE, Barcelona 1996

ANCIENT DOMES ON SPAIN

J.M. Izquierdo 8emaldo de Quirós

In stituto Técnico de Materiales y COIIsfru cciones ONTEM AO
Monte Esquina 30
280 10 Madrid, Spaill

1. INTRODUCTION

It 's not for strange that one of the first p athology treaties known in history is
related to t he study of a shell on masonry. The moment whell, after the second
half of the 17th. century, charaderistics of matherials and their behaviou r
in a strudure, related to their capacity to whistand loads as well as to their
surrounding enviroment was known, the path to begin a serious study of the
causes why, on certain moments , structures stop behaving properly and begin
showing damages , was set. And, most important, the risk those damages meant
for the structure was beginning to be possible to evaluate.

Up to this point structural collapsing causes were, esentially:

Arcrutect's unskill, or Contractor's, as reflected even in Hammurabi's Code.

Rot of organic materiaIs, and naturaI alteration of inorganic ODes. (House

leprosy, as shown in Numbers 34.45) .

• Natural catastrophes, such as earthquakes, fires or lightning bolts .

• Foundation settlements.

Gods ' will and sin attonement (cvcn aftcr grcck ralionalism , !lee Lk. 13.4,
Silohe's tower).

But Dome behaviour resisted to fit into common behaviour patterns. Architect 's
skill was demonstrated when, shortly after putting off scaffolding, ruin didn't
show up. Employed matherials didn 't easily degradate. Not in alI cases was it
possible to apply to fundatíon settlement or natural catastrophes. And God's
will was utterly díscarded when the collapsed building was one just devoted to
His culto
J. M. IZQUIERDO BERNALDO DE QUIRÓS I Andent domes on Spain 247

Figltn } - The Clergy's Dome, in SaJamanca

Figun 2 - The lantern

248 STRUCT URAL ANALY SIS OF HI STORI CAL CONSTR UCTION S

Pigu~ 3- Cracking on Dome's intrados

photograph ta,ken frorn the inter:section af the Dome 's axis with the floar (this
inclination is common in the lantern s af Salamanca's buildings dated prior
Lisbon's quake, all af thero bending in the saroe direction). In the base of
the lantern, the derangement of Dome's ashlar could be seeo.

The Dome's history was not rallying at all to the technic who was to board its
study.

In 1650 the bitumen for the Dome was being made , what brings the s uspicioD
that the balf orange was in ao advanced construction paiot. Later, in 1667, the
current workmaster, Pedro Mato, S.J., charged six iron flanges for binding the
maio Dome, as some cracks had shown 00 it. Some time befare, ou year 1661,
Marto Jimnez went to Madrid looking for a solution to cover the Dome, due to
exudation shown beside previous bitumen disposal.

Prof. G. de Ceballos, on his studies about Salaroanca's Barroque, concludes

that the cracking happelled when the Dome recived the huge lantern's load.
Even lacking of the original docuroents, it seems odd to me that binding was
a result of repairing decission caused by cracking immediatly produced after
lantern erection, for that would mean that seventeeo years were spent on this
craft, wruch seems too much time specially tak.ing ioto account that only nine
 J. M. IZQUIERDO BERNALDO DE QUIRÓS I Andem domes on Spain 249

And, even then , failures happened. And failure modes, in rnany cases, were
incomprehensible. There was a generalized idea of the existance of an analogy
between Domes' and archs' behaviour, wruch were , from Etruscans, far better
known, but third rumension 's influence on shell resistance couldn 't even be
concieved. With the partial breaking of a Dome, with the fall of part of the
nervature (and even of the keystone), it was difticult to understand what has
happening without reffering to supernatural adion .

We present in this conference a summary of some of the interventions in Domes

on historical buildings we have performed across 20 years of activity in the
Rehabilitation and Pathology department of INTEMAC, during the study of
which non-traditional fador s wruch have proved to be determinant to the
monument 's condition have appeared.

2. A SINGULAR DOME W1TH SINGULAR PROBLEMS

It 's not exagerating to say that the Clergy's Dome at Salamanca is a quite
singular construction. And not so because of its dimension s, wruch are relatively
modest, with 13,40 m of internal diameter and a base thickness of 1,40 rn, as
much as for the abnormal proportions of its lantern, whose internal diamet er is
4,90 m , and his height 10 m , calling our attention when we approach the town
on road , surpassing other constructions.

When spanish Culture Ministry charged to our Institute the study of damages,
the situation was alarming. Few months before, a moulding had fallen off,
st riking one of the church's benches and not causing personal damage by sheer
luck , and all symptoms seerned to point to a possible collapse in middle terrn ,
with the irreparable lose tbis event would cause.

Stone's degradation on the outside was too noticeable for a building counting
only some 200 years old. Vast zones in wruch ashlar surface was compIetely
disagregated could be seen , as well as furrows caused by water's erosion.
Particularly, the superior coroice in the lantern had fallen off in certain zones ,
taking the balustrade with it.

The Dome soffit was affected by important cracking in the meridian sense.
The cracks , upon reacrung thole's height, tended to go through the axis of the
windows, moving then to the coIurnns with operungs of 10-15 mm .

The Jantern shows an indination that becomes perfectly noticeable from the
outside, tran slated in the inside as an excentricity of the vertical of the cirdes
of the thole's base and the lantern 's Dome, which can be appreciated in a
250 STR UCT URAL AN ALY$ IS OF I II STO RICAL CONSTRUCT IONS

F igun ~- La ntern 's inclination

F igun 5 - Iron fl anges binding t he Dome

 1. rvl. IZQUIE RDO BERNALDO DE QUIRÓS I Andell( domes 011 Spain 251

years were needed to build traosverse archs, pendentives, basis, thole and maio
Dome ( we know that principal nane was installed on 1641). Aod much more
strange are the worries for some leaks on the main Dome six years before, when
according this hypothesis Dome was still in open air on rus crOWD.

50 we think t hat cracking and specialy leaking (this ODe due to joints operung
as we will see later) both startt":d ;L<; I<.oon as the Dome was built, and they
progressed in sucb a way that in spite of covering the Dome wit h lead plates C'of
four pounds of lead every tierce on square", what , assuming that those measures
were related to Castilian pound s and sticks, brings a medium thickness of about
two millimeters), 00 1661, six years later, t he Dome's aspect was disturbing .
And Marto Jimenez exclusive worrying about leaking proves, in our opinion ,
that on that year of 1661 cracking were nothing to be wo rried about, ar the
appropiates measures would had been taken , specially taking into account that
four yers before, the Augustines Recoletas church's Dome had collapsed .

Figure 6 - Thole's cracking

Twelve years arter binding, a fur t her inspection was made, due to doubts
about Dome 's security, and 00 August 1685 Juan de 5etin was consulted. He
then performed t he first favourable report we have noti ced, warranting the
unexistance of any collap sing risk. Anyway, some other workmasters (among
252 ST RUCTURAL ANALYS IS OF HISTORI CAL CONST RUCT IONS

Pigure 7- Pendentives' cracking

those was Blas de Tamayo), disagreed with that opinion, and lead plates were
removed, considering excessive their weight, (with an astonishing criterion ,
becallse the load would be less than 24 kp j m2, if lhe interpretation of lhe
di sponible data is correct) and were sllbstitued by a wooden and shale jacket,
attaining apparent1y the unanimity of ali \Vorkmasters about the work 's safety
on 1686.

Ncarly forty years after, on 1721 , new maintenance works are necesary, and
Joaqun de Churriguera again opens the controversy, when he gives his opinion
thal in no way lead should be disposed over, patching the most damagcd areas
with sand and lime and laying a bitumiIlous seallayer over the Dome.

Our unfortunale Dome was affected by Lisbon 's earthquake on 1755, which also
damaged l he just finished towers, and on 1812, during the Spanish Independence
War, as a result of the blasl of San BIas' powder depot, suspicions about
unstability of the dome grew , and church was closed OIl 1817 after a report
by the Architect Fran cisco de Paula Vega.

At last, a great repair is made on 1845, on which we assume that existing

cracks were clogged, and a complete renewal of internaI Dome's decoration ,
higly spoiled by cracki ng, was don e.
 J. M. IZQ UJERDO BERNALDO DE Q UIRÓS I Ancil.'nt domes on Spain 153

Every unfavourable report pleaded as damages' evident guilty to lantern 's

excessive weight. Anyway, irom the beginning an apparently underrated aspect
worried US. If Dome's stability was as precarious as to alarm ao Architect
with the proffesional qualification of Churriguera, how was it able to stand the
íormidable shocks caused by Lisbon 's earthquake or Sao Blas' powder depot's
bursting , which would be perceived on Salamanca as a sysmic adion grade VII
or VIII? That ~eeUlt:tl iucongruous a.nd, looking for an answer to that qucstion
we began the study.

The Dome is built on a siliceous sandstone with high porosity and water
absorption. The lack on resistaoce observed on the stone betweeo dry and
saturated states, loosing five sixths parts of its resistance, is also a remarkable
chara.cteristic.

A first checking of the Dome's redesign with a simplified hypothesis was done,
in order to roughly evaluate the stone's stress leveI. Results, in spite of the
simplifications made 00 those evaluations, were coocludiog: Expeded stresses
on the Dome, even on sysmic hypothesis, are low and cao 't explaio the existiug
damage. Situation is net: The results of a.n evaluation made by usual methods
and with usual desigo hypothesis show ao expected behaviour and the strudure,
along the history, persists on haviog a differeot ooe. Obviusly another silent
factor had to be there.

It is known that limestooes cao present a certain swell as its hurnidity rises.
rcs a commonly underestimated characteristic of that kind of rocks because its
value is ussualy of low importaoce and does oot become determinaot for its
structural behaviour.

In our case , however, the observance of a coupIe of symptoms on monuments

built in Salamaoca on XVI and XVII centuries forced us to suspect that 00 this
particular case the expansivity of stone could have a greater importance thao
that usually assigned to, and we opted to study its dimmensional stability in
different moisture conditions.

The results were surprising: after 24 hours of water imrnersion, specimens

showed deformations that twisted and warped the original prisms, and veins
in a brownish calor sweated out , on wrnch occasionly longitudinal cracks could
be seen. Volume expansivity was eigth or eveo ten times higher than rnaximum
expected value following prior bibliography, and internaI stresses deformed 00
a nonreversible way the pieces. Stone had a dimmensional unstability really
remarkable.
'54 STR UCTURAL ANA LYS IS OF HISTORI CA L CONSTRUCTIONS

Pigure 8- Deformed shape under seism

Figure 9- Dome's stresses under seism

 J. 1\1. lZQU IEROü BER NA LDO DE QU1RÓS I ArlÇi..:n\ domes o n Spain 255

A conservative estimation on differential strains between two cansecutive

courses showed that local tensile stressess, high enough to be able to be the
actual cause of damages were generated.

The master quarriers were perfect connoiseurs, of course, of the stone's lack af
resistace 00 high moisture conditions, and 50 they nidged stone in saturated
date, and they let it cure once carved prior the placing in its pasitian af each
blockj but through storing plenty of times in open air, they could'nt avoid a
mise en place with a high moisture content.

Once the stone was placed, it progresively dries, until a dynamic equilibrium
between rain and drought periods is reached, but the equilibrium moisture wiII
aIIways be lower than primitive one, and a remaining shrinkage will stay.

We established lhat internai stresses generated by this phenomenon are high

enough to justi fy the Dome's cracking. Let 's add that dimen sional unstability
is able lo justify also the historycal behaviour of the whole structure.

First notice about craclUng comes few years after finishing the caver, when
somebody observes the bleeding of bitumen employed on water-proofing. Taking
away that bitumen and placing a more perfect impermeabilization, cracking
grows, until puting in question the whole Dome's stability.

But this behaviour is perfectly predictable starting frem the hypothesis of

dimmensional changes: as long as stone drying advances, on a process that
takes severa! years to be accomplished, first shrinkage cracks are generated, and
those cracks get \Vorse when a beUer impermeabilization is disposed, causing
an acccleration of drying and subsequently of shrinkage.

Once was the Dome flanged, cracking seemed to decelerate, but not to stop.
Explanation of this phenomenum according to our hypothesis is clear, because,
on the fir st hand stone's thermical inercy is quite different than steel's, so just
whcn the flang e's action is more necessary, ou drought season and with high
temperatures, steel tension delies because of its diferentia! thermical expansion
in regard to stone. However, on wet season stone tends to swell and steel to
shrink by thermical adion effects, with a differential strain between dome and
Range of at least O.15deformations on stone, and puts flang es out of service for
the next drought season.

Accidelltal actions (Lisbon's quake, San Blas's bursting) had not significance
enough, according to our evaluation, to produce by themselves any significallt
damage on the Dome, but when acting over elements 011 shown conditions, they
256 STRUCTURAL ANALYSIS OF HI STORICAL CONSTRUCTIONS

must produce ao increase of cracking, as well as a rearraogemeot af lantern's

basement ashlar blocks, favoured by dynamical action aloog seism.

Disposal af a new lead cover and new joiot repainting place again the Dame
on suitable condition for a new cracking process, due to moisture incorporated
duriog crack repair and on ornamentation restoration , jointly with improvement
00 externaI impt:cwt:auilization.

00 the other hand, the thole has never beeo treated against moisture, and
its behavior in froot to swelling has no difference at all with Dome's, aod so
cracking on thole can freely develope (with the cardinal structural advantage
that , as it works esseotialy in compression, needs oot to develope circunferential
stresses for a correct structural behaviour - as Dome needs- and so the effect
on this one has a greater transcendency) aod their strains can be transferred to
the Dome.

Repair we proposed to spanish Culture Ministry began by trying to give back

to the Dome its original capacity of circunferential behaviour, byepoxi grouting
(the dryest method) of cracks and unrestrained vertical joiots, on Dome, thole
and pendeotives.

lmmediatly after , flutings on the Dome's extrados shall be done by a dry

method, and high strength cold-formed (with the object thatonce removed the
yield elongation, plastic deformations 00 steel would not be of high magnitude)
reioforced bars, which were handly tightcn using steel nipples, and cOlltrolling
their elongatioo by mechanical extensometers, ulltiI a final mediurn stress of
about 100 MPa was reached.

AI last, flutings were fiUed up with epoxi mortar, and a waterproofing layer,
00 epaxi mortar with fiberglass was disposed, foreseen a ventilation chamber
betwcen stone and resin that allow gas interchange on stooe. Restoration was
made under direction of ProL Arch. D. Antonio Fernández Alba.

So, the employed method was completely traditional, and similar to tbose
previously applied. Maio differences were only 00 the optimal date for working
(at the eod of drough season) and 00 a maximum care for avaiding water supply
to the Dome, by disposal of a water- proofing cover over scaffolding aod making
every step by the dry way. As to the rest, flan ge 's capacity was not higher
to previous one's, initial stress was even lower to that reached by heating ( the
ancient methad for prestressing), and we bad only the precaution af embeding
flanges into the ashlar deep enough for avoiding an its main part t he effect af
short-cycle thermical oscilations (day-nigbt).
 J. M. lZQU1ERDO BERNALDO DE QU1RÓS I Ancient domes on Sp:ün 257

Figure 10- Dome's Jepair. New binding

Up to now , behaviour has been correct, and at least doubts about Dome's
stability (which made once believe than there was no solution but Dome 's
complete disassembly and subsequent recon struction, on an estimated budget
of about $ 12 million) are now dissipated.

3. PROBLEMS OF GROWTH

Another case on which building's hitorycal study was determinant for finding
damage's ultimate cause was Our Savicur's Cathedral, on Saragosse. Church's
damages were important enough for fcrcing substituticn cf four columns
on church's zone corresponding to 16th. century ~nlargement. Certainly,
immediate causes were on walling's poor quality, on column 's slenderness and
on mortar's weathering degradation. But a serious ammount of existing cracks
were not, on our opinion, so easily explained.

Building's history study, caracterization of matherials and damage 's pacient

inventory lead to us to establish solid hypothesis about structural behaviour of
the whole church able for justifying symptoms.

Existing chronicles of an important disaster on the beginings of 17th. century

allowed to us to justify the most alarming damages (placed exactly on immediate
25M STR UCTURAL AN ALYSI S OF I-I ISTORI CA L CONSTRUCTI ONS

Figure J } - The C lergy 's Dome at plesent time

area s to which collapse was situated by chronicles), as cracking typology exactly

agreed with that expectable by dynami cs effects during fali. Area's minutely
examination resulted on finding ancient restoration wo rks dated later than
collapse, acting as crack's seams, showing that damage's growth had becn null
or very smali along the two last centuries, allowing t o relativizate the adual
problem.

On the other hand , study of constructiol1 stages anel of successives enlargements

fr om primit ive 12th century's Romanic basilica aliow us to establish a consistent
hypothesis about damages on church 's wheel window's wali area, at the temple's
bottom.

Due to existing soil characteristics and also considering how massives wal1s and
buttrcss are, it was not only possible but very likely that lo ng term differential
settlements were pro duced between primi t ive basili ca 's walls , 011 which soil
woulel be higly cOl1sollidated along four centuries when enlargernent begall, anel
Tear walls and counterforts .

Damage's symptomatology study on walis and nerves, alIow to establish on an

approximate way t he direction and magnitude of settlements, and consequently
to bui lt a mathemati cal m o del for vault's stressing.
 J. M. IZQ U IERDO BERNALDO DE QUlRÓS I Al1cicll( domes 011 Spain 259

Figure 12- Rebuilded column on Our SII.ViO UI Cathedral (Saragosse)

Results showed that shells themselves were su bmited to moderates stresses, but
on settlement hypothesis tensile stresses on determinated areas , with a very
neat coincidence with act ually cracked areas, vould appear.

Shells stay stable after cracking, but not columns, on which high load increases
were generated , able to justify their buckling.

That's why we reported that the best solution for the actual problem on vaults
was not to act on them, but on walls foundation , and to carry on works aimed
on reestablish their continuity,when degradation were unbearable, and mainly
structural cosmetic activities. Restoration has been made under the direction
of ProL Arch. Angel Peropadre.

4. THE IMPORTANCE OF SMALL DETAILS

The fir st problem we detied during previous technical studies for t he

restorati on of Our Lady of Disperates basilica, on Valence , (proyect by ProL
Arch. Ignacio Bosch ) was to determinate on a suitable way the actual geometry
of the Dome.
260 $T RUCTUR AL ANA LY$ I$ OF HI STO RICAL CON$TR UCTI ONS

P igure 13 - Enlargement works on the Cathed ral . T he 3rc·buttres lies over an older facade.

Figu re 14 - Plant of t he primitive Romanic basilica

 J. M . lZQUIERDO BERNALDO DE QUIRÓS I A ndent domes Oll Sp:l.in 26 [

Figure 15- Plant of the Cathedral at present

Dome's layout is particulary hard as neither suffit nor extrados are accesible,
the first oue due to be covered by the internai decorated shell and the other by
the glazed tile·roof disposed.

Resistance Dome's layer was 50 defined on an inrurect way. Shape of

inner surface, which receive Palornino's Ífescoes was determinated by using
photographical photogrammetry, and subsequentially internai sheU thickness
and gap between both shells were measured by cores opened 00 the resistance
shell.

So, a sector of Dome 's cover corresponding to a nerve on ENE orientation

was disrnounted , picking up tiles and ceramical board until footing pieces and
Dome 's extrados were seen.

Then we made several cores 25 mm f through the sheet, {rom lantern's basernent
on six meters foUowing a meridian une, point where we found that rustance
between both sheets, resistance and decorated one, was 50 low 50 thick the
nervature 's oversize was (about 10/ 15 em), and we stoped due to damage's risk
on decorated sheet. Further verifications were made on upper and lower zones
of halí major axis) to confirm that gap between both sheets stay approximately
constallt on a horizontal sectioll, on a positive resulto
262 STRUCTURAL ANALYSIS OF HI STORICAL CONSTRUCTIONS

Figure 16- Cracking areas (modelling)

Followjng on , we bad to find a mat hematical medel of Dome 's geometry. At first
we tried to fit an ellipsoid , where halfaxis were those of the elliptical basement
and keyst one high from basis.

Comparison betwecn theoretical sections 50 obtained and actual measures

showed sub stantial differences , mainly on the crown cap , at about (our meteIS
from Do mes keystone, on the arca where the Jantern basement is, and 5 0 the
approachal was considered indefen sible.

After som e un successCuly t entatives, we study t he possible app roach by a surface

cf the family of bicuadra ti c ellipsoids:

and specifically we tested a parametrical fitting ou t he Corm :

 1 M. rZQUrERDO BERNALDO DE QUIRÓS I Ancit;:JlI dOlllçs 0 0 Spain 263

Figu.re 17- Actual cracks on voults

Fitting was surprisingly excellent, with differences lower than 5 em along

plotting of theorical and adual surfaces, on resistanee shed as on decorated
one, on major halfaxis as on minor one, and so that hypothesis was accepted
as the best fitting model for the Dome's geometry.

Beginning with this hypothesis , a mathematical model of both sheets, resistance

and decorated one (painted with Palomino's frescoes, whose rescue was the
restoration final objetive), hanged this one írom the other. Model had 10,614
degrees of freedom and was performed to evaluate effects of statical and
dynarnica11oads, heating and thermal gradients.

Wind and sysmic analysis was performed on a recurrent period of 500 years.
Thermal variations trough a complete year were measured by therrnocouples
placed over tiles, under them and over Dome's extrados, in the gap between
sheets, and on internaI sheet intrados.

As final results of performed analysis we found the following interesting

conclussions:

• On decorated sheet, final stresses after creep are absolutely non-admissible on

first six meters írom thole, due to unrestrained shell buckling , as on this zone
264 STR UCTURAL ANALYSIS DF HISTORICAL CDNSTRUCTIDNS

Figure 18 -QUI Lady of Disperates (VaJence)

there are not hanging rods.

• Furthermore, Dome's creep produces ellipsiod's flaUering, due to lantern weíght.

Rods which hang decorated sheet from resistance one, lack tbeir initial tensile
stress due to matherial ueep, and OH certain cases even produces punching on
internaI sheet, due to deflection imposed by the lantern weight.

• Differential sun exposal produces very high thermal variations Qver tiles (we
even mesured nearly 50 oC on the sunny face in the heart of Jannuary!), but
over the Dome differences are much lower, due to system tiles. plaster filieI-
boarcl-ventilation ducts-footing pieces higb insulating efficiency. Temperature
on Dome's exhados shows practically no short ~ term (day-night) variation and
time- temperature curve on t his point adjust well to tile's three- days moveable
mean.

However, stresses on hanging rods were extremely sensitive to diferential

displacements on both sheets due to thermal behaviour. Analogously, small
differences on temperatures inside chamber (i. e. very small differences on
hanging rod's temperatures) become on great variations on rod stresses.

• Principal modes of vibration are slightly different between both shells, on

J. M. !ZQUlERDO BERNALDO DE QU1RÓS / Ancient domes 00 Spaio 265

Figure 19- Plant and section

Figure 20- Ventilation system undeI tile Iooí

266 $TRUCT URAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

AJUSTE DE LA GEOMETRIA DE LA CUPULA

SECCION SEMIEJE MAYOR

Figure 21 - Actua.l shape vs. ellipsoida.l , parabolic & bicuadratic ellipsoide shapes

Figure 22- Dome's flattering due to lantern weight

 J. M. IZQU IERDO BERNALDO DE QUIRÓS I Anc ient domes on Spain 267

Figure 23- FIe.xuraJ moments induced by rods

a theorical unconstrained hypothesis (i. e. removing hangs). Resonance

phenomena are then possibles.

Model answers, then, t o every cracking and damage configuration actually

existing ou the Dome. On the present time, actual scale model has been built
on Valem:e's Polythecnic University, testing different repairing protocols.

5. BEHIND EVERY GREAT COMPUTER ...

Comp uter is a powerful ally and has become indisp ensable in structural design
in very few years, but mathematical models are not , in general , conceived for
modelling an ancient building, and our knowled ge a bout long.term behaviour
of traditional matherials is ussually very poor.

Neit her limes, nor plasters na r bricks used at the present have nothing in
common with those employed between 11 th and 17th. centuries. Existing
bibliography is h ardly applicable to a particular CMe. When we read, for
in stance, that certain masonry has a m edium strengt h , in compression , of about
10 MPa with stand ard deviation af about 3.5 MPa, what are we talkig about?

• WJlich was resjstance and llOmogeinity of ashlar pieces?

268 STRUCTURAL ANALYSIS or HISTORICAL CONSTRUCTIONS

Figu~ 2./- 2nd. mode oí vibIation (principal for d«orated shell)

• In wbich form does strength vary with masonry bond?

• Which were stone's stereotomy and mortar's granulometry?

• Did mortar carry pozzolans incorporated by tIle aggregates? Was it plaster OI

lime? Do we know how much ox's blood or horse's uón e was incorporeated for
makúlg it impermeable?

• How has tIle importance af lacking joint 's matlleriaJ been measured?

Al1d most important: What is the inRuence Df everyone of uppeI circumstances,

and their possible combinations, on strength and e1asticity modulusin short and
long term?

On the present state af the art, theoretical sophistication of model can result
sterile. H sct:tmj <i.L least uouLtful ~he vradical rcability of usillg matherial's
elastic.plastic diagrams based on testing made over specimens crafted with
actual matherials. The internai behaviour of very thick masonry walls is
hardly modelled by finite-e1ement sistem . Long-tcrm deformations are often
determinatcd by joines matherial lack process along centuries, wich is a well-
known phenomellon, but very dificult to evaluate.
 J. M. rZQUlEROO BERNALDO DE QUIRÓS I Andent domes 0 11 Sp.:lin 269

Figure 25 - 4th. mode af vibration (principal for resistance shell)

50, mathematical modelling is only an additional tool on our toolbox, and not
as something magical and definitive able for giving by itself com plete sollution
to our doubts. !t's too easy to feel impressed for grafic colorfull presentation
which, properIy anaized, say nothing new. We will always find a certain mode
of vibration whose shape will be coincident with actual deformation, but do the
sysmic spectrum have enough energy at that frequency? Is it principal one, or
4th, 5th or even 9th mode? Are we sure that there is not any other clearer
explanation of actual phenomena, sueh as matherial anisotropy, loss of the joint
material or local differential settlements, producing the same local deformation?

The answer is always in experíence. A practised tecrueian can obtain much

more precise results ÍTom an evaluation made on a cellulose tissue that an ali·
round theorist one ussing bours of a Ceay IV and producing mountains of paper.
Behind eveey great computer tbere is always a small bone's and fiesh' and brain's
poor human being.
 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS
P. Roca, J.L. González, A.R. Marí and E. Onale (Eds.)
Q C1MNE, Barcelona 1996

PROPOSAL FOR THE RESTORATION OF MARCUS

AURELIUS'S COLUMN

A. Guiffre
Universita Terza di Roma
Italy

S U MMÁRY

Two monumental columns have been built in Rome sincc the second century
AD: T rajan's Column and t he Marcus Aureliu s's Column. Their construction
technique is examined fi rst, a nd thell the state of damage. The two columns,
standing less then 700 m ap art, have wit hstood a li t he earth quakes that have
sh a ken Rome since they were built, but the type of damagc t hat t hey have
suffered is different . Marcus Aurelius's Column has suffered a great deal more
cracki ng thall the othe r. A debate is going on at present am ong the cxperts
regarding the reasons for this different behaviour, and in this pa per a position
is taken on the subject. The results of studies a nd experiments rcgarding the
seismic respo nse of the columns are also discussed . The restoration by Fontana
in 1590 is illust rated a nel the present state Df safety of the columns is discussed.
From all the analyses a nd t he hy pot heses previously presenteei, a p roposal for
a new restoration aimed at seism ic safety is in ferred. The condusion presents
the author' s op inion on t he proper way to study such a monument wit h a view
to restoring it.

1.- THE C ON STRUC TI O N TE C HNIQ UE

There is no doubt that because of tht: naLurc of iL ~ cons Lruction a nd its

com memorative conception Marcus Au rclius's Column must be a t trib uted to
Apollodorus of Damascus, li ke the prototype ded icated to Trajan in 113 AD.
In both colu mn s, sevc nteen marble stones, each weighing 40 metric tons, were
transportcd from the Lunigia n a (from the caves of the modern Carrara) to the
centre of Rom e. They were first t aken by boat along t he E t ru scan coast and
up the Tiber I anel thcn dragged from the Tes taccio to t he piazza Colon oa.

This operation also involved the blocks of the pedestal, even largcr and heavier ,
anel the square abacus, and the rounded top on which the bronze statue of
 A. GUIFFRE I Mareus Aurclius's Column 271

M areus was plaeed.

It was earved in a eylindrieal form on ,·. . hieh the battles and the "miracles"
of the emperor would be seulpted, and was hollowed out until the weight was
redueed to about 27t to house the spiraI stairease that reeaUs an Arehimedes
serew The massive stones \Vere lifted with the polypastos and the wooden
seaffolding, the remains of which allowed Adrasto to build the wa rder's house ,
and they \Vere plaeed in position - the last oue at a height of 36 metres -
to eonstitute the monument that Theodosius and Areadius reproduced in a
duplieate in Byzantium.

lt is an extraordinary pile of stones. The maehines used to plaee them were far
more eomplieated than the structu re that was built with them. Oistinguished
monuments , the first representa tive of the empire in expansion and the second
one of the empire in defenee. 8ut they are mereIy piles of stones simply placed
OIle on top of the other as in prehistorie dolmens. The engincers and arehitects
of today, we could indeed say the descendants (the <lrtist.ie heirs) of Apollodurus,
diseover a disarming elementariness in t.he struct.ure of these monuments. The
Roman s were able to do (ar more: the great thermae bear witness to this.
F'urLhermore, the Syrian designe r of the colunlns had built the bridge over the
Dallube, a miracle of engineering, and an example of building technique that
rC-arose in the treatise of Rondelet , and he rcpresented it on the first column.
Apollodorus understood theatre, and Trajan 's basilica built by him a few metres
from the eolumn was a great edifiee articulated with a complex equilibriull1 of
waU s, arehes and trljsses.

And yct. the columll with the spiral rUlllling through its interior up io the t.errace
011 the abacus, and the cella in the hoUow of the stones of the pedest.al , is Like
the shells of certain molJusks that are rounded on the outside anel twisted on
the inside, or the llving organisllls that conceal their complex entrails, or more
simply like a building whos e articulated function s require an organic succession
of envirouments. Unlike the monolit.hic obelisks, the column by Apollodorus is
in fad a building, OI' raLheI' a tower. An ambiguous pile of stones.

Que wonders why Apollodorus required sueh cumbersome elcments in order to

build hi s tower How ll1u eh easicr it would ha\'e been to make the walls that
form the hollo\\' cylinder with a c1assical isodomic work, with small, weU dressed
sto nes that were easy to transport, to lift and to place. The techniqlle of the
isodomon even appears seve ral times OH the marble rellefs of Trajan 's Colllmn:
t,he person carrying the block on his back, carrying it up the steps, placing it
carefully on top of the waU ... Apollodurus wa s a master of the art of building
with "small pieces" : machinesof \\'ar and siege towers were his speciality. On the
272 STRUCTURAL ANALYSI$ OF HI $TO RICAL CONSTRUCTIONS

death of his emperor, the architect from Damascus composed a treatise on siege
warfare for the successor, the difficult Hadrian, and showed his full capabilities
and his genius: "since in battle it is not possible to transport heavy, cumbersome
elements, one should be able to make the large structures necessary for laying
siege using small pieces"

The commemorative tower , the support for the illustrated volume that narrates
the campaign of Trajan, was not treated with the refined intelligence of the
slender siege towers. Seventeen rocks (p lu s the pedestal and the abacus) were
laid one on top of the other with a primitive logic to raise the symbol of victory
by one hundred feet. The staircase that leads from the cella to the terrace
could have been built, in the interior of a normal tower structure, like so many
others in the tem pies and thermae, with sto nes cut in a trapeziu m and carefully
mounted on the outer waU; instead , it was hollowed out in the stone just as
primitive man dug out hi s cave. But not with the precision of the surveyor ,
with the exactitude of the astronomer, with the grace of the engraver, without
uncertainties in t he gradien t , without errors in the line. The step with which a
drum finishes is followed by the step with wh ich the following drum begins, and
the staircase continues uninterrupted , the junction unnoticed. Even the narrow
slits that illuminate the space have a strict rhythm, with the irreprehens ible
nature of the musical instrumento An ineffable pile of rocks, rough alld refined.

And the other tower-column s, wheth er in Rome or in Byzantium, followed

the same design strictly. The only differen ce is t hat the pedestal of Marcus
Aurelius's Column is three metres higher. Seen from the Via Flaminia (tod ay
the Via dei Corso) , once six metres lower than its actuallevel, the column on its
pedestal seemed to have a va ni shing perspective , totaUy different from that of
the column placed by Apollodurus between t he two libraries, inside a cou rtyard.
Bu t it is not the image that we are di scussing here, but t he material, which is
the direct work of conservation t hat we wish to undertake. The impressively
rising image of Marcu s Aurelius's Colu mn no longer exists toclay: surrounded
by houses , in the cen tre of a square, we see it as domestic, morc similar to
the monument of Trajan than to t hat of the victo r of t he Germani and the
SOo rmo.tian s.

