You are on page 1of 32

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/249873053

Definition of Collapse Mechanisms and Seismic


Vulnerability of Historic Masonry Buildings
Article in Earthquake Spectra August 2003
Impact Factor: 1.32 DOI: 10.1193/1.1599896

CITATIONS

READS

99

584

2 authors, including:
Dina Francesca D'Ayala
University College London
184 PUBLICATIONS 815 CITATIONS
SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,


letting you access and read them immediately.

Available from: Dina Francesca D'Ayala


Retrieved on: 09 May 2016

Definition of Collapse Mechanisms


and Seismic Vulnerability of Historic
Masonry Buildings
D. DAyalaa) and E. Speranzaa)
In historic city centers the mitigation of seismic risk is dependent on the
possibility of implementing strengthening programs. Given the cultural and
economic value attached to the historic structures, however, interventions
should be tailored to suit aesthetic and structural requirements of each building type, and provide sufficient reliability of performance in future earthquakes. A simple analytical model is developed to calculate load factors associated with various collapse mechanisms of wall assemblies, and
vulnerability functions are derived. An application shows the capability of the
procedure to quantify reduction in vulnerability associated with strengthening
implementations for different typologies. [DOI: 10.1193/1.1599896]
INTRODUCTION
In the past twenty years, seismic vulnerability studies of ordinary masonry buildings
have been conducted based on a statistical approach, developing indexed algorithms for
which the crucial parameters are recognized and assigned weights, usually based on rudimental structural models, and whose validity relies on expected normal distribution of
a large number of samples (Augusti et al. 1985, Benedetti et al. 1988, Coburn and
Spence 1992, Corsanego et al. 1986). While useful to inform mitigation policies at a national level and, to a certain extent, insurance premium definitions (Campos Costa et al.
1998), their application at a local scale, of rather smaller samples, as to which buildings
should be strengthened and what type of strengthening is needed, is impaired by the peculiarities of each building and its context, not only structural but also geographical and
historical. A number of recent studies have tried to remedy this situation by including
more sophisticated structural description of the sample objects. Among these is the work
of Calvi (1999), which proposes a displacement-based criterion but confines its application to in-plane failure of masonry walls. Recognizing the difficulty of properly quantifying all parameters involved in the definition of vulnerability, and the limitation of
probabilistic approaches, Bernardini (1999) suggests the use of fuzzy logic to define
measures of damage and vulnerability, while Sanchez-Silva and Garcia (2001) propose a
methodology for damage assessment that combines systems theory, fuzzy logic, and
neural networks. Applications so far are limited to a few case studies. For the results of
such procedures to be meaningful, a reasonably realistic mechanical model of the masonry needs to be implemented, one with a relatively small number of mechanical parameters that are easy to quantify without expensive on-site testing.

a)

Department of Architecture and Civil Engineering, University of Bath, Bath BA2 7AY, UK

479

Earthquake Spectra, Volume 19, No. 3, pages 479509, August 2003; 2003, Earthquake Engineering Research Institute

480

D. DAYALA AND E. SPERANZA

Computational models based on noncontinuous material modeling and a macroelements approach have been applied to the dynamic analysis of monumental buildings by
Casolo (2000) and Azevedo et al. (2000) and to the definition of vulnerability functions
of churches by Augusti et al. (2001). The lack of out-of-plane strength has been identified both by on-site investigation (DAyala et al. 1997) and experimental and analytical
evidence (DAyala 1999, Doherty et al. 2002) as the weak link in masonry seismic capacity. On the other hand, it is recognized by various authors that the actual failure
mechanisms and lateral displacement or loading capacities of a wall within a building
are influenced by the dimensions of the masonry units, and the nature of contact and
connections to orthogonal walls (de Felice and Giannin 2001) and floor structures (Bernardini 1999).
The present study proposes interfacing simple but realistic mechanical modeling
based on limit analysis and macroelements with the necessity of producing a tool capable of both handling relatively large numbers of buildings and performing statistical
analysis.
In the following sections first the adopted mechanical model is introduced. This is
based on the assumption of dry block masonry with essentially a frictional behavior. Cohesion is also included. On the basis of this model, in-plane overturning and sliding of
isolated walls with and without applied live loads are analyzed, and using a limit-state
approach the ultimate load factors are derived. A parametric analysis is performed. Masonry buildings are constituted of three-dimensional assemblies of walls, and the outof-plane behavior of each wall is highly influenced by the type and strength of connection with the others. The collapse mechanisms of wall assemblies are then analyzed and
ranked depending on the type of damage caused and size of structure involved. Vulnerability functions are derived and a few case studies show how the procedure can be used
to define strengthening interventions.
A SIMPLE MATERIAL MODEL FOR MASONRY CONSTRUCTION
Stone blocks of regular dimensions and neatly dressed, set in very thin mortar joints
or in dry contact, represent the best possible layout of stones to ensure structural integrity within a masonry wall. Typical examples of such masonry are the Roman opus
quadratum or the ashlar stonework of Gothic cathedrals. The Latin name refers to its
morphology, characterized by overlapping squared stones, with or without thin bedding
mortar. This kind of masonry, known in antiquity as opus isodomum, is accurately described by Vitruvius Pollione in De Architectura (first century B.C.). This construction
technique is still in use in many European countries and constitutes the masonry fabric
of most buildings in historic centers.
The structural integrity of the wall relies upon a regular staggering of blocks of subsequent courses and on the regular presence of blocks laid out orthogonally to the face
of the wall, to connect the various wythes of the wall together (Giuffre` 1989). Studies
and experimental tests have demonstrated clearly the function of the diatones in ensuring the monolithic behavior of the two wythes of the wall and in improving its out-ofplane resistance, while successive courses of ortostates are staggered to improve the inplane behavior of the wall (Ceradini 1992).

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 481

Figure 1. Identification of vertical crack and acting forces.

Although the resultant masonry construction does not have any bonding agent and,
hence, no tensile strength can be developed for actions in the vertical direction, the contact surfaces between courses are characterized by frictional behavior according to Coulomb theory (Coulomb 1776), with finite friction coefficient and possibly some cohesion c (Heyman 1972). The presence of friction enables the masonry to resist relative
slip among the units. With reference to Figure 1, for the wall of total height H and total
number of courses n, subjected to the horizontal force F distributed along its height, the
maximum slip strength that can be developed by a generic unit of dimensions l, h, b
(Figure 2) and mass mlhb, within course i, is

Cihisbg

(1)

where hiih is the height of the wall above the position of the unit at level i, and the
friction coefficient. The value of Ci, calculated at each interface among courses is thus a
stepwise function of hi and of the weight of wall above the course considered. The total
shear strength (Ctot) for a wall of given height and number of courses is
n

i1

i1

Ctot Ci hisibg

(2)

In Equation 2 it is assumed that the block dimensions and their overlapping length

Figure 2. Stone block dimensions.

