1

First European Conference on Earthquake Engineering and Seismology (a joint event of the 13th ECEE & 30th General Assembly of the ESC)

Geneva, Switzerland, 3-8 September 2006 Paper Number: Keynote Address K9

MASONRY BUILDING DESIGN IN SEISMIC AREAS: RECENT EXPERIENCES AND PROSPECTS FROM A EUROPEAN STANDPOINT

Guido MAGENES1

SUMMARY

The recent development of a new Italian seismic code, conceived as a transition towards the final adoption of Eurocodes, has been an occasion to reconsider thoroughly the current criteria for seismic design of masonry buildings, their experimental and theoretical basis, their consequences on practice. Drawing from this experience, a critical discussion of design criteria is presented, with specific attention to the definition of behaviour factors to be used in elastic analysis and more generally on methods and models for the seismic performance verification of masonry buildings. Necessary and possible developments of design/assessment procedures and code provisions are presented, and research needs are outlined.

1. INTRODUCTION Research on the seismic behaviour of masonry structures is nowadays almost entirely dedicated to existing buildings and to the issues related to assessment and reduction of their seismic vulnerability. For countries with a long story of civilization, the seismic protection of masonry buildings involves also the issue of protecting the cultural heritage of the country. There is also a rather generalized negative attitude towards the use of structural masonry for new buildings in seismic areas, since most collapses and deaths in recent earthquakes are due to inadequate performance of unreinforced masonry buildings (usually non-engineered, low-quality, old dwellings). This explains why the large majority of the current scientific and technical literature on seismic behaviour of masonry is dedicated to the study of existing structures and very seldom is masonry being nowadays considered as a choice for the design of new structures. In design of new buildings the structural behaviour of masonry elements is of concern mostly in the case of infilled frames, in which however the masonry panels are not meant to serve as primary resisting elements. Considering the European situation, it is at the same time striking and meaningful that in the last two European Conferences on Earthquake Engineering (Paris, 1998, and London, 2002), out of roughly 600 papers per conference only respectively 4 (Paris) and 2 (London) papers were dedicated to masonry as a construction technique for new buildings in seismic areas, and none of these papers was dedicated to unreinforced masonry. On a broader perspective, the last World Conference on Earthquake Engineering held in Vancouver (2004) shows a rather similar panorama, however with slightly higher percentage of papers dedicated also to the design of new masonry buildings mainly due the contribution of Latin American countries. With such a context, it may look somewhat odd that in this paper relevance will be given to the issues of seismic design of new masonry buildings, including the unreinforced type. However, on one hand, the construction of new masonry buildings in European countries is far from being marginal, even in countries with considerable seismic hazard; on the other hand, several topics that will be discussed for new buildings are of relevance also for the seismic assessment of existing buildings. To appreciate the role of masonry construction in the current European panorama, it can be of interest to give a glance to the present Italian experience in constructions. According to recent data, in 2004 the total yearly investments for residential buildings was shared among new constructions (48.5%) and activities related to

1 Department of Structural Mechanics, Faculty of Engineering, University of Pavia, and ROSE School, IUSS, via Ferrata 1, I-27100 Pavia,

Italy. Email : guido.magenes@unipv.it

2

existing buildings (51.5%), confirming the ever increasing importance of assessment/strengthening issues versus new design. The total amount of structural masonry in new residential buildings is presently 10% of the total volume built per year (according to average data of 2000-2002), whereas in the mid-1980s was around 15%. Almost the entirety of such buildings does not exceed three storeys. The use of structural masonry, mostly unreinforced, is therefore still far from being a negligible issue in Italy, despite the seismicity of the territory. A much greater share of the construction market is reported for masonry in lower seismicity countries such as Germany (largely exceeding 2/3 of the residential market). The choice of a structural system in real practice may depend on many factors, of which the structural/seismic ones have a relative weight that at present, according to Italian practitioners (ANDIL, 2003), in average does not exceed 15-20%. The remaining factors are related to other issues such as architectural/environmental, cost, thermal and acoustic insulation, speed of construction, reduced weight/volume, availability of materials and workmanship, availability of product certifications,… Clearly, the relative weight of such factors can vary from country to country or even at a regional level within the same country. In particular, the importance of structural/seismic issues rises in the case of frequent or recent earthquakes, or in the presence of a strict seismic regulation. In general, elements that can favour masonry construction with respect to other techniques are ease of construction, durability, good insulating properties, fire resistance, sustainability, and sometimes aesthetics. On the other hand, the use of structural masonry determines a strong interconnection between structural and architectural design, whereas frame structures allow more freedom in distribution of internal space, and to separate almost completely the architectural and structural design processes. Other factors can also have a significant influence on the choice of the structural system, such as the presence or absence of masonry classes in university engineering or architectural curricula, or the presence of established design codes (a national structural code for unreinforced masonry was first introduced in Italy in 1987, and for reinforced masonry only in 1996): designers do not like to design something that they do not know, or know less than, say, reinforced concrete or steel. The aim of this paper is not to give a comprehensive overview of all the issues that pertain to the seismic design/assessment and strengthening of masonry buildings (which would require a whole book rather than a conference paper), but to discuss selected issues, mostly related to codes, which are believed to have important consequences on seismic design and assessment practices, with specific reference to the European context. In recent years the author was involved in the drafting of a new seismic design code for Italy, which was conceived as a transition from the old national code towards the final adoption of Eurocodes, in particular of Eurocode 8. A great part of what will be discussed in the following draws from such an experience and more generally from the recent Italian experiences and researches on masonry structures. Reference will be made to ordinary buildings only, which constitute the bulk of the building stock, since monumental buildings present several peculiar problems that should be treated separately.

2. SEISMIC PERFORMANCE OF MASONRY BUILDINGS: GENERAL REMARKS The seismic vulnerability of masonry buildings can have wide variations depending on a series of factors, most of which have been recognized for a long time, and have been codified in most modern design provisions. The same factors are used, to a higher or lesser extent of detail, to define vulnerability classes within macroseismic intensity scales (see e.g. EMS scale, Council of Europe, 1998) or risk assessment procedures. A summary of such main factors is reported in Table 1. It is known from the past experience that the observance of the conditions listed in the right column of Table 1 can guarantee a satisfactory seismic response of unreinforced masonry (URM) buildings under earthquakes of moderate intensity and can prevent collapse under severe ground shaking. Most of such conditions, albeit stated in different form, were part of the so called “rule of art” of construction in past centuries, and explain the survival of well built, old unreinforced masonry buildings to earthquakes. It is of utmost importance to recognize that towards the achievement of a good seismic response such principles are as important, if not more important, than the use of any modern analytical tool. In the modern context, many design codes recognize this principle by introducing the so-called “simple buildings” (e.g. Eurocode 8, CEN-EN 1998-1), i.e. regular buildings characterized by a limited height, with a sufficient total area of structural walls, in which the requirements of the right column of Table 1 are appropriately satisfied, for which no structural analysis or explicit safety verification is required. More generally, all modern masonry buildings must comply with a series of minimum requirements that encompass material properties, structural geometry and configuration, structural details, whereas structural analysis and safety checks are other steps of the design process.

3

Table 1: Factors influencing the seismic vulnerability of masonry buildings Higher vulnerability Lower vulnerability i) Insufficient quality of materials (poor mortar,

weak/brittle units), poor “internal connection” of masonry (uncoursed irregular rubble stones, multi-leaf masonry with no transverse connection…).

Regular and robust units, good bond and interlocking of units, masonry behaves monolithically through the whole thickness of the wall.

ii) Very slender walls (out-of-plane instability) Limited slenderness of walls; restraints to out-of-plane failure

iii) Lack of efficient connections among walls and between walls and horizontal structures, lack of structural redundancy

Good interlocking at wall intersections, presence of tie rods and ring beams at each floor (and roof) level to favour “box action”, efficient floor-to-wall connections which reduce stress concentration.

iv) Floors do not provide diaphragm action Sufficiently stiff and resistant diaphragms to provide restraint to out-of-plane vibration of walls, to increase structural redundancy and favour internal force redistribution.

v) Presence of horizontal thrusts (e.g. from roof or arched or vaulted structures) equilibrated only by out-of-plane resistance of structural walls

Horizontal thrusts are reacted by in-plane action of strong walls/buttresses or by suitable structural elements (ties, floor diaphragms…) to form a “closed” self equilibrating system.

vi) Excessive unsupported floor spans, widely and irregularly spaced walls

Limited floor spans, regularly spaced shear walls in at least two orthogonal directions

vii) High structural and non-structural masses and low material strength

Masses and weights produce a low stress/strength ratio

viii) Structural irregularity in plan (torsional effects, stress concentrations) and in elevation (inefficient load path, stress concentrations)

Regular structure, sufficient torsional resistance, regular path of forces from upper structure to foundation

It is essential to recognize that the wide majority, if not the entirety of the collapses of URM masonry buildings in recent earthquakes involved buildings which did not comply with most of the requirements that any new masonry building would have to satisfy according to the current seismic codes, especially regarding criteria i) to v) (a recent example is the earthquake in Pakistan, Javed et al. 2006). The reported vulnerability of recently built masonry buildings in moderate earthquakes in Italy had clearly more to do with the disregard of fundamental design rules due to lack of code enforcement/construction control policies (even of basic “non seismic” rules!) rather than with inherent unfitness of unreinforced masonry buildings to resist earthquake (Molise 2002 earthquake, Decanini et al., 2002). It seems appropriate for the purpose of this discussion to classify masonry buildings according to three different categories, as proposed in Table 2, where the term “engineered structure” implies that some structural analysis and consequent safety check is carried out, and “structurally designed” means that a series of structural requirements are satisfied, other than those related to structural analysis and safety check, such as on minimum material properties, structural details, structural layout and other issues related to the criteria listed in Table 1. By referring to this classification, modern masonry buildings should fall within the SD class (structurally designed and detailed), and their vulnerability should by no means be considered equivalent to the NSNE class. The case of the “simple buildings” could be in turn classified as SDNE, since no specific structural analysis is carried out, with the exception of checks on the average vertical stress and the total area of shear walls. Further considerations can be made with reference to the definition of seismic hazard when described in terms of peak ground acceleration and macroseismic intensity. In modern codes, including Eurocode 8, the seismic hazard is described in terms of the reference peak ground acceleration on rock, which is actually the value used to scale or “anchor” the spectral shapes for design/assessment purposes. In EC8 the accepted return period for the no-collapse requirement corresponds to a return period of 475 years, equivalent to a 10%probability of exceedance in 50 years. By comparing hazard maps expressed in terms of PGA and hazard maps expressed in terms of macroseismic intensity scales defined for the same return period (see e.g. Albarello et al. for the Italian territory, 2000), an attempt can be made to establish a rough correspondence between PGA intervals and macrointensity levels, as shown in Table 3 (Mercalli-Cancani-Sieberg intensity has been used until now for all

4

Table 2: Classification of masonry buildings

Not Structurally Designed (NS)

Structurally Designed and Detailed (SD)

Non Engineered (NE) Engineered (E) NSNE SDNE SDE

Table 3: Approximate correspondence between “code” PGA and macroseismic intensities.