2 .· THE STATE OF DAMAGE

Earthquakes, which are such a dramat ic issue today, \Vere not unusual in
antiquity: the reports of Pliny and the treatise by Seneca give plcnty of ev idcnce
of this, and the city of Rome suffered many times. In the republican period t he
earthquakes of Rome were registered in the annals of t he Pontifices: ct Hastac
 A. GUlFFRE I Marcus Aurclius's Column 273

Martis in regia motae"; the cuH of the lances of Mars thus constituted a rough
seislTlological observatory . Nor do we lack documents of later periods that show
damage and restoration of the monuments of Rome: in 443 AD "ceciderunt
statuae et portica noval!, and a series of lesse r damage to the st ru ctures af the
Caliseum . Again in 508 (484?) the consu l Decius Marius Venantius repaired
the a rena and the podium of the amphitheatre , but in 80 1 (dimarando Carla
Magno in Spoleto) many monuments collapsed ) and the art of restoration that
we wish to I'ecover today was fo rgotten.

As is customary in seismolagy, we look at the past to predict t he future. In

substance we have been able to clarify , taking the motions af a lest by Ignazio
Galli , that t he fo un dation s()i ls of Rome have tran smitted (and wiU transm it )
earthquakes from two different and in depe ndent seismogenetic arcas: thai a f the
Colli Albani, the site of t he ancient volcanic activity in the region, for which the
maximum intensity recorded in Rome is VII degrecs 011 lhe Mercalli scale, and
that of lhe Appenine UlI1b ro-Abbruzzese, fro m L'Aquila to Norcia, that !l ave
oftc n produced inlensities of VIII degrees in R,o me .

The mast famous earthquake that has caused damage to Roman monuments
was on September 10 1349, just after the plague and only a fc\\' months before
the ope ning of the Jubilee.

It is difficult to determine the intens ity af that fatal event: it is recounted

by Matteo ViUani, and me nlion is found af it in a letter of t he Pel rar ch . The
eart hqu a kc caused serious damage to the Bas ilica d i San Paola i di Sa n Giovann i,
anel elem olished "a part of t he noble tawe r of the army, and the tower of th e
Count , leaving in many other parts af Rome the memory a f its ruin s" The
state af const ruction in Rome 'vas anyth ing but normal: " precarious" situalions
were more widespread than the stable ones, and though the narration af Matleo
Villani couId suggest a scenario of grade IX or grade VIII , we may consider
that the dil a pidated state of the bu ildillgs seriously aggravated the efTccts. The
two columns are nat mentioned by the sources: was th is an omission or were
they left unharmed?

Structures of superimposed blocks have a very comp lex dynami c behaviour:

their sei smic stability is quite different from that of monolithic obelisks . Beyand
a certain threshold of acceleration , the latter begin to oscillale 011 th eir base
due ta the groulld motions; only at far higher (but not ver)' predictable )
leveIs of accelerat ion thall t hi s t hreshold do they lose their equil ibri um and
collapse. Block system s are alsa set into motion beyand the same threshold o f
acceleration , and they can a lso withstand far higher acceleration s than those
that trigger the motioIl, but the oscillations produced by the earthquake fila)'
274 STRUCTURAL ANALYS IS OF HI STORICAL CONSTRUCTIONS

cause t.hc shifting OI' rotation of the individual e1eme nt.s ,sy lnptolll s Ihal cO l ne
before the collapse. F'urthermore, the fr equ ency of these oscillations does 110t
depend on 1.hc eharacteris1.ics of 1.he materi ais but on t.h~· amp lit.ude of t.he
1l10t.ion: the more the mOllolith or col umn leans away from lhe vertica l, t he
lower is 1.he frequency with which it oseillates . T hc " prope r perioel of vibration"
of this st.ructure ea llllot be dcfined as for elastic structu re s, evell if a phase of
e1astic d ynalll ics can be detected duc to vcry sll1all el y namic excit at.ions that do
1101. involve the openin g of t he join ts.

Seve ra! cxperimcnts have been attempted on 1l10dels of t,he column in order to
clarify this behaviour, anel numerical analyses and other studies are underway,
thoug h the results are as yet incollclusive.

The resist an ce of such st ru ctu res to earthquakes is particula.r1y high, but difficul t
to evaluate. We can predict the value of acceleration of t,he ground that puts
into motion t,he sys tem of drull1s , anel we know tbat in orner to J"('ach collapsc
it is neccssary for t he syst.em to show peaks of acceleratioll that are fa r high er
than this value, bu t we do no1. know how much higher. Usually, t,he crossing of
this threshold Icaves irreversible sig ns on the strllctu re : shifiing af t he drums
OI' chippillg of the edges where the rotatiol1 is concent rated: these are the
unequ ivocal sings of seis m ie stress . Thc fortuitousuess of the motion s trigge red
by carthquakes d oes Bot aUo\\' us to make quantitative forecasts Oll the dcgree
of these dislo cations: they grow with thc value of the peak acceleration but also
with the number of peak s at wh ich thc accelcratioll goes beyond the threshold
value, and we do not know how.

An inexplicablc , dist urbing piles of blo cks !

Toda)' both the ROlllan columns show sueI! signs: displacement and rotaiion can
be secn bet.wee n the blocks, but. to a different extent in the two spcc imens. The
Trajan 's Column on ly show s chipping, a c1ear sign of a tilting motion that has
concent r ated the weight of the overhanging portion 011 one edge of the drum ,
and in another p osition a displa ceme nt of a fe\\' millimctres. Here a nd there we
can find vertical damage affectin g the whole height, of a block and crossing the
t hi ck ness af Lhe eyl.indri eal waU.

Marcus A urdius's Column is badly damagcd: elisplaeements of more than I Ocm,

vertical damage going through aU t he elrums (except for one, the tent h), t wo
of them running from the base to med ium height. Domenico Font ana, \Vho
restored it il1 1590 , had to say that it " was ready to coUapse" .

So mething of the state in which Fontana found Mareus Aureliu s's Column
 A. GUIFFRE' Marcus Aurelius's Column 275

could be gathered from the different prints that were made of it before that
time, but far more appeared when a young student, Marco Pelletti, staying
for a few months as a stylite on the scaffolding of the column, first under the
auspices of the Archeological Ser vice and then to carry out his doctoral thesis
in architecture, removed all sigos, alI cracks, ali sutures .

In the secolltllialf uf the ei.ghti.e:., archi.tecb, e llgiueers, lIIechanics, archeolugists,

and surveyors worked with remarkable fervor Oll the Columl1 together with the
architect GianGiacomo Martines, who directed the restoration works for the
Archeologieal Service.

The first result of this incredible survey is the comparison between the old
prints, which showed the column split into two zones, and the map of the
inserts placed by Fontana in order to make it good; the secon d is the system
of vertical fractures that brcak t he monolithicity of blocks and transform the
work into a set of "smaU pieccs". Just what Apollodurus had avoided at the
cost of enormous effort.

We have already observed, running through the unforgettable scaffolding of

the eighties, that the most spectacular dislocations affect only portions of
blocks, fragments included between the aforemcntioned transverse damage that
\Vere pushed outward by the seismie 1I10tioll while the main part of the block
maintained its contad with the underlying piece. Only block No . 10 has
undergone a complete displacement, dragging with it thc part of the column t.hat
is above it, and the survey of the dislocations has made it possible to determ ine
the mechauism associated with this, which shows that the column could easily
have collapsed. The two vertical fractures t1lat start from the two major cracks
and [Ull to the base have isolated a pilastrino alld ou this t,he portion of the
column from the tenth block upwards supported itself momentarilYl pushing it
autwards.

But why this dreadful panor.uua of fractures?

A hypothesis comes to minei spont.aneously: before the earthquake upset the

general order of the column (in fad it is a very tragic premi se that the
earthquake could have damaged the structure to sueh an extent) lightning could
have caused the two la.rge cracks t.hat are reproduccd in the prints af Giovanni
Antonio Oosio, Marco Sadelcr and Enea Vico, and t ha.t were recognized by
fvlar co Pellett.i behind the inserts of the Fontana. It would thus also have been
responsible for the vertical fraciures that run from the cracks to the base of
the column, br~aking the blocks iuto parts. In fad this situatioll of fissures
is typical of waU structures hit by electrical discharge. Arcadius's Column in
276 STRUCTURAL ANALYSIS DF J-lISTDRICAL CONSTRUCTIONS

Constantinople, surveyed by Gyllius in the middle of t he 16th century, showed

similar vertical fr adures: the lightning had hit it on June 24, 550 and the
fracture was contained by means of aspiraI of iron hoops around the Iower
drums . The vertical fracture produced by the light ning on a brick chimney is
documented by a photograph reproduced by Donghi in his Manual. 11 Poleni
describes the cracks in the bell-tower of San Marco that occurred on April
23, 1745 The effects of lightning on ancient m onuments of Rome have been
recognized ou other occasions : a collapse in the basiIica of San Paolo in 443
was attributed to Iightning by R . Krautheimer , and the obelisk of Piazza del
PopoIo was rccently damaged due to similar atmosp herie p henomena.

Furthermore, in more rccent times reports of lightn ing striking Marcus

Aurelius's Column are not lacking: 12 August 1679, Avviso di Roma, "a
thunderbolt hit Marcus AureIius's Column in Piazza Colonna" and damaged
the sculpture; 22 September 1841, "at around ni ne thirty .. lightlling having hit
the Column of our piazza has damaged its vast reliefs"

But together with the two main vertical lesions that suggest the wrath of Jove,
the fissured panorama of Marcus Aurelius's Column shows a diffuse series of
other fradures that fragment the mOllolithicity of the blocks, similar to the
ones that are found in lesser number OH Trajan's Colmon. lt is known that
marble taken from the quarry and worked tellds to crack. The pedestal of
Trajan's Column shows such fissures, but they may a lso appear in time whell
the block has been worked to a small thickness, as for the circular waU of the
tower, and in this case the cracks become authentic fradures. Perhaps the
blocks used for the monument of Marcus Aurelius were cut from a more fragile
vein than those cut for Trajan's Column. Bui the earthquakes could also have
fract.ured the drums: a slight defed in the planarity of ihe çontad can cause a
fradure when the oscillation leads the weight of the column to be applied 011
the edge sedor.

However, thus damaged, with the monolithism of the blocks broken , Marcus
Aurelius's Column could have seemed to be in a reasonable state of repair to
ihe observer: fissures have always accom panied tbe panorama of waU stru ctures.
But this incoherellt assembly was in sueh poor state that the earthquake could
easily destroy its integrity.

Was it clairvoyance or seismic science that led Apollodurus to not spare

prisoners' shoulders to pile up giant, monolithic blocks of stone, rather than
cOllstructing the walls with t he ephemeral technique that he used in the military
campaigns? Trajan 's Column, evidently saved from the lightning, only suffered
the chipping mentioned above as a result of the earthquake. Most of the drums
 A. GUIFFRE I Marcus Aure1ius's CoJumn 277

are complete and only in one, in which we note two vertical fractures that isolate
a portion of the block, do we observe a small displacement, a partial dislocation.

~1ar cus Aurelius's Column would havc beco saved írom the earthquake, like
Trajan 's Columo , if it had not been previously fractured.

Recent st udies of the foundation soils of Rome leave no doubt about whether
the location of Marcus Aurelius 's Column couId be the si te of an amplified
seismi c action A large port ion of the urban su rface, alongside the Tiber lies
on the alluvial basin of t he river , and this is where Marcus Aureliu s's Column
is locatcd 1 wh ereas Trajan's Column stand s on firm er ground outside the edge
of the basin. It is therefore definitely possible that the two coIurnn s suffered
the earthquakes of Rome with different inten sities. Or rathe r , it is definitely
possible that Trajan's Column, placed in a firmer location than that on which
the greatcr part of the city was built, suffered lesse r seismic actions than those
that cau sed grade V III damage to the urban buildings.

Marcus Aurelius's Column is certainly more damaged , but what effect could a
grade VI II earthquake have 00 th is haughty pile of blocks?

Although the quoted st udies on the nature of the foundati ons soi ls aliow us to
foresee the prese nce of low frequencies in the seismi c motion of the a Uu vial basin
011 whi ch Marc us Aureliu s's Column is lo cated, recent experiments con firm that

these frequell cies are flot sufficient to break up an organic whole of complete
blocks. However , a stronger dynamic action than that suffered by Trajan 1s
Column would 1I0t have produ ced in a. colum n that was stiU wh ole the typical
vert. ical fraciures caused by üghtlling, but in stead chippin g 01' ge neral shifting
betwcen Lh e blocks, sig ns that always characLerize th e e ffects of earL hquakes on
such struci ures .

Certainly, wc do not have documents that st ate ho", t he cra cks in the marbl e
that wc see were caused, we have no docum entary evid ence of the a ppea ra nce
of the crac ks and t he vertical fra ctu res that. are ty pi cal of light ning, and we
have no doc ume nts that state how the earthquake di scolln ected the fractur ed
drums that were replaced by Domenico Fontana. Vve can onl)' put forward a
hypothes is that we co nsider to be reasonable.

Perso naUy, wc are inclin ed to state that the great.er damage sufl"ered by Marcus
Aureliu s's Column than by Trajan 's Column is not due to ali earthquake of
a greater inten sity (though we do agree that it cauld have bee n of a greater
intcnsity ) but to the fraciures that ran throu gh the blocks.
278 STRUCTURAL ANALYSIS OF HI STO RICAL CONSTRUCTIONS

T his statement is not without consequcll ces , as we will see, on the future of t he
column, or at least on our aHitude towarel s the future of the column.

3.- T H E R ESTO RAT I ON BY F O NTA N A

It was perhaps the earthquake of 1349 that maele the fissureel blocks collapse.
We are forecel to base the hy pothesis 011 well -known events: the ollly previous
earthquake recorded was that of the begiuning of t he 9th eentury , which was
ce rtainly very se rious , but how could the colurnn have supporteel the second
one if it had already beclI seriously damaged by the first?

Fontana undertook the restoration for Sixtus V , and fi nished it in 1589. His
intention was to restore the column to its or iginal splendour. He filled the cracks
with new pieces of marbJe an d on the new surface he had scenes of war sculpted
in imitation of t hose that ha d been int.errupted by the gaps. Accordi ng to hi m
(and we have no reason to doubt hi s wo rel ), many portions of marble stood out
a paim anel a half from the column , anel hc had to reinsert them using levers anel
winches . He was not always successful: Marco Pelletti locateel some pieces of the
original marble on which the bas- relief shows the same 15th- century technique
as the added portions. Unable to reacljust the protruding pieces the Fontana
decided to recarve that stone eliminating the protrueling part , so the thickness
of the cylindrieal walI at that point was reduced to only 30 em. The redudion
in thickness, tho ugh not 50 serious, is general: the internai surface of the hollow
eyli nder, whieh delimits the spi ral , was recut to remove the ste ps resulting from
the dislocation of t he pieees pushed outward by the seismic moveme nt. Thus ,
the dislocations go unobserved for the visitor who goes up the spira l st aircase.
Even in the interior lhe 15th-century restoration restored t he column to its
original appearan ce.

Fon tana completed the work by inserting a great number of metal b races to
conned the aelded anel relocat ed pieees.

lIow are we to judge the L:UlTeut st atc of the st ru cture? lt is true that t he
introduction of the mi ssing pieees has provid ed points of support th a t are
indispensable to the strength , anel that the replacem ent of t.hose that had
been sh ifted has returned the strud ure to how it was before the ea.rt hquake
almost destroyed it completely: apparently whole hllt. seriollsly fra. ct ured. The
llllmerous metal cramps with whi ch Fontana united t.he edges of thc fract.lIres
havc to a eertain l'xtent restorcd the contin uity of caeh bloek.

Aftcr F'ontana's rc sto ration of the column , it su ffer ed the earthqllake of 1703 ,
 A. GUIFFRE I Marcus Aurelius's Colurnn 279

which was famous beca use it was 50 destructive in VAquila and in No rcia (it
was obviously the same seismogenetic mechani sIll that. had causcd t he olle in
1349) and because the damage it caused to the areade of the Coliscull1 is often
ment.ioned. In fact , however , its effects were BOt. so serious in Rome. An
eartbquake was also report.ed in 1812, whose effects ou t.he Co li seum Icd lhe
aulhorities to commission the restoration by Stern, but this alie was also fIot
seri ous. In substancc the rest.ored columll has not yet 8u ffe rcd a ::Icrio tl s tcst of
it.s integ rity, and we do not know what would become of it in ali eart bquake
like that Df 1349.

To conc\ude: if ollr hypothesis is accepted, that is, that the CO I UIllIl was sc riou sly
fissured before the mediaeval earthquake, and 1 his earthquake did no more than
break up the fragments, it can bc observed that in cOll1parison with then the
column has today a few cracks less and a few braces more; its cond ition is beBer
but the state of fi ssur ing its substantially the same.

If our hypothes is is not accepted, anel it is believed thaL it was the 1349
earthquake that fractured the column anel reduced it to the state in which
Fontana found it, it musl be agreed that today the column is in a dramatically
more vulnerable slate thall it was before that evcnt, of which WI> fear the returno

\<\Te cannot look at this uncertain, pre ca rious pile of blocks without a certai n
anxiety.

Fontana was a technician of great repute. and it is indeed not necessary to quote
the occasiolls on which hc had the opportunity to delllollstrate this. However ,
his work on the COIUIllII, in oue way or in another, follows a logic that we would
not repeat today.

Let us reconsider hi s work on the column: his valuable repairs, the re.assembly
of the disconnected pi eces, the clamps holding together lhe fissures.

Let us first state that today we are forceel to admit that we are unequal to
lhe tas k of large wall caost ructionsi there is 110 modern architect ar engioeer
who would have the courage to undertake a project \\'ith the same structu ral
solutions as t.he large wali structures of the past, made of stone or brick. Today
we do not depend OIl our knowledgc of building, as did Domenico Fontana,
hut on the respo lls e of the mechanical models with which we schematize the
str ucture. Anel when we lack the mecharucal l110dels and the correspo nding
algorithms as a basis for the construction of the waU , we are helpless.

The structure of the column, as we have already said, is elementary but

280 STRUCTURA L ANALYS1$ OF HISTOR1CAL CONSTRUCflONS

mechanically complex: the response of the models that we have beeo able to
formulate - and only in the last few years - is imprecise. The modern engineer,
who anyway does not know the most up-to-date models, would not think twice
about perforating lhe marble from top to boUom and inserting metal bars to
fonu a vertical reinforcement through the whole height of the column. In this
way it would be possible to borrow the model that we use for reinforced concrete
structures and to estimate - correctly ar incorrectly - a Ilumerical answer.

The choice of Apollodurus , singula.r, ineffablc, disquieting, wou ld lose its

meaning.

Indeed, an engineer whose name has not been passed down did not think twice
about it. In the restoration programme of the 1950's it was planned to insert
four vertical metal bars, of which fortunately no trace is to be found: a sensible
su perintendant was able to prevent the boring.

Today, and real1y Iam referring to the last three or four years , it is beginning to
be recognized that the mechanical basis of the structures of a period, even if not
yet formulated, exist and justify (and how could they not justify?) the century-
long duration of these works. There follows the equally obvious consequence
that in this new light modem restoration work aims to avoid disturbing the
iutrinsic mechan.ics of the architecture of the past.

But this is not enough: the buildillg customs of a period have a cultural
siguifi call cc that is totalJy equivalent. to their formal sensibility. It could
perhaps be added that without an understanding of thcse customs, the work
of conservation that now , at the end of the millenllium, has been recoguized as
being the indispensable legacy to pass on to the next millenuium cannot but be
blind and disfiguring. As in fad it was from the postwar period onwards, and
is still , due to the cultural inertia that inconceivably characterizcs this sedor.

The mechanical choice of Apollodurus, to build the column as a pile of blocks,

sllould no" be undervalned. Its efficacy against earthquakes has today been
demollstrated by modem algorithms. Though these have not yet reached the
point of making a fnll evaluation of the resistance, it has been observed with
stupour in the spectacular experiments 00 the vibrating table. BuL this had
already been demonstrated by Trajan's Column: it is unharmed due to intrinsic
mechanical ()11ality, rather than due to the a1.tcnuating: virj,ue of its subsoil.

Fontana did 110t appreciate, coulcl not appreciate , the significance of the great
blocks. Por him the column was a tower: a circu lar walI that was original1y three
palms thick with in its interior aspirai sta.ircase l.ike the a li e of the Vatican Dome
 A. GUIFFRE I Marcus Aurelius's Column 281

which was completed in the same year.

Anel the circular struct u re that Fontana saw in Marcus Aurelius's Columll ,
50 minute1y fractured , was truly a \VaU. He thu s fiUed the gaps as if he
were restoring a walJ , by inserting the stones without paying attention to th e
origiual positio ns of the drums, without worrying that the inserts straclelled the
superilllposed blocks and nega.tcd thcir individuali ty.

Today we woulel not do this. We woulel not do it because the mechanics of Lhe
structurcs of s uperimposed blocks has a possiblc , though not yet ex haustivc,
formulation whe n t he surfaces of contact are Hat . Bu t above ali we would not
do it for the sirnple anel incontrovertiblc reason that Apollodurus wauld Hot have
do ne it. Apoll od uru s piled th e blocks one Otl to p of t he othe r, with pe rfectly
plan ed col1tact surfaces, ensuring the hori zontality with the preci sio n of the
plumb-linc , each drum bcing a co ndit,ion for the layin g of the following oue. If
t he drums are fractur cd , we must repair them a s far as possiblc , 110t glue th em
togcthe r or eugagc th clII with each oth er .

lf we had Lo carry ou t today Lh e rcsto ratio n that Domenico Fontana co ncluded

in 1590 we wauld fill the gaps in each drum wit h ston es cu t so as to adh ere
t o the fra.cture d edges, restorin g the integ rity of the block whilst. respeding its
original geometry. We would be preparcd to g lue the new stolles to the edges
of j he old ones with ton g h resiu s, always respccting the individuality a f the
individual drum. We on ly need the chcll1ists t.o a ss ure us that 50 ye ars from
1I0W the glu es wiU not lose t heir effect iveuess, anel that. they wiU !lot beco tne

cviden., thus disruptillg the appearance like the iron placed by Balanoz in the
thirties bet. weell the marble blo cks of t.!te Acropoli s of Ath ens.

T he beUe r we are at g luin g the pieces together , the more we will retum t he
co!urnn to it s original state that C i111 be see n in Trajan 's COIUIllI1.

But to da)' wc mu st not repeat the restaratioll af Oo meni co Fon tana, wh ich is
also th e objcct of collscrvatioll. It is t,herdore necessary to follow othcr road s.

4.- PROPOSAL FOR AN ANTI-SEISMIC RESTORANTION

In an ideal t emporal vicw, Marcus Aure Liu s's Col urnn represents the present,
Trajall 's Calurnn the past, and Arcadiu s's Column an undesired but possible
futur e.

Trajan' s Colum n is in fad in the st ate in which Marcu s Aureliu s's C olull1l\
282 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

was before "something" reduced it to the st.üe tha! allowed the earthquakc to
devastate it. In this stale Arcadius's Column remained for a long time, but
success ive earthquakes led it to the state fram which it was mercifuUy saved by
the demoHshers in 1717.

Today Marcu s Aureliu s's Column is halfway bctween these two positions.

The restoration by Fontana, we observed, does n01 resolve the precariousness

iníroduced by the fi ssuring that split almos! ali the drum s af the column . A
serious earthquake would find the structure no longer constituted) as it originally
was, af large su perimposed block s, bui af small pieces that are only weakly
connected by metal braccs. And what is more , with the two unfortunate paraUel
fissures that isolate a vertical portioJl of the externai waU from the tenth block
to the ground.

Fontana's braces may be effective to avoid the smaU disconnections in the higher
blocks, which are the result of smaU eartl! motions but they can not restrain the
mechanism that has aUowed the tenth block to move outwards, and al most to
coUapse. The braces fastened from the exterior beco me di scon nected easily if a
tilting motiOll comes in to actiou.

Ao earthquake like the one that devastated the column wauld toclay finei it
a little more in order, and perhaps it would devastate it less, but a seriou s
earthquake is always a threat. Anel ir the new eart hqu ake did 110t cause the
collap se, it could lead the column to a similar state of collapse that. leel to the
demolition af Arcadius's Column .

Therefare, if we wi sh ta avaid the Cuture represented by the monument of

Arcadius we must at Ieast prevent the mechanism triggereel of[ in 1349 from
bei ng reactivated.

Arcadius's Colurnn was secured with a sp irai iron hoop írom outsidc, a
disfiguring but effective intervention. Wc can fallow this examplc warking on
the interior .

Let us imagin g a metal spi ral that runs along the internai waU of the tower
follawing the graclient af the staircase, in fact a clouble spiral cOllstitutcd of
two bars of titanium placed ove r and under t he staircase, against the internaI
surface af the \VaU. These bars cross aU the ffactures that split the blocks anel
in correspondence of each of the fra ct ures a fine pi n , conuectecl to the sp iral
through the thicknes~ of the wall anel with a hook a few celltimet res long, secures
lhe twa eclges af the fracture from the outsicle. All the fragments are thus linked
 A. GU IFFRE I Marcus Aurelius's Column 283

to a continuous stru ct ure whi ch conneets them to each other from the inside. It
is no langer possible for ane a f the pieces to be projected outwards leaving the
adjacent pieces in th eir place. That is to say, a signifi cant degree of solidarity
is achieved that improves the overall behaviour greatly. The externaI hooks
are small , and they will secure the sculpted stone by means a f bronze plates
that make a continuous contact of maximum effectiveness without affecting the
marble.

The only drawback of this intervention is the continuity of the spiraI through
the drums. This cont inuity mu st be interrupted at each le vel. The metal bar
will t herefore be cut at each joint bet ween d rum and drum, t hou gh its fun ction
as a conn ective e1ement is full y maintained.

5.- CONCLUSIONS

It is hardly necessa ry to stat e that the propasal described ab ove is only a

criterion , an approach to use in order to res ume the studies that were interrupted
at the end of t he 80s, and with the same attention to start to evaluate from
t he ins ide of t he colum n the exact positiou of the proposed metal co nnectors.
Bu t in concl usion, we must again consider , in the light of the intervention, the
dilemma of the cause of the damage found by Fontana.

If on Septe mber 10 1340 the earlhquake found Marcu s A ureliu s's Column whole
as it was in the original project of ApoUodurus, and with it s violence amplified
by the soft alluvium, and with the fr eq ueney of its peaks of a ecele ration , it shook
and broke the column so mu ch that it threatened its overalI in teg rity, then our
p roposal is absolutely useless. Do e ar two metal bars, scrupulou sly connected
to ali the fragments of rocks t hat we today find in this tragic Aotonioe mosai c,
are very far from reco nstituting the lost monoüthicity. And if th e earthquake
once shook and broke the columo wheu it was in perfect co nditioll , it wiII totall y
demoli sh today's column , despite our titanium spirals .

If it we re tru e that the earthquake that hit Rome shor tly after the epidem ic of
t he plague found Marcus Aureliu s's Col umn healthy a nd after a few second s left
it in ruins ) today we would have no hope for the column: we ean not stop the
earthqu a ke. A healthy economic policy would advise avoiding expenditure on
usel ess restoration works, and would limi t the care to a periodi cal cleaning. A
healt hy policy of civil protection would dose th e square on which the fragments
of the crow ning glory of the philosopher emperor would be t hro"'lI.
284 STRUCTURAL ANALYSIS OF HISTO RICAL CONST RUCT IONS

Bul \vc do no! bclicvc that lhe eart. bq uake is capahlc of so m uch , despite
lhe amplificati ons anel lhe harmful frc quen cics thai lhe floods af lhe T iber
confe r OIl it j we do 11 0l be lieve that OH lhe day before lh e J ub ilec p roclaimcd
by Clcme nt. VI the gloriou.'; pile af bloc ks a ssembled hy the emperor Marcu s
Aurelius Anlo ninus was u nex pcctcdly t. rall sfonned into a p recar iou s pile af
s toncs. No, it had becl\ al read y u ndermined by p revious a ccidents, even by
prc\'iou s earthquakes. Onl y bccause af thi s s ub st antial d egradat ion could lhe
cart,hquakc ove /' comc it. in its fragmclLt.ed statc.

Anel no\V , ir WI:' takc st.c ps to mitigatc th is statc a f d eg radat ioll, ir wc restore
some af its for me r in teg rity, ir we contraI lhe weak point. frorn t he tenth drum
dowlI t.o t.he ha.se which threatc ns iL s st.abi lity\ Wl:.' call havc good hopes for its
fuiUl'e a nd cxorcise the s pcctres of Arcadiu s .

In th is case, a hcal t hy po licy o f wllservatioll of the cultural heritage s hould

ta,ke prornpt adion to p ut into pr;tcti ce t lle experielltes ilm! rcftcction s gai ncd
ill the 80s. At Lhai time , lhe collHlln, covc n'd ill scaffo ldin g: \Vas s u b jeded to
a t.horougb st.udy ihat. cO ll s idcrec! a li ii5 as pe cts, ali it.s frag lllcnt5, it5 efrecti ve
co nsis tcn cc, the realily o f iis stOTlCS and t,h eir posi t ions, lhe ll1ech;~nisllls of
dalllagc <:\lld l he p os sibil ity of it bcill g rl' p e ; ~ted . EilCh sy lJlptom wa s considered ,
cach picce of infonll;tt. ioll wa.s co rrdaicd , in ord er to rdivc the particular
Od yssey of thi:> colullln .
 A. GUIFFRE I Marcus Aurclius's Column 285

NOTES ANO REFERENCES

1.- Giangiocomo Martin es, La struttura deUa Colonna Traiana: un 'eser citazione di
meccani ca alessandrina , in " Prospe ttiva" N032 , January 1983.

2.- A literary homagc to Marcus Aureli us tltrough the cust odian of the column
appointed by Septimus Severus is in : Mana Teresa GlUJfre , La veglia di Adrasta, Ed .
Studia Tes i, Pord enone 1986.

3.- Ant onino Giuffre - Fabio Ortolani , Le colonn e coclidi testimoni dei l erremoti di
Roma, in A. Giuffre, r. .t omunenti e Terremoti : aspett i st atici dei restauro, rvlultigrafica
editri ce, Rome 1988. pp . 55-64.

4.- ApoUodore de Dam as, Les P oliorcét iques, in "Revue d 'études greques" 1890, pp.
230-281, translation by E. La cosl e ofth e t ext publi shed by M. C. Wescher ( P aris 18Gi).
The following pass age is signifi cant : C'est ain si qu 'on peut , avec un petit nombre de
pi eces de bois de faibl es di m en sions, faír e une grande tour, d 'une hauleur égal e à cell e
du rempart ;.. p. 261.

5.- For lhis reason lhe drawin g a f Marcus Aureliu s's COl tllnJl shows a sl.i ght variation
from Trajan's Column: the tapering between the fir st and the Jast drum , whkh can be
noled in the founder column , is hi ghly reduced in lhe second : it is a correction of the
effect of perspective pr oduced by the view fr otll th e base.

6.- Palumb o A. L'osc illazione delle lance de Marte: meta fora o rudimental e rilevatore
di scosse?, in Guidoboni E. (ed.), I terrimot i prima dei Mille , Istituto Nazionale di
Geofis ica, Ed . SGA Stor ia- Geofi sica-Ambiente , Bologna 1989, pp . 122-123.

i.- Catalogo delle epigrafi latine r iguardanti i terremot i, in Guidoboni E. (de.), op. cit .,
p. 148

8.- Ant onin o Giuffre, I Terremoti di Roma, in Giu frr e A., Monumenti e terremoti :
aspetti statici dei restauro, Multigrafi ca Editri ce, Rom e 1988, pp. 47-54.

9.- Ignazio GaUi. I terremot i nel Lazio , Stab. Tip . "P io 5tracca" , VeUetr i 1906. An
essa)' by Rodolfo Lanciani , Segui di terremoti neglj edifiz i di Roma antica , in Bullettino
della Conuniss ione Archeologi ca Comunale, Rame 1917 , was also af great u tilit)' in the
study of Roman earthquakes.

10.- Th e m ost important dat es are: 801 , 897(?), 1349, 1703. On this argument see
also: Di ego I\!lol.in and Emanu ela Guidoboni : Effetto font i effetto monwnenti a Roma :
j terremoi t dall 'antichità a Roma, in Guidoboni E (ed.) op. cito pp. 194-223.
286 STRUCTURAL ANALYSIS OF HI STO RICAL CONSTR UCTIONS

11.- It is often called "the earthquake af Petrarch" , who wrote af it in several letter s
providing indications af the whole that agree with those recorrled by Matteo Villani ;
Francesco Petra rca , Familiares , Book XI , Lett . VII (ft om Piacenza, 11 June 1351).

12.- Matteo ViUani in: Muratorius , Rerum Italicarum Scriptores , vol. XIV vaI. 46. B .
A di scussion on the difference between the VIII and the IX degree af the MCS scale is
found in:

13.- Giuffre A., Vulnerability af historical cities in seisrnic areas and conservat ion
criteria , in Proceedings af the congress "Terremoti e Civiltà Abitative", Re me 1993 , in
print .

14 .- Renato Masian i, Seismic assessment af ohelisks and columns, in Structural

Conservation af stone masonry, International Techni cal Conference, Athens 1989,
l CCROM, Rome 1990, pp . 165 , 172.

15.· For a prismatic solid the acceleration caused by the start of the earthquake is
a/ g= S/ H , where S is the minimum dimension of the base and H the height .

