2003137

Bulletin of Earthquake Engineering 1: 37–82, 2003.© 2003 Kluwer Academic Publishers. Printed in the Netherlands.

Seismic Vulnerability of Historical Constructions:A Contribution

CARLOS SOUSA OLIVEIRADECivil/ICIST, Instituto Superior Técnico, Lisbon, Portugal (Fax: 00 351 218418200; E-mail:[email protected])

Received 20 November 2002; accepted in revised form 26 February 2003

Abstract. Earthquakes are known to be natural hazards that have affected tremendously historicalconstructions. Unfortunately, as far as earthquake impacts are concerned, there are no world statisticsto compare the suffering of populations or of the building stock and their evolution in time, with thedamage inflicted to the stock of historical constructions. Lately, a great effort has been placed onengineering developments: (i) to better understand the seismic behaviour of historical constructionand (ii) to assess the benefits of different techniques for reinforcing these structures. However, a greatdeal of discussion is still going on the type of reinforcement that should be applied, how effectiveit is and how much it costs. Research is needed for helping in these decisions, by providing a moreprecise framework in this field. The aim of this review is to make an overall insight on some of theavailable methods for assessing seismic vulnerability of historical constructions and on how to usethem in the case of occurrence of an earthquake. Given this occurrence, the objective is to minimizethe effects of aftershocks, avoid hurried demolition made under extreme pressure and help shore-upparts in risk of falling. The final aim is also to help in the definition of strategies for the repair of thedamaged patrimony, or as a measure to prevent damage in future earthquakes for the most vulnerablecases. The paper is illustrated with the presentation of several examples published in the literaturewhere the author participated.

Key words: civil protection, historical centres, historical structures, modelling, monitoring, monu-ments, retrofitting, seismic behaviour

1. Introduction

The protection of historical built heritage from natural hazard has been the objectof many studies and discussions and Carta di Veneza (1964) has defined the mainprinciples on how this protection should be done in order to keep its most authenticvalue.

It is well known how much this heritage has suffered from the action of earth-quakes, Figure 1, where three different cases are shown. The first, Figure 1a, repre-sents the damage to a church in the 1755 Lisbon earthquake, the second, Figure 1b,the damage to a masonry church in our times and the third, Figure 1c, the damageto a modern mosque.

To minimize further destruction under future seismic activity it is necessaryto reinforce the existing structures that are more vulnerable; but this action al-

38 CARLOS SOUSA OLIVEIRA

Figure 1. Typical damage to monuments: a) the Lisbon earthquake of 1755; b) the 1980Azores earthquake; c) the 1999 Izmit earthquake.

ways causes some perturbation to the authenticity of these structures as far as thehistorical evolution of their design is concerned.

The main purpose of the engineering analysis of these structures is to obtainthe best solution that minimizes the intervention and, consequently, to balancebetween the need for safety and integrity, and the need for preservation of theoriginal structure and tissue.

There is only one way to look into the problem, which is to analyze structure bystructure, understanding its initial construction and its evolution in time under theaction of all aging ingredients, both from nature and from repairing, including theearthquake impacts.

In this overview we look at several different cases which gave us a fair amountof experience and try to see where science can help in decision making. We tryto generalize from these examples into the vast world of monuments in order toforesee the possibility of applying a simple tool to measure their vulnerability.This methodology has been explored with some success in the field of the stock ofhouses and has produced results for detecting the housing categories more prone tothe earthquake threat. Observational aspects, carried out in field missions after theoccurrence of an earthquake, form an extended basis for the method.

SEISMIC VULNERABILITY OF HISTORICAL CONSTRUCTIONS 39

Figure 1. Continued.


The latest evolution within this context is the Hazus methodology, developedearlier in the USA. Hazus 99 (1999) has launched a great campaign to assessvulnerability of various types of the US constructions merging observation withnon-linear structural analysis. However, historical structures in the US cover a nar-row recent time interval, not contemplating the most common monuments existingin the rest of the world, namely in Europe and South America. So, it is of greatimportance to develop the necessary tools to establish vulnerability functions forthese historical constructions.

Fragility of monuments has been recognized throughout the ages so that dif-ferent cultures have developed their own reinforcing techniques. The use of ring-beams, wood and steel ties, wooden frames inside walls and vaults, gravity exteriorwalls (‘counterfort’) of various forms and sizes, and replacement of heavy ma-terials by lighter portions are among the most important measures for structuralretrofitting of old structures. The same can be seen in the evolution of the layout ofthe structure as function of materials and technology.

Through the presentation of several cases, which summarize the experiencegained in the last 20 years, the following items are addressed in this review work,some in extent, others just mentioned:• Definition of Cultural, Architectural and Historical Heritage: built patrimony;monuments; historical constructions and historic urban centre;• Typology, structural categories;• Causes for severe damage: soil settlement; too much load; lack of bracing;• Causes for slight damage in parapets, obelisks, colonnades, etc.: lack of bracing;• Material properties: mechanical characterization;• Analytical studies: definition of mechanisms;• Experimental studies;• Definition of Seismic Loads;• Techniques for reinforcement;• Codes for retrofit;• Policies for prior reinforcement vs. posterior repair;• Vulnerability analysis/assessment: development tools for rapid screening of vul-nerability from geometrical plans and construction materials;• The role of Civil Protection and Emergency Preparedness;• Policies for mitigation.

Cultural heritage is seen in this paper as the set of historical constructions with aremarkable fabric, including monuments and historical centres. Even though theypresent completely different features in geometry, size and connectivity amongthemselves, there are still many common problems to be solved. One of them isthe construction material which is essentially the same, even though monumentsin general present better and more homogeneous standards. The second similarityhas to do with the way the connections between elements are performed. Third, isa topological problem derived from the fact that monuments are very often locatedadjacent to buildings in an historical centre.


The geometric scale gives the most important dissimilarity, as most monumentsare, generally speaking, of greater volume than historical centres. This scale factorhas great importance, not only due to the larger loads present in the structure butalso due to the drift of frequencies of vibration towards lower values for the largerstructures.

Other types of structures with historical interest due to its function to the societyare also included in this definition. Old bridges, aqueducts, chimneys, and alsoconstruction from the industrial period can be considered.

We will dedicate most of our discussion to historical construction includingmonuments, but nevertheless, whenever of interest, mention is made to historicalcentres.

1.1. THE WORLD DAMAGE PANORAMA FOR HISTORICAL STRUCTURES AND

MONUMENTS

Unfortunately, there are no world statistics of damage inflicted by earthquakes tothe stock of monuments. Also, no detailed catalogue of monuments exists, withlocation, typology, damage suffered along the times and the repair/reconstructionprocesses involved. Exceptions to this are a few good examples observed in recentearthquakes where information has been collected and sometimes is even availablein the internet (DISEG, 1999; DGEMN, 1999).

This world statistics would be of most interest because correlations with seismicintensity, type of structure and soil foundation could be better established. The typeof repair could also be considered.

To emphasize the problem at the world level, first we are mentioning a few casesfrom the last quarter of the XXth century with an important impact on the historicalpatrimony and, at a later stage, we dedicate a more detailed analysis to what hashappen in Portugal.

Antígua, the capital of Guatemala, was particularly affected by the 1976 earth-quake, which destroyed the Cathedral and caused a variety of damage to manyother churches. In the same year, the region of Udine in northern Italy was alsoaffected seriously by another earthquake (Doglioni et al., 1994) and the same hashappened in November 1980 in south Italy during the Irpinia earthquake.

In June 15, 1999, a magnitude Mw = 6.7 earthquake with epicentre in the stateof Puebla, Mexico (Mercalli modified intensity - VIII) caused important damageto a group of around 600 churches, a few as old as those of the XVIth century(Jiménez et al., 1999). The observed type of damage consisted essentially in theopening of longitudinal cracks along cylindrical domes, due to the lack of supportfrom the lateral walls. The bell towers were also very damaged with the forma-tion of diagonal cracking in the zone interacting with the front wall. The fall ofornamental blocks was generalized. In June 23, 2001 a large event (ML = 8.4),near the coast of Peru, also caused extended damage to monuments, namely, in theMoquegua Cathedral (EERI, 2001), located more than 100 km from the epicentre.


Acceleration records in this region indicate Peak Ground Acceleration (PGA) of220 and 295 cm/s2, respectively, for components N-S and E-W.

Curiously, in spite of the large damage caused in Mexico City during the Sep-tember 19, 1985 earthquake, to the stock of buildings of reinforced concrete struc-tures with heights between 8 and 15 stories, the monumental structures did notsuffer anything. The spectral content of the ground motion constitutes the mainreason to explain such difference in performance.

But it was the Umbria-Marche (Italy) series of earthquakes in 1997 that mostimpact caused to monuments in the last years. It spread damage in the centre ofItaly (4 main events with magnitude Mw of the order of 6), causing large effects inmore than 3350 churches in the region near Assisi.

The earthquake series caused damage to 1815 churches in Umbria (D’Ayala,2000) and to 1450 in Marche, without too many total collapses but with importantlosses in the historical heritage due to highly developed crack patterns and partialcollapses. About one third of these churches had already undergone some formof strengthening (transversal ties, longitudinal ties, ring-beam, concrete roof, orcombinations), and damage patterns were highly dependent on the existence ofthese elements. It was observed that the efficiency of the technique used is alsovery dependent on the type of damage involved and this requires much furtherresearch.

Damage caused to the historical heritage by these earthquake series is verywell described in several publications, and a great deal of good information onvulnerability of monuments can be gathered from this important experience (seeLagomarsino, 1998).

The level of vulnerability shown by churches and other monumental struc-tures under seismic action is in general higher than the vulnerability of traditionalhousing due to large open spaces, lack of horizontal thrusting elements and largeslenderness existing in those structures. This may cause an overestimation of theintensity evaluation of ground motion for the cases where only large monumentalstructures are analysed.

1.2. SOME STATISTICS OF DAMAGE IN PORTUGAL

Damage to the built heritage in Portugal has very close links with the history ofearthquake occurrences. Actually, the reverse is also very true: the strength of anearthquake is measured by the damage caused to the built patrimony, and can besomehow compared along the times because this built patrimony has been aroundfor many centuries, surviving many earthquakes.

Figures 2 and 3 and Tables I to V present an account of the damage inflictedto monuments and historical housing during few of the most severe earthquakes inPortugal: in Continental Portugal – 1755; 1969; in the Azores – 1980; 1998. It canbe seen that a large portion of the monumental stock has been heavily damaged. Anextreme good similarity between the percentage of affected monuments in the 1755


Figure 2. Damage to monuments and important buildings: Lisbon, 1755 (total 402).