PGA/g MCS EMS 0.-0.1 up to VII up to VI

0.1-0.2 VIII-IX VII-VIII 0.2-0.3 (IX -) X VIII-IX 0.3-0.5 XI-XII IX-X

macroseismic intensity assessments in Italy). Although it is not possible to establish a clear correspondence between MCS and the more rigorous European Macroseismic Scale levels (EMS, Council or Europe, 1998), as shown by Spence (1999), the last column of Table 3 suggests a rough correspondence, which however produces also good agreement with that reported for instance in (Tomaževič, 2000) regarding the relationship between PGA intervals and EMS levels. The levels of EMS can be also considered equivalent to the levels of MMI scale. Referring to the criteria of EMS, which defines a series of vulnerability classes with specific reference to the type of vertical and horizontal structures, a new SD unreinforced masonry building would be classified as vulnerability class C. It could be argued whether well built and engineered (SDE) buildings could belong to the higher class D (of lower vulnerability), but this does not affect significantly the discussion. According to EMS, a level VII would correspond to a situation in which only a few buildings of class C would suffer damage of grade 2 (slight structural damage, moderate non-structural damage); a level VIII EMS would see many class C buildings suffering damage of grade 2, a few of them damage of grade 3 (moderate structural damage and heavy non-structural damage, with failure of individual non-structural elements such as partitions or gable walls). It should then be inferred that on the base of the past observational experience and of the safety levels accepted in codes, the behaviour of SD unreinforced masonry buildings should be considered adequate with respect to the ULS (no-collapse) with design PGA up to 0.2, and an accurate design and construction of low-rise SDE URM buildings should be possible also for PGA up to 0.3 g. Needless to say, although the history of masonry design and construction has seen examples of multi-storey masonry buildings well beyond the limit of three to four storeys (Hendry, 2002), these typologies are unsuitable in new design for seismic resistance, unless reinforcement is used (reinforced or post-tensioned masonry), or unless the seismic hazard is very low (PGA < 0.04-0.05g).

3. SEISMIC DESIGN OF MASONRY BUILDINGS: THE CURRENT APPROACH European national codes on the seismic design/assessment of masonry buildings have shown until recent times a rather heterogeneous, apparently even contradictory approach to the problem (Vratsanou, 2001), especially regarding the seismic design load levels, which show wide variations from country to country, not necessarily consistently correlated with different levels of seismic hazard. In fact, the seismic assessment process is made of different steps that cannot be analyzed independently from each other: for instance, the definition of the seismic input depends on the methods and criteria that are being used for the analysis (linear, non linear, modal with response spectrum…) and safety checks (strength criteria, deformation criteria, allowable stresses or limit state approach…). Most codes are based on a strength approach, closer to the traditional way of thinking of the practicing engineer, despite seismic assessment being conceptually better described by a displacement/deformation approach. Also, masonry design practices are still characterized by a rather high level of empiricism and are strongly influenced by the local traditions. The adoption of Eurocodes is an important occasion to rationalize the design/assessment approaches but extreme care should be paid in codifying the procedures to avoid inconsistent, contradictory or unrealistic results. In particular, in such rationalization made according to quantitative, engineering criteria, following a common language as for other structural typologies, it

5

is essential to interpret correctly all the basic elements of the design and safety assessment procedures to adapt them properly to masonry structures, and it is mandatory to verify that the analytical procedures do not produce results that contradict experience and experimental evidence. 3.1 Methods of structural analysis and safety check For seismic design or assessment of buildings, modern codes (EC8, ASCE-FEMA 356) consider four main methods of structural analysis: linear static (or simplified modal), linear dynamic (typically multimodal with response spectrum), non linear static (“pushover”) and non linear dynamic. When dealing with masonry buildings, the experience of past earthquakes shows that the performance assessment or safety check, and the consequent structural model, should consider at least two different problems: the response/resistance versus the so-called “first damage mode” mechanisms, which involve usually out-of-plane damage and collapse mechanisms, and the “second mode” mechanisms which are associated to in-plane response of walls. Clearly, during an earthquake both out-of-plane and in-plane response are simultaneously mobilized, but it is generally recognized that a satisfactory seismic behaviour is attained only if out-of-plane collapse is prevented and in-plane strength and deformation capacity of walls can be fully exploited. In the Italian experience of the recent years there has been a tendency to classify the “first mode” mechanism as local mechanisms, in the sense that they are usually associated to the local response of structural elements/macro elements (Figure 1) which could in turn generate a global collapse but can be studied in first instance without recurring to a global structural model of the whole structure. A global model of the structure is usually needed when the resistance of the building to horizontal actions is provided by the combined effect of floor diaphragms and in-plane response of structural walls (although in some cases simplified partial models could be used). Since in the design of new structures the structural conception and details should guarantee that the in-plane strength of walls could be exploited without out-of-plane collapse, attention has been paid in the literature mostly to methods of global analysis.

Figure 1: Examples of first-mode “local” damage mechanisms (left, from D’Ayala & Speranza, 2003) and global response mechanism (right).

3.2 Available models for global analysis in practical applications, an overview Dealing first with the problem of global analysis of the structural system, attention will be given here to linear methods and nonlinear static methods, which are the ones that can be used in common practice. The last ten years have been characterized by a significant progress in nonlinear methods of analyses of masonry structures, to the extent that now a rather reliable nonlinear pushover analysis of buildings is a real possibility also for practice. Considering the problem of seismic design/assessment of masonry buildings, the need for non-linear analysis had been recognized in Italy and Slovenia as early as in the late 1970s, after the 1976 Friuli earthquake. In 1978 (DT2, 1978) and 1981 (Ministero dei Lavori Pubblici, 1981), recommendations on seismic assessment, repair and strengthening of masonry buildings were issued in Italy, suggesting the use of an equivalent static, simplified non-linear assessment method which had been proposed and developed in Slovenia by Tomaževic (1978). Such method, which has undergone several refinements in the subsequent years (Tomaževic, 1997 & 1999), is based on the so-called “storey-mechanism” approach, which basically consists of a separate non-linear interstorey shear-displacement analysis for each storey, where each masonry pier is characterized by an idealized non-linear shear-displacement curve (typically elastic-perfectly plastic with limited ductility). The conceptual

6

simplicity of the storey-mechanism method and its adoption by the Italian recommendations were fundamental in its diffusion among professionals, and the method has been extensively used in Italy since its first introduction in code provisions. However, the simplicity of the storey-mechanism approach, is paid with a series of limitations which may restrict its application only to some classes of buildings (Magenes and Della Fontana, 1998) not exceeding two, arguably three storeys. The need for more general methods of analysis has stimulated the research on the subject and analytical methods have made significant progress in the last decades, particularly in the field of finite element analyses (Calderini & Lagomarsino, 2006, Lourenço, 2002). Still, despite such progress, each model has a range of validity which needs to be understood with care, and the use of such tools requires high expertise, and in many cases can be applied to problems that are limited in size; therefore refined nonlinear finite element modeling does not constitute yet a suitable tool for the analysis of whole buildings in everyday engineering practice, especially when considering the task of designing/assessing ordinary two- or three-storey residential buildings. For this reason, several methods based on macro-element discretization have been developed, requiring a low to moderate computational burden. Considering the simplest possible formulation, it was felt by the author that several basic ideas of the “storey-mechanism” approach could be used and extended to a broader range of validity, maintaining concepts and idealizations that are familiar to the engineer and obtaining results that can be compared with those of more sophisticated analysis. Following this idea, a nonlinear method based on an equivalent frame idealization of multistorey walls was developed and implemented at the University of Pavia (Magenes and Della Fontana, 1998, Magenes, 2000, Magenes et al., 2000). The method was developed from the consideration that the distribution of internal forces at ultimate is basically governed by strength of members and by equilibrium. If a sufficient plastic deformation capacity in the piers is assumed, their initial elastic stiffness is therefore not as important as the definition of suitable and sufficiently accurate strength criteria, and simple bi-linear (elasto-plastic) or multilinear formulations (Figure 2) can yield effective results, also when compared with more refined nonlinear f.e.m. analyses or experimental results.

γϕ

i

j

θ = ϕ + γ i i

i

θ = total chord rotation; γ = shear deformation; ϕ= flexural rotation

Figure 2: Equivalent frame modeling of a masonry building and idealized multi-linear in-plane shear

force vs. shear deformation behaviour of a masonry pier element or a spandrel element failing in shear. When an equivalent static analysis is used within the context of seismic assessment, the strength and deformation properties of the members must be defined with reference to the experimental envelopes obtained from cyclic tests on wall elements, in particular regarding the ultimate deformation capacity (expressed in terms of chord rotation or drift capacity, θu). The ultimate deformation capacity of masonry is governed mainly by shear failure mechanisms, for which conservative estimates of θu =0.4-0.6 % are suggested by recent codes (EC8 part 3 and OPCM 3431). Such values are based mostly on available tests on solid brick or stone masonry walls, for which a larger experimental data base is available. For new typologies of masonry (e.g. with thin layer mortar and

