Engineering Structures 36 (2012) 210–219

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Seismic assessment of a historical masonry tower with nonlinear static

and dynamic analyses tuned on ambient vibration tests
Angelo D’Ambrisi a, Valentina Mariani a,⇑, Marco Mezzi b
a
Dipartimento di Costruzioni e Restauro, Università di Firenze, Piazza Brunelleschi 6, 50121 Firenze, Italy
b
Dipartimento di Ingegneria Civile ed Ambientale, Università di Perugia, via G. Duranti 93, 06125 Perugia, Italy

a r t i c l e i n f o a b s t r a c t

Article history: The paper deals with the dynamic characterization and the evaluation of the seismic performances of his-
Received 14 April 2011 torical masonry towers and bell-towers that are particularly vulnerable with respect to seismic actions.
Revised 16 November 2011 The main purpose of the study is to investigate how the seismic assessment of such historical construc-
Accepted 2 December 2011
tions is influenced by the level of refinement of the adopted models and by the type of the performed
Available online 5 January 2012
analyses. The effectiveness of some provisions contained in the Italian guidelines for the seismic risk
assessment and mitigation of cultural heritage for this kind of structures is also evaluated. The study
Keywords:
is performed with reference to the medieval civic tower of Soncino (Cremona, Italy), that is representative
Historical masonry tower
Seismic assessment
of the considered class of masonry constructions. The dynamic characteristics and the mechanical prop-
Dynamic identification erties of the masonry tower are evaluated through a dynamic identification with ambient vibration tests.
Ambient vibration tests Analytical models of the tower characterized by different refinement levels are calibrated on the results
Pushover analysis of the performed dynamic identification and used to carry out nonlinear static and dynamic analyses. The
Nonlinear dynamic analysis performed analyses evidence the very conventional nature of the behaviour factor q for this kind of struc-
ture. Moreover from the analyses it results that the q values suggested by the Italian codes overestimate
the actual q values. The results obtained with the nonlinear dynamic analysis are in a good agreement
with those obtained with the nonlinear static analysis. However the nonlinear dynamic analysis provides
levels of displacement capacity lower than those resulting from the nonlinear static analysis and allows
to consider the significant effect of the higher frequencies on some of the response parameters, as the
base shear. The vulnerability of the tower is also assessed according to the Italian guidelines for the seis-
mic risk assessment and mitigation of cultural heritage using the results of the performed nonlinear static
analysis.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction seismic actions. An effective seismic assessment of such structures

The protection of historical constructions against seismic ac-
tions is of strategic importance in Italy considering the richness
of its architectural heritage. Most historical constructions are ma-
sonry structures that were not conceived to resist lateral forces, in-
deed the old design concepts mostly focused on the effects of
gravity loads and did not provide for adequate lateral resistance
and ductility.
Among the constructions characterizing the Italian architectural
heritage, masonry structures with a predominant vertical dimen-
sion, such as towers and bell-towers, are very common. The limited
ductility of the masonry combined with the slenderness of these
towers, that behave as a vertical cantilever fixed at the base, gen-
erally provides a rather brittle structural behaviour [1]. Therefore
these constructions are particularly vulnerable with respect to
⇑ Corresponding author.
E-mail address: valentina.mariani@unifi.it (V. Mariani).
can be achieved only through nonlinear static and dynamic analy-
ses once defined suitable finite element models.
Historical structures cannot be investigated through invasive
tests due to their preservation needs, in these cases the dynamic
identification represents a valid alternative tool to define accurate
numerical models. Moreover slender structures are particularly
suitable for this type of analysis due to their sensitivity to dynamic
vibrations, that produces more clear signals. For these reasons in
recent years many masonry towers, such as the San Nicola bell
tower in Valencia [2], the bell tower of the Monza cathedral [3],
the Hagia Sofia bell tower [4] and the Saint Andrea bell tower in
Venice [5] have been investigated through dynamic identification.
The results from the in situ campaigns allow to define reliable fi-
nite element models [6–9] that constitute an essential tool for
the definition of possible strengthening interventions, combining
both seismic retrofitting and preservation.
The main purpose of the present paper is to investigate how the
seismic assessment of historical masonry towers is influenced by

0141-0296/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2011.12.009
 A. D’Ambrisi et al. / Engineering Structures 36 (2012) 210–219 211

the level of refinement of the adopted models and by the type of

the performed analyses. The effectiveness of some provisions con-
tained in the Italian guidelines for the seismic risk assessment and
mitigation of cultural heritage [10] for this kind of structures is also
evaluated. The study is performed with reference to the medieval
civic tower of Soncino (Cremona, Italy), that is representative of
the considered class of masonry constructions.
The actual constraint conditions of the tower and the mechan-
ical properties of its constituting materials are defined using the
results of a dynamic identification based on the records from pre-
viously performed ambient vibration tests. The nonlinear static
and dynamic analyses are performed on a simplified numerical
model, made up of beam elements with plastic hinges at the end,
and on a more refined 3D nonlinear model. The obtained results al-
low to predict the seismic behaviour of the tower and to define
suitable strategies for the seismic performance enhancement of
the tower according to the provisions of the current Italian guide-
Fig. 2. Plan view of the tower.
lines for the seismic risk mitigation of cultural heritage [10].