Figure 3. Damage to monuments and important buildings: Azores, 1980 (total 218).

and 1980 Azores earthquakes derives from the fact that, even though typologies arequite different, their construction was still similar. For a better description of thesetwo events see Pereira de Sousa (1923) and Kendrick (1956) for 1755 and Oliveiraet al. (1992a) for the Azores.

It is also interesting to mention that the same structure has been damaged indifferent earthquakes. This is the case of the Lisbon Cathedral (Sé de Lisboa),greatly destroyed in 1531 and 1755, and the case of Igreja da Conceição in Horta,destroyed in 1926, partially rebuilt with concrete ring-beams and greatly damagedin 1998.

2. Typlogies and Types of Damage

Many researchers have worked in the past on the characterisation of historicalmonuments having in mind the structural behaviour under seismic loads. Textsby Giuffrè (1992), Kirikov (1992), Mainstone (1993), Doglioni et al. (1994), Meli(1998), Barbat (2001), Macchi (1998), Tassios (1988), etc., are just a few docu-


Table I. Lisbon earthquake – 1st Nov. 1755; Damage tobuildings in the city (average MM intensity – IX).

Total number of existing buildings 20 000 %

Heavy damage 12 000 60

Collapse 2 000 10

Table II. Earthquake 28th Feb. 1969; Damage tomonuments (average MM intensity – VI-VIII).

Total number of affected churches 185 %

Moderate/Heavy damage 20 10.8

Cost of repair: 50 000 Euro (year 2000) per church;30 day repair time.

ments from last decade. Many more could be added to this list. The work doneon the aftermath of the Umbria-Marche earthquake under the sponsorship of theGNDT (National Group for Defence from Earthquakes) in support to the Commis-sioner Delegated for Cultural Heritage (Brovelli et al., 1998), is a great example ofwhat should be done in relation to monuments.

A simple classification was essayed by Oliveira et al. (1992b), when building adata-base related to the Azores earthquake of 1980. In that case, several categoriesof churches were considered based on size, such as cathedrals, large churches,chapels, very small chapels. Other category was directed to convents, and anotherto palaces of different sizes.

For each category, a definition on the type of material, the type of structural lay-out, and structural elements, roof type, etc., were also included. Main dimensionswould complement the information on geometry and mechanical characterization.

Another type of information includes the existence of changes, repairs made,etc. The 1980 Azores data-base contains also a measure of costs to repair thedamage.

Damage to monuments is of great variety and, consequently, difficult to putdown in a simple matrix. The most frequent types of damage in churches, by in-creasing order of importance, are: (1) damage to the towers, sometimes with theirfalling; (2) separation of the main peripheral walls, by rotation in relation to thefoundation line; (3) cracking of these walls due to in-plane shear forces; (4) partialor total collapse of corners; (5) crushing of vertical elements due to high axialforces, in columns and walls; (6) damage in regions outside the main structure.

The nature of these damages depend on many variables such as the presenceof connectors to restrain walls, the state of deterioration, the type of soil, the typeof seismic action including the presence of vertical components. Crespellani and


Table III. Earthquake – 1st Jan. 1980 in the Azores; Dam-age to buildings (average MM intensity – VIII).

Total number of existing buildings 27 000 %

Heavy damage 15 000 55

Collapse 5 400 21

Table IV. Earthquake – 1st Jan. 1980 Terceira/São Jorge, inthe Azores; Damage to monuments (average MM intensity– VIII).

Total number of existing monuments 359 %

Slight damage 54 15

Heavy damage 58 16

Collapse 50 14

Uzielli (2001) show clearly that damage of monumental structures has a lot to dowith the soil type where the structure is seating. Seismic spectral content as well asthe time evolution of the strong motion records are of most critical importance tothe seismic performance of the structure. These aspects depend very much on theproximity of fault source, of the presence of soft soils and topographic geometry.During the 1998 Azores earthquake, it was observed that vertical components ofground motion played also a significant role in the observed damage of not onlymonuments but also of masonry walls in buildings, due to the fact that frictionamong masonry units (elements) was quite reduced, allowing the dismantling ofthe fabric of the walls.

Besides the causes above referred, for which the lack of bracing is one of themost important, other causes of severe damage are soil differential settlements andthe presence of too high loads in elevated locations. The performance of arches andvaults depends essentially on the behaviour of the corresponding bedding. When

Table V. Earthquake – 9th July 1998 Faial/Pico, in theAzores; Damage to monuments (average MM intensity –VI-VIII).

Total number of existing monuments 27 %

Heavy damage 16 59

Collapse 3 11


these beddings settle, the arches cannot take too high load, forming hinges, whichwill eventually lead to collapse when a mechanism takes place.

Causes for slight damage in parapets, obelisks, colonnades, etc., are essen-tially dependent on the absolute acceleration and duration of seismic motion atthe location of these elements.

Other types of important monumental structures are old masonry bridges andaqueducts. They have not suffered much damage in past earthquakes. Two typescan be identified for earthquake resistance: slender bridges or aqueducts with longspans and bulky bridges. They have completely different behaviours due to theirdifferent flexibilities, masses and, consequently, frequencies. Not much statisticalinformation exists on these structures. We will look into the analysis of the slenderaqueducts. Damage to the bulky types is mainly due to lack of confinement oflateral walls on top of the arches.

When dealing with the stock of buildings, the main categories used in the abovementioned data-base for the Azores were again the limit states no damage, slightdamage, moderate damage, heavy damage or partial collapse, and total collapse.Different parts of these structures could undergo different types of damage.

The existence of such a data-base for making the inventory of monumentalstructures, as an updated green card, is of prime importance to keep up the state ofdeterioration of these structures, helping, in the event of an earthquake, the teamsin charge of determining the ones requiring more attention. In fact, information onthe material already available for a given structure, the existence of prior studiesand who did them, repairs operated, etc., will speed up a more correct diagnosison what to do to prevent further damage. Using data from damaged structures in alarge number of cases will permit to determine correlations of damage with groundmotion characteristics and type of structures. Actually this was done in the Azoresand in a much more detailed way in Umbria-Marche (DISEG, 1999).

3. Modelling Techniques and Experimental Calibration: Evolution in theRecent Times

As is many problems of structural engineering, there are several ways to analyse thestructural integrity of monumental structures. We can classify them into several cat-egories as follows: simple geometry linear 1-D cases; complex geometry linear 2-Dand 3-D cases; non-linear 1-D case; non-linear complex 2-D and 3-D cases. Withinthe most interesting and sophisticated non-linear models we can refer the ones withdiscrete blocky elements, the ones using plasticity models and the theory of failuresuch as the Coulomb criteria, or the models which can deal simultaneously withsmall and large displacements with the same accuracy (Extended Distinct ElementMethod; Meguro and Hakuno, 1994; Meguro and Tagel-Din, 1997). Other modelsdeveloping constitutive equations for friction between blocks (Lourenço, 1996) andfor interfaces (masonry joints under shear) (Oliveira and Lourenço, 2001) are alsomerging the theory of inelasticity into the description of more heuristic models.


Experimental techniques, using in-situ measurements based on ambient noiseto obtain the frequencies of vibration of the structure, have been utilized to cali-brate the models. These techniques are fundamental tools for this verification butthey lack of information on the structural behaviour under intense amplitudes ofvibration.

Complementary studies have been done in laboratory, by testing parts of struc-tures under large amplitudes. These studies are essential to get a proper charac-terization of material properties in the non-linear range. Also the determination ofdamping associated to different materials is of most interest.

Influence of type of masonry, the shape, size and geometric arrangement ofstones, the presence of mortar connecting different layers of stones and the form ofconnecting elements at the edges of walls, columns, vault footings, as well as theconnections between timber elements and masonry walls, present great uncertain-ties on the overall performance of the entire structure.

On top of these aspects, historical structures suffered interventions along thecenturies, with alterations in several of its members which not always contributedto a better performance. Increases in weight at the roof level, demolition of partsof the structure, opening of passages in weak zones, and the normal degradationprocess from aging, are among the most common cases of structural changes weak-ening the structure. These changes in mass, stiffness and strength properties aresometimes aggravated by the occurrence of past earthquakes which caused partialor total damages in parts of the structure.

Research has evolved tremendously in the recent past in different directions,but it is not yet possible to say that the problem of vulnerability is solved. Manyadvances in structural mechanics with better modelling have been made. But thecomplex modelling of these 3-D non-linear structures at an ‘atomic level’ is stillfar from being achieved.

It would be of most importance to develop a screening procedure to identifythe structures which are clearly more vulnerable to the earthquake action. Thisscreening procedure should rely on a minimum number of requirements capableof informing about those capacities. These should include geometry and materialproperties, in a detailed or coarse description. For the first case, a good survey andthe testing of a few materials are required. In the second case a general layout,thickness of walls and material properties taken from a generic observation mightbe enough. A good example of the second procedure is the survey made of allhistorical monuments in Portugal (DGEMN, 1999), consisting in the digitizing ofall existing plans, drawings and façades.

Early models were very simple, evolving to more complex when geometry in-formation, materials characterisation and software availability could accompany.Beam elements were considered in the analysis of structures with a 2-D and 3-Dclear element type. The most remarkable example of this technique is the study of‘Aqueduto das Águas Livres’ in Lisbon, made in late eighties with beam elements(Figure 4). This structure was later studied with discrete macro elements in a 2-


D (Azevedo and Drei, 1995) and 3-D analysis (Sincraian et al., 1998; Sincraian,2001). This last case was done for just a pillar of the Aqueduto (Lemos et al.,1998). The complexity of the entire structure was summarized to a single pillarmade of an outer blocky type structure filled with rubble masonry.

All evidences in the different performed analyses, point out to that in the longi-tudinal direction the structure can stand without great problems PGA of the orderof 1 g, whereas in the transverse direction the PGA values cannot surpass 0.3 g.

A great deal of information was gathered for this structure, from the geometryto the material properties as obtained from boreholes made at various locationsand ultra-sound prospection, and confirming the historical appreciation made fromexisting information.