16.· For a useful review of the state of the art af thi s argument , see: Ugo Andreaus and
Giuliano Augusti , Meccani ca delle colonne e delle costruzioru a bloccru lapidei : stato e
prospettive degli studi , in Pact 32, 1991, "Paes tum . étude de cas de vulnérabilité du
patrimoine" edited by Giuliana Tocco Sciarelli , pp . 67, 126 .

17 .' In a well·known experiment, if an ob ject resting on a flat surface is tilted and then
released, it fir st osciIlates slowly and then with incr easing frequen cy ul\til it stops.

18.· A review of the experimental work carried out in the Laboratory of the Department
of Structural Engineering and Geotechnics of the "La Sapienza" University of Rom e,
and a debate on the problem of the m ech anical similarity of scale mo deIs in the sector
of structures comp osed of blocks can be fOruld in: Giuffre A:, St udies in progress on
the seiSrnlC behaviour of the imperial columns in Rome , Rapp orto N 5, 1986.

, 9.· Sec:ret Vati ca.n archive - Arehivium Areis Armatio B3 - Libro di tutta la sp es a
fatta da N.S. Papa Sisto V aUa colonna Antonina e Traiana; transcription by OrneIla
Sforza and Maria Vittoria Zaccheo for the Archeological Superintendance of Rom e,
1984.

20.· Marco PeUetti was comntissianed by the Areheological Superint endance af Rome
to study the marble inserts by Fontana in the m onument o He then devot ed rurnseLf
to rus dadoral thesis, studying , interpreting and drawing aU the di slocations in
correspondenee with the fractures , st arting from the base and slawly reconstru cting
 A. GUIFFRE I Marcus Aurclius's Column 287

the current configuration of the column. The rapporteur, the author of these notes,
informed rum at the beginning of the purpose of the unusual survey alld finaUy , on
March 21 1989, presented the astonisrung but unequivocal results to the Doctorate
Commission . The survey by Marco PelleUi is tota11y different from the one carried out
by Piranesi, whom the Romans saw for months hanging in a basket on the side of the
colurnn . In the 17th century the figures of the bas-relief were studied wtih great care,
but in 1989 attentian focussed on the constructlOnal aspects and the reqtti r ement of
restoration. Establisrung whether the fissUI es betweell two blocks were cau sed by the
movement of one ar of the ot her , or of both, requires the same attention and the sam e
extension a f the field of observat ion that Marco Pelletti must have used to distinguish
the 15th century sc ultpure of Sitia Longa fr om the imperial one.

21.· See the entry "Colonna di Arcadio" in the E.A .A . (Treccani) . Dongh.i O., Manuale
dell 'Architetto (1906). Appendix , chapter on "Si curezza dei Fabbricatj", B, p. 128,
Turin , reprinted 1935.

22 .- Poleni G. , Memorie Ist oriche della Gran Cup ola dei Tempio Vaticano, Padua 1748 ,
Book I, Chap. XIV "DeUe cause esterne de' danni neUe fabbriche ; 125, co l. 75: "n
primo urto deI fulmine feri I'angolo posto verso Scirocco Levante .. Ma .. veniremo ai
gran colpo di quel fulmin e neU'angolo .. verso I'orologio, ... nella quale cantonata il
fulmin e fece una terribile squarciatura .. estendendos i in una lunghezza per pendi colare
non minore di piedi ven eti novanta". Also in the bell -tower of San Marco lhe lightning
had two areas of impact and pro duced cracks that were prolonged vertically towards
th e base. which refers to the Liber Pontifi cares (I, 239), see Guidoboni E. (ed.), op.
cit ., pp . 201.

23.- For these quotes see Romano P.: La ColorUla di Mar co Aurelio, in L'U rbe , Rome
[ 955.

24 .- An in-depth study of the surface variability of the se ismic movement in the cit)' of
Rome has beeo carri ed out by A . Rovellj , D . Molin , L. 1hlagnini, A. Caserta, Variability
of elamage pattern in Rome: combination of SQu rce anel local effects, present ed t o the
5th lnternat ional CO lúerence on Seismic Zooati on , Nice 1995.

25.-A later stud)' explicitly addressing the effects af the ea rth quake on the two calumns
(though we are in partial disagreement \\'ith it) is E. Boschi (et al.), Resonan ce of
Sub surface Sediments: an Unforeseen Complication for Designers of Romao Colwnns,
Bulletin of the Seismological Society of America , Vol. 85, N01, Feb. 1995 , pp. 320·324.

26 .- A very recent experiment being developed with the IZIIS lab oratory of Skopj e 00

the initiative of the ltaüan Ist ituto Nazionale di Geofisica has shown that the se ismic
actioll may produce th e cross wise fra ctures that split the blocks The m odel const ructed
288 STRUCTURAL ANALYSIS OF HI STORI CAL CONSTR UCTION S

o n a sca le a f 1:6 with a material tha1 reprodu ces Lhe chara ct eristi cs a f lhe marble at
a reduced scale \\Ias submitted lo very violent dy nami c actions up l o accelerations a f
0. 2g, with variati ons in lhe frequ ency content a f lhe t es t accelerogr alll 50 as t o foUow
lhe frequencies a f osciUat ion a f lhe column . Several crosswise fr a dures app eared in lhe
blocks, but no vertical ones and never more than two per block. These were fa cilitated
by defec t s a f fiatness in lhe contact between lhe blocks, bllt aft er 3nother ten slresses
that caused violcnt opening a f the joints, and di spla cement a f lh e t op that taken to lhe
scale a f lhe m onum ent rca ched 70 em , the model remained intact , as Trajan 's Column
is t oday : with some chipping , with crosswise fra ctures in the bl ocks, with a general
di splacem ent in lhe eleventh block , but witho ut the devastation that characterized
Marcus Aureliu s's Column whcn Fontana b egan t o res t ore it.

28.- The di splacemeuts between lhe rocks of lhe fr otlt column s a f t he Eretteo of the
Acropoli s in At hcns and in lhe Parthenon produced by lhe eart hquake of . are well-
kno wll , a nd the studi es of dynami cs confirm t he ari sing of t his mechanism as an etrect
of seism ic a ction .

29.- However , we canuot but agree O ll the fa ct that the fluvial basin , certainly lhe reason
fo r th e amplifi cation a f lhe seismic m ovement , transmitted it with a. fr eQuency content
that was capable of cau sing the resonan cc of th e co hunn for the folla wing reasons: a )
the cllaracteri sti c fr equellci es a f a basin ma.y be shown only by direct experimcntation
caused by the earthquake, as happened in Mexico C ity, anel not by numeri cal calculation
that does not take into a ccount the complexity of loc al conditions; but even whcn we
place our faith in th ese fa recasts (illdeed ther e are experim ent.al dat.a th3t confirm
these result s), thi s second reason cann ot be ovcrloookeel ; b) t.he column composed of
drums is very far fr om having an elasti c b ehavíol1f : the stro nger lhe seismí c actíon
is, anel lhe m ore il pra du ces oscillations that opell lhe joínt s bet ween t.he blocks, lhe
lower the fr equency of response is and lhe dynami c behavi our m oves away fr om lhe
resona nce. It can in 110 way be attempled to predi ct a rcsp onse wi t h eval uations c f th e
resonan ce. The fr equency that is measured as au efrect of vcry slight en vironmental
a ctions (linked t o the characteristi cs of lhe material and t o l he nal ure c f t he land
0 11 which the foundation rest s) has notlting to do wit h the fr equency of t he strong
osci llations that however depend on lhe geometry of lhe column aurl ou the amplitude
c f t he di spl acem ent . Hc wever , it i5 truc that thc low frcqucncics that wcre presenl in
the site of Marcus Aureliu s 's COIUIllll and not in that of Trajan 's Column increase lhe
responsc lo some ext enl .

30.- Sin ce the earthquake a f 801 secms to have been of consid erable int ensíty, our
illuslrati on could lead us to sustain that on that dat e Marcus Aur eliu s's Column was
t ruly wholc, like that ofTra jan , and the occasion that shattered the bl ocks was betwecn
801 anel 1349.
 A. GUIFFRÍ I Marcus Aurelius's Column 289

31.- Plundering wh at remained of the Setlizoru o.

32 .- The dam age to th e Colli seurn reported 0 11 Ih is occasion , ofl en quot ed , shows t he
terrible st at e a f lhe masonry r alher t han l he violence a f lh e earthquake. On anoth er
accasiol1 , in th e mid- I ilh century, there are report s a f collapses in the Colli sewn , but
no menti on is made a f earthquakes: l he coHapses tb er efore occurred spontaneo usly or
dlle to plundering . See Emanuela GuidobOlÚ , op. cit.

3.- They have al ready played this role in earthquakes sub sequent to ! 590, earthquakes
that though weak cauld not bul contain the dangerous fr equencies revealed by the
schalars af th e lstituta Naz ionale di Geofisica .
 STRUCTURAL ANALYSIS OF H ISTORICAL CONSTRUCTIONS
P. Roca, j.L. González, A.R. Marí and E. Oiiate (Eds.)
o CiMNE, Barcelona 1996

THE SPIRES OF BURGOS CATHEDRAL

L.M. Ortega, J. Perelli and J. Alberruche

GEOCISA
Los UI1/10S de lerez 10
Cos!ada, 28820 Ml1drid
Spai/l

SUMMARY

The studies and actions carried OUI by GEOC ISA in Burgos Calhedra! ca n be
divided inlo two different phases: the firsl, started in January 1991 , was basically
foc used on the instru mentation and evolution 01' the damages detected in the
Cathedral. specially those concerning to lhe spires.

The seco nd phase invo!ved a search for documents and hislorical records relaling
to previous interventions in the cathedra l. a Ihorough invenlOry 01' srires damages,
and the deve lopme nt of a structura l ana!ysis model of Ihose spires 10 eSlimate lhe
tensio nal slate of lhe stone, as well as 10 evaluate lhe possib le causes of lheir
deterioration. The characteristics of lhe monument most sig nificant materiais and lhe
possible reslOratio n techniques were also investigated on a schematic real scaJe
reproduclion of one of the floors of the sp ires , built in the laboralory.

The slruclura l analysis of lhe spires was carried ou l on two models : one for the
sto ne struClu re iso laled and a second one which incorporated lhe SlOne togelher witll
the interior metallic structure. under the act ion of selfweighl. wind. Iherma l
difTcrence between both materiaIs and corrosion of lhe stee! Slruclure .

The results indicated that lhe performance of lhe isolated slOne strucl ure is
co rreel: however lhe eva lualion of the complete sl ruclu rc (i.e. slOne and steel
Slructures working together) revealed eXlremcly high st ress leveis caused by
corros ion, which exp lained lhe damage lhal has been recorded and observed in the
Calhedra l.

As a result 01' Ih is analy sis a dccision was takcll to sepa rare the two slruclures.
 L. M. ORTEGA c( aI. I The spi rcs of Burgos Cathed ral 29 1

I. INTRODUCTION

Burgos Cathedral dates from the XIII Century. It was buill on lhe site ofaXI
Century Romanesque CathedraI erected by King Alfonso VI of CastilIe [IJ.

The Cathedral is composed by three naves; the central one higher and wider
than lhe lateral ones. A triforium was buih in the XIII Century along the main nave
above lhe poinled arches which lead to lhe lateral naves, and on lhe transep! as well;
it was laler rebuilt in the XV century. The cimborrio over lhe transept is an
octagonal structure which is divided imo lWO floors on both inner and outer faces by
surrounding balconies.

The spires were built by Hans of Cologne belween 1442 and 1458 in (he same
style as the French towers of lhat time. They were erected over the aneiem blunt
lowers. They have lhe shape af an octagonal pyramid with multiple open work s
organized in nine of the ten leveis on each of the pyramid trapezoida l faces. Over
the socle there appear nine diminishing floors, exaclly lhe same for both spires
except fo r lhe lraceries of the third and fourth leveis. Ar the tep of this lrunealed
pyramid (here is a shon cantilevered moulding along lhe whole perimeter supponing
a balcony with lhe shape of an oclagonal prism decorated wilh open works . The
slraight Iines of the pyramid edges are continued above this balcony, marked by the
corhels at each levei, which delimit trapezia with six circular va ids, whieh finish at
lhe apex with Ihiek diminishing mouldings. The overall height from lhe base o f lhe
soc le is 26.175 melres, with a distance of 7.50 metres between opposite edges at lhe
base [2J (Fig . I)

From a structural poim of view it is possible to distinguish three principal

elements: eighl nerves according to lhe pyramid edges which are connecled te eaeh
other by horizontal elements, defining differem trapezial modules in whieh eaeh
lateral face of lhe pyramid is divided, and differem traeeries in eaeh floor delimiled
by lhe lWO previous elements. The global dimensions and lhe open work designs
ereate an ensemble of significam slenderness.

According lO Marcos Rico [21, damages were already detecled one eenrury after
lhe ir eonslruction, earry ing out mayor repair works which were later eontinued in
differenl oecasions (dismantling the slatues of Saint Peler and Sainl Paul which
crowned the spi res and reparatian af bath spires) . In August 1887 damages were
produeed as a result af a lightning wh ich fell over lhe right spire; a Restoration
Project was prepared by Ricardo Velazquez Bosco in 1889 to repair those damage s
[2][3][4J .

Deterioration problems eont inued , and between 1927 and 1930 Apráiz built a
metallic structure inside each spire (Fig.2). It is an octagonal truncated pyramid ,
wirh eighl main nerves parallel te the stone Struelure pyramid edges situated at 20
em distance from them. Those nerves were connecled with each olheI' by rigid rings
at different leveis [2][3]. The stone and steel structures were joined by filling with
conerete the gap between the edges of lhem; the space between the rolled secti on
292 STRUCTURAL ANALYSIS OF I-II $TOR ICAL CONSTRUCTIONS

FIG. 1. - View of lhe sp ires of Burgos Calhedral.

 L. M. ORTEGA cl 0.1. I Thc spires of Burgos Co.lhcdrnl 293

bars which fonned lhe elements of melallic slructure was also filled with concrcle,
without being clearly detennined whether this concrele fiJling was carried oul
completel y when the metallic structure was built or at a later date.

FIG . 2 .- Stee\ struclure inside lhe spi res.

2. MONITORlNG OI' TIIE SPIRES

In January 199\ , following a rcquest from the InSlitutc for Conscrvation and
ReSlOration Df Cultural I-I critagc from the Ministry Df Culture. GEOCISA :-' Iarlcd a
periodic Illonitoring Df thc basic instrumcntatioll installed at Burgos Cathcdral. in
o rder 10 observe lhe evolution of the damagcs prcviously delcctl'd. panicularly
damagc to the spires.

At Ihal time , the spires pn:scl1tcd a scrious Slalc oI' dcterioralion: a large llullllx:r
of ftssures and cracks could bc observed from lhe interior. in lhe traccrics iI!) wdl
as in lhe horizontal c!clllcnls connecting lhe: edge nerves Df lhe spires (Fig .3). There
had cven been severa l poims in wh ich ashlars fragmcllI!' had looscned and falten. and
a bulging and deformation of lhe perimetral melallic rings reinforcing lhe :-.rire:-. ai
three different le veis was evidcnt.

In order to monitor and study lhe si tuation. instrument<llion was arrangcu

consisting in rcference Ihrcadcu stud s for lhe I11ca!-> urClllcnt of relalÍ"c 11l0"t:l1lcnt
bctwccn opposÍlc diametral poinls in lhe spi res (bet\Veen opposÍle point::. of lhe ~to llt.:
294 STR UcrURAL ANALYS IS OF HISTORICAL CONST RUCTIONS

structure as wcll as between opposite po ims a f lhe metallic st fucrure) by means af

a tape extensometer; 3nd refe rence pl ates for lhe measuremem af crack width
varialions wit h a mec hanical ex tensometer (Fig.4).

FlG. 3. - Crac ks in lhe opeo work traceries and hor izonta l elel11e nt s. Metallic hoops.

FIG . 4 .- Instnlmemation in lhe South Spire al mid-Ieve l.

 L. M. ORTEGA ~( aI. I The spircs of Burgos C;llhedral 295

The firSl phase of this monitoring was rnade by request of lhe Ministry of
Culture, taking 16 readings of the instrurnenled points periodically between 1991 and
1992 . In each measuring campaign. at approxirnale monthly intervals, IwO readings
were taken of each control poinl , moming and aftemoon. in order to evaluate the
influence of lhe daily temperature changes on lhe registe red measurements.

Afler finishing this first monitoring period, lhe graph s of evolution of

convergences and c rack width measurements (Fig.5 and 6) showt:ú (I. ~ ignifi cant
thermal influence on lhe crack width varialions, clearly noriceable both lhrough lhe
variat ions between lhe two daily measurements, and through lhe evo lulion throughout
lhe whole lime perí odo The largest opening va lues of the cracks were produced in
summer time. with a lendency to diminish during lhe autumn and increase during the
spring . Many of lhe observed cracks. in lhe both lefl and right spires . showed.
besides lhe variarions due to thermal influences. moverncnts of a slructural naturc
with cJcar remanem open ing movemenls at lhe end of an annual cycJc.

On the contrary. crac ks instrumented in orher areas of lhe Ca lhedra l (dome.

Papamoscas room. cJ oisler, covcred baluslrade. main lerrace , chamber of vaull s.
lriforium). presented residual val ues close 10 zero after an annual cycle 01'
measurements. which a ll owed lhe assumplion Ihal these move ments we re exclusively
of a thermal nature .

A cons iderable lhermal influence was a1so detecled during lhe evo lulion of lhe
measurements of conve rgence, reaching maximum val ues in lhe rnonth Df AugusL
The relali ve movements observcd belwccn díametrally opposed po ints of lhe melallic
struClurc showed a clear tendency lO relurn (Q negligible val ues when the annual
thermal cycle ended . 011 lhe contra ry, a clea r movemelll of radial expa nsion wa s
detecled in some of the measurements of the relalive movcment s between the sto nc
sl ructure. spec iall y in lhe intermediate leveI of the righl spire . The net remanem
va lue of lhe re lative movement between oppos ite points of lhe stonc structure was
0.40 mm in the left spire, whil sl in the right spire lhe $a mc phenomcnon was
detected , but even more clearly rnarked. reaching a re sidual va lue ofaboul 0.95 111m
after a cycJ e of more lhan a year .

This tendency towards radial expansion was in agrcemem with lhe obse rvations
made when measuring, delecting clear signs of corrosion in lhe contact belween lhe
mctallic struc ture alld the concre le filling lhe gap be lwcen both SlrUClu res. Thi s
corrosion cou ld be the main agem causi ng that radial ex pansioll and the cOllsequem
rema nent opening.

Approxi matel y two years after fi ni shing this first Illonitoring campa ign . and
wit hin lhe agreement for lhe restoration of lhe Cathedral belwee n lhe Council 01'
Castille ;.IIld León and lhe Mini stry of Culrurc. new measuremems were taken which
confirm the lendencies already perceivcd by lhe measurements of the first campa ign
(Fi g.7 and 8).
296 STR UCT URA L ANALYS IS OF HI STORICAL CONSTR UCTIONS

SPIRES OF BURGOS CATHEDRAL

MONITORING OF CR ACKS
~
-~
E

-=-
~

E
8
2-
CJ
z
Z '!M-
"'"
W
,~
"-
o ~

'~'+C~~~-.---.~--r---~--~~-r---'----r-~
~n9' M~'! A~ 9 ' ~n 9 1 _ 91 ~~91 ~ . gl ~ 91 h ~~ Aw~ ~y n

DATES

~.- P2 -+- P4 P5
-a- P20 - )(- P24 - 4- P27

FIG. 5. Monitoring or crack s in lhe spircs. Firsl campa ign .

SOUTH SPIRE OF BURGOS CATHEDRAL

RADIAL MOVEMENTS ANO CONVERGENCE

:j
_00
E =_
E ~
o ~

2. 1!.(l
, . : __ 1" ; ....
Z
O
'00
, . " t - ,t r ..-. -,
• - .- -- -- ..--- o

Ui
z ,- ;----1--.---. .- -
W
f-
X '00
W ,~

1-_- P-5 + P-6 '. P- 7 1

FIG . 6. Radial l11()vClllcnl s and co nvergence in lhe Soulh Spire. First campaign.
 L. M. ORTEGA ct aI. I Thc spircs or Burg.os CatheJral '297

SPIRES OF BURGOS CATHEDRAL

MONITORING OF CRAKS
'~ OO-;-,---------------------~

E O
JOOi
E
8
S-

'"
~ "'001 f"
:
~ ,:j
'~l :/./ ._~
,=
/lO" 90 Jun .91 0.. _91 Jul-92
-
Jon 93 A.ug .9J
==
DATES

- - P2 --+- P4 - - P5
-----e--- P20 - )(- P24 - A-- P27

F IG . 7. Monito ring of c racks in the spires. Average da il y values.

SOU TH SPIRE OF BURGOS CATHEDRAL

RAD IAL MOVEMENTS AND CONVERGEN CE
~,

' !- -
'x -
;)5' --'

Ê JOC_
E 250..,
.•... - .--
o
si
~'i
'5'l .-- .. ,
,.•.....
........

z .(X._
: '. : • . : : 1 .• . • .:- -: ·: ·-···· .-.. 0-'
O
in
z
~~1
,- ,--..- -,.,- ------------
; - - .:

'.-
W 5·) -
f--
X
W
.~-

200J
,sô-
.)(\;.
tJo o 9O Jun 9 1 0... -9 ' Jul 92 J a n 93 Se p 94
DATES

1--- P·5 + P-6 - - - P-7 I

F IG . 8. Radial move ments and conve rgence in the South Sp ire. Ave rage dail y
va lues.
298 STRUCTURAL ANALY5 1S OF H1STOR1CAL CONSTRUCTIONS

3. STRUCTURAL ANALYS IS

3. 1 Modelli ng

Co nsidering lhe resu lts oblained , it was decided lhat simple Sl!llClUral mode l
would be crealed with lhe basic aim of evaluati ng qualilalively lhe infl uence on lhe
spi res of an expansion force due to lhe corrosion of lhe metallic structure in conlac!
Wilh lhe concrcte presenl between it and lhe stone structure.

At lhe same time, the model was also intended to analyze several specific
aspec Is such as:

• The differences belwec n lhe strcss stales induced in lhe Slonc slruclure
by lhe wi nd fo rces, depending on whether lhe collaboration of the
internaI metaJlil: structure exislcd or l1ot, in order 10 confirm lhe
possib ililies of lhe Slone struclUrc a lollc lo suppor! its sel f-weight and lhe
wind forces.

• The importance thal lhe diffcrem Ihcrmal rcsponse of both materiais

(stee! and stone) cou ld have on lhe behaviour of lhe spire.

A mode l was developed 10 fu lfil these goa ls, materializing the differe nl elemcnls
of lhe stone strucrure by mea ns of ba rs and discreriz ing lhe socles by plate finite
eleme nts (Fig.9).

'"
,.,

'Il .

,1\\

SI ' • oP o- P lO 1 (lIf~ 18 o.)

""1 ,,,'""O !"."',... Df P IE " . I
O( "

flG. 9 .- StruclUral 1l10dcl.

 L. M . ORTEGA et aI. I Thc spircs af Burgos C:J.thcdral 299

The geomelrical definilion of lhe model was based on lhe informalion availab le
oblained afler searching in different sources. particularly on lhe drawing ma de by
lhe Archilecls Ricardo Velázquez Bosco in lhe 1889 Restoralion Project and Marcos
Rico in 1981 . 101m coordinales and member geometrical characlerislics of lhe medel
\Vere dctermined with the aid of these drawings. Once lhi s process was completed.
lhe dimensions of lhe elemcnls forming lhe Slructure were cheeked 011 s ue .

Ali the bars \Vere assumed to be prismatic of constam transverse section. and lhe
10p pinnacle of lhe spire was modelled using eight bars converging on lhe vertex of
lhe spire pyramid. \Vitll a total volume equal 10 Ihal of lhe real pinnacle . The se
simplifiealing assumptions seemed to be reasonable and in no way di slO rting .

Severa I analyses \Vere perfonned o n Ihi s initialmodel. focusing on lhe followin g

aspccls:

• Slresses on lhe SlOne spi re due to ils selfwe ighl. WilhoUI laking into
account lhe co llabo ralion of lhe melallic SlrUCIurc .

• Stresses induccd on the stone spire by lhe wind. withoul laking imo
aecoun! lhe co llaboralion of lhe melallic strue ture.

• Slresses induced on lhe spi re by lhe wind. assuming lhe collaboralion of

lhe melallic slructure.

• Slresses induced on the slOne spire as a result of the exisling c,xpansion

forces duc to lhe corrosioll in lhe co ntacl between lhe concrelC and the
mctallic strueture. assuming the eollaboration of lhe latter.

• Slresses induced 011 the slOne spi re by ti Ihermal difference of 10 oC

belween bolh Slructures.

Considering the difficulty in eSlimaling lhe deformabilil)' eharaelerislics of lhe

Slone. lhe analyses in which bOlh SIJuClures were collaborating. were repeated usi ng
very different "alues for lhe 1110dulus 01' clasticilY of lhe SlOne fabrico varying
between E = 6 .000 K g/c Jll ~ and E = 60 .000 Kg /c l11 ~.

The wind was inlroduced app1ying concemraled ho rizontal loads OH lhe joims
of lhe slructural 1l10del. The following assumplions \Vere adopted 10 oblain Iheir
val ues:

• Dynamic wind pressure: 125 K g/ m ! . co rrespo nding 10 a wind velocilY

01' 161 Km/ h.

• Spire \Vali wilhom openings.

• Three sil ualions \Vere consi dered:

300 STRUCTURAL ANALYS IS DF I-I ISTORICAL CONSTRUCTIONS

Wínd pre ssure acting over an area eq ual to the surface defined by
the mid -plane of the spi re and perpendicular tO ir. with aeo lic
coefficients equa l to 0 .8 for pressure and 0.4 for suction (Fig.
10a).

Aeolic coefficient according to the Spanish standard NBE-AE-88

(Fig. IOh)

Aeolic coefficie nt accord ing to the study made in [51 and [6]
(Fig. 10e).

+O.&p

.. ~
+ O.1Ip

-o .." .......
~

~ )!,;~
(a)
-" n - o.." - o.."

+O.1Ip

,oi'·/;!'---- --':'- 'o

•
t ,
1
•
(b)

,
•,
"
(o)

FIG. 10.- Wind hypolheses .

 L. M . ORTEGA Cl a!. I Thc spi rcs of Burgos Cmhcdral 301

The corrosion expansion forces were modelled by means of a n imposed sIrain

a n lhe bars cannecting both slructures. Since separatia ns a f abaut 1.5 em had
acrually been measured between the cancrele and the metall ic structure due (O the
thickness a f the corrosive products, ao iocremem af I cm io lhe leoglh af lhe bars
which cannecled ba th structures was assumed in the made!.

The caJcul alioos of these models were perfarmed using lhe program STAAD-III
Version 18. 0 from RESEARCH ENGlNEERS .

3.2 ResullS

Cooside ri ng lhe modelling undenaken , the anal ysis of lhe results has been
mainly focused an lhe Slresses due to axial fo rces aCl iog 00 the ba rs. which is
log icall y lhe predom inam load. even thaugh in ce na io siruati a ns sig nifica nt beod ing
stresses cau ld appear, poss ibly mag nified by lhe simplifications inherent to lhe modcl
when substituting in many cases curved elements by straight bars.

The results obtained in lhe analysis which are included in lhe graphs carrespond
to E = 6 .000 Kg/cm 2 fo r lhe masonry. represe nting lhe max imum stresses for the
co rrespo nding bars of eac h levei ; e.g . on each leve i three groups of bars were
co nsi dered corresponding to lhe pyramid edges (marked in lhe graphs as V). 10 the
hori zo ntal connec ling elemenls bet ween them (ma rked in the graphs as H). and
finall y those representing lhe Iraccries of lhe open works (ma rked in lbe graphs as
C). If lensioned and compressed bars exist simultaneously in one single group, il has
bee n decided to represem on!y lhe maximum tensile stresses (F ig. 11 to 14).

The aim 01' the study 0 1' the stone model was to chec k lhe ori ginal slructural
performance of lhe spires of Burgos Cathedra l. wilhoul lhe ex isle nce of the metallic
slruclure which was added later.

Compress ive axial forces were induced in 1\\'0 diffe rent types of ele melll s due
(Ose lfweight (pyramid edge bars aod lraceries bars) whereas lhe horizontal elemellls
were subjected to smaIl tensile stresses.

The maxi mum stresses due lO selfweight induced by the element axia l forces
\Vere 7.5 Kg/cm Z (compress ion) in the pyramid edges bars . and 2 .3 Kg/c m~ (Ie nsion)
in lhe hori zo ntal elemems aI lhe rop of the socle (Ihese cou ld be affected by the
soc le's ow n rigid ilY), whereas in the rest of lhe horizonta l eleme nts lhe maxi mulll
stress was always be low 1.0 Kg /cm1 .

In lhe combinali on of se lfwe ight plus wind, lhe absolute va lue 01' the slresses
we re hi gher fo r ali lhe groups ar bars , reaching maximum values of 10.9 Kg /cm 2 in
compress ion and 3.3 and 2 .4 Kg /cm 2 in te nsion. fo r lhe same bars as mentioned in
the precedi ng paragraph .
302 ST RUCTURAL ANA LYSIS OF HISTORICAL CONSTRUCT IONS

SPIRE OF BURGOS CATHEDRAL

STONE STRUCTURE

BARS
+ WIND
i· SELFWEIGHT _ SELFWE IGHT

FIG. 11.- Stone slructure mode!. Maxim um stresses in groups Df bars. for
selfwe ight and selfweight plus wind hYPolheses. E = 6.000 Kg/c l11~ .

SPIRE OF BURGOS CATHEDRAL

STONE AND STEEL STRUCTURE
00---------------------------------------------
ó- , ,

~ 6~ ç v

f :~c. . ~-'"-~" ~~~

.'1, I i , i

BARS

1_ SELFWEIGHT _ SELFW EIGHT + WIND

FIG. 12 . Slone and stee J strw.: tures model. M ax imum Slrcsscs in groups 01' bars.
fo r lhe selfwe ight and selfwe ighl plu s wi nd hYPol heses. E = 6.000
Kg/cm 2 ,
 L. M . ORTEG A r.:t :tI. I Th\! spi res of Burgos C:tlhcdral 303

SPIRE OF BURGOS CATHEDRAL

STONE AND STEEL STRUCTURE

,I ,
I

BARS

,ijl SELFWEIGHT _ SELFW. + TEMPERATURE

FIG. 13 . Stone and steel structures mode l. Maximull1 stresscs in groups of bars ,
for lhe selfwe igh t and selfweight plus thennaJ difference hypolheses. E
= 6.000 Kg/cm' .
SPIRE OF BURGOS CATHEDRAL
STONE ANO STEEL STRUCTURE

.
,~'-'....,JII~.--.lI--.~--
,

1· ~

BARS
_ SELFWE IG HT _ SELFW + CORROSIO N I

FIG . 14 . 5lOne and steel structures model. Maximum stresses in groups of bars.
for lhe se lfweiglll and selfweighl plus corros ion hypotheses. E = 6.000
Kg/cm'.
304 STR UCTURAL ANALYSIS OF HI STOR ICA L CONSTRUCTIONS

As indicated above, compress ive axia l st resses were oblaincd in lhe bars
mode ll ing lhe ope n wo rk 's traceries elemenls, although they we re co nsidc rabl y lowe r
than those obtained for the pyramid edge bars.

Figure 11 summarizes the maxim ul11 ax ial stresses obtained for lhe di fferem
lypcS of bars on eac h leve I. for two cases; due (O selfweig hl a nd due 10 lhe jo int
action of selfweight and the wind hypOlhesis whic h is most unfavourablc for that
specific group of eleme nt s.

In lhe model induding lhe melalli c slruClure , the maximulll axial slresses
obtained for the jo int act ion of selfweight and wind are, logically , lowe r than lhose
obtained for lhe pure slOne model (Fig . 1I and 12).

The maxi mum slresses reached in th is case have bcen 8.8 Kg/cm ! fo r
co mpression. and 2.7 and 1.5 Kg/cm1 for lel1sion, 0 11 lhe groups of bars already
mentioned.

Simil arl y, lhe introduc tion of a the rmal diffe rel1ce of 10 oC does 110 1 produce
particularly significa nt changes 011 the stresses induced by se lfwei ght. slighll y
inc reasing lhe moderate ax ial stresses prev ious ly obtained. On lhe conlrary . the
loading hypothesis in which the effects induced by co rros ion are taken iOl o account.
inc reases co nsiderably lhe ax ial forces on many ba rs, particularl y those modell ing
lhe horizontal elemeOls, where lhe maximull1 tensile stresses reach 36 . 1 Kg/cm2 .

As it coul d be ex pected , lhe co llabor<ll ion of lhe steel Slructure causes a

sig nifi ca nt rcduct ion in the maximum horizo ntal displacement , obtain ing movement s
vary ing betwee n 45 % and 85 % of lhe result co rrespond ing to lhe stone mode!.
depending on whet her lhe minimum or maxi mum modulus of elasticity is used .

Maxi mum stresses obtained for lhe different types of ba rs in th is complete model
(stone struclure plus metallic structure) a re shown in Figure 12. in both the
selfweight and se lfweig hl plu s wi nd hYPolheses (considering the most unfavourable
wind hypot hesis in each case) .