482

D. DAYALA AND E. SPERANZA

Figure 3. Normalized curves for total shear strength Ctot .

are constant, at least within a course. However, the result can easily be generalized to
less regular masonry styles and more realistic cases by identifying for each course interface the minimum overlapping length si , which will also locate the surface of slip.
Figure 3 shows the variation of the ratio of Ctot to the total weight Wtot of a wall, for
different ratios of block height to walls height and ratios of overlapping lengths to total
length of wall. It can be observed that the greater the ratio of block height to wall height,
the lower the value of Ctot /Wtot , so that increasing the number of courses and reducing
the number of blocks in each course increases the total shear strength of the wall.
The variation with height of the total shear strength implies that the occurrence of
vertical cracks in reality is unlikely, while the application of a uniformly distributed
horizontal force F along a wall, of height H and length L, will result in a diagonal crack,
assumed with origin in the lower right vertex of the panel and identified by a generic
angle c , which it forms with the vertical axis. For a monolithic panel (Figure 4a), the
friction resistance acts along the surface of the sloping crack line and can be defined as
H sin c2
CcWcsin c
g
2 cos c

(3)

where W(c) is the weight of the portion of wall identified by the angle c for a unit
thickness. C(c) is a monotonically increasing function of c , for 0c/2. If the
same wall is formed by stone blocks (Figure 4b), with reference to the block dimensions
previously introduced,

s
L
barctan , parctan
h
H

(4)

are the angles identifying the block shape and the panel shape, respectively. Then the
total shear strength developed on the crack assumes the following expression:

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 483

Figure 4. (a) Monolithic wall with diagonal crack, and (b) stone block wall with diagonal
crack.
n

Ctotc Cic

(5)

i1

which can be specialized in relation to the shape ratio of the panel, staggering ratio, and
position of the crack, as follows:

for cminp,bCtotc

for pcbCtotc

1s
H2bg
2h

(6a)

1s 1
1
1

L2bg
2
2 h tan c tan p tan c

1
for bcpCtotc tan cH2bg
2
for cmaxb ,pCtotc

1
1 1

L2bg
tan p 2 tan c

(6b)

(6c)

(6d)

In Equations 6a and 6b, Ctot is a function of the geometry of the wall and shape ratio
of the unit and independent of the crack angle (equivalent to a vertical crack). This
means that, whatever the position of the crack, for cb , the total shear strength is
constant and equivalent to the portion of a triangular wall identified by the angle b and
the height H of the wall considered (see Figure 4c). For cb , on the other hand, the
Ctot becomes independent of the block shape ratio. In Figure 5 the adimensionalized ratio of Ctot to the total weight Wtot of a panel of dimensions H, L is plotted for two different ratios of L/H and two different ratios of s/h, so as to include all cases described by
Equations 6a through 6d. The curves obtained are also compared with the curves for a
monolithic wall with the same geometric dimensions. The friction coefficient is assumed
constant, 0.4. Ctot /Wtot is always greater for the opus quadratum than for the monolithic panel, showing that the latter, while it can be safely assumed as a lower bound,
underestimates the wall capacity. For panels with s/hL/H, the initial value is constant

484

D. DAYALA AND E. SPERANZA

Figure 5. A dimensionalized ratio of shear strength to total weight of panels with =0.4.

up to tan(c)s/h and then increases. For values of L/Hs/h, Ctot /Wtot is constant up to
tan(c)L/H and then decreases down to a minimum for s/htan(c), for which Equation 6b reduces to Equation 6d. These are also the points of intersection of curves with
equal panel ratios and different block ratios. The value of c for which the minimum
ratio Ctot /Wtot is obtained is independent of the geometric ratio L/H.
In assessing the equilibrium of panels subjected to horizontal action in their plane,
the aim is to define the collapse-load factor and the mode of collapse, either sliding or
overturning, as shown in Figure 6.
The diagonal crack, inclined to the vertical of a variable angle c , shears the wall in
two portions. It is assumed that the lower left-hand portion does not participate in the
failure. The other portion, depending on the value of c , can assume a triangular, trapezoidal, or rectangular shape.
The assumptions at the basis of this approach are as follows:

Figure 6. The two mechanisms considered: (a) sliding, and (b) overturning; (c) mixed mode.

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 485

Dead load and horizontal equivalent force are applied at the center of gravity of

the panel and are expressed as a function of the gravity constant and the horizontal acceleration, respectively;
The masonry walls are simulated with a system of rigid bodies, articulated by
hinges, whose geometry is defined by the failure mechanism;
The horizontal static action is defined as proportion of the body mass, by means
of a factor , corresponding to the ratio a/g, between the lateral acceleration and
the gravitational acceleration. This refers directly to the mass of the body rather
than its weight.

The sliding mechanism is governed by

s
Ctotc

c tan c h

(7)

where (c) is the mass of the mobilized portion at incipient failure. This equation is
only valid for tan cs/h, beyond this limit s.
With respect to the overturning mechanism, the collapse-load factor o assumes different expressions in relation to the quantification of Ctot(c) seen in Equation 6, which
highlights the fact that, due to the staggering of the blocks, overturning can only take
place if sufficient sliding also takes place so that a free horizontal surface equivalent to
the length of the wall can be identified:

tan2 c
o(cminp ,b
opcb

s
h

(8a)

2 tan c
tan c
2

(8b)

tan2 c3 tanc2 tan p2 tan p 2 tan c


obcp

3 tan2 ctan2 p

s
h tan p
tan c
(8c)

ocmaxp ,b

tan c tan p3 tan c2 tan p


3 tan2 ctan2 p

(8d)

Note that Equation 8c reduces to Equation 8a for tan ctan p . Note also that for
the angle of crack greater than the staggering of the blocks, the overturning load factor
is independent of the friction coefficient, as the mechanism is purely rotational (Equations 8b and 8d). From Equations 7 and 8, ca/g having defined as the collapse-load
factor, depending on the ratio o /s , three domains of existence for the solution can be
identified:

D1cOS safe configuration

(9a)

486

D. DAYALA AND E. SPERANZA

Figure 7. Domains of existence defined by the functions o(c) and s(c) for (a): L/H0.4;
=0.4; s/h0.8; and (b) L/H1.2; =0.4; s/h0.35.