7

perforated units) experimental data should be collected before relying on such deformation capacity levels. Also for reinforced and confined masonry suitable experimentally based values should be adopted (OPCM 3431 adopts 1.5 times the capacity of URM). Other works have explored and verified the suitability of nonlinear equivalent frame modeling for unreinforced masonry (Kappos et al. 2002, Roca . et al., 2005). An efficient equivalent-frame formulation, stemming from a work by Gambarotta and Lagomarsino (1996), can allow the dynamic global analysis of whole buildings (Lagomarsino et al. 2004), when in-plane only response of walls is considered. Software packages for nonlinear pushover analyses of masonry buildings have also become recently available to the public (Magenes et al., 2006, Lagomarsino et al., 2006) in Italy. It is not the aim of this paper to discuss the methodologies by which the seismic deformation/displacement demands are evaluated from the design elastic spectrum. It can be said that the criteria used for other structures (FEMA 440, 2005) could in principle be applied to the same extent to masonry structures, although some issues need to be more investigated, related to the strength- and stiffness-degrading behaviour of masonry and the consequent possible dependence of the response on strong motion duration (Bommer et al., 2004). The comparison between linear and nonlinear analysis results, which was made possible by the progress in modelling techniques, is the best way to appreciate the limits of elastic analysis when applied to ULS seismic assessment and the reason why masonry structures were the first for which at a code level a simplified nonlinear approach was felt necessary in real applications. A simple but clear example can be obtained by analyzing a simple multi-story wall with openings subjected to an in-plane pushover analysis. Figure 3 reports the comparison between the base shear distribution among piers which is obtained in a two-storey wall by a linear elastic finite element analysis and a nonlinear finite element analysis. In both cases the same mesh of 4-node, 2D, 125 mm x 125 mm plane stress elements was used (Magenes and Della Fontana 1998), and the nonlinear constitutive law for unreinforced brick masonry was the one proposed by Gambarotta and Lagomarsino (1997). The force distribution refers to a top horizontal displacement of 24 mm, close to ultimate. As it can be noticed, in the nonlinear model the shear in the left, “windward” pier is approximately half of the shear in the elastic model, whereas the shear in the right, “leeward” pier is approximately 1.6 times the corresponding elastic value. The differences in each pier are then in the order of 15% of the total base shear. The use of a less refined equivalent frame or macro-element analysis would yield differences of the same order. The exam of the nonlinear analysis shows how at ultimate, the moderate “ductility” of the wall piers (it would be more correct to speak of nonlinear behaviour) tends to a situation where forces are shared according to strength capacity, not according to elastic stiffness. In fact, the non symmetric distribution of shears of Figure 3 arises from the different capacity of the lateral piers, each subjected to a different axial force due to the overturning effect of horizontal forces.

05

101520253035404550

Pier 01 Pier 02 Pier 03

V/V t

ot (%

)

LINEAR NONLINEAR

Figure 3: Different distribution of internal forces (shear in bottom piers) in an URM wall at ultimate according to linear and nonlinear FEM pushover analyses; Vtot = total base shear.

Similar considerations can also explain why, especially for the analysis of historical masonry structures, the approach of limit analysis has been proposed and used for quite a few decades now for the estimate of the ultimate static capacity of masonry systems, initially for vertical loading only, subsequently also for horizontal loads simulating seismic forces (Como & Grimaldi, 1985, Giuffré, 1993). As discussed further on, limit analysis is particularly useful for the evaluation of first-mode mechanisms. Despite the recent progresses that have been achieved in nonlinear analysis, however, linear analysis tools and software are still the most used and familiar among practicing engineers, and particular care should be used

8

when formulating code provisions relevant to these methods. In the following paragraphs, this issue will be discussed, referring mainly to the approach of Eurocode 8. 3.3 Linear methods for global analysis and safety check In design of new structures the “reference” method for Eurocode 8 is the linear elastic multimodal analysis, of which the “linear static” methods represents the single-mode simplification allowed for structures which are regular in elevation. The design seismic loading is defined according to the method of analysis. For ultimate limit state design the loading has the form of a design response spectrum which is obtained from an elastic acceleration response spectrum corresponding to a 475 years return period, scaled by a “seismic force reduction factor” that accounts in an approximate way for inelastic response at ultimate. In EC8, such factor is defined as behaviour factor or q-factor. With such an approach, the safety check procedure is summarized in Figure 4, where it is shown that for the two-level performance requirements (no-collapse and damage control), at ultimate (ULS) the safety check consists of a strength verification, whereas for damage control (DLS) the check is made on deformation (drift) demands. According to the performance targets, each limit state is associated to a specific level of seismic action, that corresponds to a given probability of exceedance or a given return period. The seismic action, in the form of spectral ordinates, is applied to a linear elastic model of the structure and the resulting internal force and displacements are calculated. In general, for masonry structures the ULS verification governs rather than the DLS. The ULS verification in carried out by checking that in each structural element the design resistance is not exceeded according to the strength criteria defined in Eurocode 6 (CEN-EN 1996-1-1). In other words, the ULS safety requirement is not met if the shear strength or the flexural strength of even just one element is exceeded.

Linear elasticmodel of the

structure

Seismicaction

(spectralordinates)

Internal forces, stresses,

displacements/ deformations

Strengthbased safety

check

Drift-basedcheck

ULS

DLS

Linear elasticmodel of the

structure

Seismicaction

(spectralordinates)

Internal forces, stresses,

displacements/ deformations

Strengthbased safety

check

Drift-basedcheck

ULS

DLS

Figure 4: Simplified flow-chart of typical safety/performance check via linear methods of analysis (EC8).

3.3.1 The behaviour factor issue and the role of overstrength As known, the use of the q-factor tries to take into account in an approximate way the effects of numerous factors that concur in determining the inelastic cyclic deformation and energy dissipation capacity of a structural system. The behaviour factor q is an approximation of the ratio of the seismic forces that the structure would experience if its response was completely elastic to the minimum seismic forces that may be used in the design, with a conventional elastic analysis model, still ensuring a satisfactory response of the structure (i.e. the deformation capacity is not exceeded). In the European context, important experimental references for the evaluation of behaviour factors for URM masonry buildings can be found in the researches carried out in Slovenia by Tomaževič et al. (Tomaževič & Weiss, 1994, Tomaževič, 1999, Tomaževič et al., 2004) and in Italy by Benedetti et al. (Benedetti & Castoldi, 1982, Benedetti et al., 1984, Benedetti et al., 1998). With reference to Figure 5, the response of the structure is represented by a capacity curve F-d (e.g. base shear vs. control displacement), which can be thought of as obtained from an experimental envelope curve. The typical criterion to evaluate the q-factor (see e.g. Tomaževič et al., 2004), would be:

y

el

FF

q max,= (1)

where Fy corresponds to the strength of an ideal bi-linear system equivalent to the true nonlinear. With common equivalence criteria, e.g. energy equivalence, such strength is usually slightly lower than the maximum resistance Fmax but can be considered an estimate of the ultimate base shear capacity of the structure. If this definition is followed, the experimental researches referenced above, based on the interpretation of shake table tests of URM building models, report a range of variation of q between 1.5 and 2.5.

9

du

Nonlinear capacity curve

Fmax

dy dFel,max

Fy=Fu

Fel

F

Ideal elastic

Fel,max

d

Equivalent bilinear

(elasto-plastic)

Figure 5: Reference parameters for the definition of the behaviour factor q (F base shear, d displacement

at a control point, e.g. roof).

This justifies the conservative choice of q=1.5 that has been made by most codes since the 1980s until recently (the Italian draft GNDT-CNR 1985, the ENV 1998-1-3 version of EC8, 1998). In the most recent drafts of EC8 part 1 the proposed q value for URM was 1.5 in May 2000, then a range between 1.5 and 2 in the draft of May 2002, and finally a range from 1.5 to 2.5 was introduced for the final 2003 document, keeping however as recommended value the lower limit of the interval, i.e. q=1.5. In May 2003 a new national seismic code was issued in Italy (OPCM 3274). The new code had been conceived as a document of transition from the previous national seismic code, dated 1996, towards the final adoption of Eurocode 8, and to this end many elements of it had been included, among which the limit state formulation and the recommended q-values for masonry buildings. The requested strength checks were to be carried out separately for in-plane and out-of-plane action. In-plane strength criteria were considering two failure mechanisms: shear and flexural. The in-plane design shear strength of an URM wall was defined as per Eurocode 6. (CEN-EN 1996). As regards flexural strength of a wall the design ultimate moment of a section was defined assuming no-tension behaviour and an equivalent rectangular stress distribution in the compressed portion of the cross section in ultimate conditions, with a limit stress of 0.85 fd . The same criterion is also proposed in EC8 part 3, annex C for the assessment of existing buildings. In comparison with EC6, no reduction in strength due to transversal eccentricity of the axial force is considered, producing therefore less conservative results than the strict application of EC6 would give. The seismic action was defined via elastic response spectra similar to EC8 Type 1, with four different levels of PGA on rock (ag) associated respectively to four seismic zones (ag/g = 0.05, 0.15, 0.25, 0.35 respectively). The PGA on other soil type would be obtained as agS, where S is a soil-related parameter (S=1.0÷1.35) The first application of such design code to common real practical cases, following the standard linear analysis procedure (be it static or multimodal), without resorting to force redistribution, showed the following:

- with a q of 1.5 or even 2, it was practically impossible to satisfy strength safety checks for any configuration of URM, 2- or 3-storey building for agS greater than 0.1g ; in numerous cases the strength safety checks would not be satisfied even for agS = 0.05g;

- the results of the analytical safety checks via elastic analysis clashed with the past experience and the experimental evidence;

- the results were contradictory with the provisions for “simple buildings”, which would allow for instance a two-storey URM building to be constructed in an area with agS = 0.15÷0.2g;

- the results obtained via elastic analysis were in great contradiction with the results of nonlinear static procedures, which would produce results more in line with experience..