In Figures 13 and 14 a compari so n is made belween the axial stresses due 10

se lfweight and those oblained in two diffe renl hypotheses comb ining se lfweiglll plus
Ihe rmal difference and selfweight plus corros ion.

Obviously lhe d iffe rem assumplions made co ncerning the masonry modu lu s of
elasticity have considerab le influence on lhe sl resses obtai ned in this complete model.
although lhe concJ usions resulting from lhe analysis do not change.
 L. M. ORTEGA ct aI. / The spircs of Burgos Cathcdr:l.l 30 5

4. CONCLUSIONS FOR THE RESTORATION

The following main conclusions were derived from the results obtained by
GEOCISA in the structural study of lhe spires of Burgos Calhedral:

It seems to be clear, from the calculations carried out, that the resultam
stresses in the swne caused by selfweight, wind or thermal difference
between the stone and the metallic structure are reasonably low ,
independenlly of whether the latter collaborates or not.

Tensi le axial stresses are detected in certain models. basically in

horizontal and open work bars. The values obtained are low. although
this effecl cou ld sornehow expla in some of lhe darnages detected and
certain structural so lutions adopted on the spires in lhe past years (e.g.
c racking of the open works elements and metallic hoops in lhe horizolUal
elemcms).

The movements obtained in lhe analysis do nOl seem to be too high in

any situation. The maximum displacemem varies between 4.57 em for
lhe stone Slruc(Ure considered iso lated and 2.06 em for lhe model of lhe
stone and metallic struetures working together.

EXlremely high tensile stresses are obtained in lhe horizontal and

traceries elements due to steel eorrosion in lhe eontaet between metal
slructure and eoncrete. This effect is probably responsib le for the more
sig nificam structural damages which could be observed in lhe stone
st ructure . This hYPolhesi s is confirnled nOI onl y by lhe structural
analysis, but also by the results of lhe monitoring, and lhe various visual
inspections.

The principal conclusion of this analysis was to co nfim lhe co rreCl

perfornlance of lhe isolaled stone Slructure , as well as the damage eaused
lO lhe SlOne Slructure by lhe corrosion of lhe steel one. Consequently. il
was decided to isolate both structu re s cutting the concrete in lhe gap
belween lhem using a diamond wire. afler a complete resto ration of lhe
slonc structurc (scwing by fibrcglass-rcsin bars, masonry rcpointing,
cleaning and maintenance of the stee1 structure).

ACKNOWLEDGMENTS

The authors are grateful for the collabo ration from the M inistry of Culture and
lhe Junta de Castilla y LeÓIl. as well as lhe ArchitecIs in charge of lhe restoralion.
J06 STRUCT URA L ANALYSIS OF HJSTO RJCAL CONSTRUCT10NS

REFERENCES

I. lAMPÉREZ Y ROMEA , VICENTE.- "La Catedral de Burgos". H. de J.

Thomas S.A ., Barcel ona , 1906.

2. RI CO. MARCOS.- "La Catedral de Burgos. Patrimonio dei Mund o". Edited by
the author. Burgos, 1985.

3. JUNTA DE CASTIllA Y lEÓN .- "Las Catedrales de Castilla y Lcón. Un

Proyecto Eterno". Tecnomedia S.A .. Valladolid, 1995 .

4 . VElÁZQ UEZ BOSCO. R1CARDO.- "Obras de Restaurac ión de la Catedral de

Burgos". Files of the Spanish Administration at Al calá de H enares, 1889 .

5. BECKMANN, P.· 'The step-by·step approach to investi gation and rc medial

wo rk illustrated by lhe work at Y ork Minster and H oly Trinity Church
Coventry'. International updating course on structural conso Jidati on, Cent re for
the Conservarion Df Historie Towns and Buildings, Leuven, Bel gium . Leuven
University Press, 1988.

6. BECKMANN , PO Ul Y BlANCHARD, JOHN .- 'The Spirc of Holy Trin ity

Chureh Coventry '. The Arup Journal Vol. 15 N° 4, December 1980, pp . 6- 14 .
Ove Arup Patertnership , London.
 STRUCTURAL ANAL YSIS OF HISTORICAL CONSTRUCTrONS
P. Roca, J.L. González, A R. Mar( and E. Oi'tate (Eds.)
eCIMi\:E, Barcelona 1996

ROMAN AQUEDUCT OF SEGO VIA

F. Jurado Jiménez
E.T.S. de Arqllitectllra de Madrid
cf José Fentanes 20
28035 Madrid
Spain

HISTORlCAL AND DESCRIPTIVE DATA

The origin of the water conduct is located in the high course ofthe "RIO Frio" river The
dam, small tapping prey ofthe waters, is formed by granite pieces clamped with plates
ofi ron received with lead. 11 can derive near 50 liters ofwater per mil1ute .

It s buried conduet, abou t 11 ,4 km long, was piped in ! 929 through a pipeJine of cemem
of50 and 40 cm diameter, making 110 registration tanks that mark contour, a1most
completely o riginal.

It possess a first "casteHum aquae" , call ed The Stone I-Iouse, a 5,50 x 5,90 m
rectangular building of masonry \\Iith lime mortar, with a pointed vauil and a 2,25 x
3, lO m underground deposit \Vil h a depth of2 m

From this house begins lhe granile channel carvcd in "U" fonn . It crosses under the
current 5an IIdefonso highway A secti on was dismantled upon the variation ofthe levei
ofthe highway and has been 10s1.

The high sect ion begins with XV II eentury eylindrical ashlars that form an obelisk where
one ean find a parchment supported by two arms wilh lhe representation oflhe aqueducI
According 10 Somorroslro, these ashlars were found near 5an Matias hermitage, in "La
Santa Cruz" road and were pUI on this site on 1951

After 141.5 m ofhigh co nd uet (wall of 1,4 m broad and 1,4 to 3,5 m high). onearrives
to the seco nd castellum aquae, 8,90 x 7.50 m decanting building, wilh a 4,25 x 2, 18 m
and 2.90 m of deplh inner pie also covered with a poinled vault

From this point lhe si mple areade and the numbered pillars thal we alt: u~ing as referenee
begin So thatthere exists a first alignrnent of6 arches. a 25 arches sccond alignlllcnt
and a third Orle wilh 44 arches. unlil il arrives 10 lhe 75 1h pillar, where the sharper
alignment ehange is produced and it bceames a double series of archs

From the 35 110 pillar to 51 1h therc are diverse reconstruetions, one ofthem daled in 1868
according to the cxisting inscriplion in an ashlar.

Thc double arcade is composed of 43 double arches that reach more than 28 m in height
over the eause\Vay In I Oi", 108 111 and 109111 pillars a cartel appears (the "sotabanco")
where Iherc exisled lelter inscriptions (presumably eneased with lead in the granite
308 STRUCTURAL ANALYS IS OF HI STOR ICAL CONST RUCTION S

bronze Ictters). The 108 th pillar is thicker in lhe superi or part and has two niches
(original, by the inside fingerprints oflhe boisting cJamp), where two seulptures were put
on 1520.

Crossing the medieval defen sive walls (in the place called " Postigo") the conduet
cont inues in a walJ and ot her three a rches, an almost in right angle alignment change and
otber 43 .7 In long \Vali with the conduct in it s upper parto From here, lhe aqueducI's
conduet returns to be buried under the slreet s until arriving 10 lhe "Alcázar", by a way of
underground stone eha nnel s ofmore Ihan 1 km, known in many sections for lhe
excavations, and havillg served in some cases 10 condue! residual walers.

A total of 15 km of aqueducI wilh a 958 m high water conduct and wilh variable slopes,
lhe minimal slope in the central part (88 th to 1 13 th pillars) at 0,3%.
The currenl 30 x 30 cm upper channel , c10sed aI the sides with masollry, is not the
o riginal eondue!. This was lower down (rcmai ns st ill ex ist with ali assurance) and had a
section about 60 x 60 cm.

STU DIES AND INTERVENTlONS THA T IT HAVE BE EN CA RRJED O UT

SINCE 1992

Stop or the traflie or vehicles u"der the Aqueduct.

The decision was taken dircctly by the Segovia municipal govcrnment. . even though il
generales evident int ernai IraOie problems in Segovia, it has been a drastic decision to
eliminate the most probable cause ofthe current deterioration ofthe Roman Aqueduct .

SCllfToldings alld prcvcntlllivc props Ill:1CC for proteclion and inspection.

The scaffoldings were built sin ce July 27 th to August, 1992. They have been used up
until 1l0W for lhe prolection and inspection.

Delailed Visual Recognition .

An urgent inspection was ordered for lhe section between pillar 50 and lhe " Postigo"
wall, !hat is to say, the highest part, the most monumental. Said inspect ion report warned
of the dangers of fal1ing stones from detachments, local breaks and deteriorations and
other damage observed in the ashlars, wilh a cJear concentration in lhe area of the
greatest now of vehicJes.
The inspection began on August 13 th , 1992 and was ended in November ofthis same
year. Later, in March of 1. 993 lhe inpeclion was amplified to lhe scaffolding zane .

Surnrnary or lhe exisling documentation .

Asidc from the vast bibli ographical documentation, ali lhe available information on last
interventions o n the Aqueduct have been gathered . In particular, o lle must to hig hlight
the interesting photographs (obtained from the files of!he Ministry ofCulture) taken by
Laurent ( 1863) or by Cli lTord (1853), tha! have served to compare, in a detailed way, the
slate lha! lhe ashlars had !hen wlth lhe eurrent sI ate, seeing the great acceleration of the
deterioration in these last ycars.
 F. JURADO JIr..'IÉNEZ I Rom:m :J.qucduct or Scgovia 309

In effect, it is easy to appreciate the delerioration that has rounded lhe corners of the
pillars 50 that now it is difficul t to see the carved li ne ofthe pillars' corners, that is easily
observed in lhe photographs of a century ago .
Whal are, then, lhe factors Ihat have appeared in these last one hundred years to
deteriorate lhe stone more Ihan in lhe eighleen previous centuries? The answer 15
evidently: the traflic ofvehicles and the atmospherical pollution.
Then, Ihese t\Vo imporlant causes are emphasized as the way to investigate lhe origi n of
lhe current degradation Ihat we already observed in the aqucduct visually .

PolIUlion and in-mission s meas uremenl

At first , in October of 1992, measurement is taken from a mobile unit, and, si nce January
o f 1993, a fixed unit Is installed in coordi nalion with the town hall of Segovia.
Climatological paramelcr measures were taken as well as magnitudes based on the
seasonal cycle.
The traffic was already slopped under lhe Aqueduct, therefore the obtained graphics
co rrespo nd to a situation less contaminated by lhe passing vehicles than thal which
existed before July, 1992, when this decision was taken . In spite ofall, it can be
observed thal , in winler monlhs, Ihere is a clear increase alllong lhe emissions of, for
examp le, carbon mo noxide and sulphur dioxide, which Illeans thal lhe housing healing
has a great influence on the pollu lion .

Bi ologica l d eteri ora lion a nalys is.

From March lo July of 1993 the Microbiological Leclure ship of Alcalá Universily ha s
been conducling sludies 011 lhe microorganisms lha! by biochemical and biophisical
actions possess the capacity to deteriorate granile. Microorganisms that they have
isolaled and identified and Ihat , witholll doubl, co ll aboralc with other higher organism s:
vcgelables and animais (birds, over ali) .
An ab normal levei of microo rganisms was observed, very high for granile.
II is evidenl Ihat microorganisms ofthe sulphur cyclc and those of lhe nit rogen cycle
have lhe possibility of convcrti ng elements lha! exist in lhe environment into acid, due to
lhe almospherical pollution, assaulting Iherefore lhe granite.

Sludy of lh e foreig n ma teri ais to l he Aqu edu cl (used in th e res tora ti on of

1973).
As hislorical doclIrnenlation we had lhe inlervenlion Ihal loo k place in 1973 by the
engineers, Fernández Casado and Aurel io Ramirez, widely described in the book "Thc
AqucdllcI of Segovia, lhe survival of a hydraulic work" of Aurelio Ramírez, asidc from
having lhe vo lllnlary collaborali on oflhis same aulhor.
I lowever, lo prove lhe 5COpC of thC5C works and lhe stat e of conservatiol1 and / or
del erioration ofthe materiai s and products lhat were used, thi s study was cOlldu ctcd and
endcd in March of 1993 .
Among other things, sample were taken oflhe Iransvcrsc "boltings" made wilh brass
bars and resins, as wel! as the zo nes thal were made wilh cement grout injeclion s. Al so.
samples were taken ofthe walerproofing ofthe upper channel.
As cOll sequence oflhis sllldy we observed Ihal lhe Slal e ofl he brass boltings, lhe epo;.;i
resi n and lhe channcl waterproo fing , was in peffecI condilioll. In particu lar, Ihere is no
Iype ofmicrofi ssurali on in lhe face o flh e grallit e in lOuch wilh lhe resin, which gives lIS
an expert "mallomeler" , submittcd to lhe outdoors during Iwenly years, so Ihat ifwe
310 STRUCTURAL ANALYSIS OF 1I1STOR1CAL CONSTRUCT10NS

need to relurn 10 make a "suture" of this kind, we can use the same material with the
guarantee of ils unalterabilily
As for lhe venical cement grout injections, fortunalely we found few remains. We say
fOl1unalely, because Ihi s is an issue Ihat we should criticize afthe 1973 reslOralian : in
lhe errort to till in lhe spaces among the sla nes in arder lo produce a certain monolithic
eITec! Many perforations were tOlally cmpty : we do no! know iflhe grour flowed and il
did not ever arrive la filllhe perforalion or if. with time, il has been degradare 50 Ihat il
has becn lotally 10st.
We agree therefore with lhe transverse boltings Ihal \Vere made in 1973, since the only
thi ng what was intended with this was to join lhe stones Ihat had splil and acles as a
hobble among lhe Iwo ashlar leaves ofthe upper part oflh e Aqueduct, however, as we
have alrcady said. we do 001 agree with lhe injecting afthe cemenl groltt to produce an
agglomcralion among lhe ashlars, that is in contradiction with the conception oflhe
Aqueduct dry-stone (wilhout mortar) In addition, lhe sal! resi ducs Ihal the cement
Icaves 011 lhe s!onc when is wetted, results in deterioralion .

Preventive protection.
Belween January and April of 1993 prevenlive protections were placed (through grey,
plastic nels) in some Monument zones where dangerous detachments could take place,
so Ihal we could conducl ali oflhe s:udies unlil lhe reslIlts indicated lhe solulions to
take.

Graphic elevalings lllld inilial photogralllmetry.

A photogrammetric clevation of lhe scaffolding zone began in July of 1992, did nOI
resuit tOlally satisfactory, due mos! probably to lhe short time frame alowed before
pUlling :..Ip the scaflold ing .
However, we have aceomplished a complele mapping ofth e Aqueduc! using, in part,
previous sUlveys, aereal maps ofthe city ofSegovia and our own measmemenl s.

Three-dimensionat CAD I'hOlogram melry :lIId graphic data bases.

Before lhe evident need for a whole detailed pholOgrammelry of lhe aquedu ct, around
lhe end of 1.993 lhe Lalorre &Cámara team was commissio ncd with lhe field projects
and lhe subsequenl survey 01' an ordered threc-dimensional photogrammetry of lhe whole
AqueducI
This performance is all-important to define, by one way, each and every one of the
ashlars Ihat form lhe monumenl, serving as a graph inpul (and outpUI also) for lhe data
bases thal wc are accomplishing on damages evalllation, analysis, etc ..

St utl lcs ou the granitic Slonc of lhe Aqueduct.

This topie ~s one ofthe aspects of greater imponance. Si nce em! of I. 992, invesligati ons.
sampling and measurements lhrough ultrasonic velocity transmission, physieal and
chemical tests, etc have been conductcd .
Ali ofthese studies are collecled in lhe works (September of 1.993) 10 determine lhe
causes and type of Aqueducl granite alterat ions
We proceeded to take twenty-l\Vo ashlars samples and a series ofsimilar granite qllarry
samp lcs
The majori,y ofthe Aqueduct Slone is bulk grain grani le, lhe 10wesI quality and lhe
caSiest !o carve
 F JU RADO JIMÉNEZ I R0l1131l aqueducI or Scgovia 31 [

Com parison belween existing Aqueduct granite and similar granite extracted rrom quarry
shows lhal lhe fir st has a mean compression strength orabout 230 Kp I cm 2, while lhe
2
second Is around 800 kg I cm and also have similar densil y va lues
In acceleraled obsolescence tesls we observed Ihat, what most assauhs granite is acid in
lhe enviro nment, that Is to say the sulphuric acid , aRer lhat , the ice and thaw cyclcs, and
laslly, in an almost petty leveI , the warming-cooling and lhe dampness-dryness eITect
Therefore we find, in Ihis case, a granile highly sensitive lo pollulio n In lhe samples
rrom quarry is practical1y impossible to produce any deterioratiu n wilh similar
obsolescence trials.
One or lhe questi o ns ofgreater interest, in order to tesl lhe stale orgranite aI any time,
in a nondamaging manner, has been lo delermine lhe correlation between lhe ultrasonic
speed Iran smission and lhe SlOne densily In thi s way, we madc measurements in 400
ashlars. so that no\\' this "lherll1o meler" is calibrated and gives us the state of a
degradation particular ashlar, putt ing si mpl y two little terminai s over the sto ne .

Inn er-SCO I)e visi o ns.

A funda men lallopic 10 stu dy was lhe si ate of the joinls belween lhe ashlars. even lhe
slate ofthc inn er Aq ueduct, what was in the interior ofthe pillars, over ali in lhe lower
zone ofhigher pillars (and greater section).
In lhe beginning we ma de our own geomelric hypothesis ofinner di stributi on ofashlars
and o upo n taking sorne stone cylindrical samples, we saw thal the interior was also
granite ashlars with a similar quartering 10 lhal which wc had supposed.
Nevert hcless, to loo k in side lhe ashlars in lhe pillars, we introdllced an " inncr-scope"
with a TV canunera , that alows 10 save on video tapes the film o rthis inner walk .
We saw then thal the interior ofthe Aqueduct is even more deteriorated thall lhe Ollter,
just due lo lhe facl ttlat lhe races oflhe ashlars are nol submitted to washing and,
therefore, they have a more rugged surface. At lhe sarne time it is foreseeable Ihat it may
have a higher levei ofmicroorganisll1s, since lhe ultraviolet rays do not impact directly in
lhe sllrface On lhe other hand , lhe very opened size ofthe boards in some cases permits
exislence o f fowl nests, wood chunks. canes. vegelable elemenls. etc., and material drag
by lhe water and the air
It can be concJuded that it exists an inner degradation similar aI least lo the ouler one It
is interesting to observe this degradalion since lhe leve of con lact arnong so me ashlars is
not lhe originally designed becausc losses o f volume. many ashlars contacts are
accomplished in a liule co ncrete surfa ce ando in some cases, Ihi s drag mat eria l is doing
lhe work of an agglomerate alllong sto nes, like a pad for the Sllpport

Studies 011 the Slru c ture of the Aquedu ct.

The study on the slructural behavior orthe Aqueduct is being carried oul through three
method s. o ne is a vectorial analysis through equilibrium, made with a program from ou r
own offi ce. o lher is a procedure using finite element models and a possible third way
could be lIsing block s \Vith rubbing procedu re, similar whit the Aqueduct , that is ashlars
in tOllc h wilh each olher and without material in the boards, and submitted to rubbing
Through computer veclOrial analysis we first tested the behavior or Aqueduct under its
own weight. looking then Ihat it has a quile proportio nate design pillar areas co rnpared
2
to the upper weigh l. 50 slress is praclically constant around 6 Kg I crn . Abou! eol ic
transversal action, they do not produ ce practically any defreclion to lhe result ing of
312 STRUCTURAL ANALYSIS OF HJSTORI CAL CONSTRUCTIONS

inte ri or stress, so that thcy virtually inerease in a 20% his value, around 7 Kg / em2
(always tal king about globa l stress means).
We have also used this model to simulate fraet ures and to know which materiaIs would
be in volved in a possible local eraek. For this ropic there were simulated fr aelures in
pi ll ars and areh stones
In case ofa localized fraelure in a pillar, Ihere are simply stress concentrations Ihal
multiply inilial mean value by 3, wil hou l coming dose 10 compression failure average
values measured in Irials.
In ease ofareh Slone cracks, lhe local fracture in lower archs would nOI imply a ehai n
fali reaclion of ol her eOll tiguous pillars, since lhe weight over this zone is already e noug h
to slabilize lhe difference of thmst produeed between the remaining a reh and lhe split
one.
We can ' l say lhe sallle for the upper arch stones, because ifthe upper water channel
did n't exist as structura l elelllent that can transmitte the areh thrust (Iike the keyslone), a
rupture would suppose tha! the conliguous pi llar would fali , tike a "easlle of playing
cards" effect
The computer veetorial model has been used also to lesl an inlroduction of a possible
poinl load in lhe channel, like a liu le Illaintenance vehicle running over lhe upper channel.
We have tesled lha! IIS maximum supported point load value is around 2,75 tons .
Finally, atways using lhe same analysis model, we have lesled max imum supported poinl
load value over lower archs, lhe slate ofall asymmetrieal pillars, the introduelion of
post-stressed bars, etc .. In conclusion, ali those susceptibte interven lion si tuations or
possib le slIbscquent states.
We have also studied a 30·model oflower tympanum, using a standard fin ile elements
program ("SAP.90"), simply la deteet lhe worse Slress point, where mptures would
ilppear (combined slress according Von Mises erilerion), and we delecled Ihat the
grcaleS\ stress was jusI abovc lhe base areh, aI central sbaft or symmelry axis.
Rcducing even more the scale ofanalysis, we also proceeded to study, with three-
dimensional finite elements (also Ihrough lhe "SAP.90") a model ofan ashlar with a
board above and wilh non-continuous supports in it s base, due to lhe holl ow produced
bya supposed vo lume loss in lhe board and 10 the diffcrence in support plane belween
\wO eonliguolls ashlars . T he conclusions in Ihis ease lell us thal , for compression Slress
around 10 on lhe upper pan , so me prceariolls suppo rts sueh lhe ones simulated, can
provoke Icnsio n Slrcsses 30 ti mes bigger Obviusly this slress would start cracks as those
\h{l \ can be apprecialed in large pan oflhe ash lars, sp lits Ihal go fro mjoint tojoint ,
Ll"ll ss ing lhe ashlar
Tllc Slructllres Departm enl oflhe E T S.A M has collaborated in lhe slruclural analysis
through o ther heha vior Illodels. In lirst place Ihey have accomplished a therma l
bc ha vioral slud y of the Aqueduct as a global element, lhe condusions are curious sinee,
globall y considering the Aqueduet, eith er there is nOI any Iype of deformalion and simply
the c;ompression stress among ashlars inc reases unlil6 Kg / em2 (it ' s not a high value but
indicales Ihat thermal expansion produecs stress similar in quantity to own weight) or, ir
there wercn'l any Slress increase and (hcre weren 't any accommodations among slones,
there \Vould be ver)' high tran sverse deformalions (unlil 100 em), something we have not
dctccted neither have Illeasured .
In Ihal scn se, tran sve rse di splaccmcnt measurement s and rclati ve displacelllen l bclween
Lontiguoll s ashlars. tI)' "to hunt" Ihis accommodation due to Ihermal efforts .
 F. JURADO JIMÉNEZ I Roman aqueducI of Scgo\'ia 313

Also the Structures Depanment has investigated (with ANSYS program, through finite
elements and fracture s theory), lhe hi story o f craeks, duc to lhe tension eoneentrations
tha! are prodlleed arollnd the jo ints among ashlars, local thermal behaviors and
dynamieal analysis .

Move m ent s m ea surClIl ent.

Us ing lhe Illovemcnt c heek point s implanted in Ju[y 1.986, within the projects al so
aecompli shcd by the company GEOC ISA, there have been taken measurements \Vith the
same mClhodology in September 1. 992 and Mareh 1.993 .
In December 1993 !here have been made some measuremenls throughout the day with
different lemperatures: global movements (in a 30 m length) and [ocal movements (in 40
em, Ihat is lo say, reI ative movement s between conliguou s ashlars) .
Truly, we had to find a lime ofthe year with bOlh a cold temperature and imponant sun
insolat ion . We wanted to make difTerent measurc rnents trieing to find some type of
movement indiealing us th e thermal ex pansion or the re-aceommodation between ashlars
Wailing fo r the final results, it does seem Ihat we have deteeled adjustment movemenl s
between ashl ars because of lhermal variations
This would be lhe o nly rcason thal justifies slalements like "lhe Aqueducl is an
a rticu lated strueture" and making it monolilhie. can sensibl y alter its struetural behavior.
In terms ofweight and wind aetion, this statemelll has no sense, because the stresses are
mini mal and preeisely. 1110no lill1 isI11 favo rs their uniform Iransmission. However thal
doesn ' t happen with Ihermal expansions ifre-aecolllodation betwecn ashlars is possiblc,
)ince Ihi s issue scattcrs movemenl s and redu ccs shoring up stress.

Vibra li oll s ll1 ea Sllre l11 f ut.

Vibrations transmission at lhe base or the Roman Acueduct were measured o nce again .
Measures of lbe same kind were taken in 1986 bul now we have measured in different
di rect ions, to be used as a input in a subsequent dynami ca l analysis .
On the one hand a truck j umped from a given heighl beside the Aqueduet (this tesl is
normally used in bridges) to teSI hoVo' the transmission oflhe nearby traffi e vibral ion
affecls lhe Aqueducl stru cture ..
On lhe olher hand , 10 bc able to measure nearby works vibration transmi ssio n, a
jaekhammer bealing soil beside lhe Aqueduct was used .
The result ofthese measurement s is Ihat there are almosl no nOlieeable vibrat ions in lhe
acueduct since lhe leSI signal had 10 be amplified grealely 50 thal, only noises eould be
heard .

REST ORATlO N IN I' IL LARS # 67 , 68 AN I) 69

As a continuation oflhe studi es being conduetcd, lhe Castilla-León goverment , decided

10 inlervene in piJlars # 67. 68 and 69, in arder to have complete kno\Vledge over lhe
Aqueduet diagnosis and also lo have true experience in the restoration ofthe ashlars .
In first place, lhe u1traso nic tran smi ssion speed was delermined, for each and every one
o flhe ashlars. That is a tolally objeetive and eomparable measure for the fulure state 01'
eaeh pal1ieular slone
314 STRUCTURAL ANALY$IS OF IIISTORICAL CONSTRUCTIONS

DilTerenl c1eaning proccdures were experimented according lo the existing different dirt
layers and crusts. sulfates. powder and pollulion deposits, graftities. e tc., studying its
cfficicncy. yield and the possiblc aggression ofcach procedu re.
I>roccdures "bolting"split ashlars. " micro-suturing" cracks and "seali ng" fissures, were
studied.
Difl'erenl types of a rtificial granite were experimentcd to make local repairs as well as
tests of"sheathing" in ashlars with great loss oflayers
Differenl superficial consolidation procedures, depth injections as well as water-
repellenting procedures were studied in order to lest its future behavior.
The treatment for the waler channelllppcr zone and the OLlter masonry were a lso
sllIdied
Ali of Ihat was complemenled \Vith some laboralory essays so that lhe initial intervention
proccdures rcmained c1early delined

RESTORATlON IN PILLARS # 102 TO 110.

Thc C}ucslion was 10 eliminale any possibility offalling granite chunks from detachings,
restraining the ashlars degradation process, making onl)l thase indispensable
inlervention s 10 guarantee lhe union and consolidation ofashlars wilh craek zones.
avoiding as much as possible non necessary actions lhat would be later irremediable. The
following operations were aceomplished ·

the rneasuremcnt ofultrasonic transmission speed in ali ofthe ashlars, which helps 10
discriminate and loeate lhe more deteriorated stones and take note ofa "zero" reading
for each ashlar,
2 manual drag and extraction ofloosencd material alllong thcjoints ofthe ashlars, in
order to c1iminale the aggressive acid lllediulll thal means the mixture with organie
material,
non abrasive controlled cleanliness ofal! outer surfaces. through "micro-pulling"
proeedure (silieon micro-projeetion brushing and simultaneolls aspiration) and
subsequcnt alcoholated water washing (biocide efl'eel), in order 10 adjust externaI
aspect and reslrain subseq uent deteriorations due to micro - organisms and dirt
dcposits~
-t fissure scaling through pure epoxi resin injection (with identica l formulation 10 lhe one
lIsed in 1973, with a proven good behavior). so that a progression IOward a
subsequelll break is avoid, since the wcller that gets inside aels as a cradle whcn be
frozen~
5 the micro-slIwring of crackcd slOnes, returning the unity to each ashlar, through 10
nlm diameter perforations and among 10 and 30 nllll deep Introduetion ofslainless
steel threaded bars of 5 111m diametcr and cpoxi rcsin injeclion ~
6 lhe "bolti ng" of split kcy-ashlars. lhrollgh passing perforalion of 150 cm deep and 40
111111 diameter. introdllcticlI1 ora fibrc-glass boll and epoxi resin injeclion;
7 the irnprovement and regularization ofboard all10ng arch stoncs Ihat prescnted a more
criticai si tuation and of minor contaet, Ihrollgh malwfaclurc in situ of a rcsin tlagstone
not adhcred to any ashlars (using paramns as a demoulding lilm),
~ the fixing oI' ~lIperficial flagslones in damaged ashlars whcnever possible.
 F. JURADO JIMÉNEZ I ROlll:l.n aqueduct of Sego" ia 3 15

9. lh e consolidat ia n ofsandy ashlars Ihrough epoxi resin deplh injections in bulk until il
reaches lhe slo ne surface,
IO.lhe sheathing ar volum e comp leting (th rough manufactured artificial slone in situ, and
connected to lhe origi nal ashlar) in some ashlars \Vil h great loss percentage of il s
original dimcnsio ns,bei ng Ihat csential for the stability afarch ar pillar;
11 .lhe removal a f inject ia n ca nnulas and eliminat ia n af resin remains, that ha ve been
previously im pregnated su pcrficially \Vith ground granile to even the externaI aspect ,
12.lhe revisio n af masonry Ihal surrou nd s lhe channel, la avoid subsequent fa li of rubbles
and cleanliness of channel interior, discovering possible exit s a r tappings in stone
pieces.

The subsequent fo llow·up in lhe futur e of lhe deterioratio n evolu tian o f ali the ashlars
(through new measureme nt s ofthe uhra son ic Iransmission speed) willlcll us, in each
case, the opportunit y ofaccom pl ishing new interventians

Madrid , Oct ober 1995

3 16 STR UCT URAL ANALYSIS OF HI STORICAL CONSTRUCTIONS

\": ~~
'", ~.,; . ~
,:, )
'{l'

l ,.
ia •

:,f~

-
F. JURADO JIMÉNEZ I Roman aqueduct or Segovia 317

fN
o
~
G
9
o
•
§
~

.~
~
&l
o

~
\c •
1 18 ST RUCTU RAL ANALYSIS OF HI STO RICAL CONSTRUCTIONS

;" '\'
\
 F. JURADO J1MÊ:-JEZ I Rom;lIl aqucc.luct or Scgo\ia 319

F( loqrilfía il~rE'a rl(> la cludild dI-' Seaovia, dividida en do!' " iJ,rtps por cl Ar'ueduo to.
:'\20 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCT IONS

d
Ii 'II
i
I
,I

:..i
~
.!i!
.~

i "• "'.""
I ·.
~

Ii ~

~
i, "•
I
•
I E 0,
•
II ·• ~

'o
5;.
111
I "'..l•
I! •
i I ""cc
=_-l~
~

~
\:J
 F. JURADO JIMÉNEZ I Rom:m 3CJ.ucdUCI or Scgovia 321

=--~--=-=---~~----'I--------'------
-='=-.~ .

,
~

, '"
~

, i
Ó
E
1\
~,
'0
:ç
"•
~
E
"t..
-"
ó
•
~
~
"-
,o
;} ~
~

_J .', J
- -
.122 STRUCTURAL AN/\LYS IS DF 111STORICAL CONSTRUCTIONS

V~a dei NurdurIO ~ la pIaza ()lt'I'llal to! /355/867 Úlarldo rIIÔ$ orropado par /as tdi/l(atlQflf:l. tmrrgio (1)11 ~ all~rg pot tnenII drI CQSm:I
la~ mlubPdcJ SOpl'1lOl st1à cl,m,fIOdam /973 IRt'pfOdoc{Oo" M 45125 m, MI arrh'llI dtoJ u...,ml. jgtMadG /IOf ti U'nlslmelM Culrurg!