D2

O
1 overturning failures
S

(9b)

O
1 sliding failures
S

(9c)

D3

Among the possible values of c and c , the limit state configuration is the one for
which

ccminO ,S

(10)

The couple (c ,c) uniquely identifies the portion of wall that fails and the type of
failure. For slender panels with ratio L/H1 and s/h1, the failure occurs for overturning, with cb . The load factor decreases with the increase in slenderness. However, if
s/h1 but greater than the friction coefficient and also L/H, failure might occur by
sliding even in panels with L/H1. For a given angle of friction and of the geometric
ratio of the panel, the staggering will affect the extension of the safe area in Figure 7a.
For geometric ratios of the panel L/H1, failure occurs for sliding, according to the
Coulomb criterion, if the ratio s/h, and min, with a crack inclined at any angle in
the range between cb and the horizontal, otherwise failure occurs for overturning of
a portion identified by an angle cb and a load factor s/hmin (Figure 7b). These
results compare well with experimental evidence (Ceradini 1992) as shown in Figure 8.
The role exerted by staggering ratios is highlighted in Figure 9 showing load factors
curves for a slender panel (L/H0.4) with 0.3 and different ratios s/h. For s/h
L/H, the minimum load factor corresponds to the condition s/htan c/2L/H and the
actual value is independent of and L/H, although the extent of the safe domain reduces
with s/h. The case s/hL/H identifies a singularity, as the equations defining the domain

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 487

Figure 8. Comparison of analytical and experimental results (=0.55).

of existence of the load factor are Equations 8a and 8d, where for tan cs/h, the first
presents the positive additional term /2. However, for increasing values of s/h, the
angle of crack equals b and, hence, the load factor, increases. For s/h2L/H, the minimum load factor is only slightly smaller than the friction coefficient, while the angles of
crack for sliding and overturning coincide so that either can take place. For greater ratios
of s/h, failure occurs by sliding, the load factor is equal to the friction coefficient, and
the crack angle will be inclined at any angle between cb and the horizontal.
Figures 10a and 10b generalize the previous results, obtaining load factor curves for
variable values of (s/h, L/H) and (s/h, ), respectively. In the first diagram, a higher value
of friction coefficient has been chosen to better highlight the relative influence of panel
and unit ratios in the overturning mechanism. For very slender panels the influence of
the ratio s/h is negligible. For increasing ratios of L/H, the load factor increases linearly
up to s/hL/H, and then tends to the asymptotic value minL/H for s/h (monolithic
case). Panels with ratio L/H1 initially show a linear dependence of the load factor
from the staggering ratio, however, as this increases, the load factor levels off to the
value associated with L/Hs/h, beyond which it would resume the asymptotic behavior,

Figure 9. Functions of c(a/g), for different ratios s/h (L/H0.4; =0.3).

488

D. DAYALA AND E. SPERANZA

Figure 10. Influence of staggering ratio s/h, wall shape factor L/H, and friction coefficient on
collapse-load factor c : (a) constant 0.6, and (b) constant L/H0.6.

were it not for the fact that this yields values greater than the friction coefficient and,
hence, curves level off to this value and sliding occurs in preference to overturning.
Finally, the influence of the friction coefficient is shown in Figure 10b. Initially the
behavior is linear with increasing slope as s/h increases; beyond s/h0.4, the initial
slope is common but the curves level off at min0.5 L/H for s/hL/H, while for higher
values of s/h, the behavior becomes independent of the friction coefficient for 0.5
and the maximum load factor is associated with Equation 8d and tends to L/H for s/h
, as already noted. Typical values of friction coefficient for dry stone masonry or
weak mortared unit are usually measured in the range 0.30.6, within which the diagram
in Figure 10b shows a marked sensitivity of the load factor to the actual staggering ratios.
IN-PLANE FAILURE OF WALLS LOADED BY HORIZONTAL STRUCTURES
Traditional masonry structures can be described as a three-dimensional network of
orthogonal walls creating cells that resist lateral load through a box behavior. The horizontal structures traditionally are made of timber beams or joists spanning the shortest
distance between walls, so that only one of the two orthogonal alignments of walls is
load bearing while the other sets role is mainly to provide out-of-plane restraint to the
bearing walls.
The roof structure is usually similar to the floor structure, with a number of beams
running from eave to eave and rafters spanning in between, or if there is an internal
spine wall, rafters spanning from this to the external walls. In the following, the effect of
horizontal structures on the collapse-load factor function is analyzed.
A uniformly distributed vertical load qr at the top of the wall, simulating the roof
vertical action, and q f , simulating the floor action, at n levels along the height of the

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 489

Figure 11. Uniformly distributed vertical loads of roofs and floors in a wall with n stories.

wall, are assumed (Figure 11). These are the resultants of the dead and live load per unit
surface times the half-span of the floor structure. Of these only the portion directly bearing on the part of wall identified by the crack will participate in the mechanism, contributing a vertical resultant and a horizontal resultant equating to its mass. For simplicity, it is assumed that all levels have the same interstory height, and stories are numbered
from top to bottom.
Assuming cp , the weight of the roof and the total weight of horizontal structures
acting on the portion of wall identified by the variable angle c are
n

H
Q f q f tan c ni;
n
i1

QrqrH tan c ,
if cp , then

QrqrH tan p

H
Q f q f rH tan p
tan c ni
nr
ir1

(11a)

(11b)

where

rint n

tan ctan p
tan c

(12)

The nth term of the series has no contribution as the horizontal structure corresponds

490

D. DAYALA AND E. SPERANZA

in this case to the ground floor. Hence, in the case of n1, Q f 0. The weight of the roof
has been considered separately as the load per unit length might be different and also it
might be bearing on orthogonal walls.
The added weight on the horizontal courses associated with the roof and floor structure will also increase the total shear strength. The total contribution is

For cb CqrQr

s 1
;
h tan c

Cq f Q f

s 1
h tan c

(13)

where Qr and Q f should be specialized to Equations 11a or 11b, depending on whether


the angle of crack is smaller or greater than the angle of the panel diagonal p , respectively. Defining qrq f khs , where is the density of the masonry, hs is the interstory
height assumed constant, the thickness of the wall is assumed as unit and k is a nondimensional constant, on the basis of Equations 11 to 13, it is then possible to write adimensionalized expressions for the overturning load factor o :

tan2 c 1
o(cminp ,b

s
3k 2k
2k
1
2 ni 1
1 ni
n n i1
h
n
n i1

3k 4k
tan c 2 2 ni
n n i1

obcp

3k 2k
tan c 1 2 ni
n n i1

3k 4k
2 2 ni
n n i1

(14a)

(14b)

Analogous equations can be derived for the angle of crack greater than the angle of
the panel diagonal. The load factor relative to the sliding mechanism is instead:

Ctot()CqrCqf
WQrQ f

(15)

Figure 12a shows the functions a/g for a wall with the same parameters as the
one in Figure 7a, with uniformly distributed loads, simulating the action of the roof
(qr2 KN/m) and of two intermediate floors (q f 2 KN/m). For this particular set of parameters, the resulting load factor o-30.279 is actually smaller than the one for the
self-weight of the wall alone (o-00.294, see Figure 7a). Moreover, the critical crack
occurs for an angle of 30, rather than c40 associated with the case of self-weight.
However, increasing the number of floors actually increases the load factor, and for n
4 (three intermediate stories) the load factor -40.299, greater than the one associated with the self-weight (Figure 12b). Figure 13a and 13b generalize these results by
assuming different values of s/h, q f , qr , and n.

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 491

Figure 12. Functions o and s for (a) L/H0.4, s/h0.8, =0.4, n3, qr2 KN/m, q f
2 KN/m, and (b) L/H0.4, s/h0.8, =0.4, n4, qr2 KN/m, q f 2 KN/m.