No appreciable improvement of the situation was observed resorting to force redistribution made after the linear analysis, within the specified code limits (which were identical to Eurocode 8, as discussed further). It was mandatory to understand the reasons for such a contradictory situation, reconsidering carefully all the different steps of the procedure depicted in Figure 4. The first consideration to be made is that according to the linear elastic procedures, the “ultimate” state of the structure corresponds to the attainment of the strength capacity in at least one structural element. Such state does not correspond to the ultimate capacity of the system. With reference to Figure 5, the force Fel represents the base

10

shear at which the first element would reach its strength capacity (shear or flexural) according to a linear elastic analysis. Even for unreinforced masonry elements, a limited deformation capacity into the nonlinear regime is available, sufficient to allow the system to sustain an increasing seismic load, beyond this “elastic” limit, by increasing the forces on other structural elements. The ultimate strength capacity (Fmax , or Fy in case a bilinear idealization of the response is used) is reached for values of base shear that are higher, often much higher than Fel. It appears evident how, for masonry structures, as done for other structural typologies, the definition of the behaviour factor q must take into consideration an overstrength ratio (OSR). A correct definition of q would be then:

OSRqFF

qFF

FF

FF

qel

y

el

y

y

el

el

el ⋅=⋅=⋅== 00

max,max, (2)

where the basic value q0 , which for URM buildings would assume the accepted, experimentally based values of 1.5-2.5, is multiplied by the overstrength ratio OSR = Fy/Fel (note: in Eurocode 8 and OPCM 3274 such a ratio is defined as αu/ α0 with symbols of similar conceptual meaning). The evaluation of the OSR for masonry buildings can be made resorting to experiments and to numerical modeling, via nonlinear static analysis. It is expected that the OSR should depend, with a varying degree of sensitivity, on many factors, the most relevant can be envisaged as:

- structural configuration and redundancy; - assumed spatial distribution (pattern) of seismic forces in elevation and in plan; - in-plane rigidity/flexibility of diaphragms; - modeling assumptions regarding coupling among shear walls (flexural coupling provided by r.c. ring

beams or by masonry spandrels…); - modeling assumptions regarding collaboration between intersecting walls, and structural role of walls

orthogonal to the seismic force; - modeling assumptions regarding strength criteria (e.g. shear resistance)

Since several of these factors are related to modeling hypotheses, a consistent evaluation of the OSR cannot rely on interpretation of experimental results alone. 3.3.2 Numerical evaluation of overstrength The availability of sufficiently reliable models for the nonlinear pushover analysis of whole structural systems allows to evaluate the overstrength and its dependence on model-related parameters. Recent research carried out at the University of Pavia, Italy (Magenes, 2004, Magenes & Morandi, 2006), in support and assessment of the new national seismic code OPCM 3274 focussed on typical Italian structural configurations of new unreinforced and reinforced masonry buildings. Buildings from one to three storey for unreinforced, and from two to four storey for reinforced masonry were considered, characterized by a rather simple plan configuration and regularity in elevation, with a total area of shear walls ranging from 3.5% to 7.5% of the total floor area in each principal direction. Figure 6 shows three examples of the nine plan configurations that were considered. The common solution of mixed r.c. floor slabs with hollow clay elements was assumed, in which a reinforced concrete topping of 40÷50 mm allows the assumption of a rigid diaphragm behaviour. Both the previous Italian seismic code and the new OPMC 3274 require that in new masonry buildings all walls at each floor level be connected by means of continuous r.c. ring beams having a depth equal or greater than the floor slab thickness, with a minimum longitudinal steel reinforcement of 800 mm2 and transverse reinforcement of 6 mm closed stirrups at maximum 250 mm spacing. Each building was analyzed with static linear and nonlinear methods. Since the geometry and the layout of openings allowed an equivalent frame modeling of the structure, the nonlinear analyses were carried out with the SAM II method (Magenes et al., 2006). The choice of an equivalent frame idealisation, whether with or without flexural coupling provided by r.c. ring beams, is very suitable since it is a common option in real practice also for linear analyses, and allows to use of the same geometric model for both linear and nonlinear analysis. For each building, a nonlinear capacity curve has been calculated, from which the equivalent bilinear curve and the corresponding ultimate base shear Fy could be obtained. The capacity curve and the bilinear idealization are characterized by the same secant stiffness at 0.7Fmax and by the same areas under the curves. All elements were supposed to behave elastically until any of the code failure criteria was met, assuming typical mean values for the material strengths. The value of Fel was therefore obtained from the same pushover analysis, corresponding to the point at which the first structural element met a failure criterion (shear or flexure) and entered the post-

11

elastic branch. Figure 7 reports the histogram of the values of OSR (Fy/Fel) that were obtained for the sample of two- and three-storey unreinforced masonry buildings with the following hypotheses:

- the strength of the materials is the same for all buildings; - the diaphragms are assumed to be rigid in-plane; - the flexural coupling between masonry walls is provided only by the r.c. ring beams, with the minimum

required steel reinforcement; - the seismic forces were applied separately along the two main orthogonal directions of the building, i.e.

every building produced two different overstrength ratios; - seismic forces were lumped at floor levels and were applied with a transversal accidental eccentricity of

±0.05L where L is the floor-dimension perpendicular to the direction of the force

- the control (roof) displacement was increased keeping the “modal” pattern used for static elastic analysis, assuming a linear modal shape.

Twenty building configurations make up the sample of Figure 7, giving two OSR each (one for each of the two orthogonal directions). As it can be noticed, the range of variation of the OSR is extremely wide, with a minimum of 1.21, a maximum of 3.91, a mean of 2.4 and a standard deviation of 0.62. All buildings have at least one direction with OSR ≥ 2.0.

1670

1369

Plan n.1 - Unreinforced or reinforced masonry - 2,3,4 storeys

Wall thickness: 25/30 cm

r.c. beam

r.c. column

1320

1125

Plan n.5 - Unreinforced masonry - 1 or 2 storeys

Wall thickness: 30/35 cm

700

375

660560100

1230

Plan n.8A - Unreinforced or reinforced masonry - 2 or 3 storeys

Wall thickness: 30 cm

350 1550 3502250

Figure 6: Examples of plan layout of the buildings analyzed at the University of Pavia.

Overstrength ratio (OSR) - Two- and three-storey URM buildings

0

0.5

1

1.5

2

2.5

3

3.5

4

1 2 3 4 5 6 7 8 9 10 11 1213 14 1516 17 1819 20 21 2223 24 2526 27 2829 30 31 3233 34 3536 37 3839 40

F y/F

el

Figure 7: Calculated OSR values for 2- and 3-storey unreinforced masonry buildings

Clearly, the main factor that produces the variation of the OSR in Figure 7 is the geometric configuration (plan and number of storeys). It is however expected that OSR could depend to a variable extent also on other factors such as those listed in section 3.3.1. By introducing simple changes to the models, i.e. transforming the r.c. ring beams in truss elements, it was possible to eliminate the flexural coupling, to reproduce the extreme condition which is usually assumed for easier calculations and that has been used for many decades to analyse buildings subjected to wind loading. Such a model could be consistent with a situation in which r.c. ring beams are

12

substituted by e.g. steel ties with no flexural resistance, and no structurally effective masonry spandrels are present. With such modelling hypothesis, lower but still significant values of OSR were found. At present, for a reduced sample of cases the mean value is 1.7, with a minimum of 1.2, a maximum of 2.5, and a standard deviation of 0.3. The analyses on the complete set are currently under completion, but it can be seen, as expected, that by reducing the level of structural redundancy of the system the OSR tends to decrease. It is certainly striking that in Figure 7 the great majority of the calculated values is higher than 2, sometimes greater than 3. A high OSR simply tells that the elastic analysis would predict “failure” of a structural element for a level of base seismic shear that is largely lower than the ultimate strength that the structural system can provide. For URM buildings, the OSR can reach such high values since in great part of the cases, consequent to the force redistribution that takes place as the strength capacity of different elements is progressively achieved, the internal force distribution at ultimate will differ substantially from the “stiffness-proportional” distribution that is obtained with a linear elastic analysis. Similarly, OSR ratios were also evaluated for reinforced masonry buildings, in which the minimum code reinforcement (vertical and horizontal) had been introduced. Such minimum requirement consisted of a geometric horizontal reinforcement ratio of 0.04%, a vertical reinforcement ratio of 0.05% with not less than 200 mm2 placed at each wall end, in the proximity of each opening and at each intersection between walls, and anyways with a maximum spacing of 4 m. The results produced OSR ranging from 1.2 to 2.5 with a mean value of 1.83 and a standard deviation of 0.35. All buildings had at least one direction with an OSR ≥1.59. The presence of a minimum reinforcement decreases in average the OSR and reduces the scatter with respect to the unreinforced cases. Clearly, in the case of reinforced masonry the amount of reinforcement which is provided in each wall can affect significantly the strength distribution and therefore the overstrength ratio, whereas in URM buildings strength allocation depends almost entirely on geometry and vertical load distribution. The case of minimum reinforcement ratio can be considered appropriate for the typologies used in Italy, with limited, small diameter horizontal reinforcement laid in bedjoints, and generally low vertical reinforcement ratios (the limited shear reinforcement and consequently limited shear strength does not justify high flexural reinforcement). If the typical reinforcement percentages of r.c. wall structures were approached, it is expected that the OSR would approach the value typical for such structures. All nonlinear analyses have confirmed the tendency towards an internal distribution of forces according to strength and not according to initial elastic stiffness. What observed here regarding the limits of linear analysis is however common to most structures with a moderate inelastic or nonlinear deformation capacity and should warn not to put too much emphasis on linear multimodal analysis as a “reference method” for masonry structures, in analogy with what discussed by Priestley (2003) for reinforced concrete structures. Considering the great inadequacy of linear elastic models to describe the ultimate behaviour of masonry building structures under seismic loading, for low-rise buildings (up to three-four storeys) it is also questionable that a linear elastic multimodal analysis would better predict the global capacity of a masonry building than a conventional nonlinear pushover analysis, even for irregular configurations. 3.3.3 Experimental evaluation of overstrength A recent re-examination of shaking table tests, presented by Benedetti (2004) basing on tests carried out in previous campaigns (Benedetti & Castoldi, 1982, Benedetti et al., 1984, Benedetti et al., 1998), leads to conclusions that are substantially in line with what obtained numerically. The tests were performed on up to three-story, 1:2 to 1:1 scale models of unreinforced stone and brick masonry buildings or walls, representative of common Italian existing typologies. The criterion to estimate the condition corresponding to Fel of Figure 5 was to check the first significant crisis/cracking in masonry walls, which would typically show for table motion intensities lower than those corresponding to an ultimate/near collapse condition of the building. Such a criterion is the only possible one for shaking table tests in which no direct measurement of forces/stresses is possible in structural elements. Also for experimental results a scatter was found in the estimated parameters, but the final conclusion was that, “the level of reduction of elastic forces for URM, well organized and able to maintain their box-type global behaviour, is at least of 2.5÷3.0”, obtained as the product of a basic q0 of 1.5÷2.0 multiplied by an OSR ≈ 1.5. It would be desirable to extend such re-examination also to other experimental researches carried out so far, including also static full-scale testing of buildings such as Magenes & al., 1995, Yi et al., 2006. Considering the different criteria by which the OSR were estimated numerically or experimentally, it is remarkable and somewhat reassuring that, though on a limited number of configurations and typologies, and noting the numerous factors that can influence the overstrength, different researchers have come, through different independent approaches, to similar conclusions regarding overstrength values and regarding a consistent definition of the q-factor for URM, which can exceed 2.5-3.0.