V,S/a di.'! Aruo.voclu dndr Ja nHlrglJa ell 1868 1867 la Ja d/'r..dlillc plc14 dtol Al.ogu..'p! i* putdr 00sen0u UI! alaW< dtrg/k ..., Ja. arulas M PIlas Ydwr
los /wJo,>dlO. /30 1IrIOI ~ mas,~ Ilkprodur.1OII di> 4.Sâ5.m dtolotrh,ooMJ I ,l1Ift'llr.III"~III1I,,,.....,1 M,n~leooM Culluro)
 F. JURADO lIME 'EZ I Rom:lIl :lqueducl of Sego,,!:!. 323

AlwdcJS \. :>oW',"_1 ,~"j,jl",I(JII",I,·. ,k wl<l jJllu(kl .. \,,~'(ÍUfl,' (O" Mp_W' dt' s<11o,~s l'IlIb~ l· "Ol."hk> o.
bu~,i.>1 (Ju l"fllm," ,i.> ml(JC1'it'l<! 1"I'ltl'Qn~'!rf(l

PIGnogr~",*, Iosrotntf«llOIIdus",
197J~tfC1lrllllOOn.:lr"'.."....flI_f\IfII/'
""trIO<
<If~/o.p
'* ""'Wf*Tru' rM.. ·rnfd>1"O' d
lrm roNlh~
 324 STRUCTURAL ANA LYSIS OF HI STOR ICA L CONSTRUCTIONS

- . ...il!...1q) I CAJA ' PX,P

I ! J T-~
." ","-
,o p

"p
r
L
'l
'o P

r
~,

I~~--

I "P - --{=E:J ~9H

T
" P 8

"
~p
1}2~

6}L
I y.p

"

1---+;'-, 5 o6 H
I ~,

"
r p
I }~

71 p ,,. ,
.3 601H

~ '}lE..
Y4P •
;---
I
1~P
N IV[L AC rUAl OCL
T(RR[t<O ( 0 00)
, !
'-:.!: c;:.:.= _ .1"
·i- --~"'"
4-- ,,
,,,
D.mtflSo->::ts dti Acwau..f(I. QWlórl.-,. t<I potI de lIOOl. ."'0 fM 5t9I'"
tll,~o r....,1o$ ft •..andtz CG\OIXI IIU!I,GtOUJl f Woedo IkIlrtwo
"1\(..,0,'-100 H""lOl>U$i"II f \jllll'loI
 F. JURADO J IMÉNEZ / Rornan aqucdutl of Scgovia 325

,o o 60
I t
__----1-_
I

r
-

_
~.,
N

'o
:[
I-
: l~
. 10 ., o
"

-r--]l

:r I
I
I
I~

" o
e

J'IOflO'CIO!~1 'l'Ult! dI!! t.:~~uctode~'IO. tlUI llaWlO' ,tcl,zoda r'«l>d> nlftl.~"'1l'I /(ltoso'am.!trict.l Como lO' putd.. COn'IjXlICr. k>s "'~, tllOfe$ <k f..,lIIndd ÚlJOdo
rn la:! n"J Ja! t,Ot\l.I<tfIQ!;-s. $/lb.-rlOdot"1 /o'ar,oda IUpeno> fdl"'''HIt~lI4Il1>r W'IJ:or(" OI"{lio,Jo ('ri p!ftI
326 STIWCrURAL ANALYSIS OF HJSTORICAL CONST RUCT IONS
 F JL;RAI)O JIt\lEt'\EZ I ROJll~ln aqut:ducl 01' ScgO\'ia

. ,~ ..

T-S'

F'Oloqrafía de uno de los testiqos extraídos para anallzdl

1 estado de los bu lonados con la tôn y res I na de 1973.
(>
328 STRUCTURAL ANALYSIS DF HISTDR ICAL CDNSTRUCTIDNS
F. JURADO JIMÉNEZ I Rom:m :lC)ltcduCI or Scgovia :129

•, ,!
> ,
•• ,,,
9 <
: 00 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

Endoocopio ul,lirodo pc'" in!pKCionor d inlerior dt las pilo:s tnlre 10$ Jiflorn_ (Fologro/iII adido por GEOCJSAj

I.... de • grc!:odo a Irnt.ts dei tndosropio. dorr&e 50! percibe uM IIormigo mio ;.mlo mire Jillo.es. (FOIogrofio miida
por GEOC1SAJ

Microfologrofio de grnnilo Ulrvido «I AruedOCIO. $( obs(n.o mual dt pIogiorlaso 011=00 YfiJIJrodo. ~IdespaIO porésiro ror! _ 'l"J'
do de- fI/frIIción; !riotilG y mo5COIIÍlg fFOIografill mruida dt:! nlooio dt lo pird", ef«luodo por GEOCISA).
 F JURADO JJMÉ NEZ I Rom:lJl :lCJ"edIlC{ of Segovin 331

Hestitución Eotogramétrica en tres dimensiones .

~n ST RUCT URAL ANALYSIS OF HISTO RICA L CONSTR UCTI ONS

I'IIA 'U'

• •
\ li
o "---\
~
,' :-Ti
~

I
~
'~ ·
'"
o',

"~
5"'"'~
1-
"" - Õ'_

~.
- i

I , I
" - [t '
- .\ / .
! 'IJ';l-T
1 I
I
" C;-' '

I
I _. , .; I
, I '
, '
, ~
,
o o
'. . I
..<>.-' J,'
~
I
- -
O
" ., I .
~

I' , 1-)
-/

I FI J

a
" ~"'j
:..-' ..:...~ . I .
. ..../.. ' .
b I(-~" . \
._ r--.'--' ' f, ,.

.
O
r2' ,.
I. . , I

.
~ . ~ ~ ,

Im
~c " u 1/1(\1'
 01

",.
C

8
~
zm·
rn
N

'"o
""
o
,.5
~
~

~
2,
,g'"
o
<
s·

Es tudl 0 del çomportaM iento estru ctu ra l medlante Análisis Vectorlal Automático , método disenado ~
.~

es pec lalmeOtE po r nuest ra oficina (expuesto a partlr de entonces en múltiples simposiumJ. .~

 .~

~
A

~
-;
2
n
-;
c
"r>
>
z
>
:;:
~
;;;
O
-n
J:
;;;
...;
o
"
~
r
n
O
z
'"
-i

"
C
n
-;

~
~

Med1 d nte prog ramas de Elementos Finitos se ha estudiado el carnoortamlentO térmico. el çomportamlen~0
ante vibraciones y el p r oceso de fo~ación de qrietas y fracturas locales en slllares.
 F JURAIJQ J IMÉNEZI ROIll;lI1 aqllcdutl of Scgovia 335

z
S
•o ~
«:
;; :::..,• hl~
O ~
c:; ~ ~
W ~ ~
~ ~
rf)

~
~
W
Cl ~
O :111 8 ~
b ~ ~
U :l :l
:::> z
~
Cl 8 u
w •z •z
:::> 2 2
U
«: I
I
:\36 STRUCTURAL ANALYS IS OF HI STOR ICA L CONSTRUCTIONS

La fotoqrameLria t ridimensIo na l está asociada a una base de da tos en la quc

cada si Ilar (id,'nLificado biunivocamente con ocho d í gitos) tielC su propia
ficha donde sc anotan cnsayos, intcrvencioncs, estado cn cada ilpoca, ..
F. JURADO JlMÊNEZ /ROIll:1Il :1quedUl':\ 01' Scgovi:1
,,8 STRUCTURAL ANALYSIS OF HI STORICA L CONSTRUCT IONS

8
..: ~
5: >
O
(,)
"-l
~
<fl ~
"-l
Cl
"
~

O fll •~
f-
U
:J ffi
Cl ~

"-l
:J
a•
U
..:

:-.. \1..... ~
lo •
~.. .' •
 'c"
'"
»
g
3:
'",
z
I:J
8'
g
D
.o
~
~

'"
~
n
,"o<
Õ'

'~
~
Además de anal~zar el compor t am1 ento ante las a ccio nes hab1tuales. se ha estudiado la poS1ble
inc1dencia o efecto de posi ble s roturas parciales .
'"
340 STR UCTURAL ANALYSIS or I-IISTORICAL CONSTRUCnONS

l ' llJ\ S 11 ') , 1 14 Y. 1 1 -' , DONDE AUN NO S E HA INTE RVI::Nl OO .

E; l'; n~l.:es dr io
comple t a r 1<1 pr i mera l l l t e r v'-' I lc i ól1 lIrqc nt e en
t o da la zo na monllm" lII. ,\ I (h~ l Ac uedu c:to , e ll I n c on cc rniente
ü timpi e 7.a , r;o s i do'1 y COII Rn lidaçi o n C's .
 STRUCTURAL ANALYSlS OF HI STORlCAL CONSTRUCTIONS
P. Roca, J. L. González, A.R. Marí and E. Oiiate (Eds.)
© C1MNE, Barcelo na 1996

TECHNIQUES IN STRUCTURAL INTERVENTION FOR THE

REHABILITATION OF "LAS GOLFAS DE LA PEDRERA"

R. Brufau
Ull iversitat Politewica de Catalullya
Calle Herzegovilla 25, fntlo.
08006 Barcelona
Spaill

When , a few 11l0n lh s ago, I \Vas invited to parti cipate in a Seminar about lhe
Structura l Analy sis of Historie Buildings, we agreed Ihat lhe lille of l1ly paper
wOllld be 'Techniqlles of Restora tion and Rci nforcement ', Wi lh the lInderstanding
that I would not give a general description of techniqllcs anel analylicalmethods, bllt
of the manner in which I understood, subjectively, them to be applicabl e dependent
UpOIl lhe circumstances presellled by each particular case(l). Arrhat lime J had just
won the contract 10 strengthen the structu re of the upper part of La Pedrera, a well
known building by the architec r Antoni Gaudí (1852· 1926) built in Barcelona at the
start of thi s century and that ha s beco me lhe edifice mOSI symboli c of Modernist
Calalan ArchiteclUrc.

Accepting tha! there was a ccrtain di stance o f this from a nd lhe or igi nal title of
lhe papel', bllt being confrontcd \Vith a complexed analysi s to be undel1aken for lhe
cxeculion of the new projecL lhar wou ld involve cOllsideralion of a diverse range of
oplio ns of illlervention, for me lhe Iheme acqllired an exceptional inlerest. Thi s
interesl was intensified by lh e deba le stirred up in lhe mcdia wlth respecl 10 lhe
rehabil italion of 'I as Golfas ' , or thc lofl(2), o f Ihi s building. For Ihi s rcason I
proposed to lhe organisers of the Seminar tha! they wOllld let me concelllrate my
paper on the anal Ylica l mcthod s uscd dllring the progress of this operalio n ..

The Pedrera, a1so known as Casa Milã, was constrllctcd between the years 1906
tO 1910, as a private hOll s ing promotion. silualed in Barcelona ai lhe cross road
between lhe imponant Ihoroughfare Passaig de Gracia and the Calle Provenza. Its
'Golfas'. lhe same ilS its lerraced rooL we re only used fo r the locilli o n of se rvices
for lhe c:..'omm unily rt'"s idenl in the resl of lhe building.

The greatest individual qualily of the 'Golfas' is 10 be founel in its eleganl

srructural (·onception. thal is based upon 265 diaphragm ardes made up of panitiotl
masonry(3). The parabolic elevar ion of Ihese arc hes adapls in each case 10 lhe fonn
of lhe particular terraced roof above. as much in lheir heighl ( of between 2.8 and 3
mctl'e s) as in the di stance belwcen their bases (of between 3 and 9 metre s). In plan
they are distlibllted about the IWO principal wells in lhe building. and al so have the
additi ona l orientalion about six big interior stair towers. TIle arches pal1icipate in an
extrao rdi nary exprcssive dialogue relaled in lh e sing ular range of connection s
belwcen (hem in Ih ese. Ihm define s a sp,l('e that is absolu lel y unique and
unrepealable in Catalan Arch itecture.
:142 STRUCTURAL ANALYS IS OF HISTORICAL CONST RUCTlONS

The arch system cansi sts af a number af variatians: The first af which is found
in their thickness; in the majority af cases lhe arches have been eanstrueted Ou! af
4.5 em. sol id brick, howeve r in lh e cases where Gaudí considered it necessary he
doubled ar even tripled them(4). In nearly ali lhe cases the arch was inirially fonned
from bedding planes of lhree consecutive ring ~l ayers of bricks reac hing roof leveI.
The body work was lhen completed with horizontal layers of lhe sarne Iype of
brick. A seeond variatian is faund in lhe horizontal djrection thal lhe arches project;
euriollsly many of them do nol define a straight line, but in slead , each in turn,
progressively angles at lhe key Slone so as to 'poiOl ' la lhe nearesl slair tower, to
whieh Ihey are slruelUmll y subordinate, they do lhis withaul laosing their identily as
an areh. The co nsequence af thi s movement in lhe arch plan was that Gaudí
introduced vertical partitions thatjoin up ali the arch keys, a 'spine' that established
at lhe su me time a way of absorbing the transverse compress ion forces that result
from these frequent chan ges in lhe arch plano Relative to lhe dimension of the angle
in each arch the Ihiekness of lhe vertical parlition was changed. an effeet tilat gives a
spec ial sense af order and cont ra i as Ihey dislribute the horizontal comp ressi ve
forces lhal are calTied by lhe lotaI arch syslem.

A third variation is foulld in the shape that lhe arch draws. Ir in the majority of
cases they describe a parabol ie or chain cU lVe. one cannOI speak of a geo l11etric law
in Iheir elevalion. Iheir indi vid ual form is defin ed by lhe demand s infli cted on the
arch by the grcater splendor in the funclion of th e flat roof above and from the
distributian, in plan, of the inner well s of the bui ldin g.

The location of the archcs is variable as well. wi lh a di stance betweentheir bases

of nOI grcater tilan 90 ccntimelers. bul with changes often motivated 011 one sidc by
lhe freq uenc y of the windows, anel on the olher by lhe struct ural demands of
COllcentr.J.ting large nllmbers of lhem in lhe areas where the arch system had to carry
the greates l compressivc forces. As a result of the cireu it plan, detenllinate aches
join in groups of 2. 3. 4 , OI' 6 at thci r bases, separating oul as they travei upwards,
wh ile olhers stm1 off at a distance apal1 join logcther aI their key co inciding Ihis with
their eonnect ion 10 the stai r towers. ahogether makin g up an in leresling ' palm leaf'
designo

One aspecl tltat does not cltange in lhe whole con struction of lhe ' Golfa s' is Lhe
texture and qllality of the brick lIsed. The general conslruc ti an of the bui lelin g is of
slone columns and steel beams. In sto ne pill'IS. dependenl upon the individual use,
up 10 five different types of limestone have been detec ted. Where as, the upper pari
of lhe building is of struclU ral masonry, wilh only one type of brick(5); lhe same for
the arch rings, as for lhe body work of lhe arches, as for lhe vertical partiti ans and
lh e cu ler coverings af lhe patios. Thi s aspecl confillns the idea Ihal Gaudí destined
lhe 'Golfas' as a zone 10 hOllse servi ces la the re st of the property. The total
absc nce of finishings OI' superficial Irca tmcnts manifesls Gaudf's wish to exe mplify
lhe simpli city of this spacc.

In the yeal" 1954 the 'Golfas ' of lhe Pedrera suffered an importam remodelling.
with the formation in their interior of 13 mezza nille domestic flats, arranged 0 11 Iwo
leve is wi th large areas of twO storeys that took ad vanta ge of lhe height generollsly
 K. BRUFAU NIUBÓ I Las Golfas de la P\!drcr:l 343

provided by the 10ft. giving these rooms the certain characleristic of being parI of
only one atmosphere.

The remodelling was contracted to lhe architecl Barba Corsin i(6). Barba applied
a very different interpretation of building. defined by a more cuh vision, to lhat of
Gaudí, in that he smoothed off textures and only took advantage of the overall
exisling formo I-Ie perforating the roof with 13 new chimneys thal coexisted with
greater o r lesser s uccess to thase projected by Gaudí. he increascd lhe number and
width of the exteri or openings by creating new windows and widening some of lhe
old ones. adapling lhe anes at lawest leveis inta doorways for access to the
perimeter balcony aI lhe lower levei af lhe 'Golfas'. In order to increase the useable
plan area in lhe apanments he cons tructed mezzan ine Ooors of slender thickness
with slee l beams su pported by the covered up arches. and to increa se the storey
height cut away any origina l arches thal ITIolested him. evcn e liminaling those Ihal
did nOI adapt tO the new arcllitectural func tion, substituling them with olhers newly
constructed in convenient posi lions.

If you lake inta account the 40 years lhal have passed since Barba's remodelling
lhe space; rehabilitation that almosr removed ali trace of a gaudian manifestalion of
a unique treatment af exposed brickwork and a great sing le space(7). Ihal had
broken il up ioto 13 small arcas. thc walls neatly covered with pla ster, tiles and
woodwork .. and also lhe reduced amount exislent pholographi c evidence of
Galldí's image. il is reasonable 10 suppose thal a good number of generalions of
architecls do not know tbe pOlenlial beallty hidden benealh lhe covered arches and
lhe splendor o f a space sect ioned by the horizontal CUIS caused by lhe mezzanine
floors.

In thi s contexllhe auilude of the Entity lha I owns lhe propeny is undcrstandable.
in thal they are dedicated to the recovery of whal is already known as 'Espai
Gaudí· (8). The preliminary agreements taken belween the Enlily lhal owns the
propcrty and the local Adminislrations which have urban responsibilily for the
Pcdrera. agreed Ihat it would be convenienl if lwO of the thin ee n apanmenls
constructed by the archilecl Barba Cos ini would be conserved. with lhe desire 10
maintaill the memory of a nOlable intelvention. thal in ils day was even considered
an excellenl arcllilcctural exarnple.

The reconverting of lhe 'Golfas'. ioto the future Gaudí MlIseulll and lhe
recuperalion of the Oal terraced roof above as a pllb lic spacc, bolh areas 10 be open
to the Illany members of the public that dai Iy visit the Ped rera, inlroduced a new
aspect in lhe slrengt hcn ing of Ihi s 10ft arca. There shou ld be a load capacity as
defined by lhe relevam codes of practice cancerned with the loadings lha! are
innicted by whal is now tcrmed as a public meeling space. This loading is
increased ir yOll take imo account lhe rype of lourism visits envisaged. Ihm is guided
grollp lours witll large numbers of reople confined into sllla!! arcas of 0001" space.
The requircd load capacily Ihat was the basis of lhe original slnlctllra l design for lhe
upper le ve is of tlle Pcdrera. calculated fram leSIS executed willl res pec t 10 lhe flaor
section. indicale a capacity af approximately 400 Kg/m 2. 111e self weighl of lhe
floor was assessed as 200 Kg!m 2, tllat leaves 200 Kg!m 2 capacity for the finishi ngs
and live load an amounl complelely insllfficient(9) for il s new archi teclural purpose.
344 STRUCTURi\L ANALYSIS OF I-II $TOR ICAL CONST RUCTIONS

Wilh respecllo lhe roor. lha I wOllld cOlllain lhe lofl area as a public space, Ihis
brought lhe necessity to provide adcquale thcrmal characlerislics to the area, lhis
would mean the doubling of lhe facing brick , -in both the vertical anel horizonlal
directions-, 10 obtain an internai gap for the cli mate condilioning. Thi s gap would
bc used for a trip le purpose; 10 contain the climate insulalion panels. for lhe passage
of lhe roof drainage syslem. anel as a struclural elemcn l for the stabilisation of the
arch syslem, an aspccI to be commented lIpon later 011 in this paper. The result of
Ihi s would be a celtain increase in lhe self weighl that would exaggeralc even funher
lhe inadcquacy of lhe load capacity of the floor benealh lhe arch systcm itse lf.
Taking into accollllt, in the final analysi s, tllat lhe installalion of lhe
impcnnea lisatiol1 of the tcrraced roof would funher increase th e self wcighl of this,
with it, the inevitable increase in its service Joad, draw$ us 10 I'cpeat Ihc prem ise
init iatcd by Gaudí himself that brick arches shou ld be complete ly structurally
revi seel if they are to be adapted for a new pllrpose.

Whcn an architect is pUI in the position of strcngl hening a blli lding classified as
"of Hi storic Cultural Patrimony". and what is more symbolic of a definite era, they
are co nd ilioncd and initially confined in Ihei r dcc isions by limil s thal lhe bllild ing
ilself im poses; on the onc hand. in tcnns of possible const ruct iOIl, and on the other
by the cvidcnt expression of ccrta in rulcs thal make up its sty le. Some or other of
Ihese conditions are defined in the hi sto rie pcriod from whieh lhe architeeture
comes.

In the firs l place, an arehilect c<ln conclude Ih al they C,lI1I10t confront these
conel ition s without causing an incompatibililY wilh, preciscly. what makes up
essc ntially thm which, in some cases, they should be restoring, -if a blli lding has
su ffered the distortions of manipulalions by someone, or st rcng th ening by
so mcbody clsc. -if lhe bllilding had been able la survivc faithfll1to Ihemselves,
despite difficlll lies over a long period of time .

But !bis is nOI the only valid argument, in Ihat. from the Illomcnt Ihat he is
co ntracted with lh c rehabi litation, lhe architecl can suppoS<! that their c1cclion is
based on accepting , on lhe part of lhe propeny owners, their 'mallner' of \Vorking
and projcct design ing. Thi s can be eleve loped until the point thal the rcason for lhe
contract ca n be conveniently deri ved from lhe acceplance thallhe arc hilect brings
thei r professiona l knowledge and will incorpora te Ihe ir desig n experience in their
intelvention, thus cnabling a more perso nal anel subjectivc altitude less dependem
upon condilions thal are slrictly hislorica l. Relcascd from historie responsibililies
thc arch ilecl can 're inlcrp rel ' lhe bui ld ing. T his is lhe rOlltC Ihal reputed
professionals(lO) have taken \Vith a certa in frcqllency, somctillles with success.
I-I owever, there is an cvidcnt risk lakcn when lhe projcct designer, 11l0tivaled by
thejr desire to manipulale lhe buileling, lakes irrc versiblc dccisions whose exeç ution
does not enablin g fllture reslOration of lhe original csscn tia l charactcris ti cs of lhe
building.
 R. BRUFAU NIUBÓ I Las Golfas de la Pcdrcra 345

Thi s is lhe COllh!X I Ihal frames lhe projecl to re habilitale of lhe 'Golfa s' of lhe
Pedrera. The i1l1porlant mani pulatio n that the arch itect Barba Cosini executed in
1954, appears 10 de ny, under initial examinalion, lhe option of a resloratian af lhe
10fl area as a fUlut"e Gaudí museum. Given the implications af thi s, vo ices stalled 10
so und in ali directions:

-The importanlman ipulations of Ihat in 1954 the arch itect Barba Corsin i
ulHJ c: rtouk makes il maleriall y impassib le to restore the gaudian
e lemen ts ..

-Barba Corsini' s apartments represents a unique and appreciable era in

lhe hi slory af Catalan archileclure and as such shou ld be preserved ....

-li is very impollan l 10 maintain some apaJlments as a recard of lhe work

of Barba Corsin i..

-Maintaining same af the apartments wi Jl impede lhe total and complete

vision af a great space designed by Gaudí.. .

-The express ive force af lhe gaudian arch iteclure, manifes! in this space,
cannot be sacrificcd by the partial maintenance of architecture af much
lesser interest.. ..

- Befare the stan a f th e reslaralion pracess a greal number a f the

apartments were faund abandaned and in a state of deleri a ratio n. Those
Ihat are still in service are used for storage rooms ar stockrooms for some
a f the offices below .... What ever lhe purpose th ey deal with an obsalete
usagc ....

-Th e rec ord af lhe Barba Corsini aparunents make s up parI of the
'memo ry' of the bu il ding, occupying a position of honor in lhe future
Mu se um of lhe Galldí space prese nting them Wi lh lhe ane ntion Ihal Ihey
de serve as a represenlation of a bygone archileclure, bringing wilh Ihe m
photogra ph ic material, models, furnilure etc ... , bUI never mai ntaining
them in their actual latal state ..

-There is no comparison, Barba Corsini is not Gaudí..

-Gaudí's work reconslrucled is not lhe same as havi ng the authentic

work by Gaud í.. ..

Ali thesc conjeclures I ha ve read ar listened to during lhe first months since the
start of the process af restorati on. As lhe alllhor of lhe reslOrati on projecl, frol11 lhe
momenl lhal lhe propeny owncrs gave lhe go ahead, I had as the ultimate priarity
lhe lotai rec uperat ion of the space proj ected by Gaudí. Despite lhe prev ious
agreement lO mailllain two of lhe thineen apaJ1me nts, I should really concentrate my
attention on olher aspec ts of lhe project:

- Once lhe apanlllents had bee n dismantled, plaste r and fini shings
removed, it was revealed thal more than 60% of the ori ginal arches had
becn muti laled du ring the 1954 rehabilitation , in most of the cases cutting
346 STRUCTURAL ANALYS1S or HI $TORICAL CONSTRUCTIONS

away one of the arch rings to gain height in the flaIS wilh two storey s. In
other eases they had been cut again. up to 50 em. in one case to allow the
entrance of the kitchen furniture or wardrobes. The cuts were rarely
undertaken bearing in mind the maintenance of symmetry. causing lhe
unbalanc ing of lhe arch. 11 was counled Ihal nOI less Ihan 40 arches had
had the ir Ihree rings removed and in one case ali Iheir load resistant
funcl ion was confined 10 lhe masonry lhal made up lhe arch body work.

- Another important con sideration is the way in which lhe mezzanine

loads. generated by lhe double slOrey fia is, was reso lved by Iransmitting
the forces in to lhe surrounding arch iteclure bearing in mi nd that there are
no load baring walls or even paJ1ition walls in ali lhe open floor plan.

The suspicions wilh respeel to th is llIrned oul to be right, in Ihal lhe steel
members in t!tese mezzanine seclions were directly sUPPOl1ed by lhe thin
wall arches. entering head on in some cases and in others obliquely. In
very few cases it was observed that there had been installed elements 10
spread out lhe loads between a few arches suppons. The anti -rusl
protection of these small metal profiles was in scarce evidence, this factor
had caused problems of degradation of the end s of these beam s. thal had
resulted in lhe c racking of lhe arches il1lo wh ich Ihey had been plugged.
due to lhe expansion force of lhe sleel as il oxidised. If we suppose (hal
lhe profiles had been designcd -prabably to minimisc Iheir outline in the
tota l seclion - withoul imposing conditions of tension, il expl~ins the
bending thal has beeo detected in some of the mezzanine leveis tbat have
caused an accentuation in lhe provocalion of structural degradation at the
connection between some beam ends and lhe supporting arch.

- Many arc he s have an aspcct af profuse cracking(ll). This cracking is

found as Tllllch in the lhe arches thal receive the loads of lhe spl it leveis,
as those that have suffered the remova I of some or al1 of Ih eir rin gs. or
have suffered some other cul. Cracking was especially nOliced in the
ardes Ihal had becn cut imo in an asymmetrical way .

The form described by Ihi s group of cracks manifested a range of

struclural reac tion, fram lhe type of crack thal indicates lhe exposure to
excessive compressive forces, a loss in lhe arch height and exeessive
craeking on one af its the two sidcs. To cOlltinll olls longitudinal cracks
in the comJ110n monar layer bctween the .uch rings and the body masonry
providing evidence that eaeh Slnlctural elemen l was doing a diffcrent job.
Al so cracks were revealed in the tran sverse pal1iti ons of lhe bu ilding face
thesc are attributed to lhe excess deflection of lhe base Ooor. benealh lhe
arch system. tha! had taken on more load Ihan it was designed for.

Especially. adding to this tl1e d iscovery of horizontal tension cracks in

some arches. with a separation of up 10 4 111m between lheir cdges and
with import ant tangential sliding in lhe direction af the particular arch,
Ihis prese nls evidence of structural loading far removed from Ihal
OIiginall y intendcd by Gaud í.

· One of the Barba Corsini apat'lments. 10 be precise the mOSI significant

.lJ1d Olle of lhe two which will presumable be conserved, was recently
renovated to ael as an affice spaee dllring lhe restoration wo rk s,
rcpaint ing the \VaUs and structure. Considering lhe liule lime thal elapsed
since lhe rcnovation. iE was worrying to obselve the s[art and ad vance of
 R. BRUFAU NIUBÓ I Las Golfas de la Pcdn:r.t 347

some cracks. So rec.:ent1y a series of 50 experimental tests were carried

out in Ihis and lhe olheI' lO be conserved flal. wirh results that plainly
confinn those of the earl ier analysis of the struc.:lure after lhe dismounting
of the rest of the apartment units.

-Ali of the load baring elements should be reassessed, starting with the
tloor bases. that will be adapted for lhe new load system. fol1owed by the
mezzanine leveis of the apanmenlS that are to be c.:onserved. clearly
inc.:reasing their section dimension, and finishing with lhe arches
themselves, tha{ have been effected by lhe intervention of 1954.

-The general stability of the 10ft storey is thoughl lO be guaranteed by the

straleg ic location of lhe six stair towers tha! channel lhe transmission of
horizontal forces to the principa l structure. What ever the case a
continnation in respecI to this will be necessary. presenting lhe provision
of the option of extending until the lower floor some of lhe middle size
ventilation chimneys, lhat in {heir day were supported their.

Undcrstanding these conditiol1s, the first difficu1ty encountered has been to fix
lhe force limits that can be cOl1sidered acceptable for the masonry design lhat will
make up the slim wall arches. Dcspite, having lhe laboralory analysed strength
parameters from test samples taken Oul of the strUClure itself, with the considerable
influence of lhe [954 inlerventiotl and the alteration tha! this inflicted 011 the general
aetion of lhe whole unit, it is extremely difficu!t to relate the craeked slate of the
element with the pennissibte tensile force of its material. Also tllere are no reliable
resu1ts available for the delennination of the longitudinal Young's 1110dulus of
elasticity given that its detemlination is based on data, from extracted samp les, in
whiell there is not sufficiem confidenee so as 10 allow ea1culalions to be based upon
it.

Confronted witll (bis situatiol1 we have opted to verify the structural behavior of
some of the ardes -individually and togetllcr- that have no! suffered any dalllage
during the length of nearly 90 years Df life, despile having been lllmilated OI'
redueed in !heir number of their arch rings. Designing, in accordance with the
relevant codes, for horizontal wind aetion combined willl the maxirnum vertical
forces Ihat they have undergone in lhe length of their hi slory.

Techniques based UpOIl finile elemelH method have been uscd to obtain the
loading conditions mos! unfavorable for cach of the arches modeled. Once having
colleeted the data from a significam l1umber of them estublishing the values of
maximum eompression force that ead peace eould SUppOll before cracking lOok
place. These res ults have, in each Case, a faetor lha! re lates [hem to lhe individual
areh height. thus the effeets of slenderness were analysed in the load resistance of
the arches(J:!). This allows lIS to calculate the real resistance of each slim walled
arch, subjected to the most unfavorable conditions, as a functiol1 of its heighl and
thickness. Understanding how 'resistance' tO lhe ultimate compressive force, is
established at whatever point, as a re;:lction of that pan of lhe element.
348 STRUCTURAL ANALYS IS OF HI STOR ICAL CONSTRUCTIONS

From lhe strength values af many af lhe uncracked arches, obtaining

representalive values for each cne af lhe intervals af slenderness. These values
were taken as reference points for fulure structural analysis. They were used for lhe
structural design af lhe rebuilt arch section, subjecl to lhe proposed increased live
load added to lhe gravily loading, which had beco lhe la rgesl load Ihey suffered
before.

During lhe first months lhe project. wh ile lhe white and clean 'p lastered' image
offer by Barba Cosini's apartments still reigned, we debated and tested up to five
sol utions for lhe slrenglhening af lhe arches, creating an 'in situ' made l 011 a rea l
arch af each aplioo.

- Th e first tWQ oplions involved lhe use af metal sec tions of small
dimensions , eonlinuous on the front edge, following about the d irector
line of lhe arches. The first, wilh a CPN-80 seclion connecled lO lhe lwo
arch rings found in lhe majority of lhe arches with lhe dismounling of lhe
old apartments. The second, with a 85x 10 mm. plate with a number of
lateral bars of small diameter Ihal are fixed in IwO small layers of high
resistance mortar, of 2 em. thickness, radial to both sides of the exisling
brickwork for their whole lenglh of the arch shapc. These two so lutions
would both involve lhe addilional finishings based on paim or plastering.
bul have lhe advantage in Ihat Ihey only make use of lhe arch ruins,
resuIting from the 1954 inlervenlion , wilh little rebuilding . The
maximum disadvanlage was Ihat they increased lhe sl im walJ arch
dimension, by up to 10 CI11., totally disfiguring the original refined
hal1110ny invoked in lhe space by GaudI.

-The Iwo following options do not make use of melai profiles due to lhe
difficu lty of adapting Ihem to the existing arch shape(14), but instead
involve lhe emp loymenl of high strength mortar, lon gi ludinally
reinforced, pou red into thin plate moulds Ihal are easily adapted 'in situ'
to lhe specific curve of each an:h. The difference between them lies in lhe
way in which Iheir shape works with lhe rest of lhe masomy arch. These
IWO oplions require, as for lhe IWO before. finishings of plaster and paint
10 hide lhe texture of lhe morrar, which was incompatible with lhe
existem brick work. Compared to the IwO metal profile oplions, ir was
found that the final thickness of these mortar opt ions is a lot less. in Ihat
one af them reached a Ihickness af 65 cm .. Also averall advantage was
that these procedures could be directly appl ied 10 lhe arches after the
demolishing of lhe apal1ments.