In the present study, windows have been quantified by subtraction of the load associated with the window void, and assuming that the crack will continue at the same inclination before and after the opening. Substantiation of this assumption can be found in
damage surveys for houses with flexible floor structures (EERI 2002) as well as in discrete elements modeling (Baggio and Trovalusci 1998).

Figure 13. Load factor functions for a wall with L/H0.4 and =0.3, with variable number of
floors: (a) s/h1.8, and (b) s/h0.4.

492

D. DAYALA AND E. SPERANZA

FORMULATION OF THE OUT-OF-PLANE FAILURE OF WALL ASSEMBLIES


Besides the in-plane failure, the external masonry walls of a house will typically be
subjected to out-of-plane mechanisms. The way in which these will develop depends on
the quality and strength of the connections with the other elements of the structure, party
walls, internal load-bearing partitions, floors, and roof structures. If the structure has not
undergone strengthening, it is assumed that the only means of restraint to overturning
exerted by other elements onto a particular wall is governed by the friction of the contact
surface, and this will give rise to different types of failures, highlighted in Figure 14.
However, if the building structural capacity has been improved, by means of introduction
of ties or ring beams, then usually the simple overturning is prevented, while mechanisms relying on arch effect develop as discussed in the next section (see Figure 16).
Direct observation of crack patterns recorded in post-earthquake damage surveys
(DAyala 1999), together with the available literature on reconnaissance of stone masonry buildings earthquake damage (Karantoni et al. 1997, Tomazevic et al. 1999),
yield to the conclusion that often the most recurrent failure mechanism surveyed is one
of overturning of the street facade. This, however, most commonly occurs involving portions of the party walls as well. The extension and number of the party walls involved
will depend on the nature of the connection of these with the facade, giving rise to six
possible different types of failure, as shown in Figure 14. Mechanism Type A assumes
that no connection is present at the edges of the wall, or this is insufficient to generate
restraint by the party wall. Mechanisms B1 and B2 will occur instead of mechanism
Type A when the level of connection is sufficient to involve, beyond the facade wall,
respectively, one or both party walls into the overturning, due to sufficient length of
overlapping between elements common to both walls. These mechanisms develop by the
occurrence of a diagonal crack along the party walls and a horizontal hinge on the facade. The angle of the crack c in both mechanisms B1 and B2 is optimized by means
of the in-plane formulation discussed previously. In the case of connections with only
one party wall, the mechanism of Type D is also possible, for which only a portion of the
facade is subjected to overturning and the party walls are not involved directly in the
mechanism. This may also occur in the case of strengthening devices not uniformly distributed on the whole facade. For the calculation of this mechanism it is assumed that the
crack inclination is defined by the diagonal through i stories, the thickness is constant
and averaged over the height of the portion involved, and that torsional friction along the
sloped crack line is not considered.
The mechanism characterized by the overturning of a corner around a horizontal
hinge placed at any story level and orthogonal to the corner bisector plane, mechanism
Type C, will only occur when at least one corner of the building is free, without any
adjacent structures, hence, for buildings isolated or positioned at the end of a block.
Here it is assumed that the angles of crack fi and si , respectively, affecting the facade
and the orthogonal wall, result from the minimization procedure seen in the previous
section for in-plane failure.
Mechanism Type E is considered when due to the window layout, and depending on
the type of construction, there might be solution of integrity within the facade plane
leading to partial failures. It is assumed that the portion of facade subject to collapse is

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 493

Figure 14. Geometric parameters and applied load for overturning mechanisms.

conditioned by the width of the side piers, which define the distance between the column
of openings and the closer vertical edge. Therefore, if the windows are aligned vertically
and the width of the side pier is greater than the width of the window and if the edge
connection is active, the vertical crack can occur within the facade itself. This may occur
on one or both sides of the facade.
For all mechanisms it is assumed that there are sufficient units laid orthogonally to
the plane of the facade as to ensure integrity within the thickness. It is also assumed that

494

D. DAYALA AND E. SPERANZA

every mechanism can involve any number of stories up to the whole facade, hence their
formulation does not assume a fixed position of the cylindrical hinge but performs for
each mechanism an optimization through minimization of the associated load factor. The
lower-bound approach is adopted to calculate for each mechanism the failure load factor
that will assume the general format:

xCy1
y2

(16)

where is the total mass of the portion of facade, orthogonal walls and live loads involved in the mechanism, C is the resultant of the restraining action caused by friction,
and x, y1 , y2 are the distances of the point of applications of the resultant forces from
the cylindrical hinge.
The detailed formula for each mechanism is shown in Table 1, while the models
adopted for each mechanism with the relevant geometric parameters are shown in Figure
14. In the formulas, j is the number of stories involved and the story level at which the
hinge is positioned; hs the story height assumed as constant for a given facade; k is a
parameter expressing the imposed load of roof and floors in terms of the self-weight of
the walls; and r is the number of masonry courses within the height jhs . Other relevant
symbols specific to each formula are defined in Table 1.
Equations in Table 1 are written in an adimensionalized format, using the same symbolism as the previous section and reduced to the essential geometric parameters. The
load factors are a function of the slenderness T(h)/H of the walls, and of the restraining
action due to friction, through the dimensions of the unit, the friction coefficient, and the
staggering between courses. Different behaviors will also result depending on whether
the portions involved in the mechanisms are or are not bearing walls. In the following
the variation of the load factors for each mechanism associated with the different parameters is discussed.
The staggering ratio together with the value of friction coefficient influence the occurrence, load factor, and extension of the mechanisms via the level of connection of the
overturning wall with the orthogonal walls, so that any analysis of their relevance is
strictly linked to the level of connection assumed at the edges. If there is no connection
the only mechanism possible is the overturning of the whole facade (Type A). However,
when the edge connections are present, the collapse of the whole facade is prevented,
while the number of stories from the top that will overturn becomes a function of the
staggering ratio. In Figure 15a it can be observed that for increasing ratios s/h, the number of stories to collapse will be gradually smaller, while c also increases. The analysis
is carried out on a typical facade up to five stories high, with an interstory height of 3.00
m and an average thickness of 0.80 m. The resultant slenderness is 18.75. A friction coefficient =0.4 is used and live loads qrq f 1.5 KN/m are considered.
On the same diagram is also plotted the resultant behavior for mechanisms B1 and
B2, which involve either one or two party walls. In the first case, the increasing staggering ratio produces a translation of the curve in terms of load factor but only a slight
shift in terms of critical number of stories involved. In the second case, when both party
walls are involved, as the value of the ratio s/h increases, the load factor curve of B2 is