In-situ dynamic analysis with ambient noise done in late eighties (Oliveiraet al., 1991), before digital signals became popular, was able to identify severalfrequencies and corresponding modal shapes. These frequencies were confirmedfrom the analysis of the record obtained during the occurrence of a magnitudeMb = 3.5 earthquake with epicentre 10–15 km away, recorded at the top of onepillar. Spectral analysis of this record clearly shows the existence of several highermodes, but the first two modes, even though present, were not greatly excited dueto the high frequency content of this small magnitude event. (see Annexe 1 formore details)

Other cases of bridge-type analysis were done in the bridge of Ribeira Grande(Oliveira et al., 1994) and in the Aqueduct of ‘Aqueduto Água da Pratá’ (Drei andOliveira, 2001a). In both cases, experimental in-situ measurements were alwaysused to calibrate the analytical models. For the first case, a simple representationwith beam elements was used, while in the second case a more sophisticated shellelement was developed (Figure 5) (see Annexe 2 for more details).

The application of simple linear models for the analysis of churches was alsoperformed in the early nineties (Gil and Oliveira, 1994 and Guerreiro et al., 1990).

The use of even more sophisticated representations of structures, with 3-D de-tailed finite element discretizations has been developed in very recent years. Ex-amples of illustrations in the case of churches are given in Lourenço et al., (1999)and Moreira et al., (2001) (Figure 6) for linear global analysis and by Lagomarsinoet al., (2002), involving some kind of non-linear analysis.

The Discrete Element Method has also been used to analyse the behaviour ofstructures under large displacements and results seem very promising (Sincraianet al., 1999). A light-house of blocky masonry (Figure 7) and a statue of cut-stone(Figure 8) were analysed under strong shaking showing good similarities betweenthe observed pattern of damage and the output of the analytical response (Oliveiraet al., 2002). ‘Aqueduto da Amoreira’ was also analysed (Drei and Oliveira, 2001b)comparing a linear 3-D model with a more refined discrete element technique (seeAnnexes 3 and 4 for more details about Figures 7 and 8).


Figure 4. View of the ‘Aqueduto das Aguas Livres’ in Lisbon. Computed shapes for the firstthree transversal modes of vibration from in-situ measurements (Oliveira et al., 1991).


Figure 5. Finite Element Model of ‘Aqueduto da Agua da Prata’ in Evora (Drei and Oliveira,2001a).

Figure 6. Detailed Finite Element Model of a church in Pico, Azores (Moreira et al., 2001).

Another technique which cannot be ruled out in the context of structural safetyof monuments is the instrumentation of these structures with strong motion equip-ment waiting until records from earthquake motion are registered.

The technique has been applied in cases of buildings, bridges and a few casesof geotechnical structures, with very interesting results. It allows to calibrate mod-elling and to check all hypothesis of non-linearity. By similarity, it is supposed to beone of the most efficient ways to monitor the performance of historical structures,which have a very complex behaviour. Unfortunately, there are so far very fewcases of reference, as described below.

One, not yet publicized, occurred in the Blue Mosque during the Izmit earth-quake of August 17, 1999, where several records were obtained clearly showingan increase of natural periods which tend to recover partially after some time (M.Erdik, personal communication, 2001).

According to Meli et al., (2001), the Tehuacan earthquake (Mw = 7), June 1999in Mexico, caused damage to about 1800 historic buildings, most of them colonial


Figure 7. Lighthouse of Ribeirinha: a) elevation and plan; b) response obtained with the Dis-crete Element Model for different input multipliers, July 9, 1998 earthquake (Oliveira et al.,2002).

temples and convents of the central states of Puebla and Oaxaca. The example ofthe Mexico City Cathedral, where several strong motion instruments were placed,is of great value to validate design procedures and determine the main character-istics of seismic response. The recording of a few earthquakes, even though of notgreat amplitude, was good enough to understand the most important features of itsbehaviour, stressing the weaker points of the structure.

Also, in the aftermath of the Umbria-Marche earthquake crisis in Centre Italy,September/October 1997, the Holy Monastery and the San Francesco Basilica inAssisi were monitored for earthquake action by automatic systems with 25 ac-celerometer transducers (Menga et al., 1998). During a period of 7 months in 1998,380 events were recorded, 50 of which with PGA above 1 cm/s2. Maximum PGA’sof 100 to 200 cm/s2 were recorded at top levels. From this data it was observedthat for larger events the amplification in terms of PGA’s from base to top was ofthe order of 2.4 whereas this amplification increases up to 4.0 for lower intensityvalues, indicating the increasingly dissipative capacities of this type of structurewhen excitation level increases.

Another example, not of comparable interest, was obtained at a traditional 3-story masonry house within an antique XVIIth century block of buildings (Fig-ure 9), which was monitored with 3 strong motion instruments, one at each story,and recorded a few magnitude 3 to 4 earthquakes producing PGA at the ground


Figure 8. Statue in Horta: permanent configuration after the July 9,1998 earthquake fordifferent input multipliers and the real situation (Oliveira et al., 2002).

floor in the order of 0.03 g. (Figures 10 and 11), and showing a clear amplificationat the top level (see Annexe 5 for more details).

Models based on collapse mechanisms have also been used to derive vulner-ability functions, both for historical centres (Spence et al., 2000) and for monu-ments (D’Ayala, 2000). This implies the definition of prior patterns for collapsemechanisms, establishing minimum energy concepts.

Several studies were performed on shaking tables with small scale models orparts of structures. Pseudo-Dynamic testing were made on real scale structures.One interesting example was performed in ELSA/Ispra (Pinto et al., 1998) on anarch of blocky masonry to test analytical modelling in case of large displacements(Figure 12). Geometrical as well as material non-linearities were analysed at lengthand results are very promising in confirming the accuracy of sophisticated non-linear modelling (see Annexe 6 for more details).

With all these testing devices and techniques, it is greatly desirable to have inthe near future good and simple tools for analytical evaluations of seismic perfor-mance, such as the ‘pushover’ technique, capable of define a capacity curve for acomplex structure such as an historical construction. The examples presented byLagomarsino et al. (2002) go in this direction.

4. Mechanical Characterization of the Masonry and Structural Elements.Testing

Only in recent years testing of old masonry elements has begun. This together withthe fact that a large variety of different masonry materials and arrangements do


Figure 9. Modelling an urban XVIIth century house in Azores: the GZCAH.


Figure 10. Ratios of PGA of observed records at GZCAH Terceira Island Azores – earthquakeMarch 4, 2000 (black – 3rd level/ground level; dark grey – 2nd level/ground floor).

exist, creates a great stimulus for a more correct characterization of the dynamicproperties of these materials. The influence of the quality of mortar and its state ofdeterioration is also critical.

Under moderate seismic loading, masonry structures develop non-linear be-haviour due to cracking of mortar, which creates additional difficulties for theinterpretation. Even higher difficulties arise in the cases of strong loading, up tothe collapse, with the formation of large deformations, including mechanisms andinstabilities.

Many researchers have developed in the past modelling for the analysis ofhistorical structures, from simple masonry material to individual elements, to com-plete structures. A few were already cited in section 3. Two techniques were usedfor this purpose: (1) vibration measurements of simple structures for which it waspossible to determine, from basic physics, the modulus of elasticity of the materialunder small to moderate amplitudes; (2) performing in-situ testing of parts of wallsor other construction elements, to determine non-linear behaviour and the corre-sponding mechanical properties. The first technique could also be used, as seenpreviously, in connection with other more complex structures.

For simple material testing conventional laboratory testing is recommended.Ultra-sound techniques are also of great help in homogeneous bodies.

Other techniques performed elsewhere for large deformations include the useof flat-jacks or similar devices and the laboratory testing of non altered parts orelements.

Equivalent moduli of elasticity for the initial state of deformation were ob-tained from natural frequencies in experimental low-strain dynamic testing. As asummary of results one could say that a poor quality masonry in a church tower(Oliveira et al., 1991) led to E = 5 GPa, while in a bridge (‘Ponte da RibeiraGrande’) with good basalt but poor connecting mortar or no mortar, E = 5.1 GPawas a good value (Oliveira et al., 1994). At the Águas Livres Aqueduct, the scle-rometer gave an equivalent E = 13 GPa for the harder limestone, 5 GPa for thealtered limestone and around 1 GPa for the interior rubble.


Figure 11. Analytical model versus observation (earthquake March 4, 2000): superpositionof responses (unit: cm s−2) 2-nd story level, transverse and longitudinal components (black –recorded; grey – model) (Emasonry = 0.506 GPa; Etimber = 11.0 GPa).

From testing different types of traditional masonry plastered with poor aggre-gates (thickness of 66 cm and density of 18 kN m−3), we obtained a modulus ofelasticity E of 0.3 to 0.5 GPa, and of 0.2 GPa for the case of dry irregular rubblemasonry (Guedes et al., 1999). For more rigid elements, such as ignimbrite stone,tested in laboratory by ultra-sound techniques, E = 7.5 GPa. Timber used in theroof has an E varying from 10 to 12 GPa.


Figure 12. The COSISMO project (Pinto et al., 1998): a) test set-up; b) deformation patternand corresponding damages.

Table VI. Typical Modulus of elasticity, E, for different traditional old masonry.

Individual elements Blocky good masonry Blocky intermediate Poor masonry

Up to 100 GPa 20 GPa 2–5 GPa 0.3–0.5 GPa

In relation to strength properties information is even more scarce than for theE values. For a poor masonry the yielding compression in bending presents valuesvarying from 75 to 150 kPa and ultimate shear from 10 to 20 kPa (Guedes et al.,1999) for large distortions.

Other examples can be given: In a recent study (Kaya et al., 2002) of the Suley-maniye Mosque in Istanbul, a comparison between computed and in-situ measuredfrequencies has recommended values for modulus of elasticity E = 2.3 to 3.3; GPa;ν = 0.2; γ = 20 kN/m3 for stone and bricks, and E = 0.6 GPa for mortar. Thesenumbers agree to a certain extend with the values obtained with non-destructivetests of ultrasonic nature.

An analysis of the Monastery of Jéronimos, in Lisbon made of outer goodblocky limestone, mixed with rubble stone, have led to E = 2.7 GPa; ν = 0.2;


γ = 23 kNm−3; tensile stress ≈ 0.1 MPa; compression stress ≈ 2–2.5 MPa.(Lourenço and Mourão, 2001). On the other hand, Marble of Parthenon lead toE = 70–80 GPa, tensile stress ≈ 7–8 MPa; compression stress ≈ 80–90 MPa(Angelides, 1998).

5. Vulnerability Assessment

Vulnerability studies are essential tools to solve a large group of problems amongwhich we can enumerate the ones for preparing emergency planning for civil pro-tection. Great efforts have been given to the housing stock of buildings and, inparticular, to the segment of historical buildings forming the historical centres.In the following we refer to some work devoted to these centres, as they canconsidered as relevant parts of the monumental tissue and because a great dealof information, data and experience can be used in these monumental structures.