13

Clearly, further investigations are needed to clarify how the overstrength can be related to structural solutions and details that can vary from country to country. For example, the previous Italian seismic code as well as the new OPCM 3274 enforce the use of continuous reinforced concrete ring beams as described in the previous section, where EC8 is very generic in this regard, specifying only that “the connection between the floors and walls shall be provided by steel ties or reinforced concrete ring beams”. Simple steel ties as opposed to r.c. beams offer lower flexural coupling, affecting the structural behaviour and thus the overstrength. It would seem appropriate that both the basic q0 and the overstrength ratio be defined as a Nationally Defined Parameter (NDP), separately for specific structural typologies, including solutions with improved ductility/robustness which could be used in higher seismicity areas. On the basis of the above considerations, in the subsequent update of OPCM 3431(2005) the q-factors for new unreinforced and reinforced masonry buildings were corrected as follow:

- Unreinforced masonry building; regular in elevation q = 2.0 αu/α1 - Unreinforced masonry building; non regular in elevation q = 1.5 αu/α1 - Reinforced masonry building; regular in elevation q = 2.5 αu/α1 - Reinforced masonry building; non regular in elevation q = 2.0 αu/α1

where αu/α1 is the OSR, for which the following values were suggested:

- Single storey unreinforced masonry buildings αu /α1 = 1,4 - Two or more storey unreinforced masonry buildings αu /α1 = 1,8 - Single storey reinforced masonry buildings αu /α1 = 1,3 - Two or more storey reinforced masonry buildings αu /α1 = 1,5

3.3.4 Drawbacks of the linear methods based on the q-factor, role of force redistribution The recognition of the necessity of the OSR in masonry design/assessment, which has been introduced in the revised version of OPCM 3274 (OPCM 3431, 2005) was undoubtedly an important step to rationally explain and find a solution for the inconsistencies found in the application of the code. Nevertheless, the choice of a specific value of OSR, even for the same homogeneous typology of masonry buildings, does not overcome completely the intrinsic problems of the linear methods of analysis. Considering the homogeneous class of two- and three-storey buildings whose OSR are given in Figure 7, the choice of a single conservative value, be it the minimum or a “sufficiently conservative” percentile (e.g. 1.8 as proposed in OPCM 3431), has the consequence that in the wide majority of the cases, in which the OSR is much higher (e.g. 2.5 or 3), the design seismic action will be much higher than it should. For such configurations, the use of a default conservative OSR could be so penalizing that the strength safety checks can never be satisfied, even if the quality of materials, the structural configuration and details, the total amount of shear walls clearly show that the design should be safe. It is very useful in such a situation, to resort to a redistribution of internal forces after the linear elastic analysis is carried out. Redistribution consists of reducing the internal forces (shears) in the walls that would not satisfy the strength verification, increasing consequently the forces in other walls in such a way to satisfy global equilibrium requirements (i.e. rotational and translational equilibrium of each storey). This possibility was considered by design codes, including Eurocode 8, since the very first drafts. However, the limits to the redistribution have been so far so strict to make redistribution almost ineffective in many practical problems. The current limit given by Eurocode 8, which was adopted by OPCM 3274, states that the shear in any wall is neither reduced more than 25% nor increased by more than 33%. The origin of these limits dates back to 1985 or earlier (CIB, 1985), at times in which the experience in nonlinear analysis was quite limited, and seem to recall criteria originally developed for reinforced concrete structures. As shown in previous discussions, the problem of elastic analysis is that it does not provide a correct distribution of internal forces with reference to ultimate limit state, and that differences with more accurate nonlinear analyses tend to be much higher than the limits imposed for redistribution. A typical, recurrent case in seismic analysis of real building configurations is to find that a wall carrying a limited, sometimes negligible percentage of the total interstorey shear does not satisfy the strength check because the vertical load on it is low, and strength in URM walls is mainly determined by geometry and vertical load. Quite often a reduction of 25% of the shear in the wall does not solve the problem at all. In reality, such wall would simply provide a lower shear resistance and other walls would carry the shear in exceedance needed for equilibrium. A more rational approach would possibly be to allow a redistribution which could allocate shears following the available strength reserve of the walls, approaching what would be the result of a nonlinear analysis. These considerations have led, in the revised OPCM 3431 of 2005, to an increase of the limits, stating that the variation in shear in each wall should not exceed the largest value between 25% of the shear in the wall and 10% of the total interstorey shear. In the case of non rigid diaphragms the redistribution is allowed only among

14

coplanar walls connected by ties or r.c. beams (in such case the interstorey shear is evaluated considering only the contribution of the coplanar walls). 3.4 Out-of-plane (local) response and safety assessment The issue of out-of-plane stability of walls subjected to seismic excitation is strangely not well put in evidence in Eurocode 8, to the point that the seismic loading is not clearly defined and the engineer would have to find his own way to a safety check, resorting for instance to the seismic loading defined for non structural elements (as suggested in Tomaževič, 1999). It can be said that strict slenderness limitations, minimum thickness requirements and appropriate structural conception and detailing (rigid diaphragms and efficient floor-to-wall connection) can guarantee in most cases the prevention of out-of-plane driven failures; however, on one hand such slenderness and thickness limitations are Nationally Determined Parameters that could vary significantly from country to country, on the other hand out-of-plane stability is an issue also for “secondary” seismic elements and non-structural partitions, which may not comply with such limitations. Recent experimental and theoretical research (Doherty et al., 2002, Griffith et al., 2003) have confirmed that out-of-plane response and stability of walls under seismic excitation, when ultimate conditions are considered, is more a matter of displacement demand vs. displacement capacity rather than a strength issue, as first pointed out by Priestley (1985). The problem is quite complex (Figure 8), requiring the evaluation of: - seismic demand on walls considering the dynamic filtering effect of building and diaphragms, and the dynamic response of wall - strength of wall against out-of-plane forces and relevant mechanisms of resistance - out-of-plane displacement capacity of walls

Earthquakeexcitation at footings

In-plane shear wallsresponse filters the ground motion and transmits to floordiaphragms

Floor diaphragm responseamplifies accelerations and transmits excitation to out-of-plane walls

Out-of-plane wall

Shakingmotion

Parapet wall

Earthquakeexcitation at footings

In-plane shear wallsresponse filters the ground motion and transmits to floordiaphragms

Floor diaphragm responseamplifies accelerations and transmits excitation to out-of-plane walls

Out-of-plane wall

Shakingmotion

Parapet wall

Figure 8: Simplified conceptual representation of out-of-plane seismic response (Doherty, 2000, adapted

from Priestley, 1985)

h

h/2

Δ

Δ/2

P

W/2

W/2

P+W

H=ΔW/2h

H=ΔW/2h

Mid-height Displacement

App

lied

Lat

eral

For

ce

Rigid bodyCalculated curve

Fu

ΔuΔ2

F0

Figure 9: Post-cracked out-of-plane force-displacement envelope of an URM wall in vertical bending.

f=F/h equivalent pressure

Rigid body Real semi-rigid behaviour

15

Most of the past research dedicated to wind loading has focused mainly on strength capacity, which can come from three possible sources: vertical compression, apparent flexural strength in one- or two-way bending, thrust (or arching) action. Considering the simpler one-way, vertical bending condition, in an URM wall the apparent flexural strength is due to vertical tensile strength of bedjoints or bricks, whichever the lesser. The attainment of cracking, which incidentally could develop already under service loading, due for instance to eccentricity of vertical loads, does not imply necessarily collapse, and could be seen as a damage limit state. In post-cracking regime, lateral resistance is provided by the presence of vertical compression, and could be sensitive to geometric second order effects. The behaviour of the subsystem is close to elastic nonlinear, with moderate energy dissipation. An appropriate safety assessment should be based on the evaluation of the main characteristics of the lateral response, depicted in Figure 9, namely initial stiffness, maximum force, displacement at static instability. Proposals are available for one way bending (Griffith et al. 2003), research is still under development for cyclic two-way bending (Vaculik et al., 2006), in which friction and cohesion of bedjoints and tensile strength of bricks play an important role in the maximum force capacity and also on the post-cracking hysteretic behaviour. The code approach of OPCM 3431 explicitly requires the designer to evaluate the seismic demand in the form of an equivalent static out-of-plane force proportional to the mass of the wall (and of any fixture rigidly connected) according to the expression proposed in EC8 for non-structural elements. This includes an approximate estimate of the filtering effect and possible resonance between the fundamental period of the building and the fundamental out-of-plane natural period of the wall. The seismic force is in turn reduced by a behaviour factor of the wall element, which for primary elements is assumed as qa = 3, whereas for non-structural walls a qa = 2 would be used. Such force-based approach is clearly very rough, and it is deemed not to produce consistent results, especially since it is based on the initial elastic properties (periods) and it assumes a constant q factor, independent of the displacement capacity, which is a size-dependent quantity (i.e. it increases as the thickness of the wall increases, even if the slenderness of the wall remains the same). Since the safety check is carried out in terms of strength, namely bending moment, second order effects affecting the component of the resisting moment which depends from axial load (Mu,N) could be taken into consideration by reducing the first order resisting moment Mu,N multiplying it by a coefficient φM ≤1.0, which accounts for slenderness and axial load ratio. A conservative estimate of such coefficient has been evaluated by Morandi and Magenes (2006) for the case of simply supported walls in one-way bending. In general, it is felt that considerable further research is still needed before an appropriate understanding of the problem and reliable assessment procedures will be achieved. 3.5 New typologies of masonry, implications on seismic design As discussed in the introduction, issues that affect the choice of a construction system or material go beyond structural requirements. The building industry has seen and is still seeing a continuous evolution of the masonry products (units, mortars…) and systems. In general, the experimental information on seismic behaviour of the more traditional masonry systems, such as fully mortared clay brick masonry, is far from being exhaustive. The seismic behaviour of more recent typologies such as those using thin layer mortar or non mortared vertical joints with tongue and groove, which allow faster construction, still needs to be thoroughly investigated, especially regarding the potential brittleness of failure under in plane cyclic shear, which may be not strictly correlated with the good performance in terms of strength under vertical compression. For this reason several experimental research project are presently being carried out throughout Europe, concentrating on the development of masonry construction systems which can balance different technological requirements still producing a satisfactory seismic response, both for unreinforced and reinforced masonry (da Porto et al., 2005, ESECMaSE, 2005-2007, DISWALL, 2005-2007). The diversity of such systems can also lead to different values of the meaningful parameters for seismic design with respect to more traditional systems: besides shear strength, shear deformation capacity under cyclic/dynamic loading and q-factor values for elastic design.