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 495

Table 1. Mechanisms and load factor equations for overturning failures

496

D. DAYALA AND E. SPERANZA

higher in the first part, corresponding to the upper stories of the facade. However, in this
case, the ratio s/h does not produce changes to the number of stories involved in the
collapse, as the load factors curve reaches its minimum in correspondence of the whole
facades height in both cases. For small value of the staggering ratio, the most critical
mechanism among A, B1, and B2, is B2 for any number of stories, while for high staggering ratios, mechanisms A and B1 show load factors smaller than B2 at the top floor.
It is also worth noticing that mechanisms D and E have very similar behavior for
both cases of staggering, while for s/h0.3, E has the smallest load factor corresponding to five stories mobilized, and in the case s/h1.2, D has the smaller load factor corresponding to two stories. Both mechanisms perform, in general, worse than mechanism
A, so that it can be expected that these types of collapse will be more common than the
whole overturning of the facade. Finally, it can be noted that while for low values of
staggering ratios, all overturning mechanisms for any number of stories have load factor
lower than 0.4, for higher values of staggering, most load factors, except for B1, B2, and
D, are greater than 0.4 and, hence, in these conditions, sliding will occur in preference to
overturning. For mechanism Type C involving the corner failure, the influence of change
in staggering ratio is rather modest. While for low values the worst load case is associated with the whole facade, for increasing values of s/h the load factor results independent from the number of stories involved, and for s/h1.6 the load factor curve shows a
minimum for n2.
The dependence of the load factor from the presence of quoins, i.e., effective connection between the facade and the party walls is highlighted in Figure 15b where curves
for four overturning mechanisms in absence, presence of one or two connections, are
plotted in terms of load factor versus number of stories involved in the mechanism.
Mechanisms B2 and C are independent from the presence of quoins as in both cases it is
assumed that the cracks develop within the framework of the walls and, hence, they have
not been included in the graph. In the absence of edge connections (=0), there is no
substantial difference between the four mechanisms and the lowest load factor is associated with the whole facade. The presence of one connection (=1) improves very
slightly the behavior of B1 and D, for which the worst case is still the whole facade,
while E shows an increase in load factor and a shift to three stories as worst case. The
best behavior is shown by case A for whom the two-story condition is the worst scenario.
When both edge connections are present (=2), the most critical mechanism becomes E
with the smallest load factor at the upper story, while the mechanism that most benefits
from the connections, and therefore is most unlikely to occur, is Type D.
INSERTION OF STRENGTHENING DEVICES: OUT-OF-PLANE
MECHANISMS BASED ON ARCH EFFECT
In order to prevent the occurrence of out-of-plane overturning of the types seen in
the previous section, metallic ties were traditionally used to anchor the facade to either
the party walls or the floor and roof structures. Their efficiency relies on a uniform distribution over the width of the wall and on the presence of anchorage plates, which reduce the shear stress at the interface with the masonry units and mortar. In the last decades, following the earthquake of Friuli and Montenegro, in southern Europe various
strengthening methods have been designed and tested both in laboratory (Tomazevic

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 497

Figure 15. Out-of-plane mechanisms: (a) influence of staggering ratio s/h, and (b) influence of
edge connections on load factor and surface extension.

1996, Benedetti et al. 1998) and by successive earthquakes (Spence and DAyala 1999)
aimed at preventing the out-of-plane drift of the upper portions of the walls. Besides
modern versions of the traditional iron tie, the performance of reinforced concrete or
reinforced masonry ring beams laid out at the top of the walls, under the roof eaves, or
at floor levels, so as to create a continuous hoop around the four external walls, has attracted particular interest. While from a theoretical point of view the continuous ring
beam represents a more efficient lateral constraint than the discreet insertion of ties, in
order to be effective it needs to be properly anchored to the masonry of the walls so as
to prevent relative sliding and rotation (Triantafillou et al. 1997).
In the present section two mechanisms are developed for the facade obtained by considering the presence of strengthening devices: the vertical arch mechanism and the
horizontal arch mechanism. The first one occurs when due to the presence of ties or ring
beams, vertical strips of the facade tend to deflect out of plane, being restrained at bottom and top (Figure 16). The analytical model simulates the presence of ties by identifying along the height of the buildings two locations for which lateral movements are
constrained. When the mechanism is triggered, a third hinge forms along the facade, in
an intermediate position to the first two and identified by a horizontal crack. As the cylindrical hinges simulate the containing action exerted by strengthening devices, it is assumed that the two extreme hinges are positioned at floors and roof levels only. The procedure iterates to find the position of these two and the intermediate one corresponding
to the lowest load factor. In the case of ring beams, it is assumed that these will exert a
rotational restraint for the wall cross section and, hence, a bending capacity for the cross
section is added to the arch effect capacity (see Table 2, equation associated with mechanism F): the bending capacity is represented by the second term at the second member
of the equation, in which in presence of ties the terms containing T j will be equal to 0.

498

D. DAYALA AND E. SPERANZA

Figure 16. Geometric parameters and applied load for mechanisms with arch effect.

The horizontal arch mechanism occurs when the facade span is rather wide, and internal bearing walls exert none or very little restraining action. In this case, the behavior
of the wall subjected to lateral action can be assimilated to a horizontal arch. The failure
is characterized by a central trapezoidal portion, which, under the effect of the lateral
action, tends to displace outwards from the facade plane rotating around the instantaneous centers of rotation 1 and 2 (Figure 16, mechanism G). In order to be triggered,
the mechanism requires the formation of one cylindrical hinge (C) along the vertical
symmetry axis of the facade and two horizontal sliding rollers (A, B), associated with
cylindrical diagonal hinges, placed at the bottom of the portion of wall under examination, at the level of the horizontal crack (see Figure 16). Because of the outward displacement of the central hinge, the edges A and B are forced to move along the horizontal. In order to be triggered, the mechanism must satisfy the condition according to
which the resistance to the outward displacement, developed by the edges of the facade,
is overcome. This resistance has been assumed as the force required to trigger the sliding
of the triangular edges along the bottom horizontal layer (Figure 16). It is assumed that
the thickness is constant along the height h j and that torsional effect of friction on the
cylindrical hinges is negligible.

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 499

Table 2. Mechanism and load factors for failures based on arch effect

For the mechanism Type F, the value of the load factor depends upon the parameter
governing the position of the intermediate hinge k, and this in turn is governed by the
presence and magnitude of imposed load q f and qr .
In Figure 17a the load factor is plotted against the position of the intermediate hinge,
for a wall of slenderness 18.75. The most salient feature is the stabilizing effect of vertical imposed load. Curves are compared for the case of walls of width 1 m and 5 m
without imposed load from horizontal structures, and with a load of 1 kN/m. In the case
of no imposed load the minimum load factor is associated with a position of the hinge
coinciding with the upper edge of the wall (1/=0), while when load is present the load
factor shows a minimum within the height of the wall ((1/=0.3) and a greater load factor (0.45 as opposed to 0.105). As it can be noted by comparing the curve with q f
0.4 kN/m and 1kN/m, the value of the load factor and the position of the hinge depends on the magnitude of the load and tends to move downward as the load increases.
The effect of intermediate stories is also clearly outlined, with a substantial step upwards
of the load factor, so that in reality the position of the hinge will be always confined
within the upper story. Figure 17b shows the variation of the position of the hinge for
increasing values of the magnitude of the roofs imposed load. This tends to the asymptote 0.45 for qr, as it can be proven by differentiation with respect to and limit of
equation F in Table 2. Curves in Figure 20 also show the influence of the staggering ratio
s/h (=0.3,1.2,2) and of the presence of quoins (=0, no quoins, =2, 2 quoins).
In mechanism G, apart from the direct proportionality of to , the most influential
parameter is the ratio L/H. Figure 18a highlights how the angle of the inclined crack i
(Figure 16), for which the minimum load factor is obtained, increases with the ratio L/H,
while the load factor decreases. For values L/H1.33, the minimum is associated with

500

D. DAYALA AND E. SPERANZA

Figure 17. Mechanism F: (a) load factor patterns in the function of the variable , governing
the position of the intermediate hinge k; and (b) position of hinge and minimum load factor as
a function of qr magnitude, s/h, and connection to party walls.