5.1. DEFINITION OF GROUND MOTION

Ground motion is the process linking the vulnerability to damage. It has alreadybeen referred the importance of a correct evaluation of past actions to correlatethem with damage occurred, as well as to predict, within the lifetime of the struc-ture, the motions that can act at the foundations. This process requires the analysisof the hazard and of the frequency content and time evolution of motion.

Besides the standard analyses, in recent earthquakes (Azores, 1998; Izmit, 1999;Chi-Chi, 1999), two new observations, besides the soil strata characterization, be-come clear in near field: (1) the fling; and (2) the directivity effect. The first onehas to do with the tectonic deformation from the fault slip. The second with therupture process which causes a constructive interference in the direction of ruptureand destructive interference in the opposite direction. These two effects are veryimportant in the behaviour of the monuments and need to be explored. They havenever been included in any code of actions.

Another topic of great importance in the near-field, already mentioned, is thepresence of large vertical accelerations which are very critical for masonry struc-tures as their influence on the friction forces acting on the contact surfaces maybecome crucial.

The ground motion in terms of demand-acceleration-displacement-spectrum to-gether with capacity curve specific for a given historical construction will definethe preliminary response of the structure in terms of displacement, and thus es-timating its performance for that ground motion. This will constitute a first andsimple approach to a more ‘scientific’ evaluation of the seismic performance ofhistorical constructions in analytical/experimental terms.


Figure 13. Fragility functions for historical urban centre (adapted from Oliveira et al., 1992).

5.2. HISTORICAL CENTRES

Several historical urban centres have been studied in the recent past in order todefine the most critical structures and to help in the design of emergency planning.Usually these centres present a large concentration of population during the workhours and their vulnerability is higher than the one in areas with more modernconstruction.

In Portugal various cases have been the subject of study. The first example wasdeveloped within the TOSQA Project (1994–1996), an EC/Environment projectcontributing to the understanding of seismic behaviour of historical urban centres,with application to Alfama, in Lisbon (D’Ayala et al., 1997). Using similar method-ologies, the cases of Évora and Cascais were essayed (C. S. Oliveira, personalcommunication, 1994, 1996). More recently and using more sophisticated analysis,Lisbon, Angra do Heroísmo and Faro were also subjected to this type of analysis(Dias et al., 2001).

It consists in classifying buildings by typology class and attributing vulnera-bility functions to each class. The classes were based essentially on age, numberof stories and presence of similar adjacent buildings, and vulnerability was takenfrom experiences of past earthquakes, conveniently adapted to the situation understudy. Application of Hazus 99 methodology is now in progress, but great care hasto be exercised prior to use it in the framework of old masonry structures. That iswhere the analytical methods referred in sections 3 and 4 can produce promisinginformation.

The inventory of the building stock was obtained for all these cases from fieldwork. Other more expedite techniques are being studied. The use of the Censusinformation has also been applied in studies for a more coarse analysis.

To calibrate vulnerability and fragility functions, data from past events is es-sential. The cases of Azores and Umbria-Marche are of great importance to boththe historical centres as well as for monuments. In Figure 13 we show the fragility


curves for the old masonry housing in Azores, obtained from the data-bank withmore than 3000 cases. For Italy a great effort has been placed in this area of knowl-edge; Giuffrè (1993), Carocci (2001), D’Ayala and Speranza (2002) and Sarà et al.,(2002) are just a few researchers dedicated to it.

5.3. MONUMENTAL STRUCTURES

The experience from the GNDT studies in Italy (Doglioni et al., 1994), togetherwith the information gathered in the Umbria-Marche earthquake, led Lagomarsino(1998) to propose a classification of damage to churches into several categories,Figure 14, creating a damage index. This researcher also presented a vulnerabilityindex based on the potential level of damage in each one of the 16 situations ofFigure 14. Each situation is called a macroelement that can be damaged separatelyin a church and has been identified in many past earthquakes as common patterns.(The number of macroelements has been increased lately to 18 for a better descrip-tion of mechanisms that may occur). Of course they do not appear independentlyone from another, but for the time being, simple average of all situations is alreadya large step for an overall assessment of their seismic behaviour.• Damage index: is a number between 0 and 1, which measures the average levelof damage to the church;• Vulnerability index: is linked to the propensity of the church to be damaged bythe earthquake.

While the damage index is an average of the different damages observed ineach one of the 16 possible mechanisms that can be formed during an earthquake,the vulnerability index is also an average measuring the potential for activatingcollapse mechanisms, independently from the seismic action. As Lagomarsino(1998) states, the first index is particularly useful to estimate the seriousness of theoverall damage inflicted to a church, and the second one can be used for preventionattitudes, prior to the earthquake.

These two indexes are a first tentative to address the vulnerability of monumentson a descriptional basis. Merging with analytical modelling will give better supportto the method.

Correlation between vulnerability index and damage has been obtained by(D’Ayala, 2000). Figure 15 shows the differences of performance of various typesof elements and collapse mechanisms. And Figure 16 proposes a global vulnera-bility function for monuments.

Recent works by Doglioni et al. (2000) have developed the concept of mecha-nisms in macroelements in a much more refined form, detailing the types, includingfurther elements and subdivisions and considering specific cases.


Figure 14. Schematic representation of damage mechanisms in churches (Lagomarsino,1998). Each macroelement defines a possible damage mechanism in a church, as observedin past earthquakes. Damage indexes and vulnerability indexes are overall measures ac-counting for the various mechanisms and their degree of damage (observed⇒damage;potential⇒vulnerability).


Figure 15. Vulnerability of monuments for different parts and mechanisms (D’Ayala, 2000).

Figure 16. Vulnerability functions (adapted from D’Ayala, 2000).

6. Proposal of Possible Reinforcing

6.1. EXISTING DATA

Strengthening masonry buildings to improve their resistance to earthquakes hasbeen practised throughout the history, particularly in those areas where damagingearthquakes are more frequent. The use of external gravity pillars to balance walls,steel ties at floor and roof levels connecting the external walls together, steel staplesaround corners or connecting large stones, are among the most common practices


in some historical areas (Spence et al., 2000). More recently other techniques wereessayed, including peripheral beams and ‘reinforced masonry’ (Lizzi, 1990) thatconsists of a dense series of steel bars grouted in small holes across the masonrymaterial.

However, until the last 10 years, there were very few cases in which old ma-sonry buildings, strengthened to improve their earthquake resistance, had beentested in subsequent earthquakes (Spence et al., 2000). Recent European earth-quakes in Kozani, Greece (1995) Umbria-Marche, Italy (1997) and Azores, Por-tugal (1998) have provided a very much larger data-base of damage informationon such buildings in Europe, which can be used to review the effectiveness ofalternative approaches to strengthening.

6.2. REINFORCEMENT OF MONUMENTS IN CONTINENTAL PORTUGAL,TERCEIRA, FAIAL AND PICO

The experience in Portugal in the last 30 years can be summarized as follows.In 1969 (Continent), for the earthquake which damaged quite a number of

churches in southern Portugal (see section 1), the repair consisted essentially inapplying a ring-beam running at the top of the walls just below the roof, or atmid height if walls were too high. This reinforced concrete tie had a minimumnumber of steel bars, just to ensure connectivity at the top of those walls. The sametechnique was widely used in connection to the 1973 earthquake crises that affectedPico and São Jorge islands in the Azores, Figure 17.

It is worth noting that in the Azores, after the 1926 earthquake in Faial, repairactions used the concept of steel rods to tie peripheral walls. This technique gavequite good results as observed in the subsequent events.

In 1980 (Azores), the repair philosophy for repairing the monuments and ofthe older building stock as well, was in a first phase to insert inside the walls astrong reinforced concrete (r.c.) frame system which would take almost all theseismic load. Inclined r.c. slabs made the roof, Figure 18. In a second phase, theintervention became more soft with the introduction of more gentle r.c. and steelelements.

In 1998, contrarily to the policy recommended for the reconstruction of thebuilding stock (Ravara, 1980; Carvalho et al., 1998), no indications were providedto the designers. Consequently, the strategy for each monument was decided by theengineering designer.

For solving masonry weakness in its wall plan, recent techniques for confiningthese walls were developed using jacketing of masonry. This kind of intervention,which has been widely used in the Azores after the 1998 earthquake, is made byshotcrete and light steel net reinforcement, Figure 19a, in both interior and exteriorwall surfaces, with connecting ties through the wall. In zones of aggressive weatherconditions, stainless steel has to be recommended for prevention of early corrosion.


Figure 17. Typical reinforcement practiced in Portugal in the period 1960–1970.

Another technique to bind the blocky masonry is by using steel ribbons, permittinga good confinement of the entire wall, Figure 19b (Dolce et al., 2001).

The damage data produced in the historical centres, Assisi (EMS intensity VI),Sellano (EMS intensity VII-VIII) and Nocera Umbra (EMS intensity VIII), by theUmbria-Marche earthquake sequence, was examined by D’Ayala (2000) in orderto analyse the influence of strengthen and no strengthen policies and, among thefirst, which techniques would produce better results.

This data, analysed for a group of 5 typical macroelements, shows that the per-formance of churches which had undergone some type of strengthening is muchbetter than those without any improvement; and the ones with seismic strength-ening performed even better, Figure 20. Even though the use of tie-rods did notproduced an improvement as good as that of all seismically strengthened build-ings, the behaviour still was much better than that of churches with no seismicstrengthening at all. This last statement was confirmed in the Azores earthquake


Figure 18. Schematic layout of reinforced elements used in the period 1980–1990: a) planview; b) vertical cross-section.

of 1998, for buildings damaged in 1973 and 1926 and retrofitted with tie-rods andconcrete ring-beams.

These techniques have been supported by the results of laboratory studies anddesign procedures which indicate that they should be able to improve performance,even if not to the level of resistance required for new structures; and regulationshave been developed (e.g., in Italy) to permit improvement strategies to be em-ployed on a local basis. New technologies, including the use of damper devises inidentified locations, light materials to replace heavy parts, plastic reinforcementsto be applied in zones of stress concentrations, etc., are recommendable if usedwithout aggression to the historical patrimony.


Figure 19. Recent solutions proposed for confining masonry walls: a) light steel skin framing;b) stainless steel ribbons applied to masonry (Dolce et al., 2001).