~ The fifth oplion, which was the one finally se1ecled. was based upon lhe
restoration of lhe space as Gaudí intended it. Recon struct ali lhe arches
unlil they had their original shape and Ih ick ness, based upon abundant
photographic documentation collec led . A brick fabricalor wus brought in
to lhe project to study the texture, form, colar, and dimcnsions of lhe
three bricks used by Gaudí. wilh lhe result Ihat we cou ld imiwte lhem as
accurately as possible. Resistance tests were carried out on lhe 'new'
brick with satisfactory results. Also mortar was found that would be
adequate for each pasition in lhe new so lution; as a fUllction af the
necessary sU'ength and ilS workability.
R. BRUFAU NIUBÓ / Las Golfas de la Pcdrcra 349
350 STRUCTU RAL ANALYSIS or HI5TORICAL CONSTRU CTION5

UnderSlandi ng ilS fUllll'e si luation, eac h arch had lo be redesigned so that it co uld
res pond lO the ncw demands made upon it, thi s mean t lhe increase of va rious
q llal ilies:

a). Resloring, by rebuildin g lhe three arch rings, th e open arca designed
by Gaudí.
b). Increasing Ibe qualily of lhe arc h s tfllcture. by adding, in each case a
minimum of one rin g layer af fab ri cated brick and mortar af hi gher
strength th an the ori ginal materi al. In each case, the new brick work
ensured a characteri stic resistance Ihroughout of 120 Kg/cm 2 , obtained by
the use of mortar af a higher resislance.

c).Rebu ildin g, with qual ity monar. the jainls betwcen the exi ste nt arc h
rings and lhe bOdy work so as to enSllre that lh e arch reacted to lhe
innicted loads as a single element.

d ). Posilianing in the see lion af ali the arches, a co ntinuai 12 mm .

diameter rc inforeing bar, located in lhe mortar belween tbe fir sl and
second layer of th e arch rin g. In order to be able lO locate lh is bar a new
brick des ign was devel oped wilh a ehanncl in one of it s bedding planes .
The reinforcement bar was 'enveloped' in the can nel by high capac ily non
shri nking mortar used in the beddi ng plane. Th is bar gives the g ll arantee
a f absarption af the hypathelical, but improbable, tensilc forces that could
appear in the arch lInder abnormalloading(l)).

e). Redllc in g lhe slenderncss af cach arch by inco rporatin g in il a

transverse patlilion madc of lhe same Iype of mason ry, located pamllc llo
lhe face of lhe building. stabilisi ng the third, inner. arch ring . By Ihis
innovat ion achieving an increase in lhe active radius of cllrvature af the
arch. and co nsequently ca using a reeluclio n in il s slen dcrness. Tbi s
pan ition enabled lhe possibil ily af an air gap lhat served a trip le fllnction:
hiding the passage of waste and drainage pipes fram lhe roof levei,
containing the air condi lioni ng mechan ism and pane ls. anel aeting as a
thenllal filter.

Definitc ly accepling this so lution, and know ing lhe gealllctric dimensio ns and
slrength of lhe materia is, the next sle p was 10 apply certain lests on lhe bondage
sLrenglh between (hem. To this end lhe folla wing procedure Ims bcen followed:

t. Simple analysis of lh e arches: Using the sal11 e ana lytical method, as for
fi nite element s. employed in lhe init ial tcs ting of lhe arches tha! had no! sllffcred
cracking. Ir is verified Ihal, with the ncw loading systcm, tensio n does nol cxceed
aI any poinl in the arc h, [hase impased by lhe slimness af lhe ori gina l desi g no (hi s
imposed tension was consi dered permissible. In thi s way aceepli ng the arc h as a
valid structu ral e lement. If lhe max imum ten sion exceeded those in lhe orig inal
des ig n lhe thickness was doubl eel, f irslly wilh li laye r of 4.5 em. Ihic k brick,
repealÍn g lhe tcSI with the new dimensio n(15). In the case lhat it was still found 10
be insuffic ien t increasing the thi ckness with another laye r of brick, whic h would
Ihen be funher testcd.

In these lests il wa s not laken inlo accaunt, cJearly being a favorab le fa cI, lhe
redllct ia n in slenderncss caused by the add itionallransversc partitioning of lhe areh
 R. BRUFAU N IUBÓ I LlS Golfas de b Pedrera 35 1

thal practically immobilises lhe mast interior arch ring. This factar is Icft as a
gcncrous increase in lhe coefficienls of sufety far each arch.

2. Analysis of the angled arc hes: The arches lhat bend ofr when lhey
encaunter the vertical 'spine' panitian, required a special study of the disu'ibutian af
forces tha! is praduced al this point. The campression forces concentra ted at the
arch key should be channeled by lhe partitians towards the stair tower, to which the
anglc ·points'. Init ially this involved a spadal analysi s; çonsidering arch panition s
as lineal elements concentraling lheir line in the centre of gravity of the lhree arch
rings. On application af lhe loading relevant to these elcmem configuration, lhe
transvcrse forces that shollld be conducted by lhe partilions was deduced.

3. Design of lhe upper parlitions: Knowing lhe forces lhal these partitions
had 10 carry, bec;:\lJse of lhe previous spacial analysis, their design carried oul.
Their height is always 45 em., corresponding to lhe height of lhe th ree layered arçh
rings Ihat joins them. BUl their particular thickness has to be calcu lated exactly,
permitting lhe following dimensions: 4.5, 4.5+2.5, 4.5+4.5. 2.5+4.5+2.5,
4.5+4.5+4.5, elc .... TIlis grading is important for the adju stment of thi s partilion to
horizonlally transfer the load. It is an aspecl thal, becallse of its visual atlracl ion,
became of grealest interest out of ali lhe complex framcwork of nerves and ribs lhat
define lhe space. It is in lhe execution of Ihis feature thm we have concentrated the
grea lesl part of our creative slrenglh looking for conceplual expression 10 suit Gaudí
and hi s collaborators.

A focal view of these panilians is found in the arcas where they depict the bends of
the arch systc m. attempl to visually read (hem as if it mcans understanding the
isostatic lines of lhe rooL divincly passion:He.

4. Analysis o r lhe stair lowers: Knowing the reactions at the conneclion of

the arches 10 lhe towers, procecded to test lhe good load carrying behavior of l!lese
to guaranlee lhe slubili ly of lhe arch system. especia lly wi lh respecl 10 the
hypothesis of high wind forces. In plan lhe lurge size of Ihese towers. diamelcr 3.5
Ill .. and thcircircular shape, favour the guarantec of good reaçtion lo whichevcr the
dominant wind direclion is found to be. ens urin g lhe same behavior in ali the
buiJding.

5. St ud y of lhe base Ooors: lhe new loadings obliged lhe sys lematic
reinforcement of ali the base floors. Tbus compi ling a mixed com posilc section by
Jaying a mesh reinforced concrete layer on lOp of lhe ori ginal steel beams below:
The di stribution of adequate number of connec lors belween each part of the sectio n
guaranteed a composile reaction under loading of the componem materiais. The
Joad is shared belween a 6 cm. reinforced concrele layer. calTying lhe compression
forces and lhe steel prolile. calTying lhe tension forces.

Wilh respect 10 the strengthening of lhe Ooors. ir is enough to mention lhe use of
10w den si lY concrele, 1500 Kg/m 3. achieved by the use of light weight gravei and
addilives previously sllldied in the laboralory. This lightweight was necessary to
reducc lhe general self weight of the upper slorey.
352 STRUCTURAL ANALYS IS OF HI STOR ICA L CONSTRUCTIONS

6. Treatrncnl of existing cracks: In those arches and partit ions where sm all
cr'dcks had beeo detected. lhe procedure was lhe injecLion of fluid epoxy-resin based
mortars. When lhe cracks we re af grea!er importance, or lhe arc h had loca l
buckling, ir was decided to dismounl lhe zone of lhe e lement effccted, immediately
following on with adcqualc reconstruction. Ir the arch presenled a grave state with
general buck ling it was opted for complete dismollnti ng af lb e arch and it s tota l
reconstruction.

Tcchnical lcam for lhe excc ulOC af lhe projccl of Slrcnglllcning a f lhe Golfas af lhe Pcdrcra:
Architcct Dircc lors: Francisco Asarta and Robcr1 Bruf:lU.
Tccllllic:11 Architccts; Enrie Mim alld Ramón Garriga.
Collaboralors: Gcmma BusqueIs. Joan Ramón Blasco, Kcnichi Kurokawa.
ConslruClion com p;:m y: Ciosa A legreI .

•
 R. BRUFAU NIUBÓ I Las Golfns de la I\:drcra 353

APPENDIX

(I). In lhe firsl place, I thought of presen lin g lhe melhods o f analy sis used in
srrengthen ing of fou r d ifferent rura l churches. dat ing from between lhe 12th and
17th Centuries. *CampoITe!s in Huesca, AJcolea de C inca. Balag uer and Sarroca in
Lleida-, ali of whom had been effected by grave stabili sa tion problems. that in three
of the fOllr cases, thal put (hem io a position of immi nent collapse. I was interested
10 explain Ih al wl lh the small iOlcrventi o o of po st-Ien s ionin g of certa in stone
elemcn ts, ~co n ce ntraled in particu lar arcas of key struclural imponance, and w ithou t
leavi ng a publicly vi sib le trace-o il was possible (O reac h a new lens ile Slale,
compa lible \Vith lhe ex isting dencctions and crack in g and with the reall oad bearing
capacily of the stnlClure.

(2). The popular na me 'The Gol fas of lhe Pedreru' refers to lhe Cala lan word
'Golfes' which signifies lhe 10ft space below the roof of a building, th at has ilS
principal characleri stic in th e double fun c lion as slOrage roam and as an .mic thal
provides climali c insularian for the flals in lh e upper storcy. Additionally, in thi s
particu lar case Gaudí des igned for a wash hou se fo r the fl ats. with easy access to
lhe magnificent n at roof, pl"ovid in g in lum a clothes drying area.

(3). Given lhe diffe ring base length and thal the varying vertica l heighl in each
case, lhe shape of the each arch shou ld also change. It is interesting to know of the
melhod used to shape Ihese arches: The maxi mum heigh l was cha lked 0 11 a central
vel1i cal measure Ihat is marked on a wal l. a hari zontalline ar the levei af thi s central
height is then marked hal f lh e base leng rh away on each side , from these two
pos itions a chain is Ihus draped in such a way that its maxilll1l111 ampl itude tOllch es
grounel levei at lhe ce ntral ven ical marker. The chain Curve is lraced 01110 lhe wall
suc il Ihal lhe carpen ler ca n build al"ch fonnwork il1side the shape, (hi s fO llmvork is
finally invened in the location of lhe proposcd arch so th at lhe three 1l1;lsanry arch
rings ca n be built abOLll it s shape. Gauelí applied this ' inv ersio n ' techn iq uc to
convert the sign of a pure ly tensite chain shape il1lo thc pllre ly camprcssive struclure
of the bri<,'k arch.

(4). The popular be lief (hat Guudí ba sed hi s prin cip ai s af construc lio ll o n
exceptiona l intuition cou ld be true in certain case s, but not with respect 10 hi s
approach (O load bearing SlruClllres. Amongst hi s habitual collabo ra lOr:; figure
some expe rts in struc tural analys is and cons lruclio n, s uc h as the Bayó brothers,
Ja ume (1874*1 96 1) was lhe Professor of mater ial strength in lhe Esc llcla dt:
Arqultectura in Barce lona, while losep (1878~197 1 ) was lhe buil der of lhe most
oUlstandi ng warks of lhe maSler. Other frequ cn t catl aborators with a knowtcdge of
material strength are lhe architects SlIgranes and Canaleta. Cel1ainly lhe Peclrera is a
building noted for its fin e perceplioll of the struclural work 10 be cani cd ou l byeach
of its componen l elements, exemplified in lhe we l! known slructural design af the
base of one of th e interior wells: a magnificem riveted plate g irder struclure in the
fann of 'bicycle whce l', this etemel1l has been res lored anel strengthened.
354 STKUCTURAL ANALYSIS OI' HISTORICAL CONSTRUCTIONS

(5). Gaudí used for l he constructio n af lhe who le storey. hand mude reddish cl ay
bri ck with an excellent fi ring. S efare lhe start af lhe strengthening project some
tests were carried oul on lhe fabricated brick lIsed by Gaudí. Nane af lhe resu lts
gave an ultimate compressive slrength lesser Ihan 150 Kg/cm2, while lhe morta r
presented a strcngth near to 60 Kg/cm 2

(6). Fnmcesc Barba Co rsini was parI af lhe 'R group', a well know n Illovemen t
that developed in Cata luiia du ring Ihase years imposing a notable influence in lhe
fonnu tion af new generations af archilcclS Ihat followcd on in lhe 1960's.

(7). llle serviceable arca af lhe 'Go lfas" as desig ned by Gaudí was 1275 m2, in
one so lo space thal is only inlelTupted by the six sta ir towers that give access to the
terraeed roor. -Iowers that were convencd in to the six sculptura l focal points of the
plaee- a sccnario th at is only disruptcd by lhe neeessary hOllsing of the operating
mac hinery fo r the blli lding's centrallifl. a usage that it is 10gi<.-allO presume to be
outside that entailed in Gaudí's original general conccption.

(8). Catalan name, Iranshued as' Gaudí's space'.

(9). T he aelual relevam Spanish requirement fo r public meeting areas and Ihealers
wilhout fixed seating, is a design load of 500 Kg/m2. There is absolutely no doubt
Ihat the o riginal structural design of lhe 'Golfas' and tcrraced roofs of the 'Casa
M il à' was not inlendcd at any time to be used as an area for the congrega tion of
numerOllS Illcmbers of the public.

(lO). Ir is easy to locate the arehi tec t Ba rba Corsini himsel f and his 1954
intervent ion in the Pcdrera in thi s dassificatio ll.

(11). More than 100 arehes were effectcd. 10 a grea ler or lesse r degree. by
crac king.

( 12). The arches have been grouped de pcndenl upon their gcometric slendelllcss.
considering Iheir lenglh, the effccts of buckl in g, as the distance between the the arch
bases an el the positiol1 of the upper transversc vertica l partitioll, that usually
coi ncides wit h the arch key.

( 11). Ae ll1a ll y lhere are a number of examples of arches effecled by lcnsion

c racking. due to lhe infliction of loads Ihat they were nOI original ly designcd for,
caused by unsymmet ricul Illut il alion of lhe arch shape. Some of the crack widths
have reacheel 4 mm., with ho rizontal movement towards the exterior of the
building. Indicating the effeetive collapsing from lhe in side of the origina l
infrast ructure. Note the difference between thi s and Gaudí's des ign based
experimentation of material qualities, commenled on in appendix (3).

(14). In all the 'Golfas' no two arches with the same shape have becn found. There
Il1l1S1 have been 265 cu rvature 'tempomry form work ' conslnlcled in lhe workshop ,
onc fo r each areh. th is means a difficult execution of the first twO oplions.
 R. BRUFAU NIUBÓ I L~s Golfas de la I\!drcr~ 155

(15). Guudí's initial projeet was based upon the usage af hand !nade bricks af 4.5-
5 em. Ihi ckness, but also useel, oceas ianall y in some laeati ans a slimmer briek af
2.5 em Ihiekness. af lhe same texlure anel qualily.
 STRUCTURAL ANA LYSIS DF H ISTORICAL CONSTRUCTIONS
P. Roca, f.L González, A.R. Mari and E. Oiíate (Eds.)
© ClMNE, Barcelona 1996

PREVIOUS STUDIES OF THE "CASA DE LOS BOTINES DE

LEON"

J.L. G onzález Moreno~Navarro

E. T.S. Arqllitectura de Barcelona
Universitat PoUteClZica de Catallmya
Spain

1.- A FD R EW O RD ON T H E M ET H OD

Our approach to Architectural Restoration , and in particu lar the studi es that
have been made of the Casa de los Botines de León, are in general determined
by three cri teria. The 6rst is that it is necessary to preserve all the architectural
value of historical buildings as a sign and a dacume nt, induding the structure
af the build ing , a nd it is t herefore cssential to use its own resources as far as
possible. T he second is that, as has been clearly demonstrated in recent years ,
maintaining the materiais and e1ements of the building is precisely the best
guarantee for its conservation. The t hird is that it is imposs ible to achieve
all this without a complete comprehension of the mo nurne nt to be restored, a
comprehension that must encompass both the physical reality of the building
and th e melltality of the architectural culturc alld building tradition s in which
its creators were immersed.

But t he greatest impediment to acting in accordance with these cri teria does
not st cm from t.he laek of positive kllowleelge but from ou r own culturc anel
mentality.

Qur current teehnologieal t. raining is of litt lc use fo r the eomprehen siol1 of

m o nument s . Th c mcntality gencrated by iuuu:;il'ialization , the applieatioIl
of procedures elesig ned for builelings that have not yet been constructed , the
separatio n between strueture anel enclosure walli ngs , betwecn permallenee and
stability. and man y othe r fragrncntation s that havc permitteel the building
cu lture in o ur ce ntury t o evolve positively, prevent us from unelerstanding the
reaLity of ho\\' builelings were eon structeel in the pasto

Build ings previo us to t he 20th eentury (and quite a few of this eentury ) respo nel
t o a cu h ure bas ed o n factors that are quite ditfcrent from the current one , sueh as
t he elecisi ve presc nc e of t.he eraft trade , ar th e predominance of the wall system s
 J. L. GONZÁLEZ MORENO·NAVARRO I Casa de [os BOlincs de León 357

that respond solely to requirements of load-bearing and enclosure walling, in

which conce pts 50 out-of-date as the wall-plate and the know-how of craftsmen
are essential.

The building culture of th e p ast responds to a unitary co nception that includ es

both the parts of t he buildings and the socia l o rganization necessary for it s
construction.

If we carry out restorat ion work on a building of the past, it is necessary to use a
meth od' that neutra liz es , compensates and reorients our mentality through the
development of new standard s of reasoning that a re coherent with the buildin g
culture previous to the 20th ce ntury.

T he final objective of the method is na ne other, of course, than to obtain

sc ien tifi c knowledge of t he physical reality of the building, whi ch mu st provid e
informat ion on what it is now, how it behaves, how it wiH behave, and liow it
has bchaved in its hi stori cal past.

Now, it should be remembered that the scientific method requires preliminary

hypotheses that chan nel alI t hose activities in a precise direction to be
established before t he process of modelling a nd systemati cal experiment
Call1 IDences. Alld, as a U wor ks Oll t he topi c state, it is in this preliminary
phase af formulati on of hypot heses th at accumu lated knowledge, ex perience,
ability, imagination, in short , what can be considered as t he intuiti on a f the
rese archer, take ou an irreplaceable value.

The scientifi c method is th e " art 1> af testing all swers, but a bove aU t he " art" of
ask ing the right questia n s.

Aud precisely in this intuition lies the greatest dallge r of applyi ng to a ncieut
buildillgs a mentality that is controlled by rei nforced concrete.

To avoid this it is necessary:

1) To give priority to the unde rst anding of the building as a total unit in which
its res pan se to aU the requirem ents does not involve specialized mechani sms but
is a si ngle, all-encompass ing mecharusm.

2) Not to make the sim plification in whi ch thi s lack af specialization involves
less heterogelleity. Heterogeneity is also an essential feature of traditional
architect ure, which does not produce buildin gs with many parts of the saroe
material (as today with reinforced concrete) but rather ones in which the
358 STRUCTU RAL ANALYSIS OF HI STO R1CAL CONST RUCTION S

mu ltiplicity of Young modules is the fu ndamental charaderistic.

3) To compensate the trend to fragmentation, it wiU be necess ary to seek

procedures of analysis lhat encompass aU the variables , aU the objedives and
especiaUy ali the perceptual scales of t he means of constructioll. It must be
a process lhal allow s lhe building to be broken down inlo ils parts, down
lo the malerials, anel to be constantly recomposed unlil the buileling unil
is reconstruded in such a way that alJ lhe scales of perceplion are included
homogeneously.

At the cnd of t he firsl study we wiU have rcached lhe firsl levei of intuitive
and quaü tative knowledge, which wiU indude several hypotheses on slruclur;:t.1
behavior, the cause of fi ssures, elc. The sit uation wiU be more favorable if we
have also captu red the mentality of the bui lding culture t hrough documents
describing the specific history of our buileling, or stuelies or scho last ic notes on
these mentalities.

But there wiU rernain all infinite number of unresolved problems that we
wiU only be able to solve through a sysle malic campaign of inspection anel
experimentation.

The time has arrived lo demonstrate our "art" of asking questions.

This phase of the process wiU involve the wide range of currenl procedures
for carrying out the systematic campaig n of obtaining scientificjquantitative
knowledge , inclucling photogrammetry, endoscopy, radar tomography, a wide
range of material testing techniques, t hermography, acoustics and numerical
moclelling by finite e1ements.

FinaUy, after demonstrating our "art " of test ing answer s, we rcach the key
moment of demon strating our "art" of interpretillg them anel reordering them
in order to reach the desired underst anding of the monument , whi ch is not at
aU easy. Success wiU only depend on our own skiUs.

In facl , only a co rrect preliminary understan ding that is both intuiti\'e and
quantitative wiU aUow us to imagine minimal resLoration projects that make
fuU use of the rn on ument 's own resources, lhus preserving as far aS possible its
value as a document and its meaning. The content of the recent meeting of the
IA BSE on restoration of the monumental heritage is proof that these cri teria
have been generaUy accepted) even by engineers 2 •

All of this is clearly shown in the studies 3 of the Casa de los Botines de León
 J. L. GONZÁLEZ MORENO-NAVARRO I Ca~~ dI! los Botincs dI! LcÓIl 359

(Pig. I). The owners~ have a lways given their (uH consent for these studies to
bc carried out without any ecollomic limito It would have been worthwhi le to
do th is work on any building by Gaud í, but in this case it has proved to be of
particular interest. It was initially suspected that the building would be unable
to embark on a new stagc of its life, but the st udies led to the definitioll of a
minimal intervention strategy that hardly varies the conceptual and physical
stru cture of the building, but provides fuU guarantces of safei)'.

Pig. 1.- Casa Botines

2. - GEN ER.AL DATA

2. 1 The building

Thc Casa de los Botines was built according to the design of Antonio Gandi in
1892 for the fam il y Fernández and Al1drés. The desire of the textile me r('hant
oWllet's was to have two open gro und -floor prelllises for cOll1mercial use, the
lIlain 1I00 r for their OWII dweUing and the remailling fl oor::; for renteel property.

T!te bllildillg: has a trapczoidal groulld plan adapted to t he arca of t he available

:-;It('. "' he bllildillg IIlcasures approximately meter s
011 the main (,, (adt'. 20 a!ld
,60 STR UCTURAL ANALYS IS OF HI STORI CA L CONSTRUCTIONS

Fig. 2.- Plan af levels 2th to 5t h of Casa Botines

25 m at the sides and 35 m at the Tear faç.ade ( Fig.2 ).

The structure af the huilding is composed af the peripheral \Vali of the façade,
Df variable thicknes s according to the height, fram 1 meter to 0.15 ro, formed
by two limestone \VaU faces, the Duter oue af ashlar 20 to 25 em thick, alld the
inner ane af roughstone with a variable thickness. Due to lhe utilization af the
ground fiaor 1 Gaudí confidently designed large openings that give maximum
iUuminatioll , though this means that the piers between them must be narrow.
00 the higher flaors, howeve r , the solid part is larger than t he openings.

The different fioor slabs af the building rest on this Dutee wali. The first two
are made of timbrel vaults of two thicknesses supportcd on metal beams, and
higher floors with wooden beams and brick vaults.

ln the basement and ground floor the interior structure of the building (Fig.
3) is composed of a set of cast iron pillars that rest on the insulated fo otillgs
without any typc of bracing. The metal bcams rcst 011 stol1 e capitaIs on top of
the pillars. The beams also support the stretcher s wall that form the internaI
structure of the main floor and the higher floors.
 J. L GO~ZÁ L EZ MORENO-NAVA RRO I C:asa de los BOlincs de Lcón 361

Fig. 3.· Section of Casa Botines

The studies on Gaudí's building ~ and the study of its context show that the
material and the solutions used place it within the general Catalan building
practice of the era, and cf Barcelona in particular. This can be seen in the use of
light construction elements such as stretchers waU , cast iron piUars, brick vault
floors and wooden beams with sim pie solutions for joints and simple supports,
This produces both a fuU use of the resistance of the materiais and a high degree
of isostasy_

This is one of Gaudí' s work s that has received the least attention in the
extensive specialized bibliography. There are no studies that give any idea of
the intentions of the architect when he designed it. There are only conjectures
 362 ST RUCT URAL ANALYSIS OF HI STO RICAL CONSTR UCTIONS

00 its gothicism ar superficial and erroneous interpretations 5uch as that in this

work he was a precursor of Le Corhusier in his use af the apeo-pIao fIoor

ln fact, Gaudi merely transposed to Leóo a practice that was commoo in

Barcelona and is seeo in countless buildings of t he old city centre ar of the right
af Cerdà's Eixample (expansion area). However, this transposition was 001
completely coherent with the original, since it was a practice used for buildings
between party walls and 001 for isolated huildings. Gaudí takes a madel for
buildings between party walls located in a block, and by sim pIe transposition
and rotation of the perpendicular section to the façade he organizes ao isolated
building. The doubt about the capacity of the building to resist extraordinary
seismic movements seems to be more than reasonable6. The doubt about
whether it was a stroke of genius by Gaudí or unwitting is also reasonable.

2.2 The Studies

The alm of the project for restoration of the building 7 is to adapt it to its
new function as the headquarters of the owner's without altering the original
structure and spatial layout. It will also be attempted, as far as possible, to
restore to the original state the areas remodeled by interventions in recent years,
which compIetely impair the spatial significance of the two open-plan fioors. It
is obvious that ali this is difficult to achieve without an exhaustive knowledge
of the real conditions of the building and is structural eIements.

The preliminary studies consist of a first part dealing with knowledge about the
building and a second part defining the proposaIs for restoration. The first is
divided jnto five sections:

(1) anaIysis of the parts of the building that constitute a key element of the
behaviour of the structure , but are not sufficiently well known;

(2) a study of damage and deterioration, which is in fact not great i

(3) mechanical tests conducted to characterize materiaIs, elements and brickwork;

(4) structural analysis subsystem by subsystem; and

(5) computer studies of the general behavior of the whole building as a sum of ali
the subsystems.
 J. L. GONZÁLEZ MORENO-NAVARRO I Casa de los BOlines de Lcón 363

3.- ANALYS I S B Y SUBSYS TEMS

3.1 T he façade walls

The initial analysis of the building reveals at the outset that it s resistance
to horizontal forces, however slight , is based on the behavior of the four
façade walls. Th e isostatic layout of the gantries 00 the lower floors and the
extraordinary slenderness and lack of bracing of the upper brick walls clearly
show that its resistance to horizontal forces must be ni!. It is for thi s reason that
a complete knowledge of the façade wall has beco me a high-priority objective.

The first fundamental question is, in addition to the real internai structure or
the thicknesses of the two walls, their union and their behavior to gravitational
loads.

The outer face of ashlar masonry with a relatively regular bond , and the inner
face of rubble with far thicker joints, have very different mechanical behaviors.
The question is sim pie: do these walls act together as a si ngle unit 50 that the
building uses the whole thickness of the wall for its stabitity? If this is important
Oil the higher fioors , it is crucial in the ground fioar piers, which have the same

composition but are the olles with the smallest section and the ooes that bear
the greatest loads.

The method used is based on determining the state of vertical stress of the two
faces. The test pro ced ure used to ascertain this, the so-called fiat jacks (Fig. 4),
have been used for over 10 years, especially in Italy. However , in Spain they are
practically unknown. In additioll to the existing st ress, they give the modulu s
of deformation of the brickwork.

Four double tests have been conducted on two piers of the main façade. The
cuts were located exactly in the same datum tine both ou the inside and the
outside8 • Without going in to detail, suffice it to say that the result s have been
revealing. The fairly moderate working stress of the outer face is five times
greater than that of the inner fa ce, which rcceives ali the loads originating ffútll
the upper walls and from the fioor slabs, thus ensuring the total and absolute
union between the two faces , and making it possible to consider the whole
thickness of the wall from the point of view of buckling and resistance to forces
prod uced by horizontal action.

But , although the technique of flat jacks is fairly non-invasive, the size of the cuts
required prevents the use of this test in the ground floor piers, the aoalysis of
which is essential. To solve this problem the extensometric technique by remova}
36-1 STRUCT URAL ANA LYS1S OF HISTOR1CAL CONSTRUCTIONS

Fig. 4. - Performed flat jack tes t

Fig. 5.- Nu merica l model

of material was used 9 , complemented by laborat.ory t.ests of th e characteristics
of the rough ashlar and dressed stolle.

The procedure is based 011 the fact that a state of stress acting ou a material
remain s altered if a part of the material is removed. If through a suitable
technique the experimental re!axations in any three directions around the
removed zone are evaluated , it is possible to deduce the initial state of stress.
ln slead of requiring a cut of a cerlain dimen sion in the tende! of the briekwork ,
the extraction of material is reduced to only a holiow 3 em i.1I diameter.

As in th e upper sections, the mos t rigid part of lhe pier is the one that t akes ali
 J. L. GONZÁ LEZ MORENO·N AVARRO / Casa 1.10: los BOlino:s d~ León 365

the load , and the outer part bears a Joad 10 t imes greater t ha n the inner parto

In both cases, the st ress tests also showed that the building is (ar from the
breaking st ress.

Ou t he ot he r hand , t he comparison between the real st resses detected and the

stresses for eeast in t he calcu la tions due tu tbc c.x.il;t illg gravitational adi oo is a
magnificent proceeIure for checking the good ness of the fore cast s.

In co nclu sio n , the brickwork is sec ure against gra\ritational actiou. lt stiU has
to be tested for horizontal action.

3.2 The bl' ick walls

Th e objective and the met hods \Vere quite different from t hose used for t he
façade waUs. The main difference is the surpri sing slenderuess of the brick
waU , whi ch in t he main fioor is almost 30. The appli cati on of t he cri teria
of Basic Norm FL·90 leads us to conclud e that the waU of the main Roor
is insufficient beca use its theoretical capacity has been exceeded five t imes.
However , ab solu tely no sy mptorns of faiJure have beell found.

This is a paradoxical situation that has bec ll found very often . The strict
appLication af the NBE gives th e idea that many sound buildings sho uld have
colJapsed , though this does not mean that the norm is in a ppropriate. M uch
work remain s to be dane OH t hi s subjecl 1o .

For the above reasons, in the stud y of brick \VaUs the top priority has been to
determine the real resistance through laboratory tests with sa mples extracted
from the waUs, and the real stat e of st ress using ftat jacks.

As obtaining all these data in volves performing a se ri es of costly tests, we

mu st limit the number to the indi spensable minimum. Howcver, we mu st be
reasonably certain that thi s indi spensable minimul11 is reprcsc ntativc of nU th e
load· bearing waUs.

Therefore, before performing t he tests, we detc rmined t he degree of

homogeneity of the different brick elements that are distributed in the waUs,
and of course of the differcnt 1110rtars, whi ch are more liable to hete rogcn city.

To this eneI wc used two types of test that analyze different variables anel
tbat are cffective for checking the homogenci ty. The fir st type is based 011
366 STRUCTURAL ANALYS IS OF HI STORI CAL CONSTRUCTI ONS

petrographic and mineralogical methods l l , and the second on t he measu rement

af the resistance to the penetration.

Thc res ults show that this is a typical brickwork building of t he pr e ~ indu st rial
era in which it was constructed. The basic characteri st ics af the bricks show
con siderable variatioll , but onl)' some 8% do not have the minimum required
firing levei, and as they are distributed homogeneou sly aU over the brickwork,
it is not necessary to di stinguish several zones.

This is nol the case with lhe mortar. It s process af manufadure involves
a greatcr variatio n in the result s, and in t his case there is al50 a clear
differentiation of zones: the mortar af the main Oaor has a lower proportiolling
(of the order of 1:5) and strength than the mortar of the upper fIoors.

If we go from thc tests to direct observation , we observe that in fact the main
floor has a more irregular brickwork t han the upper ones. It is reasonable to
make the hypothesis that either the bricklayers improved their skill with the
progress of the work, or they were replaced by more skilled workcrs. At any
rate , as a consequence of this it was necessary to conduct a rather high number
of co mpressi ve strength tests, differentiating the main floo r from the resto

Thc res ult s show that t he walls of th ese upper floors have sufficient load-bearing
capacity.

On the main floor the stresses detected by the flat jack coincide ahnost exactly
with the forecasts if we suppose the load to be ab solu tely equally dist ributed
in the wali , and are far from the st resses indicated by the Basic Norm as a
consequence of buckling. It should be understood that the intention of the
basic norm is not to forecast the stresses but to warn about the lack of excess
load-bearing capacity in a situation such as the oue described. Logically, it was
found that the real stresses are indeed of the order of half the characteristic
st ress deduced from the tests. Therefore, the wali does not have symptoms of
failuTe. However, it has no reserve strength against even the slightest variation
in it s conditions of stability. Thcrc is no doubt tho.t it should be reinfor ced , but
since it has excess strengt h it is only necess ary to reduce it s slenderness.

3.3 The wooden floors

The two objectives with respect to t he horizontal structure of the upper fioors
were co nsidered consecutively. The first was to evaluate the number of gi rders
whose st ate of conservation completely prevents their use. Since the number to
 J. L. GONZÁLEZ MQRENO-NAVARRO I Casa de los Botines de León 367

be replaced is not excessive, we went on to evaluate their strength 12 •

The result s were clearly positive. The wood girders were in good condition, and
even those that were not in perfect condition but were recoverable were fully
able to bear the extra loads of the new use of the building.

If some limitation must be mentiom:J it j::; that in a few zune::; a ri::;e ::;lighily
higher than 1/ 300, (but lower than 1/ 400) is foreseeable, which would hinder
the utilization of new brick partitions . However as this possibility has never
been considered , it should not be considered as a limitation.