=90 degrees, indicating that the entire upper portion of the wall will overturn. In this
analysis the load associated with horizontal structures has been neglected.
The influence of the staggering ratios together with the presence of internal bearing
walls is outlined in Figure 18b, in which different curves of min are compared. As it can
be observed, in the absence of internal bearing walls, the increase of the staggering ratio
produces an improvement of the seismic performance only for ratios L/H smaller than 1.
Beyond this limit, the two curves are coincident. When an internal bearing wall is
present, the load factor curves tend to slide upwards, with greater increment for greater
values of s/h, as should be expected.
APPLICATION OF THE FAMIVE PROCEDURE
The analytical models presented in the previous section have been developed into a
numerical procedure, interfaced with an electronic form and database so as to obtain an
integrated package, FaMIVE (Failure Mechanism Identification and Vulnerability Evaluation) (DAyala and Speranza 2002), that allows the collection of data and online
build-up of a database on the state of conservation and structural appraisal of historic
buildings in earthquake-prone urban areas, with particular attention to the detection of
strengthening devices. The data is available for online evaluation of a facade or building
seismic vulnerability by application of limit-state analysis and collapse mechanisms approach. Using a pre-established set of decisional criteria, the choice of the most feasible
mechanism is arrived at and its associated collapse-load factor is weighed in terms of
type and extent of damage expected so as to yield a measure in terms of vulnerability.
The application presented uses data collected in four typical medieval hilltop towns
of the Marche region of Italy (Offida, Treia, Cagli, and Serra dei Conti), in the framework of a study aimed at defining their microzonation (Dolce and Speranza 2001). For

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 501

Figure 18. Mechanism G: (a) load factor patterns in the function of crack, for different ratios
L/H (Ti0.8, s/h1.2, =0, =0.4), and (b) load factors in the function of L/H for different
staggering ratios s/h, in presence and absence of one internal bearing wall.

each center a number of historic building typologies were identified and their seismic
vulnerability initially established according to the EMS 98 scale (Grunthal 1998). These
same typologies have been analyzed with FaMIVE. The purpose of the application is to
compare seismic capacity of different structural and material typologies within a homogeneous building tradition, and to analyze the effect of strengthening devices and alteration on their vulnerability level.
GENERAL URBAN AND SEISMIC FEATURES OF THE FOUR TOWNS
Offida (AP), Treia (MC), and Cagli (PU), are medium-sized pre-appeninic towns
with an average of 9,000 inhabitants, while Serra dei Conti is smaller with a population
of 3,000. Offida, Serra dei Conti, and Treia have rather similar urban form, at the top of
an elongated hill outcrop, with buildings arranged in circular parallel arrays, according
to the steepness of the slope. City walls, preserved to various extents, surround all three.
Cagli, set deeper into the Apennines, lies in the alluvial valley created by the Metauro
River. Offida was first classified as seismic in 1963, following an earthquake in 1943 of
intensity MCS 9, with the epicenter close to the town; the other three towns were listed
in the second category of the Seismic Italian Catalogue by Comuni in 1983. According
to the latest hazard map proposed by the GNDT-SSN (Albarello et al. 1999), the expected PGA with 10% probability of exceedance in 50 years, is between 0.150.25 for
the four towns; however, from the same source the maximum macroseismic intensity expected (MCS scale) are 9 for Cagli, 8 for Offida, and 7 for Treia and Serra dei Conti.
The 1997 Umbria-Marche earthquake did not produce serious damage in either of the
latter three, but was strongly felt in Cagli. A histogram of historic intensities felt in Cagli, Offida, and Treia is shown in Figure 19.

502

D. DAYALA AND E. SPERANZA

Figure 19. Seismic intensity from historic catalogues for three of the four sample towns (from
data by Monachesi and Stucchi 1997).
CONSTRUCTION TYPOLOGIES

In Offida and Serra dei Conti, the common construction method is brickwork: the
typical size of bricks is 32016060 mm with joints of 10 mm and variable level of
connection in the thickness of the wall, with two wythes of bricks and rubble infill observed in Offida, solid brickwork in Serra dei Conti. In Treia, the external wythe is commonly made of alternate layers of bricks and limestone blocks. The stones are put
through the wall thickness while the bricks are irregularly staggered. In Cagli, stonework, sometimes interlaid with brickwork, is the common building method, with an internal wythe of rubble stone. Offida shows the poorest level of conservation of the historic buildings, due to the massive introduction of reinforced concrete elements in the
masonry structures during reconstruction after the earthquake of 1943. During the fifties
and sixties there were also numerous vertical additions of one or two stories. In the other
three towns the conservation of traditional buildings is rather good. Following is a description of the most common building typology for each town, together with the main
type of alteration.
Serra Dei Conti
Buildings are usually arranged in regular terraced arrays with an original height of
two stories. A staircase placed at the center, between two slender walls, 150 mm thick,
characterizes the plan. Floors and roofs are generally made of timber, with beams parallel to the facade. The brickwork construction is rather accurate, with strong connections at the corners. The facades openings have a regular layout, with regular piers at the
edges. Building 8, in Figure 20, represents this type. The estimated class of vulnerability
for this typology, according to the EMS 98 scale is B. Major alterations to this type,
before 1983, are the addition of one story and the substitution of the original floors and
roof with reinforced concrete slabs; after 1983, seismic upgrading measures such as ring
beams and regular patterns of metallic ties. It is considered that when this occurs the
building vulnerability class is C1. A poorer typology, of three stories, characterized by
reduced width of the street front, with an irregular layout of opening and the staircase
running along the party wall, is also present. The brickwork is less regular and there are

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 503

Figure 20. Common typologies in the four towns.

no strengthening interventions. The associated EMS 98 class is A. Also in the higher


vulnerability class are a few buildings that show no connection of the street facade with
the party walls, as built between adjacent pre-existing urban cells.
Treia
Most of the dwellings within the historic center are placed on a steep slope, with up
to four or five stories on the main facade, and one or two at the back (Figure 20). Floors
and roofs are made of timber joists, spanning between party walls. The use of traditional
ties and anchors is rather common. This typology has been classified as EMS 98 vulnerability Class B-A. Buildings with poorer material framework, showing insertion of
stone layers, are also present and classified as A.
A less common typology also considered in this sample is characterized by flatter
foundation ground, with only one story difference between front and back and a larger