However, it is very important to better perceive the costs associated to eachtechnique, the degree of benefit, feasibility of application and its durability. Thesepoints are of crucial importance when interventions in many structures are needed.Also the definition of a time for intervention requires a cost-benefit analysis withall parameters available.

6.3. URGENT INTERVENTIONS

Shore-up techniques, based on well defined modelling and experimental diagno-sis, to save damaged monuments is of prime importance in the reduction of theconsequential damages.

This requires the uses of a rapid screening methodology, for which detailedequipment of observation and measurement is required. This includes a propersetting to access to some conspicuous locations such as inner sections of vaults,and measuring the changes of frequency of some important mode shapes.

7. Emergency and Preparedness

As said before, the aim of this review paper, besides making an overall insight onthe seismic vulnerability of the historical constructions, is to analyse how to use



the vulnerability concepts right after the occurrence of an earthquake. With thisobjective in mind, we should minimise the effects of aftershocks, avoid an hurrieddemolition made under extreme pressure and help shore-up parts in risk of falling.

The role of Civil Protection and Emergency Preparedness is as follows:• Prior to the event – Evaluate the most critical monuments; enquire the means andresources as preparation for intervention; get a sheet with check list for immediateintervention; acquire equipment for inspection.• After the event – Do not let further destruction; have teams prepared for rapidintervention for ‘shore-up’ and stop aggravation due to aftershock sequences. Aguideline with the best intervention techniques should become available.

Figure 21 presents two examples of actions taken in the aftermath of importantevents with the objective of reducing the impact of future seismic activity such asaftershocks. The first one shows how it was possible to save the stability of thegable of a church in Assisi after being damaged by a previous earthquake (Croci,1998), and the second the temporary bracing of a masonry tower in the Azoresexhibiting an important diagonal cracking, capable of reducing the potential forcollapse during aftershock activity.


Figure 20. Influence of type of repairing in the vulnerability of monuments (D’Ayala, 2000).15–20% more churches without any kind of retrofit suffer the same type of damage ratio thanchurches with some kind of retrofit, for most of the domain (d = 0.15 to 0.9).

It is also very important to use all available modelling techniques to determinethe capacity of a structure for surviving a strong aftershock.

Zuccaro and Papa (2002) suggest a handbook for seismic damage evaluationof masonry structures, based on a collection of examples of mechanisms that canoccur in the structure. Together with the assessment of vulnerability index it ispossible to evaluate the state of damage of the structure during a post-event survey.Doglioni et al., (2000) presents a large collection of examples with illustrationsfor intervention after the Umbria-Marche earthquakes, an important guide to befollowed by all entities devoted to this problem.

8. A Final Word

In this review paper we have tried to link analytical modelling, experimental obser-vation and measurements with earthquake performance of historical constructions,in order to devise techniques for better evaluate their vulnerability. An historicalevolution and advancements were briefly referred.

Many issues still need good answers; explanations require new scientific de-velopments; and simple and cheap interventions are needed. But the internationalknowledge has already reached a level that, if placed at the easy of professionalsand seismic code-makers, and applied without restrictions by their owners, there isa great potential for an important mitigation of our architectural heritage.


Figure 21. Emergency interventions: a) fixing the gable (tympanum wall) of church in Assisi;b) shore-up of Torre do Relogio in Horta; c) detail of b).

‘Europe’s historical structures, until the middle of this century, consists almostentirely of masonry buildings. Not only the numerous notable individual mon-uments, but also the entire historic centres of most European centres are nowrecognised as vital elements of this historical structures; and in many areas, thesebuilding groups are in need of protection from natural as well as man-made haz-ards, of which perhaps the earthquake hazard is the most destructive’ (Spenceet al., 2000).

Acknowledgements

Many individuals from various institutions have collaborated in parts of the worksreferred in this paper. Their views and ideas helped me putting together what ispresented before. A special word to Engineer J.H. Correia Guedes from DirecçãoRegional de Cultura dos Açores for his strong defence of the built patrimony.Thanks to Prof. Paulo Lourenço from the University of Minho, Portugal, for sug-gesting me this topic for development, are also due. An acknowledgement is dueto two anonymous reviewers for their interesting comments and to the Editor forhis careful revision of the text. This work was partially supported by Fundação deCiência e da Tecnologia, Lisbon, ‘Programa Pluri-Anual’.


References

Angelides M. (1998) Monuments under seismic action. A contribution to the understanding of struc-tural behaviour and to the improvement of restoration techniques. Proc. SISM-98, Workshop onReducing Earthquake Risks to Structures and Monuments in the European Union. CambridgeUniversity.

Azevedo J. and Drei A. (1995) Characterization of typical historical construction from a structuralpoint of view: Application to the assessment of the seismic behavior of the “Águas Livres”aqueduct. Report IC-IST-AI 8/95. Instituto Superior Técnico. Lisboa.

Barbat A.H. (2001) Comportamiento sísmico de los monumentos históricos. 2◦ Congreso Iberoamer-icano de Ingenieria Sísmica. Asociación Española de Ingenieria Sísmica. Madrid. (CD-ROM, inspanish).

Brovelli E., Brun S., Carassale L., Lagomarsini S., Lemme A., Patrignani, Podestà S. and Stagno G.(1998) Assessment of damage and seismic vulnerability of churches in Umbria. Recupero andConservazione, Milano: De Lettera (in italian, in press).

Campos-Costa A., Sousa M.L.N. and Martins A. (1997) Ensaios dinâmicos ‘in situ’ da portaria deSão Vicente de Fora. Report LNEC 90/97-C3ES. LNEC, Lisbon (in portuguese).

Carocci C.F. (2001) Guidelines for the safety and preservation of historical centres in seismic areas.Proceedings, 3th International Seminar on Historical Constructions 2001. In: Paulo Lourençoand Pere Roca, pp. 145–166. Universidade do Minho, Guimarães.

Carta di Veneza. (1964) ICOMOS.Carvalho E.C., Oliveira C.S., Fragoso M.R. and Miranda V. (1998) Regras gerais de reabilitação e

reconstrução de edifícios correntes afectados pela crise sísmica do Faial, Pico e São Jorge iniciadapelo sismo de 9 de Julho de 1998. Relatório 100/98, LREC, Ponta Delgada (in portuguese).

Costa A.G. and Vasconcelos O. (2001) Caracterização das propriedades mecânicas das paredes dealvenaria tradicional das casas do Faial, Açores. 5◦ Encontro Nacional de Sismologia e Engen-haria Sísmica, pp. 451–464, Ponta Delgada. Laboratório Regional de Engenharia Sísmica (inportuguese).

Crespellani T. and Uzielli M. (2001) Geotechnical analysis and interpretation of seismic damage forthe church of S. Filippo at Nocera Umbra, Italy. XV ICSMGE Satellite Conference on ‘Lessonsfrom Recent Strong Earthquakes’, A.M. Ansal (ed.), Istanbul.

Croci G. (1998) The Basilicata of St. Francis of Assisi after the earthquake of 26 September 1997.Monument98 – Workshop on Seismic Performance of Monuments, pp. K43–K54. LNEC, Lisbon.

Croci G. (2001) Strengthening of monuments under the effect of static loads, soil settlement andseismic actions – Examples. Proceedings, 3th International Seminar on Historical Constructions2001. Paulo Lourenço and Pere Roca (eds.), pp. 1167–1190. Universidade do Minho, Guimarães.

D’Ayala D.F., Spence R.S., Oliveira C.S. and Pomonis A. (1997) Earthquake loss estimation forEurope’s historic town centres. Earthquake Spectra, 13(4), pp. 773–794.

D’Ayala D.F. (2000) Establishing correlation between vulnerability and damage survey for churches.12th World Conference on Earthquake Engineering, paper no. 2237. New Zealand.

D’Ayala D.F. and Speranza E. (2002) An integrated procedure for the assessment of seismicvulnerability of historical buildings. Proceedings, 12th European Conference on EarthquakeEngineering. Barbican Centre, London, Paper reference 561. Elsevier Science, Ltd.

DGEMN (1999) Boletim Monumentos da Direcção Geral dos Edifícios e Monumentos Nacionais.CD-ROM and http://www.monumentos.pt

Dias C.S., Ferreira M.A., Oliveira M., Pestana P. and Oliveira C.S. (2001) Planeamento em zonasde risco sísmico- Cidade de Faro. 5◦ Encontro Nacional de Sismologia e Engenharia Sísmica,pp. 185–198, Ponta Delgada. Laboratório Regional de Engenharia Sísmica (in portuguese).

DISEG (1999) Vulnerabilità sismica del patrimonio monumentale, http://www.diseg.unige.it/ricerca/gruppi/vulnerabilita/GNDT.htm (in italian).

Doglioni F., Moretti A. and Petrini V. (1994). Churches and earthquakes. LINT. Trieste (in italian).


Doglioni F. et al. (2000) Codice di pratica (linee guida) per la progettazione degli interventi diriparazione, miglioramento sísmico e restauro dei beni architettonici danneggiati dal terremotoumbro-marchigiano del 1997. In Bollettino Ufficiale della Regione Marche, Anno XXXI, N.Ed.S.15, pp. 11–252, Regione Marche (in italian).

Dolce M., Nigro D., Ponzo F.C. and Marnetto R. (2001) The cam system for the retrofit of ma-sonry structures. 7th International Seminar on Seismic Isolation, Passive Energy Dissipationand Active Control of Vibrations of Structures, Assisi, Italy, October 2–5.

Drei A. and Oliveira C.S. (2001a) Preliminary study of a stone masonry aqueduct under seismicloading using linear modelling: application to the ‘Aqueduto da Prata’ in Évora. 3rd InternationalSeminar on Structural Analysis of Historical Constructions. Universidade do Minho, Guimarães.

Drei A. and Oliveira C.S. (2001b) The seismic behaviour of the ‘Aqueduto da Amoreira’ in Elvasusing distinct element modelling. Proceedings, 3th International Seminar on Historical Con-structions 2001. In: Paulo Lourenço and Pere Roca (eds.), pp. 903–912. Universidade do Minho,Guimarães.

EC8-prEN 1980-3 (2002) Design of structures for earthquake resistance. Part 3: Strengthening andrepair of buildings. Doc CEN/T/TC250/SC8/N323.

EERI (2001) http://www.eeri.org.Gil N.P. and Oliveira, C.S. (1994) Convento do Carmo Church: Dynamic analysis. Report CMEST.