4. SEISMIC ASSESSMENT OF EXISTING MASONRY BUILDINGS

The problem of the seismic assessment and retrofit of existing buildings, amongst which a great number are unreinforced masonry, has become by now one of the main topics of interest in the world of construction, also due to progressive relative reduction of new construction activity with respect to interventions on existing structures. The topic itself is extremely complex, due to the enormous variability of structural forms and materials that can be found in countries with a long history of civilization such as in Europe. Such complexity constitutes a great hindrance to a strict codification of methodologies and approaches such as may be possible

16

with new designs. Considering specifically masonry buildings, such a diversity of structural forms and materials is enormous from country to country, but first of all, such structural forms very often do not lend themselves to be approached with the same engineering criteria used for reinforced concrete or steel construction. The attempt of transposing Eurocode 8 part 3 (CEN-EN 1998-3) to the Italian reality presented a series of novelties that in part were a serious progress towards a safe and rational approach to assessment, and in part were not compatible with the reality of the problem due to the impossibility to extend concepts and procedures which would be appropriate for other types of structures such as r.c. or steel framed buildings to masonry. 4.1 Knowledge levels An important step forward was the introduction of the essential problem of the knowledge of the structure, and the conceptual definition of different knowledge levels and consequent confidence factors for assessment. However, in the rational definition of knowledge levels of masonry buildings it should be noted that in almost all real cases:

- no construction drawings or structural design calculations are available, nor are test reports - the building was built in absence of design regulations, so no “simulation of design” is thinkable - often the direct experimental measurement of material parameters is not feasible or, if in principle

feasible, completely unreliable Regarding the latter comment, it is known how an in-situ measure of shear strength of stone masonry is presently possible only with destructive testing of panels of significant dimensions, rarely smaller than 1.0 x 1.0 m, through a self-equilibrated diagonal compression testing or more complex shear-compression test procedures (Sheppard, 1985, Chiostrini et al., 2003). The conditions for the feasibility of such tests may not be always satisfied, depending on the quality and texture of masonry, on the thickness of the walls, on the number of storeys of the building, on the availability of adequate experimental equipment. The use of easy-to-perform, moderately destructive tests such as the “shove test” (ASTM C 1531-03) are conceivable only for regular brick masonry, which is only one of the many typologies that can be found in Italy (such as in other countries). In OPCM 3431 it was therefore felt essential to define specific criteria for masonry regarding the different knowledge levels. First, full geometric survey is always required, and information regarding structural details should specifically address: quality of connections between vertical walls, quality of connections between floor/roof and walls, and presence of ring beams or other tying devices, presence of structurally efficient architraves/lintels above openings, presence of elements which can equilibrate horizontal thrusts, presence of structural or non structural elements of high vulnerability, typology of masonry (stone or brick, regular or irregular units, single-leaf or multi-leaf , with or without transversal ties…). The category limited in situ testing which defines the minimum knowledge level regarding material properties, for which EC8 part 3 recommends at least 1 test per floor, has been substituted with limited in situ survey where the mechanical properties of the material are estimated after visual inspections, removing plaster in selected areas to asses the texture and the connection between orthogonal wall, visual inspections through the thickness to recognize the internal level of connection of the leaves and the ability of the wall to behave monolithically through the thickness, qualitative assay of the mortar consistency. Such recognition of the typology and quality of the material is then used to associate it with values of the mechanical parameters reported in a table, which was developed on the basis of the experimental data available on the most common typologies. In addition to the prevision of a knowledge level that does not necessarily foresee direct experimental testing, another important element introduced in OPCM 3431 is the possibility of using experimental information obtained in other buildings, once it is possible to establish a clear and proved correspondence of materials, texture, construction technique and details. Such criterion is deemed to obtain acceptable results when applied in limited areas characterized by the same building traditions. . 4.2 Global modelling and assessment Regarding criteria for global modelling and assessment, the approaches which are presently deemed to be more applicable to existing masonry buildings in normal practice are the linear methods with the use of a q-factor and the static nonlinear methods. The linear methods with the direct use of the unscaled elastic design spectrum and the distinction of ductile and brittle mechanisms, as proposed by EC8 part 3, does not lend itself to be applied to existing masonry structures, in which a moderate ductility has to be recognized also for shear failures, and a sharp distinction between shear failure and flexural failures is problematic and somewhat arguable. Regarding linear methods, the considerations made in section 3 regarding overstrength and the limits of elastic analysis are applicable also to existing buildings. However, the problems of selecting appropriate values for the

17

OSR in existing masonry buildings are magnified by the wide variety of structural configurations that can be found. As a first remark, it is essential to realize that a global analysis of the building is meaningful when presence of ties and diaphragms and appropriate connections can guarantee a global box-type action. However, diaphragms in old building are often flexible, providing a lesser degree of coupling of the walls that tend to vibrate more independently. Retrofitting interventions may lead to a stiffening of the diaphragm, but not to the extent in which it can be considered as rigid in a global analysis. In such situations, all methods of analysis currently available to designers tend to give a rather approximate picture of the response. If, on one hand, an elastic 3D model can be used to obtain a better understanding of the initial periods and modes of vibration, still the elastic strength assessment at ULS would present all the inconvenients of elastic analysis. In principle a global elastic model could be used to single out subsystems that could be analyzed in a second stage with static nonlinear models. When flexible diaphragms are present, the in-plane response of walls could be assessed by analyzing separately the capacity of co-planar systems, such for instance as a façade wall. When dealing with existing buildings, an exceedingly conservative underestimation of global seismic resistance is to be avoided possibly more than in new design. Whereas strengthening interventions aiming to increase the robustness and redundancy of the system by improving and inserting connections to avoid first mode damage mechanism are in general always positive, the increase in global strength is more difficult to achieve and it is pursued sometimes with heavy, non conservative (from the architectural point of view) interventions whose effects may be very difficult to predict, and whose effectiveness can be sometimes arguable, as past experiences have shown. Once the first-mode mechanisms are prevented, the redundancy of masonry buildings is such that an OSR of 1.5 should be considered adequate in elastic analysis (as suggested by OPCM 3431). Although further experimental and theoretical research is needed to fully support a generalized use of such a value, from a pragmatic point of view the use of such a value generates results which are more consistent with experience and with the application of nonlinear analysis procedures. The possibility of applying linear elastic methods of analysis with reasonable levels of seismic loads is important especially when complicated geometries and irregular positioning of openings do not allow a reliable macro-element discretization of the building (refined nonlinear f.e.m. analyses are not an option for most practicing engineers). As regards the shear strength of unreinforced masonry, the possibility of using the criterion proposed by Turnšek and Sheppard (1980) was introduced as an alternative to the Eurocode 6 formulation, since in Italy all the available experimental data on existing masonry, especially stone masonry, have been collected and processed with reference to such criterion. 4.3 Local mechanisms As said above, damage to most vulnerable buildings even in moderate earthquakes is mainly due the “first damage mode” mechanisms, which involve usually out-of-plane damage and collapse mechanisms. The recognition of potential first mode mechanism is therefore an essential part of the assessment procedure. A global seismic response mechanism can take place only if first mode damage is prevented. In this regard, EC8 part 3 neglects to address the problem. The choice in OPCM 3431 was to introduce a specific paragraph stating that the assessment of local mechanisms is compulsory. If the choice of criteria by which such assessment is explicitly carried out are in principle left to the designer, an informative annex is provided (annex 11.C) in which a procedure is proposed, based on the application of limit analysis. In the first damage mode, most of the collapse mechanisms are partial, in the sense that they involve specific sub-structures or components. Collapse is due, most of the times, to loss of equilibrium rather than to the exceedance of some level of stress. Recurrent mechanisms have been observed in past earthquake, and classifications and possible analytical descriptions have been proposed and are continuously being developed (e.g. D’Ayala & Speranza, 2003, Lagomarsino et al., 2004, Restrepo-Vélez & Magenes, 2005), When horizontal acceleration are high enough to trigger a mechanism, it can be assumed that the different parts can be idealized as rigid bodies, and the lateral force capacity of the subsystem can be related to a corresponding acceleration. Static threshold resistance can be evaluated through limit analysis and the application of the principle of virtual work (PVW). Earlier uses of rigid-body limit analysis in assessment (e.g. Giuffré, 1993) were made essentially on a comparative basis, to evaluate which parts of the structure are most vulnerable, and to check the effect of strengthening techniques (e.g. insertion of tie-rods, of rigid diaphragms…) on out-of-plane mechanisms, by evaluating the change of the static collapse horizontal load multiplier, consequent to strengthening. The basic idea of the assessment was to verify that horizontal load multipliers for out-of-plane mechanisms are higher than the global base shear coefficient of the building, corresponding to the strength associated to in-plane response of walls. The more recent approach of annex 11.C of OPCM 3431 proposes the use of rigid-body analysis within equivalent static assessment procedures which

18

take into account, in an approximate way, the dynamic nature of the response. The procedure can be summarized as the sequence of the following steps. First step. A SDOF mechanism and its kinematics are defined, by idealizing the substructure involved in the mechanism (e.g. a portion of a façade wall) as a set of rigid bodies which can slide/rotate, separated from each other by fracture lines. Second step The static horizontal multiplier of vertical weights α0 that corresponds to the static threshold resistance is evaluated, via the application on PVW. Third step. From the SDOF mechanism, an equivalent s.d.o.f. system is defined, characterized by an effective mass M* is the effective mass and a0* an effective static threshold acceleration corresponding to the onset of the mechanism.. Fourth step. Perform the safety check with either a “simplified linear” or a “simplified nonlinear” procedure. The simplified linear consists in checking that:

⎟⎠⎞

⎜⎝⎛ +≥

HZ

qSa

a g 5.11*

0 (3)

where agS is the design elastic acceleration spectral ordinate at T=0, q is the behaviour factor assumed equal to 2.0; Z is the height, with respect to ground, of the centroid of all inertial masses involved in the mechanism; H is the total height of the building. The second procedure consists of the application of a nonlinear static procedure based on the definition of an effective secant period of the mechanism, and the consequent displacement demand, which is evaluated from an elastic response spectrum that accounts for filtering effect of the building and the amplification depending from the ratio Z/H. Either procedure allows also the evaluation of the value of ag which corresponds to the attainment of the ultimate limit state of the mechanism. One of the main criteria for the strengthening would be to insert tying/reinforcing elements that guarantee that such acceleration is higher than the one corresponding to the ultimate limit state associated with global response. 4.4 Aggregates of buildings In urban centres buildings are often inserted in aggregates or blocks in which each structural “unit” is interacting with the adjacent. This is the result of the progressive growth of the urban tissue, in which elevations are added to existing buildings and enlargements or progressive growths in plan are made by adding structural cells or units in contact with the previously existing, so that often adjacent units share the same boundary wall. Within a similar situation, the distinction of structurally independent units is problematic if not impossible, and any structural analysis that aims at evaluating the “global” response should in principle either model the whole block or model the structural unit with suitable boundary conditions that take into account the effect of the adjacent ones. Such an approach is clearly problematic even resorting to the most updated analytical tools, and beyond the possibility of average practicing engineers. Any method of structural analysis which considers the structural unit as independent is therefore clearly conventional. It is believed that a structural code on existing buildings should consider such situation when defining the allowable methods of structural analysis, and the possibility of conventional simplified analyses should be foreseen. In particular, the aim of the global safety check should be to verify whether a sufficient amount of walls along the principal directions of the unit is present or not. Suitable simplified criteria should be validated and made available to professionals, such as evaluation based on nonlinear storey-mechanism models. Research in this respect is strongly needed. 4.5 Mixed buildings From the early 20th-century the spreading of reinforced concrete technology caused the birth of mixed solutions starting from existing structures in order to satisfy requirements mostly related to functional purposes: masonry structures subjected to internal demolishment, column insertions, r.c. staircases insertions, plan enlargements or raisings by mean of steel or concrete structures. A significant number of such buildings is present on the Italian territory and in general in European countries. Furthermore, the insertion of additional r.c. walls or steel bracings is one of the current techniques which is being used for strengthening existing buildings, especially when the “global” seismic capacity of the system is deemed insufficient. Despite the relevance of such structures, which show very inhomogeneous characteristics, very little is known about their seismic behaviour, and very little research has been carried out, especially experimental. As a consequence, codes provide little support to the

19

designer besides suggesting methodological principles. From the structural and seismic point of view several typologies of structures could be distinguished: a) buildings with structural masonry walls and isolated linear RC or steel elements, i.e. columns (disconnected from structural masonry); b) buildings with structural masonry walls connected to (in contact with) tie columns and tie beams; c) buildings with structural masonry walls and r.c. walls (typically service cores) or steel braced structures; d) buildings with different vertical structures at different floors (e.g. lower floors masonry, upper floors r.c.); Hybrid configurations are clearly possible, combining in variable extent the characteristics of the above listed typologies. Each one of these typologies presents specific problems. The scientific literature regarding the seismic behaviour of typologies a) and c) is extremely poor with only few experiments (Modena & Tomazevic, 1990). In particular an almost total lack of experimental information on typology c) is reported, particularly regarding non symmetric plan layouts. A larger number of studies, including experimental researches, are available for typology b) with particular reference to confined masonry, in which the role of RC tie-columns and beams is prominently the increase in deformation and energy dissipation capacity of the system. Some experience could be also drawn from research on masonry infilled frames, although several differences would have to be taken into account, mostly regarding the different relative importance of r.c. members and masonry panels in carrying vertical and horizontal loads. Typology d), which includes especially buildings which were raised by addition of upper storeys, has also received scarce if no attention, as a consequence of its several drawbacks from the structural viewpoint in the realization of new buildings. Although in principle only nonlinear models could give a reasonable estimate of the seismic behaviour of such structures, code makers must realize that nonlinear models for such structures still must be developed and, if available, must be validated against experimental results. On the other hand, the use of elastic analysis in structures in which the deformation capacity of different structural elements can differ by large amounts, may cause a total unreliability of methods based on the use of a single q-factor. Clearly much research is urgently needed for the development of rational and applicable methodologies of assessment of such structures, which in turn would also have a significant spin-off on analysis and capacity models for strengthening.

5. SEISMIC STRENGTHENING OF MASONRY BUILDINGS

This last section of the paper is dedicated only to a very brief discussion of general issues that emerged in the drafting of OPCM 3431 regarding strengthening criteria and techniques. The choice of suitable strengthening techniques for existing masonry building is still rather controversial when dealing with some specific problems. Some techniques that had been proposed in the past have proved not always as effective as expected during recent earthquakes, whereas several new promising techniques and innovative materials are being proposed and used before a thorough validation from the field is available. Earthquakes occurred in areas which had been previously subjected to a strengthening campaign (such as Umbria, Italy, Decanini et al., 2000, western Slovenia, Tomaževič et al. 2000) have pointed out how partial interventions or merely formally correct measures (according to a literal interpretation of codes) can be unsuccessful, whereas comprehensive interventions with high quality of workmanship and construction control proved to be effective. A typical case of ineffective and arguably counterproductive intervention was the substitution of existing wooden floors or roofs with heavier, stiffer reinforced concrete structures, whose connection with the walls was made by the introduction of “in breach” r.c. ring beams. Their effectiveness was doubtful and even negative when the ring beam was resting only on the inner leaf of a double leaf walls or in general on low strength, rubble stone masonry. The replacement of wooden floors with rigid slabs (whose aim would be in principle to increase the integrity of the structural system), has been recognized as not always necessary, and alternative techniques have been developed in recent years and are still being studied to stiffen existing diaphragms and improve their connections with walls. On the basis of such experiences it was deemed therefore that the code approach, rather than being prescriptive or recommending specific techniques, should focus more on conceptual aspects, stating clearly the objectives of the interventions and introducing an informative annex, with a discussion of the most common techniques, their correct application, grouping them according to their intended purpose: interventions to improve connections, to reduce and eliminate horizontal thrusts, to stiffen floor diaphragms, to improve the roof structure and its connection to the walls, etc… Such informative annex, which in principle should not be necessary in a design code, has the main purpose of spreading a better “culture of seismic strengthening” among professionals.

6. CONCLUDING REMARKS

The present paper has presented and discussed selected issues pertaining to seismic design and assessment of masonry buildings.

20

A large part of the discussion has been dedicated to the problem of defining a consistent level of seismic force reduction to be used in elastic analyses, emphasizing the necessity of introducing overstrength factors and redistribution of forces in order to obtain results that are consistent with experience. A revision of Eurocode 8 provisions on behaviour factors is strongly needed to avoid unjustified underestimation of the seismic performance of new and existing masonry buildings. More attention should be given to the definition of rational out-of-plane assessment procedures of walls, especially important in case of slender structural or non-structural walls. The more recent approaches discussed herein still need further developments and validations. The use of new materials/construction systems should be always thoroughly validated with specific reference to seismic response, not only to vertical load bearing strength. To this end it is suggested that specific experimental procedures and criteria should be developed and codified to assess the conformity of new systems to specific seismic performance requirement. Current ongoing European research projects are pursuing this objective. Based on the Italian experience, it is felt that several improvements should be made to the Eurocode 8 approach to the assessment of existing masonry buildings, based on the recognition of the peculiarity of masonry, especially historical masonry, with respect to reinforced concrete or steel structures. Besides the definition of rational criteria for global assessment and for the assessment of first-mode damage mechanisms, for which some proposals have been made, still needing a systematic experimental validation, an important problem appears to be the approach to buildings inserted in aggregates and to buildings with mixed structures, for which very limited research has been carried out so far. Regarding the issue of existing masonry buildings, an extended, coordinated research programme has started in Italy funded by the Department of Civil Protection, under the coordination of ReLUIS (Network of University Laboratories of Earthquake Engineering), for which large scale shaking table tests are also foreseen at the new experimental facility of EUCENTRE, Pavia. It is hoped that this paper will stimulate researchers and code-makers in the earthquake engineering community to pay the due attention to the relevance of masonry as a structural material.

7. ACKNOWLEDGEMENTS Great part of this paper draws from the experience of drafting the masonry sections of OPCM 3431 (available at http://www.eucentre.it or on request from the author) for which the contributions of the colleagues Sergio Lagomarsino, Antonio Borri, Claudio Modena, Luigia Binda, Giacomo Di Pasquale are particularly acknowledged. Thanks are also given to Nigel Priestley, Sergio Lagomarsino, Julian Bommer and Gian Michele Calvi for their kind contribution in reviewing this paper and giving useful suggestions.

8. REFERENCES Albarello, D., Bosi, V., Bramerini, F., Lucantoni, A., Naso, G., Peruzza, L., Rebez, A., Sabetta F., Slejko, D.,

(2000) Carte di pericolosità sismica del territorio nazionale, Quaderni di Geofisica, 12. ANDIL Assolaterizi (2003), Come si costruisce in Italia: indagine campionaria su 1000 cantieri., Cresme, Roma ASCE-FEMA 356 (2000), Prestandard and Commentary for the Seismic Rehabilitation of Buildings, American

Society of Civil Engineers. ASTM C 1531-03, (2003) Standard Test Methods for In Situ Measurement of Masonry Mortar Joint Shear

Strength Index. Benedetti D., (2004), Costruzioni in muratura: duttilità, norme ed esperienze, Ingegneria Sismica, n. 3, 5-18. Benedetti D., Castoldi A. (1982), Dynamic and static experimental analysis of stone masonry buildings,

Proceedings 7th ECEE, Athens. Benedetti D., Benzoni G.M. and Petrini V. (1984), Experimental evaluation of seismic provisions for stone

masonry buildings, Proceedings 8th WCEE, San Francisco. Benedetti D., Pezzoli P. and Carydis P., (1998), Shaking table tests on 24 simple masonry buildings, Earthquake

Engin. Struct. Dyn. Vol. 27. Bommer, J.J., Magenes, G., Penazzo, P., Hancock, J. (2004), The Influence of Strong-Motion Duration on the

Seismic Response of Masonry Structures, Bulletin of Earthquake Engineering, Vol.2, no.1, 1-26. Calderini, C., Lagomarsino, S. (2006). A micromechanical inelastic model for historical masonry. Journal of

Earthquake Engineering, (in print). Cattari, S., Lagomarsino, S., (2006). Nonlinear analysis of mixed masonry and reinforced concrete structures. 1st

European Conf. on Earthquake Engineering and Seismology, Geneva, Switzerland, 3-8 September 2006. CEB (1983), Model Code for Seismic Design of Concrete Structures, CEB Bulletin d’Information, no. 160.

21

CEN - EN 1996-1-1 (2005) Eurocode 6 - Design of masonry structures - Part 1-1: General rules for reinforced and unreinforced masonry structures.

CEN – EN 1998-1 (2005), Eurocode 8: Design of structures for earthquake resistance, Part 1: General rules, seismic action and rules for buildings.

CEN – EN 1998-3 (2005), Eurocode 8: Design of structures for earthquake resistance, Part 3: Strengthening and repair of buildings.

CIB (1985), Recommendations for seismic design of unreinforced, confined and reinforced masonry structures, Appendix to the International recommendations for the design and erection of unreinforced and reinforced masonry structures, Conseil International du Bâtiment.