504

D. DAYALA AND E. SPERANZA

plan size. At the ground floor there are structural cross vaults, set against the hillside and
used as storage areas. Other stories have two internal bearing walls, one orthogonal one
parallel to the facade. The associated EMS 98 vulnerability is Class B. Strengthening
and alteration works have followed a pattern similar to the one described for Serra dei
Conti.
Cagli
Within the city walls there is a high percentage of well-built and well-preserved relatively large-plan isolated or semi-detached houses. Organized usually over three levels,
this typology has large street fronts, regularly divided with the staircase running at the
back on two internal bearing walls and supported on the landing by a spine wall parallel
to the facade. Original timber roofs and floors run orthogonal to the facade and traditional ties are common. External walls are made of good squared stonework solid
through the thickness. Some have recently been completely restored as well as strengthened. The associated EMS 98 class is C1 (Figure 20). Poorer dwellings organized in
terraces and arrays are two stories high with very short spans (34 m) and made of
rubble or poorly cut stone, with primary floor structure running parallel to the facade.
These show a rather high level of alteration to the original constructive technique and
the associated EMS 98 vulnerability class is A. A third typology is found outside the
city walls, consisting of well-built two-story terraces with spans of 67 m and a staircase
in the middle supported by two thin walls, 150 mm each, which cannot contribute to
restrain the facade. Often they present well-built latter vertical additions of one story
associated with the insertion of ties at the upper levels. They have been classified as
EMS 98 Class B.
Offida
The local destructive seismic event of 1943 damaged mainly internal bearing structures and horizontal structures. Timber roofs and floors were replaced by reinforced concrete ones. Terraces are here built along the slope rather than on level ground as seen
elsewhere. The original height is two to three stories, but buildings with four or five stories are rather common. Some have very narrow spans of 2 or 3 m with floor structures
running parallel to the facade and a staircase along the back wall (Figure 20); others
slightly larger have a narrow staircase along the party wall. The ground floor of uneven
height is used for storage and is sometime vaulted. The associated EMS 98 vulnerability is Class B. Many of these buildings, especially the taller ones display ties, concrete
floors, and concrete lintels over opening and vertical additions. Not uncommon is the
case of facades not tied back to party walls. These are considered EMS 98 Class A/B.
Within the terraces, buildings with larger span (78 m) are also present, usually three
stories high with a ground floor with cross vaults supported by large pillars and a bearing wall in the middle of the plan at the upper stories. The original floor structure is
timber, and many are reasonably preserved. They are classified as EMS 98 Class C1.
DISCUSSION OF RESULTS
Initially the basic typologies for each town were analyzed and compared. While the
geometric absolute dimensions might be different with a different number of stories,

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 505

Figure 21. Representation of vulnerability as a function of slenderness.

they have same horizontal structures and similar masonry characteristics. They all have
good connections between the facade and the party walls, so that mechanism Type A is
always prevented. The slenderness (height to thickness ratio) of the portions activated by
the mechanisms are all similar, between 8 and 13, and mechanisms activated are Type E
for Treia and Serra dei Conti, Type B2 for Cagli, and Type G for Offida. Values of
equivalent shear capacity (ESC) are very close for the first three, 0.25 to 0.29, while
Offida with the lowest slenderness also shows the highest ESC at 0.37. The results in
terms of vulnerability are ranked in four classes, from Low to Extreme, defined on the
basis of correlation to damage levels observed during the 1997 Umbria-Marche earthquake (DAyala 1999). According to this classification, the vulnerability for the four
samples is of class Medium, comparing well with Class B of EMS 98. Results are
shown in Figure 21. For the out-of-plane mechanism the procedure indicates whether
and to what extent the floor structure is involved in the failure, depending on whether the
facade is load bearing and the extent of the collapsed surface, and this parameter contributes to defining the assigned vulnerability class.
A similar analysis for buildings of the same typology as above, but refurbished before 1983, with the insertion of reinforced concrete floors without ring beams, shows
virtually no change in the type of mechanism and a very small percentage increase in the
respective ESC values (about 10%), which substantially does not change the vulnerability class. However, in one case where the upper portion of the wall is not restrained by
the floor structure and there is the added weight of a vertical addition, mechanism Type
G is triggered for a very low ESC0.10, highlighting the danger of structurally unsound
alterations. The vulnerability class attribution is Extreme.
Where ties or ring beams have been implemented, out-of-plane mechanisms are usually prevented and the critical mechanism becomes the in-plane mechanism Type H. The
value of ESC for this coincides with the friction coefficient, in this case chosen at 0.40.
In some of the slender buildings, however, mechanisms Type B2 or Type F will have

506

D. DAYALA AND E. SPERANZA

comparable values of ESC to the magnitude of the in-plane one. The vulnerability for
these buildings is Low, if the reliability of the information is high and they can be rightly
associated with Class C1 of the EMS 98. However, if there is no certainty about the
efficiency of the strengthening, although present, then the lower bound of the range of
values should be taken and this would indicate a Medium vulnerability.
Finally, buildings with poor connection between orthogonal walls or poor masonry
construction were analyzed. These are usually buildings that have undergone substantial
alteration, or that have been built at a latter stage between existing adjacent buildings, or
vernacular buildings with rubble masonry and generally poor workmanship. Results for
ESC are in the range 0.100.13 and the vulnerability class for these buildings is High.
This compares well to EMS 98 Class A.
CONCLUSIONS
In the seismic vulnerability analysis of historic centers it is possible to characterize
the seismic behavior of external walls and, hence, the propensity and type of damage, by
studying the masonry construction and typological layout and by identifying the alterations, nonseismic improvement, and seismic strengthening. This study, carried out on
the buildings of four medium-sized historic centers of the Marche region in Italy, has
proven that where the construction and craftsmanship have not been altered, original
buildings show a medium level of vulnerability, which given the level of hazard of those
centers would result in moderate risk and damage. The insertion of ring beams and ties
slightly reduces the vulnerability in ordinary (by slenderness) buildings. However, these
devices prove very efficient for buildings of greater slenderness that would otherwise be
intrinsically more vulnerable (see Figure 21). A relative minority of buildings show vertical addition or other alterations, such as the relocation of openings, that can seriously
impair the seismic capacity. Indeed, most of these buildings have proven to be highly
vulnerable. The typological analysis has also highlighted the crucial role played by internal load-bearing walls properly connected to the outer shell in preventing out-of-plane
mechanisms.
The vulnerability classes had been defined in a previous work based on observed
damage from one event to essentially two building typologies. However, the application
discussed here, to buildings of different masonry typology and presence of strengthening
devices, shows that they correlate well with the EMS 98 classes. Hence, the approach
shown is capable of providing for these a quantifiable performance range in terms of a
collapse-load factor, i.e., a direct correlation to physical seismic input parameters.
NOTATIONS