Lisbon (in portuguese).Giuffrè A. (1992) Evolution of seismic codes in Italy. In Colloque International: les Systemes

Nationaux Face aux Séismes Majeurs. Ed. L. Mendes Victor, Lisbonne.Giuffrè A. (1993) Guida al progetto di restauro antisismico. In: Giuffrè A. (ed.), Sicurezza e Con-

servazione dei centri storici in area sismica, il caso Ortigia, pp. 151–188, Bari: Laterza (initalian)

Guedes, J. H. Correia, Oliveira, C. and Lucas, A. (1999) Ensaio à rotura de paredes de alvenariatradicional. 4◦ Encontro Nacional de Sismologia e Engenharia Sísmica, pp. 503–513, Faro.Escola Superior de Tecnologia, Universidade do Algarve (in portuguese).

Guerreiro L.M., Branco F.A. and Azevedo J. (1990) Análise estrutural de uma igreja do séc. XVI,Proc. 2as. Jornadas Portuguesas de Engenharia de Estruturas, Vol. II. LNEC, Lisboa (inportuguese).

Hart R.D., Cundall P.A. and Lemos J.V. (1988) Formulation of a three-dimensional distinct elementmodel – Part II: Mechanic calculations. International Journal of Rock Mechanics and MiningSciences, 25, pp. 117–125.

HAZUS 99 (1999) Natural hazard loss estimation methodology. http://www.hazus.org.Jiménez J.I., Villareal J.I., Centeno M.R., González B.G., Correa J.J.G., Acevedo C.R. and Salazar

I.S. (1999) Tehuacán, México, Earthquake of June 15, 1999. Seismological Research Letters,70(6), 698–704.

Kaya S.M., Yuzugullu O., Erdik M. and Aydmoglu N. (2002) Earthquake performance of Suley-maniye Mosque. Proceedings, 12th European Conference on Earthquake Engineering. BarbicanCentre, London, Paper reference 851. Elsevier Science, Ltd.

Kendrick T.D. (1956) The Lisbon earthquake. Methuen & Co. Ltd. London.Kirikov B. (1992) History of earthquake resistant construction from antiquity to our time. Instituto

de Ciencias de la Construcción Eduardo Torroja, y Fundación MAPFRE. Madrid.Lagomarsino S. (1998) A new methodology for the post-earthquake investigation of ancient

churches. 11th European Conference on Earthquake Engineering, Paris. CD-ROM, Rotterdam:Balkema.

Lagomarsino S., Podestà S. and Resemini S. (2002) Seismic response of historical churches. Pro-ceedings, 12th European Conference on Earthquake Engineering. Barbican Centre, London,Paper reference 471. Elsevier Science, Ltd.

Lemos J.V. (1998) Discrete elements analysis of the São Vicente de Fora model test. Monument98 –Workshop on Seismic Performance of Monuments, pp. 13–20. LNEC, Lisbon.


Lemos J.V., Schiappa de Azevedo F., Oliveira C.S. and Sincraian G.E. (1998) Three dimensionalanalysis of a block masonry pillar using discrete elements. Monument98 – Workshop on SeismicPerformance of Monuments, pp. 117–126. LNEC, Lisbon.

Lizzi F. (1990) Principles and techniques for consolidation of structures, particularly monuments, inseismic areas. Conference in: 10 anos após o sismo de 1 de Janeiro de 1980 nos Açores, Angrado Heroísmo.

Lourenço P.B. (1996) Computational strategies for masonry structures. PhD Thesis. Delft UniversityPress. The Netherlands.

Lourenço P.B., Oliveira D.V., Mourão S.C., Cóias e Silva V. and Vicente A. (1999) Análise sísmicada Igreja de Santo Cristo em Outeiro. 4◦ Encontro Nacional de Sismologia e Engenharia Sísmica,pp. 669–678, Faro. Escola Superior de Tecnologia, Universidade do Algarve (in portuguese).

Lourenço P. (2001) Analysis of historical constructions: from thrust-lines to advanced simulations.Proceedings, 3th International Seminar on Historical Constructions 2001. In: Paulo Lourençoand Pere Roca (eds.), pp. 91–116. Universidade do Minho, Guimarães.

Lourenço P. and Mourão S. (2001) Safety assessment of Monastery of Jerónimos, Lisbon. Proceed-ings, 3th International Seminar on Historical Constructions 2001. In: Paulo Lourenço and PereRoca (eds.), pp. 697–706. Universidade do Minho, Guimarães.

Macchi G. (1998) Seismic risk and dynamic identification in towers. Keynote lecture, Pro-ceedings, Monument-98, Workshop on Seismic Performance of Monuments. Joint EditionDGEMN/LNEC/JRC, Lisbon.

Mainstone R.J. (1993) The structural conservation of Hagia Sophia. Structural Repair and Main-tenance of Historical Buildings. In: Brebbia C.A. and Frewer R.J.B. III (eds.), ComputationalMechanics Publications. Southampton.

Meguro K. and Hakuno M. (1994) Simulation of collapse of structures due to earthquakes using theextended distinct method. 10th World Conference on Earthquake Engineering, 7, pp. 3793–3796.

Meguro K. and Tagel-Din (1997) A new efficient technique for fracture analysis of structures. Bul-letin of Earthquake Resistant Structure Research Center, (30). Institute of Industrial Science, TheUniversity of Tokyo, Tokyo.

Meli R. (1998) Ingenieria estructural de los edificios históricos. Fundación ICA, México D.F. (inspanish).

Meli, R., Rivera, D. and Miranda, E. (2001) Measured seismic response of the Mexico CityCathedral. Proceedings, 3th International Seminar on Historical Constructions 2001. In: PauloLourenço and Pere Roca (eds.), pp. 877–886. Universidade do Minho, Guimarães.

Menga R., Pizzigalli E. and Ravasio F. (1998) Seismic monitoring system of the San FrancescoBasilica in Assisi – Italy. Proceedings, Monument-98, Workshop on Seismic Performance ofMonuments. Lisbon, Joint Edition DGEMN/LNEC/JRC.

Moreira D., Neves N., Arede A. and Costa A.G. (2001) Análise sísmica da igreja da Madalena naIlha do Pico. 5◦ Encontro Nacional de Sismologia e Engenharia Sísmica, pp. 627–638, PontaDelgada. Laboratório Regional de Engenharia Sísmica (in portuguese).

Oliveira C.S., Lopes M.S. and Martins A. (1991) Comportamento sísmico do aqueduto das ÁguasLivres. Relatório 202/91 NDA. LNEC, Lisboa (in portuguese).

Oliveira C.S., Lucas A. and Guedes J.H. Correia. (1992a) Monografia: 10 anos após o sismo de 1 deJaneiro de 1980 nos Açores. Edição SRHOP/LNEC, Lisbon (in portuguese).

Oliveira C.S., Lucas A., Guedes J.H. Correia and Andrade R.A. (1992b) Metodologia para quantifi-cação dos danos observados no parque monumental. In Monografia: 10 anos após o sismo de 1de Janeiro de 1980 nos Açores. Edição SRHOP/LNEC, Lisbon (in portuguese).

Oliveira C.S., Gil N.P. and Fragoso M.R. (1994) Estudo Experimental e Analítico do ComportamentoDinâmico de Estruturas de Alvenaria de Pedra. Aplicação à Ponte da Ribeira Grande, Proc. 2◦ENCORE, LNEC, Lisbon (in portuguese).


Oliveira C.S., Martins A. and Lopes M.S. (1995) Seismic studies for the “Águas Livres” aqueductin Lisbon. 10th European Conference on Earthquake Engineering. In: Duma (ed.), Balkema:Rotterdam, Nederland.

Oliveira C.S., Guedes J.H.C., Lucas A. and Oliveira H.M. (2000) Estudo sobre o comportamentosísmico de edifícios de alvenaria tradicional. Proceedings REPAR 2000, pp. 467–476, Lisboa (inportuguese).

Oliveira C.S., Lemos J.V. and Sincraian G.E. (2002) Modelling large displacements of structuresdamaged by earthquake motions. European Earthquake Engineering 3, 56–71, Pàtron Editore,Bologna.

Oliveira D. and Lourenço P.B. (2001) Um modelo para o comportamento cíclico de elementosjunta Outeiro. 5◦ Encontro Nacional de Sismologia e Engenharia Sísmica, pp. 291–304, PontaDelgada. Laboratório Regional de Engenharia Sísmica (in portuguese).

Pegon P. and Pinto A.V. (1996) Seismic study of monumental structures – Structural analysis, mod-elling and definition of experimental model. Report EUR 16387 EN, JRC, European Commission.Ispra, Italy.

Pereira de Sousa F.L. (1923) O Terramoto do 1◦ de Novembro de 1755 em Portugal e um EstudoDemográfico, 4 Vols., Memória dos Serviços Geológicos de Portugal, Lisbon (in portuguese).

Pinto A.V., Verzeletti, G., Molina, F.J., Plumier C., Vaz C.T., Carvalho E.C. and Oliveira C.S.(1998) Laboratory tests on a full-scale monument. Proceedings, 11th European Conference onEarthquake Engineering. Paris.

Ravara A.P. (1980) Estudos sobre a acção do sismo dos Açores de 1/1/1980. 2◦ Relatório, LNEC,Março de 1980 (in portuguese).

RSAEEP (1983) Regulamento de segurança e acções em estruturas de edifícios e pontes. Dec. Leino. 235/83 de 31 de Maio de 1983. Casa da Moeda. Lisboa (in portuguese).

Sarà G., Barbetti G., Marsicano S. and Pintucchi B. (2002) Seismic vulnerability reduction of tradi-tional masonry buildings. Proceedings, 12th European Conference on Earthquake Engineering.Barbican Centre, London, Paper reference 508. Elsevier Science, Ltd.

Sincraian G.E., Lemos J.V. and Oliveira C.S. (1998) Assessment of the seismic behavior of a stonemasonry aqueduct using the discrete element method. 11th European Conference on EarthquakeEngineering, September 1998, Paris. Balkema, Rotterdam.

Sincraian G., Oliveira C.S. and Lemos J.V. (1999) Comportamento de um Farol de Alvenaria dePedra sob Acção Sísmica Intensa. 1.er Congreso Nacional de Ingenieria Sismica, Murcia, Spain(in portuguese).

Sincraian G.E. (2001) Seismic Behaviour of Blocky Masonry Structures: A Discrete Element MethodApproach. PhD Thesis, Instituto Superior Técnico, Lisbon.

Spence R.S., Papa F., Oliveira C.S., D’Ayala D.F. and Zuccaro G. (2000) The performance ofstrengthened masonry buildings in recent European earthquakes. 12th World Conference onEarthquake Engineering, paper no. 1366, New Zealand.