It was precisely the conclusion on the correct state of the ftoors that led to the
definitive go~ahead for the absolute conservation of the whole building.

Of course, there remain many interesting problems that we willleave for another
occasion. We must leave sufficient space for the even more interesting final part
of the work.

Having concluded the studies by subsystems, and having obtained positive

results for them aU, we rnust go on to the last levei of the analysis, in which the
object of the study is the whole system formed by the subsystems.

4.~ GENERAL ANALYSIS

4.1 Construction of the model

The creation of a numerical mo de} of a building requires a full interpretation

of its structural operation in order to assess the requirements and the degree
of det ail with which the model must be made, but above all to make sure that ,
with the unavoidable simplifications, the model is actually representative of the
real behavior of the original.

Rcturning to idcas put forward at the beginning, in this phase our

interpretations of the behavior of old brickwork and its joints, which are so
different in ali aspects írom t he structures of reinforced concrete, become
especially import ant.

In this case, this interpretation has led us to prepare a model of the building as a
whole , including alI the structural eIements that can affect its general b ehavior.

To do this, the model was built with the four façade waUs , the internaI load~
368 STR UCTURAL ANALYS!S OF HISTORI CAL CONSTR UCTI ONS

bearing waUs and the fioors oí t he b uilding. It was attem p ted to t a ke into
accoun t aspect s such as t he infiuence oí the asymmetry produced by the
t rapezoidal form oí the grou n d plan aí the bu il ding, the possibJe contribution
aí internai load-bearing waUs against horizontal adions, the capacity of the
fioors for effcdively bracin g the internaI and externai vertical elements, and the
behavior oí the joints b etween corners of íaçade waUs (Fig. 5).

This general study was complemented with simplified studies dealing wit h
indivi d ual façades, simulating app roximatcly the state of the joints of the latter
with the remai ning structural elements of t he building.

For thc analy sis it was dccidcd to use a Gencralized Matrix Formulation 1J . I 1,
wit h which it is possible to deal exactly with spatial linear elements with a
curved directrix and a variable section.

To model the two-dimensional shear stress panel or load -bearing waU elem ents ,
we u sed equ ivalent system s of linear elements as described bclow. In spite of it s
limitations in th e study of bri ckwork constr uctions , for simplicit y we accepted
the hypothesi s of linear e1asticity as a first approximation t hat was valid within
a certain margin and in the context of states of stress of moderate intensity.

T he two-dimensional vertical e1ements , including façad e walls and internalload-

bearing waUs, were dealt with as equivalcllt bar syst cms t hrough the technique
developed by K wan l~ .

With thi s method it is possible to simu late with great p recision and economy of
means th e bchavior oí two-d imensional elements that function as a wall or panel,
paying app ropriatc atten tion to their behavior under combined compressivc
stress or two-way ten sions , bending and shear. In addition to this ) it provides a
very realistic mode! of the connectioIl of t hese e!ements with the adjacent lintels
or parapets between windows.

It has thu s been possible to dispense wit h more sophisticated method s (s uch as
finite e!ements oí panels or shcet s) that required excess ive compu ter resour ces
(in time and com p uter memory ). In t he fa çade waU sections we distinguished
between the ashlar and the rough stone.

Though we recognize that in practice the partitioll S could have some st ructural
fu nction, they have not been integrated in the mode} because we fe!t that t heir
possible contributi on is not very reliable and difficult to quantify.

To reprodu ce the bracing produ ced by the fioors betwecn interna] vertical
 J. L. GONZÁ LEZ MORENO-NAVARRO I Casa de los Botines de Leôn :'6'1

elements (pillars or walls) and externaI elements (wallsL we used a limited

number of one-dimensional two-hinged elements forming ao approximately
orthogonal grid 00 each fIoor.

On the main and ground fioors, these elements coincide with metal girders
between vault s; their infiexibility is defined as that of an individual metal rib ,
and t he possible contribution of the brick vaults is ignored for safety.

Similarly, the bracing elements introduced on the upper floors represent the
axial rigidity of a group of wooclen beams of t he fioor slab, ignoring the possible
contribution of the brick vaults and the timber fioor.

The connection of the façades at the corners has been reproduced acco rding
to actual observations of the building: from the basement levei to the main
fioor levei there is a continuous reinforcement, while on the upper floors this
connection con sists of strong individual lintels. Figs. 9 and 10 show the model
that has been constructed.

4.2 Structllral Analysis

The characterization of the structural behavior of the building was reached

through the study of a series of cases in which the participation or
interrelationship of the different subsystems was introduced progressively. The
following combinations of aetioos have been studied:

I) weight and overload ;

2) weight, 60perpendicular direetion to the main façade j
3) weight, 60parallel direetion to the main façade; and
4) building reinforced under its own weight, 60overload and seismic force acting
in the most unfavorable directioIl.

The seismic actiotl was evaluated according to the P.D.S.l Seismic Norm, which
gives León a modcrate seism ic intensity.

a) Distribution of forces
Through the elastic analyses t hat. were performed , the building as a whole
showed three basic mechanisms , the combination of which gives its behaviour
in response t.o gravitational load and horizontal actions.

Firstly: the façades work as a ftat grid resting on the perpendicular façades, both
370 STRUCTURAL ANALYS IS OF HISTORI CAL CONSTRUCTIONS

in response to horizontal loads and to the effects arising írom the eccentricity
of the gravitational load. In both cases t here is an effect of support of the rear
façade on the main façade , 50 part of the effects of grid bending are aHenuated
at the expense of a slight increase in the effects of the latter.

Secondly, the façades tend to act as a gantry against horizontal actions oriented
according to their plane, collecting and trallsferring most of the horizontal forces
toward the foundation by means of this mechanism.

Finally, there is a150 a frame.type behavior in horizontal section to both

horizontal and gravitational actiou, which is a consequence of supposing that
the façades are joined rigidly through the comer lintels.

This mechanism produces strong bending in comer lintels and parapets of

externai bays.

In response to lateral actions, the internalload. bearing walls hardly contribute

to the rigidity of the whole, since they are not supported directly on the
foundation, and can follow the general movement of the building almost
without being deformed. Furthermore, t hese walls are mainly oriented in the
longitudinal direction (parallel to the principal façade) in which the seismic
effect is less criticaI.

b) Behavior in respon se to gravitational charges

It was found that the equilibrium of the façade walls against gravitationalloads
is possible because they are braced by the fioor slabs and the adjoining façades.
Indeed, if this support is removed, regardless of possible buckling of the walls,
the eccentricity oí the gravitational loads applied would tend to produce a
dangerous state oí bending in the ground fioar piers. If the support between
perpendicular façades is considered to be effective, the elements sub jected to
most forces are still the ground fioor piers , though in this case they are subject
to states of compression that are tolerable for the materiaIs of which they are
made.

c) Behavior in response to horizontal forces

In t he current configuration the building does not respond adequately t.o
horizontal forces for the following reasons. Firstly, the only rigid vertical planes
are the façad es. It cannot be considered that the internaI load- bearing walls
can contribute by introducing planes oí intermediate rigidity since, among ot.her
reasons, they are simply supported on the pillars 011 the main floor levei , ano
because oí this they do not have a rigid connection with the foundat.ions.
 J. L. GONZÂLEZ MORENO·NAVARRO / C3.~a dI! los BOlincs de Lcón :n I

Secondly, due to the nature of the strudure, the exis ting floors cannot be
expeded to form a rigid element in their own plane that is able to work as
a wide girder.

However, it can be accepted that they call transmit horizontal forces betwec n
opposite façades, acting as a sim pie bracing.

As consequence of such deficiencies J the façade waJJs are subject to horizontal

forces, which systematicaUy produce strong transverse bending effects, whether
due to the local work (fiat grid pIan at the façade levei) or the overalJ work
(frarne plan according to horizontal sectiolls).

In particular, the elernents that are rnost subj ected to bending are the corner
lintels and the parapets of externai and central bays 011 the façades oriented
transversely to the direction of incidence of the hori zontal forces (Figs. 6 anel
7).

Fig. 6.- Stress intensities under the effect of the earthquake

But it is also found that the façade waUs are suitable for acting as a grillage in
their own plane, even if they receive alI the inertial seism ic forces in questiono
372 STR UCT URA L AN ALYS IS OF HI STO RI CAL CONST RUCTIONS

Fig. 7. - Stress intensities unde r th e effect of the earthquake

In rcality non e of these effects is s ufficiently important to place t he general

safety of t he bui lcling in dan ger. However, as thi s small probability is multiplied
by a li cxtraordinarily negative value in losses due to the monumental valu e
of the building, it is reasonable to propos e a reinforcemellt systell1 t hat is
110t exccssively complex o lt is proposecl to carry it out th rou gh a sys tem
of rigidificatioll whose basic effect is to perll1it the dcsired tran sfer of forces
betwee n façad es, whilst minimizin g the lo cal gricl bending for ces and the general
frame be nd ing forces.

d ) Ideas on the reinforce ment

It is co nceivable to und erta ke a large-scale p roject wit h a view to forll1ing a
new ri gid intern ai sys tem that can indepe ndently resist t he whole effcct of t he
seismic forces .

Nevertheless, such a project would make m ore sen se jr th e build ing were loca ted
in a zo ne of greaLcr sei smi c activity. For t he Illoderate seis mic activity of León
it is possible to imagine smaller-scale projects based on a prud en t ut ili zati on of
t he resistent cap acity of t he exi st ing elemenLs .

Th e formation of a rigid d iaphragm in t he ceiling of t he secon d fioor ) co inciding

 1. L. GONZÁLEZ MORENO-NAVARRO I Cas:l dt.: los B OIIllt.:S dt.: Lc6n 373

practically with the top of th e waUs , is extraordinarily efficient for reducing

the transverse deflections of the walls , since it prevents the free deformation af
theÍr upper edge. After simulating this form af reÍnforcement numerically, it
was founel that on its own it was sufficient to ensure the carrect behavior of the
building during an earthquak e. This diaphragm must be able to work as a wide
girder , effectively tying the four façades to each other.

When the diaphragms are simlllated, the bending forces of parapets and co rner
lintels disappear due to the work as a grid ar frame. The transverse façades to
the seismic for ces are practically free of bending effects induced by horizontal
loads, but th e effects due to purely gravitational loads are very preponderant.
The transvcrse façades to the seismic forces mobilize a form of gantry effect
that they are abJe ta resist ade(llIately (Fig. 8)

Fig. 8.- Stress intensities in the strengthed structure

For the diaphragm on the second floar it is proposed to build a metal framework
usi ng bars wi t h an angular or tubular section. This framework would be totally
co ntained in a plane belaw the current Roars, reqlliring approximately tive
additional centimeters af width.

It wauld be formeel by a latlice st ructure, whose prin cipal elements \Vould

374 STRUCTURAL ANALYS IS OF HI STO RI CAL CONSTRUCTIONS

coincide with the basie layout cf the axes af the existing floors, plus the necessary
systern af cross bracing of the axes. The transverse bracing of t he laUice would
be provided by the ficor slab.

5.· CONCLUSION

The rehabili tation project collects ali the previou s recommendations and quite
a few that have not beeo mentioned , integrating them with the remaining
requirement s arising Ctom the new use of the bui ldi ng. When alI thi5 work
has been done it will still be a document af how a Catalan architect moved his
cultute, including the building traditions af his regioll, to León, as was proposed
by ali the participants in the rehabilitation operation.

However , concernill g thi5 cult me there remains a certain doubt that must be
darified.

The possess ion of such a powerful tool as the one we have describcd lcads to its
utilization beyond its initial application. In our case wc were unable to resist
su bmitting t he computer mode! of the Casa de los Boti nes to t he conditions
that might have been found if it had been built in Barcelona.

It is kn own lhat in Barcelona the seismic ri sk is considerably higher than in

León.

Now, after su bmitted it to lhe provisions of PDS-l fo r zone V-H it was found
t hal the building became considerably weaker, and collapse was probab le. Was
Gaudí aware of t he seismic differences belween lhe two cities when he desig ned
il , so t hat he did nol adopt the buildi ng restrictions of the original mo deis
of Barcelona? The queslion is left unanswered, wilh the hope that these
Preliminary Studies will also serve to encourage research into t he history of
our building traditions and mentality, a knowledge of which - and here we
n::turu to lhe beginning - is esse ntial for a correct ar chitectural restoration.

NOTES

1. An appropriate method is that applied by the Servei de Patrimoni Arquitectonic Local of

t he Diputació de Barcelona.
See, J. L. González, "La comprensión prev ia de la realidad física dei monumento . Una propuesta
desde nuestra experiencia", in IV SIMPOSro SOBRE RESTAURACION MONUMENTAL ,
Barcelona, November 1993.
 J. L. GONZÁLEZ MORENO-NAVAR RO I Cas:t de los Botines de León 375

2. lABSE SYMPOSlUM. Stru ctural Preser vation of the Architedural Heritage. International
Association for Bridge and StructuraJ Engineering. Rome, September 1993.

3. J . L. G onzález, A. CasaIs, P. Roca , A. FaJcones, C. Molins, and 1. Hernanz . Estudios Previos

para la Rehabilitación de la Casa Botines de León.
Department of Architect ural Technology I, UPC , Barcelona. The a rchitect Javier Ramos
GuaUart participated in th e iuiti,,1 definition of the ba.sie lines of thc study. Numerical
models and general methodology developed through the support granted by the In ter ministerial
Commission of Science and Technology to the project "DesarroUo y ensayo de t écnicas para el
análisis estruduraJ de const rucciones antiguas dei patrimonio a rquitectónico artístico" (SEC93-
11 60) were used . For the numericaJ calculations we used the computer resources of lhe CESCA ,
Supercomputing Center of Catalunya, and of the Computin g Cellter of lhe School of C ivil
Engineering of Barcelona

4. Caja Espana d e Inversiones, S.A.

5. T he same team has anaJyzed another singular work by Gaudí . See A. González, A. Casais,
P. Roca, 1.L. González, 51udies of Gaudí's "Cripta de la Colonia Güell" , lABSE SY MPOSlUM ,
Rom e 1993, and also A. Casais, 1.L. González, P. Roca , La necesaria comprensión previa de
la realidad construcliva dei monumento: el caso singular de la Cripta de la Colonia G uüell , in
INFORMES DE LA CONSTRUCCION, No. 427, Sept / Oct 1993.

6. T he same conclusion is reached if we apply methods of evalu at ion of seismic vulnerabilily of

buildings developed to permit a systematical but rapid approximation to the analysis of existing
buildings. Of special interest is the one explained in the monograph cited at the end of the
note. T he method is based on e valuating the buildings according to 11 variables such as t he
general organization of th e load-bearing system, th e strength of its materiais, the quality of the
horizontal diaphragms and the distance between bracing waUs. According t o the assessment
made of t he buildings of the Eixample of Barcelona, many of them built at the same time as the
Casa de los Botines, a conside rable proportion prove to have a degree of vulnerability betwee n
medium-high and high . The Casa d e los Botines would have a medium-high rating. Also see,
C. Caicedo, A. H. 8arbat , J .A.Canas,
Vulnerabilidad sísmica de edificios, Monography CIMNE 15-6 1994.

7. lts authots ate the architeds Mariano Díez Saez de Miera. y Félix Compadre Diez .

8. The flat jack and brickwork sample t est s were ma.de by t he Central Laboratory for Structures
and Materiais of the C EDEX of the Ministry of Pubüc Work s, Transport and Environment in
Madrid .

9. The extensometer tests were callied out by Mr Santiago Sánchez Beitia, senior lecturer at
the Department of Appüed Physics I of the School of Architecture of the University of t he
.176 STR UCTURA L ANALYS1S OF HI STO RICA L CONST I~ UCTlONS

Basqu c:' Cou nt ry.

10. See, Joscp Lluís Gon zález, 3. L'aparició d(' les lLo rma tives , and i. Verificac ió de la seguretat
estructu ral a It's càrregues verticais. Aplica ció de les P IET 70 i I'N BE-FL-90. Comparació amb
l' EC-6., chapters of lh e book, Manual d(' d iagnosi i in l('rvel\ ció ell sistemes estructural s de parets
de càrrega, Manual de Diagnosi - 4 , Collegi d'aparellado rs i Arquitectes T ec nics de Ba rcelona,
Barcelona , 1995.

11. 'fite miner;llogical test s were conducted by the R&D Gro up working on the Historical
Patrim ollY of the Crystallography and Milleralog)' Dep'lrtment of the Ullive rsitat de Barcelona,
direded by D. Màriu s Vendrell Saz.

12. 'fh e allalysis alld testing of th e wooden beams were COlld ucted hy staft' of the Laboratory
of the Technical College of Barcelona of UPC , direct ed by Mr . JU311 R3m óII Rosell i Am igó in
cooperation with the laboralories of AITIM ill Madrid.

13. C. Molins, P. Roca, and A. Marí, Ulla Formulació n Mat ricial Generalizada: 1 - Anàlisis
estàtico, in Revista Internacional de Mét odos Numéricos para cl cálculo y d iseiio en ingeniería
, Vol. 10, n04 (1994) .

14. C. Molins, P. Roca, and A. Ma ri, Una Formulació n Ma tri cial Generalizada: 2· Análisis
dinámico, in Revista Internacional de Mét od os Numéricos para el cálculo y d iseno en ingenie ría
, Vol. 11 , nOI ( 1995).

15. Kwan , A.K. H., Allalysis for cOllpled wall j fr ame structures by fram e met hod with shear
deformatio n allowed , Proceedillgs o r the lustitutio n of Ci vil Engilleers , Pa rt 11, June 199 1.
 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTiONS
p, Roca, J.L González, AR Mari and E. Oi'late (Eds.)
C> CIMNE, Barcelona 1996

STUDIES OF GAUDI'S "CRIPTA DE LA COLONIA GÜELL"

P. Roca
E. T.S. Etlgi flyers de Camills, CaI/ais y Ports
Univers ifaf Politecnica de Catalllllya
08034 BarcelOlla
Spai n

SUMMARY

The Crypt of the Coloni a Güell is only t he existing part of an unfinished

construction imagined and commenced to be erected under Gaudj's direct ion
during the period 1908 to 1915. Once finished , it would have been the
church of the Güell Factory's Village in Santa Colorna de Cervelló, near
Barcelona. A remarkable feature of this building is found in the use made
by Gaudi of a tridimensional funicular model to design the geometry of its
unique structural syste m . The strings in the model were materialized as a
hierarchy of diafragms , arches , and oblique pilars made of brick fabric and
stone. Due to this design, the theoretical stability of th e building should be
largely dependent upon the achievement of a funicular-type balance of forces
among ali the elements of the whole structural system.

A numeric analysis, based on a Generalized Matrix Formulation, was

performed to study in detail the today actual resist ing mechani sm und er
gravity loads for the built part of the st ructure , as a way to measure
the alteration of the funicular equilibrium and relate it to the damage
that is observed. Long term phenomena like mortar shrinkage \Vere also
incorporated in th e stucly and found very singnificative . Through the
comparison between the numeric predictions and the observed in-situ damage
a new undestanding of the real state of the structure and actual resisting
mechanism were obtained. Particularly, the stability of the st ru cture was
demonst rated as a result of the capability of t he building imagined by
Gaudy to aclapt to different states of equilibrium throughout the hypothetic
complete const ruction processo
378 STRUCTURAL ANALYSIS OF HISTORICAL CONSTR UCTIONS

Fig. 1.- The entrance to the Cr ypt under the portico

THE BUILDING AND THE FUNICULAR MODEL

The Crypt af the Colonia Guell is the anly part actually built of what was
to have been the parish church of the madel Village íounded by Eusebi Guell
in 1890. Gaudi was commissioned to design it in 1894, when he was also
working in the Expiatory Temple of the Sagrada Familia. The foundation
stone af the Church was laid in 1908. Gaudí spent fourteen years to lay
out the structure, due to which its construction was not started on until
The works were abandoned in 1914, leaving anIy the crypt and the portico
finished (Figs. 1,2). It is located in Santa Colama de Cervello, just a few
kilometers from Barcelone.

The general form of the fioor pIan is oval, with a star-s haped outline ( Fig. 2)
and it measures 26 by 63 m. The used materiaIs are mainly brick íabric and
stone. The slab rooí is supported by an eskeletaI system which consists oí a
hyerarchy oí ribs, arches and columns. Some oí the columns are oblique. The
externai walls oí the crypt itselí reveal elevations with different inclinations
in such a way that the base oí these configures the star- shaped perimenter;
 P. ROCA I Gaudi's "Cripta de l:l Colonin Gücll " )79

G O
©
O

&

F ig. 2.- Plan and elevati on of the building

the upper part of the walIs are pierced by rhomboidal window s. On the so ut h
facade there is t he portico, its rooí consisting of of hyperboli c paraboloids
made of brick masonry. This was intended as the access to the church, located
to the fia 0 r above.

The central nucleus is delimited by four basalt columns and the arches which
separate it (rom the choi r, ali of them angled in towards th e cent re of the
nucleus , and two peripheral aisles forming a do bule ambulatory, made up
of ten columns arranged in a double semi-circle around the nucleu s ( Figs.
3,4). Colurnl1s and pillars support the main brick arches from which the rib s
spring. These are also of brick, and have bee n treated in two different waySj
t hose which con verge on the two circular bosses - linked by a practically
Rat brick areh- consisting of a brick rowlock arch 15 em thick, supporting
a wali of the same thickness. The remaining ribs are waUs 12 em thick ,
J80 STR UCT URAL ANALYSIS OF I-IJSTORI CA L CONSTR UCTIONS

but raised over brickwork over brickwork arches three courses thick. The
reticulum form ed by the ribs is Hu sh at the same levei and supports the
saleTa ar masanry floor slab (Fig. 5). It consists of a first triple layer of
facing brick) with a suspe nded floor over this) formed by brick partitions,
su pporting the final double layeri pres umably, teh paving of the church was
to have been laid over t hi s.

Fig_ 3.- View of lhe cent ral columns

The limitations of the graphical or numeri cal methods available to Gaudi)

obliged to him work out hi s stereo-fun icula r st ru cture using a physical scale
mo deI. Thus) Gaudí devcloped a tridimen sional funicular model in which
st rings were used to represent actllal brick mason ry arches and oblique brick
or stone columns (Fig. 6). In fact , lhe whole arrangement of colllmns ,
a rches a nd ribs is the materiali zat ion of the funicular model which Gaudi
built and st udied over lhe ten years before ils construction was started on.
There is small doubt that the crypt is th e first large structu re laid out
based on a similar thrce- dimen sional funicular model) were funi cular lin es
illterweave in the t hree spatial dimensions. More details in regard to historie
or constructioTl aspect s may be fOllud in Casais et alo (1990) and Go nzález
et alo (1 993).
 P. ROCA I G:wd i's '·Cripta d~ la Colonia GüdJ" 38 1

Fig. 4.- Section of the hypothet ic complete building

During the Spanish Civil War, the original hanging model wa.s destroyed and
just a few photographs of it survided . However, thank s to the initiative and
the huge task of a graup of Dutch and German admirers af Gaudi , we now
have th e reconst ructed versia ll of it (Tamlow , 1989). Besides, the correlatian
demon st rated by co mparin g phatagraphs af the new model and that hy Gaudi
permits reasonable speculation about the new images af the unbuilt part.

The in cünation of the columns in the crypt was imposed by Gaudi for
composition al reasons, and it was achieved thanks ta the thrust generated by
the arches or ribs which would s upport the Roor of lhe chur ch. Surprinsingly,
the aciual in clination af those column s in th e built structure is high er than
the one observed in the funi cular model.

Thus, the co mparison between t he reconstructed model and t he adual

building arises some doubts. If the structure is in equilibrium at the moment ,
with lines apparently distant from the funicular model, ... would it be in
equilibrium once it was finished?
382 STRUCTURAL ANALY$ I$ OF HI STO RI CA L CONSTR UCTIONS

Fig. 5.- Section af the upper selera

We have herc precisely the great paradox af the crypt. It is a building which
was co ncei ved to be stable Dnce it was fini shed, but nevertheless it is stable
now , in its unfinished statc.

ABOUT THE ACTUAL STATE OF THE BUILDING

However , ane may still t hink that the structure behaves according to
the funicular madel although only partially. This may be dedu ced Crom
all ind icato r which has generally becn overlooked but is af consider able
importance when it comes to checking the possible hypot hesis about hte
behaviour of the building: the intrincate pattern af carcks in the fa cing brick
ceiling a r salera which CQvers t he central nave, and those which affect several
c f the ribs and some a f the arches.

Dc tailed analysis through direct observation and an exhau s- tive photo--

graphic survey of over 1000 images has enabled to determine the patterns of
 P. ROCA I Gaudi's "Cripta de la Colonia Güell " 183

Fig. 6.- Original picture of the hanging model elaborated by Gaudí

(inverted ).

cracks reproduced in Fig. 8, which is only an overaU view amongst all those
that were detected.

The building has not collapsed , but neither can we say it is undamaged.
The detailed inspection of the structure through dired observation and
photographic survey has enabled to determine the pattern of cracks which
is partially reproduced here (Figs. 7,8) . Those cracks could be seen not
only in the conti nu ous slab of brick masonry over the crypt, but also in the
secondary arches or ribs.
384 STRUCTURAL ANALYSIS OF J-I1 STOR ICA L CONSTRUCTI ONS

Fig. 7.- Example of cracked rib

M ET H OD OF A NALYSI S

The structural analysis of s)'stems composed of curved members such as

the arches, diaphragm s or nervures which may be found in many ancient
structures, is commonI)' carried out usi ng t he finite element method with iso-
parametric type with di splacements as unknowns. It may be seen that , due
to the deficiencies of the method in the description of the internai equilibrium
of the elements, accurate results of internai forces are only obtained when
a con siderable amount of individual elemeuts are used in thc geomctric
d iscretization. However , for structural system s composed of unidimensional
curved members, it is possible to establish analytical generalizations of
conventional matrix mei hod s based directly on exact equilibrium. Although
the practical use of these matrix formulation s was limited in the past by the
large volume of mat hematical operations required, recent developrnents in
digital compu ters make this poi nt less criticai nowadays , while t he as p ects
of accuracy and versatili ty gain renewed interesL

The analytical model used in the studies presented is directly based on

a Matrix Generalized Formulation specificaU y developed to treat ancient
 P. ROCA I G:ludi·s ··Cripl:l de la Colon i:l Gücl l" 385

'-.

Fig. 8 .- Crack pattern in arches and salera

buildings cansistin g af multiple struct ural systems with curved, variable cross
sect ion members. The formulation , initially based on the work of Baron
(196 1) for static linear analysis, has been extended to nonlinear geametric
and modal vibration analyses. Relevant aspects af the resulting method are
the fallowing:

(1) Automatic generatian af complex geometries thraughout the length af the

elemento Three cross sections, having arbitrary shapes, are to be given at
three respective points af the axial curve af each elemenL

(2) Each different cross-sectian is defined as a co mpositian af elementary

trapezoids , where each of them may be associated to a different type of
material.
 386 STRU CTURAL ANALYSIS OF HISTORICA L CONST RUCTIONS

(3) Specific devices are included to model load bearing or shear walls as
equivalent systems of linear elements according to the method proposed by
K wan (1991 ).

(4 ) Nonlinear geometric analysis based on an updated Lagrangian formulation ,

thus aUowing the treatment of cases involving instabilty phenomena of arches
ar other curved elements.

(5) Moda! dynamic analysis based ou the formulat ion af a consistent elementary
mass matrix which objectively takes iuta account the distribution of mass
and stiffness throughout the element.

Constitutive equations for brick ar stone masonry at the macro-modeling

levei are now being implemented in the general roodel 50 that an integrated
nonlinear geom etric and material nolinear analysis method will available in
a short time.

Nevertherless, the linear elastic analy sis carried out for the Crvpt, as
described in the following sections, was very informative and revealed itself
suitable to obtaind an interesting understandillg of the adual state of
equilibrium af the building, as well as to investigate the causes of the existing
damage.

Before its systematic use for the study af existing buildings, the madel,
implemented in the computer program CRIPTA , was checked through the
analysis of a series of sim pIe and multi pIe systems af curved members , for
which analytical or experimental res ults were available. The comparisans,
described hy Molin s et alo (1994,1995), showed the very sati sfadory levei af
accuracy and numerical efficiency which are achieved even far geometrically
camplex structures.

NUMERICAL MODELLING AND RESULTS

In arder to reach a better understanding of the presellt state of the structure

and also understand its actual equilibrium , it was decided to carry aut several
comp u ter analyses. The first type of analysis was based in plane stress and
was used to study the individ ual elements like ribs. Due ta time limitatians,
it is not presented here.

The second type af analysis was a global one using the above mentioned
Generalized Matrix Formulation (implemented in computer program
 P. ROC A I G:mdi's "Cripta de la Coloni:l Güell " 387

CRlPTA ). The structural members were treated as linear elements with

curved centroidal axes, as well as arbitrary cross sections, thus allowing to
model the arches, ribs, diaphragms and colurnns incorporated in the existing
structure. This allowed to accuratelly take into account their actual stiffness.
Th us, the model show n in Figs. 9,10,11 was constru cted fr om the elevation
and other disposab le inform ation. In addition, solid undeforrnable elements
were introduced to simulate the massive capitab were l:ulumns, arches and
rib s connect.

Fig. 9.- View of the numeri cal mod el

The adopted formulation made it possible to reproduce states of stresses

caused by combined axial, shear , bendin.e; and torsion for ces, and thus to
simulate possible modes of global equilibrium more comp lex to that of a
funic ular model , having also in to account the actual distrubution of stiffn ess
between st ru ctu ral elements.

The lead joints which exists at the juntion between the cent ral columns and
thi r pedestais and capitaIs, were treated alternativelly as perfectly fixed or
rotational free hinges. First, the existing part of the st ructure was studied
subjected to the vertical load produced by the weight of the eskeletaI system
af ribs , arches and columns as well as the weight of the upper slab, which
388 STRUCTURAL ANALYSIS or 1·IISTORICAL CONSTRUCT10NS

Fig. 10.- View of the numerical model (with stress intensities represented in
chromatic scale )

rests on the firsL By this st udy, the zones subjected to high tension stresses
were recognized.

The tensional stresses obtained in the different elements are mainly caused
by t hei r in dividual behaviour under vertical load, and are hardly infiuenced
by interacting forces due to global effects. Thus, the highest tension zones
appear at the joints between ribs and capitaIs or arches, and lower tension
zones appear at the middle of the s pau af ribs.

Moreover, a correlation was found between the analytical prediction of high

tension leveis and the cracks observed in the structure. However, the first
showed many other poten tially cracked zones whi ch were apparcntly intact in
the building. Thcse are interpreted as parts w hich, although cracked, do not
show an evident structural damage or which, alt hough intact , are subject to
a high leveI of stress and might be easily damaged by overloading ar altering
the present geometry of the structure.
 P. ROCA I Gnudi's 'üipta de la Cololli:-t Gücll" 389

Fig. 11.- View of the numerical model (with stress intensities represented in
chromatic scale)

THE RHEOLOGICAL ACTIONS

There are also radial cracks in the upper slab which cannot be explained by
the gravity forces. Having discounted other causes, t he only probable one is
hydraulic shrin kage .

The values shich are habitually handled to dimension the movement due
to hydraulic shinkage in masonry structures range between 1 and 7 yo 8
tenths of a millimetre for each linear rnetre of \VaU. No values have been
found which refer to sheer brickwork elemellts, since this is a constructioll
procedure which is not ordinalrily used, andan which test have not been
gellerally been carried auto However, the values could not be lawer in view
of the larger relative proportion of mortar in the section, as weU as the great
proportion of portland cement in the mortar.

In fact, the R-X difractometric analysis aí some mortar patterns showed that,
surprisingly, it consisted af almast pure Portland cement. This is against the
usual technique for building such a brick rooí ar vault at that time , since
390 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

t h e fist skin was always treated with p laster to avoid the use of scafolds and
forms.

Moreover, the measurement of the total wi d th of the cracks amounts very

closely to the shrin kage cont raction that coul d we expected for such a mortar.

Rence we can draw a further conclusion: contraction is the process which

makes visible many cracks oí gravitational origin, ar contribu tes to their
appearance when it is added to the mechanical ten sions in the cases where
tension in the ribs and foreseeable lines of fract u re in t he upper slab coincide.

F ig. 8 represents a plan of the salera which shows the gaps provided to leave
roo m for the columns of the church above, which no doubt is relevant to
the matter of shrinkage. O n this pIan , if we observe the numerical forecasts
and the present evaluations which recommend the provision of walls with
expansion joints every eight metres, and if we look for the areas wh ere,
because of t heir smaLIer cross-section and a geometry which m ight provoke
cr acking, assuming a uniform grid, we reach the conclusion t h at it was to be
expected that tensions would occur along these lines .

To aLI t his should be added the plan showing the areas where, according
to the computer mode!, the upper part of t he ribs is under tension. If
we ass ume complete adherence between the ribs and the sol era , we may
sup p ose that this tension is transmitted to the latter. Hence, in addition
to the tension dyeto hydraulic contraction we must consider the tension
due to mech anical beh aviour under gravity. Due to the fairly con siderable
correlation between the hypothesis an d reality, we may conclude that t he
radial cracks are due to contraction, while the cracks between t he head s of
the columns, the capitais, are due to a combination between contraction and
tension or flexural phenomena due to the gravity loads of the ri b arches.