angle defining bricks aspect ratio

angle between the vertical and the crack

angle defining panel aspect ratio

depth of masonry unit

cohesion

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 507

Ctot

Total Shear Strength

mass density

height of masonry unit

hs

story height

total height of wall

length of masonry unit

length of wall

collapse-load factor for a given mechanism

min

minimum collapse-load factor for a given mechanism

collapse-load factor associated with overturning mechanism

collapse-load factor associated with sliding mechanism

friction coefficient

q f , qr uniformly distributed load representing dead and live load of floor and roof
structure, respectively
s

overlapping length of masonry units

thickness of wall

total mass of walls and horizontal structures of portion mobilized by mechanism


REFERENCES

Albarello, D., Bosi, V., Bramerini, F., Lucantoni, A., Naso, G., Peruzza, L., Rebez, A., Sabetta,
F., and Slejko, D., 1999. New seismic hazard maps of the Italian territory, SSN-GNDT,
Rome 1999, web site: http://www.serviziosismico.it/PROG/2000/cartepericolosita/
index.html.
Augusti, G., Benedetti, D., and Corsanego, A., 1985. Investigations on seismic vulnerability and
seismic risk in Italy, Structural Safety and Reliability: Proceedings of ICOSSAR 85, The 4th
International Conference on Structural Safety and Reliability. (Conference code 10768).
Augusti, G., Ciampoli, M., and Giovenale P., 2001. Seismic vulnerability of monumental buildings, Structural Safety 23, 253274.
Azevedo, J., Sincrain, G., and Lemos, J. V., 2000. Seismic behavior of blocky masonry structures, Earthquake Spectra 16 (2), 337365.
Baggio, C., and Trovalusci, P., 1998. Limit analysis for no-tension and frictional threedimensional discrete systems, Mech. Struct. Mach. 26 (3), 287304.
Benedetti, D., Benzoni, G., and Parisi, M., 1988. Seismic vulnerability and risk evaluation for
old urban nuclei, Earthquake Eng. Struct. Dyn. 17, 183201.
Benedetti, D., Carydis, P. G., and Pezzoli, P., 1998. Shaking table tests on 24 simple masonry
buildings, Earthquake Eng. Struct. Dyn. 27, 6790.
Bernardini, 1999. A qualitative and quantitative measures in seismic damage assessment and
forecasting of masonry buildings, in Seismic Damage to Masonry Buildings, Balkema, Rotterdam, pp. 169178.

508

D. DAYALA AND E. SPERANZA

Calvi, G. M., 1999. A displacement-based approach for vulnerability evaluation of classes of


buildings, Aviat. Week Space Technol. 3 (3), 411438.
Campos Costa, A., Sousa, M. L., and Oliveira, C. S., 1998. Seismic risk: Methods and application to Portugal, 11th European Conference of Earthquake Engineering, Balkema.
Casolo, S., 2000. Modeling the out-of plane seismic behavior of masonry walls by rigid element, Earthquake Eng. Struct. Dyn. 29, 17971813.
Ceradini, V., 1992. Modellazione, Sperimentazione per lo Studio della Struttura Muraria
Storica, Ph.D. Thesis in history and theory of structures, University of Rome La Sapienza.
Coburn, A., and Spence, R., 1992. Earthquake Protection, Wiley and Sons, New York.
Corsanego, A., Del Grosso, A., and Stura, D., 1986. Seismic vulnerability assessment for buildings: A critical review of current methodologies, Proceedings of the 8th World Conference on
Earthquake Engineering, Lisbon.
Coulomb, C. H., 1776. Essai sur une application de regles de maximis et minimis a quelques
problemes de statique relatifs a larchitecture, Memoires de Mathematique & de Physique,
presentes a` lAcademie Royale des Sciences par divers Savants.
DAyala, D., 1999. Correlation of seismic vulnerability and damages between classes of buildings: Churches and houses, in Seismic Damage to Masonry Buildings, Balkema, Rotterdam,
pp. 4158.
DAyala, D., Spence, R., Oliveira, C. S., and Pomonis, A., 1997. Earthquake loss estimation for
Europes historic town centers, Earthquake Spectra, Special Issue on Earthquake Loss Estimation 13 (4), 773793.
DAyala, D., and Speranza, E., 2002. An integrated procedure for the assessment of seismic
vulnerability of historic buildings, 12th European Conference on Earthquake Engineering,
London, Paper No. 561.
de Felice, G., and Giannini, R., 2001. Out-of-plane seismic resistance of masonry walls, J.
Earthquake Eng. 5 (2), 253271.
Doherty, K., Griffith, M. C., Lam, N., and Wilson, J., 2002. Displacement-based seismic analysis of out-of-plane bending of unreinforced masonry walls, Earthquake Eng. Struct. Dyn. 3,
833850.
Dolce, M., and Speranza, E., 2001. Calibrazione e controllo della vulnerabilita sismica a scala
urbana: Il caso di Serra dei Conti (AN). in Proceedings of the X International Congress
LIngegneria Sismica in ItaliaPotenza-Matera.
Earthquake Engineering Research Institute, 2002. World Housing Encyclopedia, http://
www.world-housing.net/.
Giuffre`, A., 1989. Mechanics of historical masonry and strengthening criteria, in XV Regional
Seminar on Earthquake Engineering, Ravello, Italy, 1823 September. pp. 6191.

Grunthal, G., ed., 1998. European Macroseimsic Scale 1998, Conseil de lEurope, Cahiers du
Centre Europeen de Geodynamique et de Seismologie, vol. 15, Luxembourg.
Heyman J., 1972. Coulombs Memoir on Statics: An Essay in the History of Civil Engineering,
University Press, Cambridge.
Karantoni, F., and Bouckovalas, G., 1997. Description and analysis of building damage due to
Pyrgos, Greece earthquake, Soil Dyn. Earthquake Eng. 16, 141150.
Monachesi, G., and Stucchi, M., 1997. DOM4.1, An intensity database of damaging earthquakes in the Italian area, http://emidius.mi.ingv.it/DOM/.
Sanchez-Silva, M., and Garcia, L., 2001. Earthquake damage assessment based on fuzzy logic
and neural networks, Earthquake Spectra 17 (1), 89112.

DEFINITION OF COLLAPSE MECHANISMS AND SEISMIC VULNERABILITY OF HISTORIC MASONRY BUILDINGS 509

Spence, R., and DAyala, D., 1999. The Umbria-Marche earthquake of September 1997. Preliminary structural assessment, Structural Engineering International 9 (3), 229233.
Triantafillou, T., and Fardis, N., 1997. Strengthening of historic masonry structures with composite materials, Earthquake Spectra 30, 486496496.
Tomazevic, M., Klemenc, I., and Lutman, M., 1999. Seismic behavior of masonry buildings
Lesson from the Bovec earthquake of April 12, 1998, 8th Canadian Conference of Earthquake Engineering, Vancouver, B.C.
Tomazevic, M., Lutman, M., and Weiss, P., 1996. Seismic upgrading of old brick masonry urban houses: tying of walls with steel ties, Earthquake Spectra 12 (3), 599622.

(Received 18 July 2002; accepted 19 March 2003)

You might also like