Tassios T.P. (1988) Meccanica delle murature. Liguori Editore. Napoli, Italia (in italian).The TOSQA Project 1994-1996. Earthquake Protection for Historic Town Centres. Commission for

the European Communities. EV5V-CT93-0305, Brussels.Zuccaro G. and Papa F. (2002) Multimedial handbook for seismic damage evaluation and post

event macroseismic assessment. Proceedings, XXIIIth General Assembly of the EuropeanSeismological Commission. Genova (CD Rom).


Annexes - case studies

Several case-studies referred in the text are here described in more detail for rea-sons of completeness.

CASE 1. AQUEDUTO DAS ÁGUAS LIVRES (Oliveira et al., 1991)

The construction of the Aqueduct was initiated in 1739 and its inauguration tookplace in 1748. It is a 941 m long bridge-type structure with 35 arches, many ofgothic type, supporting an upper deck extending across the valley of Alcantara,Figure 4. The higher point over the valley, is around 60 m. The Aqueduct is ahollow structure made of masonry, well cut in the exterior, filled with rubble inthe interior. On the upper deck accompanying all the development of the Aqueductthere is a gallery with towers for ventilation. Foundations of the different pillarsare superficial and the geotechnical material, even though of general good quality(limestone), changes quite significantly along the valley cross-section (for detaileddescription see Oliveira et al., 1991 and 1995).

It was difficult to obtain information on the detailing of structural elements. Nodata was available, besides general drawings with main dimensions and a detailedaccount on material costs. Complementary observation is currently under way toconfirm the geometry and arrangement of the interior sections, to detect possiblecavities and their geometry, and to estimate the state of deterioration.

Studies on the construction techniques practiced in other monuments and otherregions could be of great value for a better understanding of this type of monumen-tal structure.

Damaged inflicted to the Aqueduct during the 1755 earthquake is not com-pletely well understood. From different descriptions, mild damage seems to haveoccurred to 3 out of 16 towers. The seismic MMI intensity in this area was VIIIaccording to Pereira de Sousa (1923). Inspection of the structures indicates thepresence of quite a significant number of steel bars in the towers, fixing some ofthe large stone blocks. These steel bars, presently in quite rusty condition, are ran-domly located but more frequently observed in the towers near to the center of thevalley. This indication suggests that damage occurred during the 1755 earthquakemight have been repaired using this technique.

In order to characterize the dynamic behaviour of the structure, an experimentaltesting was performed ‘in-situ’ using both ambient and made-man vibration. Sevenvelocity transducers were placed along the structure, Figure 4, and recorded themotion in transverse direction. For the ambient vibration, 10 samples per channel,2048 points at 40 Hz sampling rate, were made. For the made-man only 4 samples/ channel were needed.

Amplitudes of vibration above the higher arch were one micron for ambientand ten times larger for the made-man. The processing of data consisted on thecomputation of the Fourier spectrum for each time series, and obtaining an averageFourier spectrum for the whole 10 series.


Table VII. Comparison between computed/measured frequencies forthe first 10 modes of the ‘Aqueduto das Aguas Livres’.

Mode Frequency (Hz) Mode description

# Analytical Experimental

1 0.90 0.90 Transverse symmetric

2 1.24 1.15–1.20 Transverse anti-symmetric


4 2.05 1.90 Transverse anti-symmetric


6 2.96 Transverse anti-symmetric

7 2.98 2.90 Longitudinal

8 3.44 Transversal

9 3.87 Transversal

10 4.38 Partial transversal

This treatment was found to be very efficient in detecting the main frequen-cies of vibration in the different locations under analysis. This permitted a firstinterpretation of the three first mode shapes. Band-pass filtering the time signal forfrequencies in the neighbourhood of each natural frequency, checking the phaseof signal pairs and averaging out for many samples the results at each one of theseven locations, it was possible to determine experimentally the mode shapes (andcompare them with the analytical model, Figure 4 and Table VII).

Other multi-channel techniques, such as the use of cross-spectra density matrixand their largest eigen value, were not used because the first technique gave verygood results.

An average value of 1.9% damping ratio for the first mode was obtained fromfree vibrations produced after the made-man testing.

CASE 2. AQUEDUTO DA PRATA (Drei and Oliveira, 2001A)

The aim of this research was to provide a first simple model of the seismic be-haviour of the medieval stone masonry aqueduct of Évora, in Portugal, called‘Aqueduto da Prata’ (Figure 5). The study was carried out in two steps: the firstwas an on-site determination of the geometric and architectural features of the con-struction, rough site measurements of the resistance characteristics of the materials,and direct measurements of the main dynamic properties. The second step was afinite element linear modelling of the structure, based on the site measurements, inorder to obtain preliminary indications for the evaluation of the structural safety inwhat concerns the seismic risk point of view. The seismic load considered was thatprovided by the Portuguese seismic code (RSAEEP, 1983).


The structure was modelled using 3-D beam-type bar elements connecting 6degrees-of-freedom nodes. Only the central part of the Aqueduct was represented:15 pillars forming 14 arches, in a total of 96 bars and 97 nodes. The bars werechosen, as having uniform geometric properties throughout their lengths. Due tothe good foundation conditions all pillars were fixed at the foundation. Even thoughthis representation was not the best model for the structure due to: (i) neglect theinfluence of less longer lateral parts of the Aqueduct; (ii) failure to simulate wellthe outer and inner sections of the pillars and the arches, the model was found tobe good enough to obtain the first mode shapes and frequencies.

For simplicity, the inertia values of the bars were taken from the geometry of fullsections, considering an homogeneous material (γ = 25 kN/m3). The equivalentmodulus of elasticity obtained to fit the results of experimental testing was E =19 GPa, with good agreement for all identified mode shapes. This value is abovethe values one could anticipate from past experience, but it also reflect that one isin presence of a very well constructed structure.

The obtained results, even if coming from a simplified model, seems to indicatethat ‘Aqueduto da Prata’ could undergo serious damage in case of an earthquakecomparable to the seismic load assumed. Therefore, more refined models to studythe seismic behaviour of this historical structure, as well as to determine differentpossible reinforcements, are necessary.

CASE 3. LIGHTHOUSE OF RIBEIRINHA (FAROL) (Oliveira et al., 2002)

One case of great importance to calibrate the possible acceleration values producedin the July 9, 1998 earthquake in Faial, Azores, was the lighthouse of Ribeirinha, anold masonry construction of good quality built around 1918 on the top of pyroclastlava 100 m high, very close to an almost vertical cliff. It is located at approximately5 km from the epicentre. The structure is composed by a tower with 13.5 m highwith a thick cross-section almost constant in elevation (outside – rectangle with5.8 × 6.2 m at the base; inside – circle with 4.0 m diameter), topped by an opticalsystem, surrounded by an H-shape house, 2-storey high (Figure 7a).

The July 9, 1998 event caused extensive damage to the tower, to the house,to the outside facilities including land property walls, and to the embankment onthe southern part. Describing only the effects in the tower, one can say that the ithas undergone important deformation with the formation of large inclined crackspassing through the openings, very consistent throughout all the four sides. Thegeneral appearance is of an overall rotation in the tower, counter clockwise viewedfrom above, with peripheral enlargement at mid height. The tower was broken intoseveral solid blocks, which show relative horizontal displacements above 50 cm. Atthe top, the optical system suffered no damage (light was on for some time after theevent), the only problem arose in the two glasses of the outer cylindrical window(at opposite locations along NNE-SSW axis) which broke. In the support of the


optical system the 3 cm thick steel skirt, which connects to steel plate, cracked in anextension of 50 cm denoting again intense movement in the NNE-SSW direction.

The discrete element model (DE) was based on the simplest 3-D approximation.Only rigid blocks were used. Deformable blocks would make the dynamic analysisprohibitive. Thus, the entire deformation of the system has to be given to the joints.A number of simplifications were required to make the analysis reasonable froma computational point of view, as it is not possible to include in the model everyindividual block in the real structure. The block structure has to be chosen in such away that the main modes of deformation and failure of the system can be capturedwith reasonable accuracy.

The DE model shown in Figure 7b is composed of about 80 rigid blocks. Thetypical cross-sections are also shown in this figure. It should be noted that in thecode 3DEC (Hart et al., 1988), only convex blocks are permitted, so more complexgeometries are generated by joining blocks. The corner blocks in the present modelare represented by 2 blocks joined together. Also, the numerical representation ofa structure in 3DEC is based on blocks of polyhedral form. Therefore, the circularinterior cross-section of the tower has to be represented in an approximate mannerby straight segments.

An elasto-plastic constitutive model was used for the mortar joints, with a Cou-lomb friction slip criterion. The mechanical properties of the joints were as follows:

Normal stiffness 1000 MPa/mShear stiffness 400 MPa/mCohesion 0.5 MPaResidual cohesion 0.5 MPaFriction angle 35◦Tensile strength 0

For the unit mass the materials an average value of 2000 kg/m3 was chosen.The mass density of the top block was assumed to simulate the real total mass ofthe optical system and enclosure. The height was chosen to account for the correctcentre of mass elevation. The top block is joined rigidly to the slab block. For thejoints at the base of the model the values of friction angle, cohesion and tensilestrength were increased to 45◦, 5.0 MPa and 1.5 MPa, respectively, in order toprovide some confinement that in reality was given to the tower by the surroundingbuildings. The cross-section of the model is a square of 6.0 × 6.0 m at the base andit narrows towards the top to a square of 5.1 × 5.1 m.

The natural frequency of the elastic model was about 4 Hz. The in-situ measuredfrequency after the earthquake was about 2.5 Hz, a value slightly lower than thefrequency obtained for the damaged model (about 2.75 Hz).

For a better understanding of the phenomenon, the deformed structure (withoutthe top block) is shown for the three magnification factors (Figure 7b). The rotationmovement may be clearly noticed. Also, the large sliding and joint separation,


especially in the top section of the tower are noticeable, even if the top slab block(not shown), simulating the steel plate, provides some measure of linkage of theupper blocks. The earthquake magnified 1.5 times, with a PGA of 0.6 g, alreadyinduces some permanent displacements localised more in the upper part of thestructure, whereas the bottom part remains almost undamaged. Damage increases,with noticeable rotation components, for the factor F = 2. For the earthquake mag-nified 3 times, the damage is substantial and distributed all over the structure, withmany lower joints open.