Chiostrini S., Galano L., Vignoli A. (2003). In Situ Shear and Compression Tests in Ancient Stone Masonry Walls of Tuscany, Italy, Journal of Testing and Evaluation, ASTM, Vol. 31, No. 4, 289-304.

Como, M., Grimaldi, A., (1985) An unilateral model for the limit analysis of masonry walls, in “Unilateral Problems in Structural Analysis”, Springer.

Council of Europe (1998), European Macroseismic Scale 1998, edited by G.Grünthal, Centre Europèen de Géodynamique et de Séismologie, Luxembourg.

Da Porto F., E. Garbin, C. Modena, M.R. Valluzzi (2005). Failure modes for in plane loaded masonry walls made with thin layer mortar, 10th Canadian Masonry Symposium, Banff, Alberta, 694-704

D’Ayala, D. & Speranza, E. (2003). Definition of Collapse Mechanisms and Seismic Vulnerability of Masonry Structures. Earthquake Spectra 19 (3), 479-509.

Decanini, L., Gavarini, C. and Mollaioli F. (2000), Some remarks on the Umbria-Marche earthquakes of 1997, European Earthquake Engineering, n.3, 18-48.

Decanini, L., De Sortis, A., Goretti, A., Langenbach, R., Mollaioli, F., Rasulo, A., (2004), Performance of masonry buildings during the 2002 Molise, Italy, earthquake, Earthquake Spectra, no.S1, S191-S220.

DISWAll, 2005-2007, COOP-CT-2005-18120 "Developing innovative systems for reinforced masonry walls"; scientific coordinator: C. Modena, University of Padova, Italy

Doherty, K.T., Griffith, M.C., Lam, N. and Wilson, J. (2002), Displacement-based seismic analysis for out-of-plane bending of unreinforced masonry wall, Earthq. Engng. Struct. Dyn., Vol. 31, 833-850

DT2 (1978), Raccomandazioni per la riparazione strutturale degli edifici in muratura. Regione Autonoma Friuli Venezia-Giulia, Aprile 1978.

ESECMaSE, 2004-2007, CT-2003-500291 “Enhanced Safety and Efficient Construction of Masonry in Europe”, scientific coordinator: E.Fehling, University of Kassel, Germany.

FEMA 440, (2005), Improvement of Nonlinear Static Seismic Analysis Procedures, prepared by the ATC-55 Project, FEMA – NEHRP, June 2005.

Gambarotta, L., Lagomarsino, S., (1996) Sulla risposta dinamica di pareti in muratura, Atti del Convegno Nazionale “La Meccanica delle Murature tra Teoria e Progetto”, Messina, Pitagora Ed. Bologna.

Gambarotta L., Lagomarsino S., (1997). Damage models for the seismic response of brick masonry shear walls. Part II: the continuum model and its applications. Earthq. Engng. Struct. Dyn., Vol. 26, 441-462.

Griffith, M.C., Magenes, G., Melis, G. and Picchi, L., (2003), Evaluation of out-of-plane stability of unreinforced masonry walls subjected to seismic excitation, Jour. of Earthq. Engng., v.7, SP 1, 141-169.

Giuffré, A.., ed., (1993), Sicurezza e conservazione dei centri storici – Il caso Ortigia, Editori Laterza, Italy. GNDT-CNR (1985). Proposta di nuova normativa sismica. Ingegneria Sismica n.1/85, Inserto. Hendry, A.W., (2002), Engineered design of masonry buildings: fifty years development in Europe, Prog. Struct.

Engng Mater., no.4, 291-300. Javed, M., Naeem Khan, A., Penna, A., Magenes, G. (2006), Behaviour of masonry structures during the

Kashmir 2005 earthquake, 1st European Conf. on Earthquake Engineering and Seismology, Geneva, Switzerland, 3-8 September 2006.

Kappos, A., Penelis, G.G., Drakopoulos, C.G., (2002) Evaluation of simplified models for lateral load analysis of unreinforced masonry buildings, Journal of Structural Engineering, ASCE, 128 (7), 890-897.

Lagomarsino, S., Penna, A., Galasco, A., (2006), TREMURI Program: Seismic Analysis Program for 3D Masonry Buildings, University of Genoa.

Lagomarsino, S., Podestà, S., Resemini, S., Curti, E., Parodi, S., (2004) Mechanical models for the seismic vulnerability assessment of churches, Structural Analysis of Historical Construction, Balkema,1091-1103.

Lourenço, P.B., (2002). Computations on historic masonry structures, Prog. Struct. Engng Mater.,Vol. 4, no.3, 301-319.

Magenes G., (2000). A method for pushover analysis in seismic assessment of masonry buildings, 12th World Conference on Earthquake Engineering, Auckland, New Zealand, paper 1866.

Magenes G., (2004). Prospettive per la revisione della normativa sismica nazionale con riguardo alle costruzioni in muratura, Assemblea ANDIL, sezione "Murature", Isola Vicentina, Italy, 22 aprile 2004.

22

Magenes G., Bolognini D. and Braggio C., (2000). "Metodi semplificati per l'analisi sismica non lineare di edifici in muratura", CNR - Gruppo Nazionale per la Difesa dai Terremoti, Roma, (http://gndt.ingv.it/Pubblicazioni/Magenes_copertina_con_intestazione.htm).

Magenes, G., Calvi, G.M., Kingsley, G.R., (1995), Seismic Testing of a Full-Scale, Two-Story Masonry Building: Test Procedure and Measured Experimental Response, in Experimental and Numerical Investigation on a brick Masonry Building Prototype - Numerical Prediction of the Experiment, Report 3.0 - G.N.D.T., University of Pavia.

Magenes, G., Della Fontana, A., (1998), Simplified non-linear seismic analysis of masonry buildings, Proc. of the British Masonry Society, No. 8, October 1998, 190-195

Magenes G., Morandi P. (2006). Normativa nazionale ed europea sulle murature: Eurocodice 8 ed evoluzione dell’Ordinanza per le costruzioni in zona sismica (OPCM 3274)”, Rapporto EUCENTRE, Pavia.

Magenes, G., Remino, M., Manzini, C., Morandi, P., Bolognini, D., (2006), SAM II, Software for the Simplified Seismic Analysis of Masonry buildings, University of Pavia and EUCENTRE.

Ministero dei Lavori Pubblici, (1981) Normativa per le riparazioni ed il rafforzamento degli edifici danneggiati dal sisma nelle regioni Basilicata, Campania e Puglia, D.M. 2/7/1981, Suppl. Ord. G.U. 21/7/1981, n.198.

Modena C., Tomaževic M. (1990). A research program on the seismic behaviour of modern masonry buildings. Proceedings of the ninth European Conference on Earthquake Engineering. Vol. 5: 13-22.

Morandi P., Magenes G., (2006). Second order effects in out-of-plane strength of urm walls subjected to bending and compression. ROSE Report.

OPCM n. 3274 (2003). “Primi elementi in materia di criteri generali per la classificazione sismica del territorio nazionale e di normative tecniche per le costruzioni in zona sismica”, Suppl. ord. n.72 alla G.U. n. 105 del 8/5/2003, e successive modifiche ed integrazioni.

OPCM. n. 3431 (2005). “Ulteriori modifiche ed integrazioni all'Ordinanza n.3274 del 20/3/2003, recante ‘Primi elementi in materia di criteri generali per la classificazione sismica del territorio nazionale e di normative tecniche per le costruzioni in zona sismica’ ” Suppl. ord. n.85 alla G.U. n.107 del 10/5/2005.

Penna, A., Resemini, S., Galasco, A., Lagomarsino, S. (2004) Non-linear seismic analysis of masonry structures, 13th World Conference on Earthquake Engineering, Vancouver, BC, Canada, paper 843.

Priestley, M. J. N. (1985), Seismic behaviour of unreinforced masonry walls, Bulletin of the New Zealand Nat. Soc. Earthq. Engrg. 18(2).

Priestley M.J.N., (2003), Myths and Fallacies in Earthquake Engineering, Revisited. IUSS Press, Pavia, Italy. Restrepo-Vélez, L.F., Magenes, G., (2005), Seismic risk of unreinforced masonry buildings, ROSE Report

2005/05. Roca, P., Molins, C., and Marì, A.R., (2005), Stregth capacity of masonry wall structures by the equivalent

frame method, Journal Of Structural Engng, ASCE, vol.131, no.10,1601-1610. Sheppard P. F. (1985). In-situ test of the shear strength and deformability of an 18th century stone-and-brick

masonry wall, Proc. of the 7th Int. Brick/Block Masonry Conference, Melbourne, Vol. 1, pp. 149-160. Spence, R., (1999), Intensity, damage and loss in earthquakes, in A.Bernardini (ed.) Seismic Damage to Masonry

Buildings, Balkema, 27-40. Tomaževič M. (1978) The computer program POR, Report ZRMK Tomaževič M. (1999). Earthquake -resistant design of masonry buildings, Series on Innovation in Structures and

Construction, Vol. 1, Imperial College Press, London Tomaževič M. (2000). Seismic redesign of existing stone-masonry buildings, Europ. Earthq. Engng., 3, 59-66. Tomaževič M., Bosiljkov V. and Weiss P., (2004), Structural behaviour factor for masonry structures”, 13th

World Conference on Earthquake Engineering, Vancouver B.C., Canada, paper no. 2642 Tomaževič M., Klemenc, I., Lutman, M., (2000), Strengthening of existing stone-masonry houses: lessons

learned from the earthquake of Bovec of April 12, 1998, European Earthquake Engineering, 1, 13-22. Tomaževič M., Weiss P. (1994). “Seismic behaviour of plain and reinforced masonry buildings”, Journal of

Structural Engineering, ASCE, 120 (2), 323-338. Turnšek, V., Sheppard, P., (1980) The shear and flexural resistance of masonry walls, Proc. of the Intern.

Research Conference on Earthquake Engineering, Skopje, pp. 517-573 Vaculik, J, Griffith, M, Lam, N, Wilson, J and Lumantarna, E, (2005). "Cyclic response of unreinforced clay

brick masonry walls," Proceedings of 2005 Conference, Australian Earthquake Engineering Society. Vratsanou, V., (2001), Earthquake resistant design of unreinforced masonry buildings – A comparative study of

the National Seismic Codes in Europe, European Earthquake Engineering, no.1. Yi, T., Moon, F.L., Leon, R.T., Kahn, L.F., (2006), Lateral load testing on a two-story unreinforced masonry

building, Journal of Structural Engineering, ASCE, vol. 132, no.5, 643-652.