It its possible to draw a further conclusion: contraction is the proceS5 which

makes vi5ible many cr acks of gravit.ational origin, or contributes to their
appearance when it i5 added to the mechanical tensions in the cases where
ten sion in the ribs and foreseeable lines of fracture in the salera coincide.

CONCLUSIONS

These studies an d considerations aLIows as to present some conclusions upoo

t he actual state of the structure and also about the hypothetical complete
building. F irst, t he b uilding is stable u n der its own weight and the permanent
 P. ROCA I Gaudi 's "Cripll de Il Colonia Güell " 39 1

loads at present affecting it. This is 50 in spite of the fact that the present
loads were t he cause, at some time, of the existing damage, The overall
stability is more t han assured, thanks to t he strength of the interior columns
and the perimeter waUs.

Fig. 12,a,- Movement of the central nucleus under today's dead lo ad

Fig. 12,b.- Movement of the central nucleus under the effect of concentrated
loads simulating the completition of the building

The o btained scheme of forces at the leveI of arches and ribs keeps similar
to that of a funicular type of equilibrium , although the geometry does no
correspond to that of he funicular mo de!. It may be seen that, owing to
their much larger sectionaI dimensions , the deformations of columns and the
perimetral wall are very small in any case, 50 that the equilibrium af arches
and ribs is not affected by the fact t hat the devised global structural sys tem
is not campleted .
392 ST RUCT URAL ANA LY SIS OF HI STOR ICA L CONSTR UCTIONS

When lead joints are treated as perfect hinges, a significat ive movcment of the
central columns and upper capitaIs is obtained (Fig. 12,a) which produces
balancing axial and fl exural fo rces in the adjacent rib s. Some real effects
that are observed in t hese zones may also be correlated to such a movement
of the capitais, üke a more extenclecl cra cking in the ribs a nd diagonal cracks
in the upper slab .

The nonexisting part of the structure was simulatecl by th e hypothetic forces

that it would have caused on the existin g parto These forces were knwow
through some labels that were vi sible in the remaining phot ografies of the
original maquet t e. It was establi shed that ab solute funi cular cquilibrium
would not have been obtained for the finished building either. In particular,
t he capitaIs tend to move in ao opposite direction to that produ ced by the
d ead load of the Crypt levei itself (Fig . 12,h ). This suggest s that the theoretic
state of funi cular equilibrium is only reach ed at an intermediat e phase during
the construction of the building.

To overcome the construction difficulties, Gaudi could have im agined a design

in which a t rue funi cular equilibrium was not entirely ob t ain ed eith er for the
fi nal configu rat ion of the structure, or t he const ru ction at t he crypt leveI. The
di sturbing cffccts due to imperfect equilibrium would thus be resis ted thanks
to the confinement action of the very stiff vertical elements. The perfect
funicular equilibrium would have beeo reached at a peculiar intermediate
stage of the construction process , being any previous or laHer alteration of
it counteracted by the enlarged capacity and stiffness given to the resisting
vertical elements. This is just an speculative suggestion t o explain how the
architech could have accounted for the inherent d ifficulties of constructing a
system laid-out through a funicular model.

REFERENCES

BARON , F. , (1961)
Matrix Analysis of Structures Curved In Space} Journal of tll e Structural
Division ASCE} VoI. 87} N°. ST3.

TOMLOW, J ., (1989)
T he hanging model oí Gaudí and its reconstruction. New information
for the design of the Church of the Güell Colony. In stitut [ür leichte
Flachentragwerke, Universitat Stuttgard ..

CASALS, A., GONZALEZ J. L., (1990)

 P. ROCA I Gaudi's "Cripta de la Colonia Giictt" 393

Gaudí and the my stery of the encarnation (t he m ys t.er ics of the Crypt of the
Colonia GÜell ). Rev. Informes de la Construcción , Vo1.92, no. 408, (written
in Spanish).

KWAN , A.K.H. (1991 )

Analysis fo coup led wallj frame st ructures by frame method with shear
deformation allowed. Proc. Inst. oI Civil Ellg., Part II.

GONZALEZ, A., GONZALEZ, J . L. , ROCA, P., CASALS, A. , (1993)

Studies of Gaudi's Cripta de la Colonia GÜell. Pra c. af lABSE Symposium
on StructuraJ Preservatian af th e Architectural Heritagc, Rome.

MOLINS , C., ROCA, P. , MARJ , A. R., (1994 )

Una farmulación matricial generalizada: (f) análisis es tático. R evista In t. de
lvfét odos Numéricos en In g., VoI. lO ) N° A.

MOLINS C., RO CA P., BARBAT , A. H. (1995)

Una formu!ación matricial generalizada: (II ) anál isis dinámico. Revista Int.
de Métodos' Numéricos en Ing, VoI. lI , N°. !.
 STRUCTURAL ANALYS1S OF H1STORICAL CONSTRUCTlONS
P. Roca, J.L. González, A.R. Marí and E. Oilate (Eds.)
© CIMNE, Barcelona 1996

A SUMMARY OF THE OPINIONS PUT FORWARD DURING

THE DISCUSSIONS

P. Roca and J.L. González

Universitat Politecnica de Catalzmya
08034 Barcelona
Spain

1. . PRE SE NTATIO N OF THE SEMIN AR

This chapter aims to present some of t he ideas arosen through the technical
discussions held during the celebration of the International Seminar on
Str uctural Analysis of Historical Construcions. The event, which was held in
t he Assembly HaU of the School of Civ il Engiueering of Barcelo na on November
8 to 10 1995, was a grea t success, with more than two hundred participants
come from many countries of Europe and A merica.

The different ledures presented were organ ized in five monographic sessions. A
first series of interventions aimed to establish the methodological and analytical
bases for a const ruct.io nal stu dy of an ancient building belonging to the historical
and architedural heritage. Giorgio Macchi and D. Ant.onio González were
t he first speakers, talking ou "General methodology: T he combined use of
numerical and experimental techni ques" and "A general perspective of the
stuclies. The role of the analystsll respectively. Anton ino Giuffre then dealt wi1.h
llMethodology for analysis and in1.ervention based on historical knowledge".

The second group of pa pers aimed to present the state of the art of science and
t.echnology concerning the elements available to the technician for approaching
such studies. This sedion induded aspects such as inspectioIl) experimentation ,
characterization of adions an d numerical modelling . All the papers were
illustrated with real cases of great interest. Roberto Meli was thc first speaker,
presenting a study of the Metropolitan Ca thedral of Mexico as an example of
action that. can arrect historical constructions. Giorgio Croci then talked O Il
J' Behavior of dr)' masonry structures present ing the studies 0 11 the tem pies of
Angkor Vat in Cambodia and the Pyramid of Kep hren in Egypt". Climent
Molins talked on "Constit utive models fo r the mechanical characterization of
stone and brick masonryll. Finally, Tim H. Hughes presented an interesting
review of the studies made in the Unit.ed Ki ngdom in connection with the great
number of stone or brick arch bridges that exist in this country.
 P. ROCA nnd 1. L. GONZÁLEZ I The Disc ussions 395

The possibilities of experimental techniques in the study of historical

constructions - in particular inspection l monitoring and cont rol techniques -
were presented by Pier Paolo Rossi ancl by Rafael AstudiUo. Eugenio Oilate gave
an expla nation of the possibilities of numerical analysis techniques, presenting
as an example the stud ies carriecl out Oil the domes of the Basilica of San Marco.

Tile last grouf-I of vaf-lcn:i was devutt::J tu ut::.t:riLillg in lidai! the studies) and
if applicable lhe repairs or reinforcements , carried out on emblematic and well~
known historical const ruction s. In ch ronological arder) the speakers considered
the cases of the altics of the Casa Milão (Robert Brufau» the Casa de los Botines
(José L. González), the Charles Bridge in Prage (Jaroslav Zák)l the Coliseum
of Rome (G iorgio Croci), the Basilica of San Marco in Venice (Franco Mola and
Renato Vitaliani), the Column of Marcus Aurelio in Rome (Antonino Giuffre),
lhe Crypt oft he Güell Colony (Pere Roca), the spires of the Cathedral of Burgos
(Luís Ortega), the Aqueduct of Segovia (Francisco Jurado) , and the Tower of
Pisa (Giorgio Macchi). Finally, José Gómez gave an explanation of the Sagrada
FamHia on the site of the monument, which formed part of one of the technical
visits.

The Seminar induded a technical discussion and a round table, both of which
proved to be of great interest due to the quality of the ideas put forward and
the opportunity of pointing out some methodological weaknesses which are
sti ll today difficult to overcome in order to perform a rigorous stu dy of an
ancicllt const ruction. In addition to the already cited lecturers, José María
Ballester, Andrea Chiarugi, Javier López~Rey, Antonio R. Marí Bernat l Mircea
Mihai lescu, Manuel Melero and O. C. Zienkiewic also took part in the discussion.

The ideas put forward during these discussions are presented in the following
paragraphs.

2.~ OPENING OF THE DISCUSSIONS

As coordinalor of the first Technical Discussion of the Semina r on 9 November,

P. Roca opened the session by inviting the participants to consider the following
problems:

(1) Many models created in order to study emblematic structures - sueh as

those that have been presented in this Seminar - are based OH the acceptance of
the mechanics of a contilluous medium and on the hypothesis of linear elasticity.
To what exlent is this approach valid whcn applied to masonry structu res?
 396 STRUCTURAL ANALYSI$ OF HI$TORICAL CON$TRUCT10NS

(2) Is there any poi nt in working with more sophisticated models? (For exam pl e,
mo deis that reproduce the behavior of the mason ry with great realism, including
its different forms of failure). In order to function correctly these models require
a great amount of data on the mecharucal properties of the materiaIs. As a rule,
it is difficult to obtain these data completely.

(3) Given the unavoidable limitations of the available information, and the
limitations inherent to the methods of analysis that are u sed, to what extent
is it possible to evaluate the strud ural safety of an ancient construction ? 1s it
really necessary to quantify this safety exactly?

As the coordinator of the Round Table on 10 November, J . L. González wis hed

to consider problems that ranged from very specific aspects to very general anes.

(4) With regard to the specific aspects, he invited the congress members to
continue t he discussion ou the methods for modelling ancient constructions.

(5) With regard to more general aspects, he suggested dealing with the
importance of the dialogue between the different professionals, presenting this
issue as a key philosophical aspect due to the in creasing variety of professionals
involved in studies af historical constructions.

However , the discussions raised many more topics related to a greater ar lesser
extent to these first problems.

In this article we have decided to order the opinions gathered according

to related topics, seeking a greater linearity and coherence of presentatian.
Hawever, as a consequence, the final group of the interventians does nat
carrespond exactly to the chronalagical order in which they were farmulated.

3.- THE DIALOGUE BETWEEN THE DIFFERENT PROFESSIONALS

Collabaration between different professionals is something obligatary, since th is

type af study mobilizes an increasing amaunt af knowledgc, and it is impossible
for a single professianal to cover it ali. Furthermore, it is necessary to avoid
the isolation of each expert within his area, anel to seek a maximum dialogue
between the different disciplines.

For the praject management, t he main issue lies in the formulation of what
questions they must ask each specialist of the team (J. L. González).
 P. ROCA and J. L. GONZÁLEZ / The Oiscussions 397

The opinions put forward in this Seminar should serve to aliow those who wish
to work in the study of ancient constructions to understand the need to work in
the environment of a multidisciplinary team . In fad, this type of study requires
t he participation of ao increasing number of professionals. \Ve should be capable
of transmitting the complexity of decision-making in this regard and therefore
t he need to have a profound knowledge. Therefore, all the specialists who have
presented their projects at this Seminar have sho\\'n ao exemplary humility,
recognizing the need to obtain informatiol1 and to work wit.h collaborators of
ali types (J. L. González).

Though the importance of multidisciplinary collaboration should not be

underestimateel, it is also necessary to assess the individual work of reftection
that each person participating in the st udy must undertake. The knowledge
phase is more a phase of individual work than teamwork.

It is also important to give fuU consideration to the objective of the study:

:;k nowing" . lf the objective is approached correctly, the methods are secondary
if they are consistent with it. As A . Giuffre has stated, one must "be attentive
to the monument ", rather than to the means t.hat we have for knowing it.

As G. Macchi has recalled 1 the criticai and reflexive rclationship with the
monument by the professional is baseei ou knowledge. Those who scorn the
method guided by profound knowledge, and who prefer intuition, should be
reminded t hat at the basis of creativit.y there is always a very wide knowledge
base: being intuit.ive is a consequence. One must be attentive lo the monument
on the basis of a profound kuO\ . . ledge of the 1l1eans that. are with in one 1s
reach . The aim of t he restoration is to recover the emot.ional relationship of
t he monu1l1ent "vith the community, an asped t,hat the analyst engineer must
bear in mind when he sets about his work (A. González) .

4.- iHE ELEMENTS OF THE STUDY

The study of ancient constructions must include the following aspects (C.
Croci):

1. The role of history: history is the greatest laborat.ory on a real scale: its
interpretation can produce valuable elements for the diagnosis.

2. Direct observation: This obviously depends ou the experience of the <lllalyst. 1t

illdudes an inductive process based on the capacity to accumulate t'xperien('('
and to lnake judge1l1ents. lt adds and compares the case being si lldif'rI \Vil h
 398 STRUCTURAL ANALYS IS OF HISTORICAL CONSTRUCTIONS

otlter previous cases that are already known (which explains its inductive
nature).

3. Mathcmatical m odcl: This can be a very powcrful, vcry useful and CVCIl
indispcllsable tool. The controversy concerning its use stems exclusively from
tbe wrong use by persons who do not have sufficient knowledge of the problem
that t,hey wish to solve. The 1I10del does not represent absolute reality, but
only all app roximation. However, the knowledge that can be gathered from
this approximation can be of very great utility. I n any case, the model should
not be confused with reaüty; one should be aware of the hypotheses adopted
and be criticai with them, appraising their suitability.

4. Experimentation and inst rumentat ion. It shoulel be clearly stated that, in the
worst case, such activities can turn out to be useless as a consequence of
improper planning. This can occur, particularly, if a preliminary project of
experimental research has not been drawn up . However, if these activities are
accomplished with wisdom anel proper planning, they caIl proeluce very useful
results that provide a greater knowledge of the structure and its safety. Another
problem stems from their cost, which in certain cases can be unjustifiably high
taking into account the real needs.

The study of au ancient construction should be based o n the following premises:

(1) start from a constructional knowledge that is as complete anel perfect
as possible; (2) use sim pie analysis methods before using more sophisticated
methods; (3) d o not neglect the importance of thc "clinical eye" - the
experience of the researcher (F. Jurado).

5.- ON THE USE OF NUMERICAL AN A LYSI S METHODS

5 .1 T h e p urp ose o f numerical modelli n g

Numerical methods are one more of the different elements that are used in
studies of historical constructions. Their great value, however, lies in their
capacity to quantify, which is an as pect of great importance. In fact, rather
than knowing what occurs to the structures, what we wish to do is to define
suitable interventions, and knowledge is a condition for this. From this point
of view, numerical models can be seen as a "numericallaboratory" in which it
is possible to simulate the interventions ( E. Dilate).

In my own experience in the field of monumental restoration, I have never

decided a solution based only on the results of a structural analysis . There are
 P. ROCA :md J. L. GONZÁLEZ I Thc Discussiolls 399

filany other factors (hi storical, construction , functional...) which most of the
times have a larger influence on the decision (F. Jurado).

The use of numerical models is unavoidable due to the high cost of making
phy sical scale models. However, in order to use them it is necessary to establish
many hypotheses , towards which the analyst must maintain a criticai attitude
(M. Mihailescu).

In fact, these sludies are based on lhe scientifi c method, which involves: (1) a
preLiminary campaign of gat hering of datai (2) proposing all abstractioll, in the
form of a model; and (3) testing the representativeness of the model with respect
to reality through a new experiment. However, such a great number of variables
are involved , 50 many faclors bear an influen ce, there are so many unknowns ,
that in no case wiU the model be able to gather the whole complexity of the
problem. Nevertheless, without disregarding the imporlance of the historical
analysis, there is no doubt that numerical mo deIs are an imporlanl elemenl in
the studies (M. Melero).

\Ve wish to define light inlerventions that are as respectful as possible to the
monument. Bul this must be dane on the basis of a fuU knowledge of lhe
current state and lhe real needs of reinforcemenl, which normaUy requires
quant ifying the phenomcna. However, il must be borne in mind thal lhe choice
of an analysis method anel lhe process of preparation of a mode! involve from
lhe begin nillg a profound interpretation of the conslruction and its resistant
mechanisms (P. Roca).

5.2 The interest and app licab ility of class ica l t h e ories

\Vhene\'er lhe characleristics and si mplicily of the problem permit classical

mcthods based on the equilibrium of thrusls to be used, these are preferable to
more so phisticated techniques of analysis (J. L. González).

The theory of thrust s, as it was formnJatprl hy C;o1l1omb, is applicable in lhe case

of a sim pie or isolaled arch , but is difficult to use in the case of more complex
structures. In the case of a bultress (or a syste m of buttresses and f1ying
buttresses ) there is a priori an infinity of solulions for lhe line of thrusts, and il
is extremely difficult to determine the real solulioll. Because of this , attempting
to apply this theory to the case of complex structures is nol ver)' ralionaI. From
this point of view, it is belter to a pply the mechanics of the cont inuous soLi d
since it simultaneously envisages equili brium and compatibility cond itions (G.
~ I acchi ).
400 STR UCT URA L ANA LYSIS OF HI STORICAL CONST RUCTI ONS

It is not corred to attempt to solve with sim pie methods problems that act ually
cannot be solved by them. Nor is it correct to ignore the poss ibilities of the new
techniques of analy sis and experiment ation. Let us con sider , for example, the
case of the Cathedral of Mexico presented by R. Meu: t he problem s dealt with
t here and simulated in the analysis are fundamentally ones of cornpatibility,
which demands the use of sophistica ted analysis (J. M. Izquierdo ).

It is possible to distinguish between three possible objects of study, to whi ch

differeot a n al ysis methods correspond (J. M. lzquierdo ):

(1) For a sim pIe safety study, t he use of method s based on the decomposition
of stresses may be suffi cient.

(2) If t he study is of stability, these method s are in sufficient, eve n if equiubrium

is imposed secondarily,

(3) lf the stud y is of behavior , even simplc mode!s thí\t consider the equilibrium
secondarily are in suffi cient since it is necess ary to include the condition
concerning compatibility of deformations.

In 90% of cases th e problems to b e st udiccl invol ve cOll structi ons under the
effect of gravitational loads, and t herefore m ethod s based ou direct equilibrium
are sufficient (F. Jura do),

Any calculation method that i5 useful and proves to providc information 00 the
authentic st ru ctural behavior of t he con stru ction can be cOllsidered sa tisfactory.
On the basis of t he a vailable experience, it should be said that eJast ic calculation
is reaUy a possible form of approxim a tin g th e solut ion. However , experience
al so shows t hat , thou gh useful in gi ven sit uations, method s based on dired
equilibri u m are in general in sufficient ( A. Chiarugi ) .

In many cases, reality imposes an isostatic model based purely on the

equilibriulU (A. Giuffre ).

5.3 The validity of the mechanics of the continuous medium

Masonry is a very difficult material to mode!. In any case, a mode! that

does not have adcquate references wiU !lot be useful. The numerical model
mu st bc drawn up taking as a reference the behavioral reaüty af the masonry.
It wiU have to be weU calibrated by expcrimentation. At t hc University of
Rome ) 50phisticated model s based on cri tc ria of homogenizat ion are being used ;
 P. ROCA 3nd J. L. GONZÁLEZ I Thc Discussions 401

however I such models are still not applicable to the general analysis of large
masonry structures ( A. Giuffre).

In any case, the num erical model must be chosen 50 as to produce good
results t hat are e10se to reality. For thi s, it is useful to observe maso nry as
an assem bly of a wid e sample of individual blocks . Ali modelling must en . . isage
th e kinemati cs c f these bloch. It is thus necess ary to takt illtu accuu nt t hat the
collap se of a masonry structure (for example, under t he a chon of an earthquake )
cao be aceompanied or followed by large m ovements of it s bloeks (A. Giuffre).

On eo nsidering thi s quest ion , it is necessary to consider the existenee of different

types of mason ry, t he differentiation of whi ch is essent ial for a nalysis method s.
Th ese are:

1. Brick masonry, essentially not homogeneous and not isotropic.

2. Irregular masonry (co nglomerate), whieh is present in Roman eo nst ruct ions,
medie val bridges and other const ru ctions such as the Tower of P isa. The model s
speci fi calIy formulated for briekwork masonry are fiot valid for rubble work. The
models form ulated for eouerete can produce good app roximations in this case.

3. Dry joint work s, in whi ch the mechanies of the contin llllm is certainly not
applicable (G. Macchi).

When co nsid crin g existing constructions and mo nume nt s, th e structural

engi neer should approach his projects on t he basis of two poss ible ob jectives:

1. The knowledge of their current situation. This is a typ ical " in verse" prob lem, or
a problem of ide ntifi cation: knowing t he answer, we wis h to find the qu estio n.
To solve it , we act through the procedure of diagnosis. It is neeessa ry fi rst
of a lI to have a virtual mode!. ThllS, the models should not on ly account
for the equilibrium , but also for the compatibili ty of movements a nd for the
behavioral mechanics of the st ructure. It is a lso necessary to have a whole ran ge
of informat io n on the co nstru ction and , using this information , to improve the
model by obtain ing a progressive approximation to reality.

2. The kn ow ledge of it s safety. This problem , which is esse ntially different to the
previous one, is of t he "direct" ty pe, that is to say, it is similar to the study of
a new cons truction. It is necessary to analyze the future behavior, that is to
say, the evolution of the structure. Thi s requires the use of very sophisticated
and accurate methods. But is it really necessa ry to es tablish the safety margin
of the structure of a monument? ( A: Chiarugi).
 402 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

The model used wiU have to be adapted to the resistaot mechanism of the
structure that we wish to analyze. \ I\fhen wc formulate a mechaoism we merely
1110del rea.lity. \iVhen t his reality is oot knowll, lleither do we koow how it should
be modelled . An analysis that is not baseei OIl a mode! adjusted to reality can
lead to inappropriatc interventions (for ex a mple, seek ing lateral binding in the
example of the Antonine ColuIIln) (A. Giufre).

An elastic analysis cannot be acceptcd when the structure is not elastic. If the
problem is of a non-elastic type, the analysis method sho uld also be non-e!astic.
For complex problems, complex rncthods are req uired, but l10t fictitious and
inappropriate methods (A. Giuffn!).

5.4 T h e v alu e o f sop hist icated ll10 dels

For the mom ent, it is advisable to use models of linear elastic analysis to mode!
large masonry structures. As they deal with very large structures, non-linear
analysis models could produce impodant problems of convergence. In any case,
whatevcr the type of 1l10del used, it is fundamental to validate it by using it in
combination with experimentation (P. P . Rossi).

As more complex tools become available, there is a tendency to use them for all
types of projects, induding those t hat in fact cao be solved maouaUy. As a rulc,
it is necessary to adjust the scope of the t.ools that a re used to the needs of the
problem and to the means available. Particularly, I have seen very so phisticated
non-linear methods used in situations in which hardly any data \Vere available
(R. Astudillo).

Oue must not feel persuaded to introduze more variables than actually needed,
just because more and more powerful computers and numerical methods become
available. lf our model has 110t been calibratcd with respect to reality, we
should not be confident with the results obtained using it. Calibrating means
to reproduce strengthes, dcformatioIls and failures (F. J urado).

A further sophistication in the method of aoalysis must carry together a greater

refillcment 011 the solution. A larger study must have as a cOllscqllence a
minimun interventioIl, which mcans t hat t he usage of more soh isticated method s
should be dircctcd to elllarge our safcty, but not to cause a more heavy
intervention (F. Jurado).

The p roblems of strllct u ral analysis have reccntly undergolle a great evollltioll,
cspccially 011 account of the availability of the Method of Finite Elelllcnt.s.
 P. ROCA and J. L. GONZÁLEZ I Thc Discussions 403

The great use that is eu rrently made of this mcthod was show n during the
Seminar. Howcver 1 this ap plieation must be observed critically. especially if
il leads lo lhe pereeplion th80l its applic8otion is rclated to lhe possibi!ity of
applyi ng a strict scientific method. Thus: upon analyzing an aqueeluet under
lhe effect of l he gra.vilational ioads, the most suitable melhod see OlS lo be
thal of direcl equilibri um under lhe impased loads; on lhe other hallel, if
we wish to study this aqueduet under the effect af the tempera.ture) sin ee
there appears to be a eom patibility problem, it is Ileccssary to work with the
method of finile e1ements. Apparently (fram a scientific paint of view) we
shou ld not use two differenl 1l10dels to study lhe same structure. However,
the result may be correet from the engineerillg point of vicw. In the analysis
that. was presented , the elastic approach was the most important. Another
approximation tha1 is adopted, but was not eva luated in the diseussion, is that
of small movements. This seeond hypothesis leads to mistakes of the arder
of the 5%, which a mathematieian \. .·ould eonsider incompatible with seience.
"Engineering science" is sOInething different to strict "science". Simple, weU-
calibrated formulations mal' be suitable for reaching satisfactory enginccring
solutions, thus making it unneeess ary to use very sop hi st icated models. The
use of models of a nou-linear type was also proposed. Will ihe engineer that uses
them be able to lake "e ngineering" decisions with them? These method s are
necessary, but it is also necessary to maintain a criticai attitude toward their
use. The technique of finitc clcments is without doubt a powerful too1. But
science is some thillg clse. 1t is Ilecessary to redefine w1lat we understand by a
"solution", since its goodness depends on the context: an unsatisfactory solution
for structuralm echanics may be a good solution for engineering Illcehanics (F.
Mola).

The use of so phi sticated and precise method s would have a ver)' significant
effect on lhe eost of the intervention, that is to say, the cost of an incorrect
intervention would be avoided (A. R. Marí).

Ali interesting way of overcoming the limitations of a linear e1astic analysis can
bc given by the use of a " non-tcnsion " model. The non-linear analy sis based 011
a model of this type is not difficult but it could provi de a very precise resulto
He would have liked to see examples of this type of analysis during the Seminar
(A. R. Mari).

5.5 The evaluation of the safety of ancient constl'uctions

lt is possible to obtain an idea about the safety of these structures using

experimentation anel numerical analysis in eombination. Beeause they make
•
404 STRUCTURAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

it possible to control stress distributions and movements, experimentation and

monitoring are a fundamental tooI for controlling working conditions and the
maintenance a f the safcty of these structures (P. P. Rossi).

One must be extremely cautious about using the results of the analysis to
value the safety of an ancient construction. lt is advisable to combine various
methods, calibrat ing tbem through comparison with other known cases, and to
control the behavior of the structure by monitoring it. In particular, in order
to assess the safety, it is preferable to use sim pie modcls d irectly based on the
equilib rium (R. Meli).

As is statcd in Section 6.2, the safety study would require an analysis of the
future behavior of the structure, that is to say, its evolution in time. It also
requires very sop histicated and accurate methods that would in particular be
capable of reproducing the collapse of the struct ure ( A. Ch iarugi) .

5.6 The training and experience of the analysts

The person d irecti ng the studies wiU require a great deal of experience. AIso ,
we must insist that the teehnieal staff must not be the victim of their "faith
in numbers" ar their faith in computers. The studies should be based on the
knowledge and experienee of the analysts (L. Ortega).

Seienee (ba.sed on rationality) should not be confused with mathematies a nd

numerical models. I n fact, the objective is to use with the best benefit aU the
numerieal information, and this benefit depends on the reliability of the models
anel of the results produced. HO\\rever, evcn whcn the models do not represent
su itably the reali ty, usefu ll information can be adquired if the designer is able
to interpret them (G. C ro cci) .

are far removed from qualitative reasoning and írom expenenee, therefore
offering little reliability (G . Croei).

The su bject Mechanical Surveying should be included in the training of

st ructural engi neers ( A. Chiarugi).

Until ten years ago the main tool for understanding monuments was experience
and qllalitative reasoning. Numerical met hods are often of little use, but at
times they are ext remely IlsefuL It would have to be determined in each case
whether the results are better or \Vorse than the result s that can be obtained
through experience and qualitative reasoning (M. Mihailescu).
 P. ROCA and J. L. GONZÁLEZ I Thc Discussiolls 405

The technician must use familiar techniques that he really knows how to use.
He should fiOt use the computer to petform calculations of a st ructure that
he does fiOt know how to calculate by hand. He can thus make a. subsequent
criticai analysis of the general results and of the hypothesis employed (J. M.
Izquierdo).

5.7 Relation s hip between history and structural analysis

The st udies must not be based purely 011 consid erations of a histori cal type.
Indeed , t he fad that the st ructure has remained stalldin g for hundred or
thou sand s of years does 110t meal1 that it wiU be safe in the present. We should
recaU th e cases of the Campanile of Venice, which coUap sed in 1902, 800 years
after its cOl1struction, or of the Civic Tower of Pavia, whose recent coUap se - in
20 seconds after more than 800 years of life- occurred witho ut any fiss ures
having been previously detected. In this last case, nurnerical analysis rnade it
possible to understan d very clearly the reaso ns for its collapse.

Thu s, it must be stated that histori cal perman ency is no guarallt ce of safety
on its OWI1. Though this of course depends on the type of construction (mauy
existing co nst ruction s may be elose to coUapse), as a rule, allalysis can be
collsidered as Ilecess ary (G. Macchi ).

Thc hi storical ev iden ce must be used to calibrate th e Illodel , and the model
must be capable of rcproducing the historical evidence. The diagnosis consists
in understanding what has occurred. If fi ssu res are detected (as in Bruneleschi 's
dome) it is necessary nrst of all to st udy th eir cau se - to make a diagnosis-
usi ng the best analytical procedures. It is necessary to individualize the
mechanical hi sto ry, which means modifying the model iteratively during its
preparation) reprodu ci ll g in some form the history of the construction) until the
model co rrespond s to reality and predicts the obse rved damage. The mode!
wil! have to predict what ex.ists in reality. Thus, for exam ple, it was possiblc to
ullderstand the principal fissure system of Brulleleschi 's dome by employing the
conditions of equilibrium and compatibility from a continuous medium mode! ;
when these nssu res were introduced in the model, it predided the second famil y
of nssures that e.x.isted. Thanks to aU this , a [ar better knowledge of this
construction has been obtained (A. Chiarugi).

IdeaUy, analyses of an evolutionary type should be accomplished) that is to say,

workin g with mct hod s capable of reprodu cing the cons tru ctional hi sto ry and
the hi story of repairs and restoration s (A. R.. Marí).
 406 STRUCTU RAL ANALYSIS OF HISTORICAL CONSTRUCTIONS

5.8 Final comments ou numerical modelling

In relation to the preparation of large numerical models, we must separate

different situations Of differentiate betwecn different leveis of detail:

(3) General models: as a rule these are very useful for this type of study. There
importance lies Hot 50 much in that they are bettcr than the hypothesis of linear
c1asticity, bui Lha! they provide a good understanding, and suitably account for ,
the state of continuity of the structure.

(b ) Local models: used in order to study a criticaI part of the struct ure with
great refinement and detail, once the parts have been identified through t~e
previous general analysis. In these mo deis it makes sense to use more refined
and sophisticated modelling tcchniqucs.

(c) Models for dynamic studies: their use requires knowing a great number
of propertics that are not always availablc, or must be obtained by direct
measurement (R. Meti).

As a rule, it is advisable to eitber combine different types of methods that

consider leveis of different detail (the local or general levei [R. Melli]), or
correspond to differcnt theories or leveIs of sophistication in the analysis
(cla5sical methocls based on the equilibrium of thru5ts , calculation as a
continuous medium ... [J. L. González]).

Much is said about the cost of numerical models. Obviously, it is necessary

to aclapt the means to the magnitude of the prob!em and to the available
resources. However , it should be remembered t hat the cost of the preparation
and utilization of a numerical model is as a rule low wi t hin the general process
of stu dy in comparison with the cost of an experimcntation and monitoring
campaign (P. Ro ca).

First of ali, it is necessary to decide what one wishes to achieve through the
numericaI study. In fact, tbe purpose of studyi ng an al1cient structure is to
understand an existing reality, anel it therefore requires the use of methods
baseei on what are called second-o rder systems in information theory. It is not
a question of unclerstanding how one must design the struct ure so th a t it will be
stable, but of und erstanding what happens to the existing structure. The most
customary calculation methods (based OIl first order systems within infonnation
theory) may t herefore be totally unsuitable.

Furthermore, since second-order systems are oftcn very unstable , which can
 P. ROCA :md J. L. GONZALEZ I Thc lJi ~(" lI ~s ions .07

lead to very largc variations in the results, it is necessary to emphasize the

!leed to work with methods that are highly controllable and wel! adapted to
the questions considered; it is necessary to establish a strong contrai over each
calculation phase. (J. M. lzquierdo).

A sensitivity analysis would in many cases compcnsate for a lack of knowledgc

<l.IIJ strength uf the llHl.terials (A. R. r-.'Iarí).
cf thc lIlechau..i cal jJl"UjJcrtic:.

6. F I N AL C ONSID ERAT I ON

There is obviously a consensus on the main problcms. The conclusions of these

discussions should serve as a guide to those who wish to cnter the world af
studies of ancient cons1ructions_ The conclusions lI1ust convey the complexity
af the tapics discussed (J. M. Ballester).