CASE 4. STATUE IN HORTA (Oliveira et al., 2002)

The statue of Manuel Arriaga is located downtown Horta, near the harbour at 500 mfrom the Observatório Príncipe de Mónaco. The total height of the structure isaround 5 m and it is composed by a pedestal of about 3 m height and a 2 m tallstatue, anchored to the pedestal by means of a metallic plate. The pedestal is madeof basaltic blocks with no mortar in the joints. The effect of the earthquake onthe statue was a rotation, clockwise seen from above, Figure 10, with a maximumdisplacement at the top joint of 5 cm.

The 3DEC model is composed of 14 rigid blocks (Figure 8). The top block sim-ulating the statue is rigidly joined to the one representing the metallic plate, so thatthere is no relative displacement between them. For the joint located between thestatue and the pedestal high values of friction angle, cohesion and tensile strengthwere used in order to prevent the top block from falling apart. The cross-sectionof the pedestal is 80 × 80 cm with a 90 × 90 cm block at the base. For the jointsbetween the basalt blocks, a value of the friction angle of 45◦ was assumed in thestudy. For the unit mass of the basalt stones a value of 2800 kg/m3 was chosen,while for the top block of the model a value of 1000 kg/m3 was used to simulatethe actual mass of the bronze statue.

As the case of the lighthouse, the elasto-plastic constitutive model was used forthe mortar joints, with a Coulomb friction slip criterion. The mechanical propertiesof the joints were as follows:

Normal stiffness 1000 MPa/mShear stiffness 400 MPa/mCohesion 0Residual cohesion 0Friction angle 45◦Tensile strength 0

The three components of the Faial earthquake obtained at the ObservatórioPríncipe de Mónaco were scaled by factors of 0.4 to 1.0 (Figure 8) to examinethe influence of the PGA level on the response of the structure.


The rotation movement was clearly observed in all the cases as an effect of theabove mentioned input without applying any rotational component at the base ofthe model. A permanent displacement at the end of each run was noticed, especiallyat the upper part of the pedestal, well in accordance with the damage observed inthe real structure, from a qualitative point of view.

In what regards the magnitude of this permanent displacement, it can be saidthat an input ground motion corresponding to some 70% of the registered one leadsto a displacement value at the end of the run similar to the one measured on themonument after the Faial earthquake. Also, the deformed shape of the model at theend of the run is almost identical with the damaged structure.

CASE 5. THE XVIII URBAN HOUSE IN TERCEIRA (Oliveira et al., 2000)

During the January, 1st, 1980 earthquake, the urban tissue in Angra do Heroísmowas severely affected by the earthquake. The structures that constitute this tissueare old masonry buildings two/three story high organized in rectangular blocks.They are very typical of the construction practiced prior to mid XXth century. Themain structural elements are two parallel front walls, one facing the street, theother to the back, and two perpendicular walls at the lateral edges, forming a box-structure. Inside these peripheral walls there are partitions in thin wood or brickelements, and floors are made of wood beams.

Front walls are of better and thicker rubble masonry than the edged walls, butthey exhibit a large portion of openings for windows.

The understanding of the behaviour of these types of historical structures un-der the seismic loads is of primary importance in respect to define measures forretrofitting. One example was selected for a detailed analysis using linear methodsand comparing the results with in-situ testing and with real recorded earthquakeweek motion. The GZCAH building is a traditional 3-story masonry house withinan antique block of buildings (Figure 9).

It has a length of 14 m along the street and 11.5 m to the interior. Front wallsare 0.66 m thick at the base and progressively decrease to 0.45 at the top. Par-titions are 0.15 m thick in lower stories and 0.10 m in the upper stories. Frontwalls are profoundly opened by windows and doors which give space to shortbalconies. Foundation soils are of moderate consistency and this building sufferedmild cracking during the 1980 earthquake.

The fact that this building is located in a steep area, interacting with the ad-jacent structures made it very difficult to succeed in the analytical modelling. Inreality, frequencies of vibration depend very much on this interdependence and theresponse for a given event with records at the ground floor and on the two otherstories are not well reproduced by the model (Figures 11). However, it is very clearthat the building clearly amplifies the motion from bottom to top (Figure 10) asis was consistently observed during different events with PGA at in the order of


0.03 g. Modelling was made using masonry properties as obtained from testing ofcomponents.

CASE 6. THE COSISMO PROJECT – SÃO VICENTE DE FORA (PINTO et al., 1998)

The COSISMO project, a joint research project on seismic analysis/assessmentof monuments was a first attempt to contribute to the required progress in thefield of old masonry monumental structures. The research programme includedthe following main tasks:• Dynamic characterisation of a representative monumental structure, the SãoVicente de Fora monastery in Lisbon, by in situ testing and numerical modelling;• Laboratory testing at the reaction wall of the European Laboratory for StructuralAssessment (ELSA) of a representative model of part of the structure at full scale,which enabled the calibration and/or development of non-linear numerical modelsto be used for predicting the earthquake response of such structures;• Development and calibration of non-linear and equivalent linear models appro-priate for high intensity shaking;• Assessment of the seismic vulnerability of the Monastery, using the developedand calibrated models and appropriate seismic hazard characteristics.• To investigate the applicability of some retrofitting solutions and techniques formonumental structures.

Pseudo-dynamic and cyclic tests on a full-scale model of the São Vicente façademodel (Figure 12) were carried out at the ELSA laboratory of the Joint ResearchCentre and aim at the characterisation of the non-linear behaviour of limestoneblock structures under earthquake loading. Local and overall stability of the stone-blocks, including columns and arches were assessed for large displacement am-plitudes. Furthermore, stiffness and equivalent damping ratios are evaluated fordifferent deformation levels. The research programme on monuments aims at thecalibration of refined non-linear models to represent local behaviours and at thecalibration of equivalent linear models (stiffness and damping) to be used for thevulnerability analyses and safety assessment of such a type of structures (Pegonet al., 1996). It is also attempted to investigate the effects of retrofitting tech-niques based on the classical ties, in particular in what concerns post-tensioningand eventual additional damping characteristics.

The monastery, designed in an austere style, is a typical example of the Por-tuguese architecture in the XVIth century. It still has its principal parts surroundingthe cloisters. In addition to the high artistic level, representative of the Portuguesefine arts of the XVIIIth century, the monastery of São Vicente de Fora reflects andrepresents important phases of the History of the country.

From the architectonic/engineering point of view, it represents the typical mon-ument of Lisbon, where limestone block masonry columns and arches forming aresistant structure are harmoniously combined with stone masonry bearing wallsand ceramic dome floors/roofs with beautiful frescos painted underneath.


The monument survived to the catastrophic November 1st, 1755, Lisbon earth-quake, which destroyed the city and killed more than 10 thousand lives. However,today some of the effects of the strong ground shaking are still visible there. The2 m thick south external wall of the monastery became curved (mid-span disloca-tion of about 40 cm, the same happened to the west end-side external wall and theEast end-side edifice, where the Pantheons are presently located, collapsed. A quitedetailed description of the damages to the monastery during the 1755 earthquakeis available. Hence, prediction of the damages to the monastery using the presentmodelling capabilities calibrated on the basis of the experimental results obtainedby in situ tests and by laboratory tests is therefore a challenge.

The test model was constructed using materials and construction techniques(stone blocks arrangement) similar to the prototype. It is a plane structure withthree stone block columns, two complete arches and two external half arches. Theupper part of the model is made of stone masonry. Three millimetres thick mortarjoints were assured during the construction. A photographic view of the model anda schematic representation of the test set-up and model dimensions are shown inFigure 12.

Concerning the model it is noted that it contains very typical components of themonumental construction. In fact, block-stones assembled columns supporting thearches, or vaults at different storey levels. The cloister ground floor open and theupper parts filled with stone-masonry is a common architectural/structural solutionin Lisbon also adopted in the reconstruction after the 1755 earthquake.

For example, the down town main place in Lisbon (Praça do Comércio), re-constructed after 1755, is completely surrounded (except the river side) by col-umn/arch facades very similar to the one adopted for the full-scale model. Thephysical model has been defined taking into account the following aspects: it shouldrepresent typical monumental constructions, should be representative of the con-structions (realistic, in terms of materials, scale and stone arrangements).

The question of using a scaled model has been raised many times by the teaminvolved in this project. It was aimed at a realistic model to which simple andrealistic boundary conditions could be applied.

Several tests were envisaged for this model; the initial dynamic characterisationtests in order to obtain frequencies and mode shapes and evaluate damping for verylow deformation levels (microns), initial stiffness tests, two pseudo-dynamic testsfor earthquake intensities corresponding to a low and a medium value of returnperiod and final cyclic tests with pre-defined displacement histories. These lasttests permitted to investigate the influence of boundary conditions, namely the post-tensioning forces in the horizontal ties.

From the dynamic characterisation tests and the stiffness tests were obtainedinitial stiffness, which, in conjunction with the required initial frequency of themodel, dictated the mass to be used in the pseudo-dynamic tests. It is noted that forthe pseudo-dynamic tests a one-degree of freedom system (1DOF) was considered.


Two pseudo-dynamic tests were initially envisaged, corresponding to a moder-ate and a high earthquake intensity. In addition, two types of earthquakes wereconsidered, because two earthquake scenarios should be considered for Lisbon(far-field and near-field – RSAEEP, 1983) with rather different energy content).

The tests on the São Vicente façade model have shown the deformation capacityof the column-arch system commonly used in many monumental structures. Driftsof about 2% were imposed without any loss of the load carrying capacity. Further-more, the model has shown important dissipation characteristics due to crackingand friction. Cracking appeared at the interfaces block/masonry in the arch zonesand at the interface between the upper part of the columns and the masonry. Suchdissipation characteristics for important deformation levels exist thanks to the tiebars (pre-compression forces), which allow to develop the required confining ofthe upper masonry part of the façade model.

It is therefore expected good deformation and dissipation characteristics ofthese type of structures if a rationally distribution of ties at the floor levels exists.Design and practical application of these tie systems using new analysis tools andconstruction techniques are under investigation.

The entire structure of São Vicente de Fora Cloister was subjected to a prelim-inary analytical model to check to safety of the structure under seismic loading.This model was calibrated for the initial stiffness with in-situ measurements ofvibration (Campos-Costa et al., 1997), and the recommendations from the pseudo-dynamic full test of the façade should be included in the final model (Lemos, 1998).Unfortunately, definite results are not yet available.

2003137

Documents

historical structures

lisbon earthquake

historical centres

izmit earthquake

earthquake impacts

azores earthquake c

existing structures

seismic behaviour1