2. The medieval civic tower of Soncino

2.1. Geometrical survey and historical overview
The medieval civic tower of Soncino (Cremona, Italy) is a histor-
ical masonry construction that characterizes the landscape of the
small town located in the northern Italy. The tower was erected
in the 12th century even if some historical sources date the first
phase of its construction at the end of 10th century [11]. Initially
the tower was 31.5 m high but in 1575 it was heightened up to
41.8 m, adding a bell chamber on its top. After the 1802 earthquake
the tower has undergone several interventions that gave it the
present shape (Fig. 1).
The tower has a square plant with a 6 m side; its brick masonry
walls have a thickness varying from 1.55 at the base to 0.98 m at
the top. This variation creates five thickness discontinuities along
the height in correspondence of the wooden floors. At the height
of 6.7 m there is a barrel masonry vault, while at the height of
35 m there is a reinforced concrete slab. The tower is adjacent to
relevant public constructions, including the town hall, up to the
height of 11 m. These constructions are constituted by masonry
walls in the east–west direction (X direction) and by frames with
masonry columns and concrete beams in the north–south direction
(Y direction), on the southern side only (Fig. 2). The adjacent con-
structions affect the level of constraint of the tower and can not be
neglected in its structural analysis.
Fig. 1. North–west view of the medieval civic tower of Soncino (Cremona, Italy).
The tower shows good preservation conditions, although it pre-
sents some superficial cracks and damages due to the aging, the
atmospheric exposition and the past earthquakes (Fig. 3). The ma-
sonry used to heighten the tower in 1575 shows a higher degrada-
tion with respect to the masonry of the original construction.
Although the tower does not have serious structural damages, its
preservation against potential future earthquakes is of primary
importance.
2.2. Seismic hazard
The area of Soncino is a seismic prone zone. Historical cata-
logues list a M = 5.7 earthquake occurred in the area on May
12th 1802: all historical sources indicate Soncino as the most
damaged locality, with an assigned MCS intensity of VIII. In the
occasion of this earthquake also the tower was damaged. Accord-
ing to the current Italian code [12] the site is characterized by
the hazard curve reported in Fig. 4 in terms of PGA at the bedrock
ag versus the return period TR. The PGA at the bedrock with
a 475 years return period is ag = 0.119 g. For the site effect an
amplification factor S = 1.8 corresponding to a soft subsoil condi-
tion can be assumed.
2.3. Experimental campaign and dynamic identification
The civic tower of Soncino has been the object of an experimen-
tal campaign performed by the Milano IDPA-CNR Institute based
on ambient vibrations tests [13,14]. The tests were performed
using four recording stations: one, TS0, placed at the base of the
tower and three, TS1, TS2 and TS3 placed along the tower at the
height of 16.5, 26.8 and 35.2 m, respectively (Fig. 5).
The aim of the performed structural identification was the eval-
uation of the Young modulus E of the masonry and of the con-
straint effect given by the adjacent constructions to the tower.
This evaluation is performed, as a first approximation, modeling
the tower with a cantilever beam having the same geometrical
characteristics of the tower and a constant Young modulus. Two
different hypotheses of restraint are considered: (1) the tower is
fixed at the base and completely free along the height, without
any interaction with the adjacent constructions; (2) the tower is
fully restrained up to the top of the adjacent constructions. The
matching of the obtained analytical modal shapes with the exper-
imental responses at the recording stations is controlled (Fig. 6). As
evidenced by the results reported in Fig. 6, the best fitting between
the analytical modal shape and the experimental evidence is
212 A. D’Ambrisi et al. / Engineering Structures 36 (2012) 210–219

Fig. 3. Crack pattern in the external façades of the tower.

Fig. 4. Seismic hazard curve at the site.

Fig. 5. Position along the tower of the recording stations.

obtained with the model that includes in a simplified way the

effect of the adjacent constructions.
The tower is made of different types of masonry realized during Fig. 6. Comparison of the obtained analytical modal shapes with the experimental
its different construction and restoration phases, as shown in Fig. 7. responses at the recording stations TS1, TS2 e TS3: (a) model (1); (b) model (2).
 A. D’Ambrisi et al. / Engineering Structures 36 (2012) 210–219 213

Fig. 7. Construction and restoration phases of the tower.

With reference to the construction phases two values of E should The values of the fundamental frequencies and the normalized
be estimated: one for the masonry of the original construction modal shapes at the recording stations in the X and the Y directions
(E1), the other for the masonry used to heighten the tower in obtained from the tests are:
1575 (E2). To this purpose the signals recorded during the test have 8 9
been analyzed in the frequency domain according to the procedure < 0:0659 >
> =
shortly described in the following. Xexp;1x ¼ 1:08Hz Uexp;1x ¼ 0:3881
>
: >
;
For each recording station and for each component of accelera- 1:0000
8 9
tion the power spectral density (PSD) function has been calculated.
< 0:0594 >
> =
The peak values of the PSD function identify the fundamental fre-
Xexp;1y ¼ 1:11Hz Uexp;1y ¼ 0:3854 ð3Þ
quencies of the structure. As an example Fig. 8 shows the PSD >
: >
;
curves of the accelerations recorded at the TS1 station in the X, Y 1:0000
and Z directions. The slight difference between the two fundamental frequencies,
The peak value of the PSD function at the frequency Xk allows to although minimal, evidences the higher constraint effect in the X
calculate the root mean square of the recorded acceleration  x2k
€ direction given by the masonry walls of the constructions adjacent
through the following relation: to the tower.
Considering the symmetry of the structure and the indepen-
€x2 ¼ PSD€x ðXk Þ  DX ð1Þ
k
2p dence of the two fundamental frequencies a 2D finite element
model of the tower has been implemented to identify the parame-
where DX = 0.0038 Hz is the resolution of the PSD curve. Approxi- ters E1 and E2. The identification process is based on an iterative
mating the first modal shape with a sine curve, the modal displace-
procedure that performs the modal analysis until the objective
ment dmod can be calculated as: function
pffiffiffi
dmod ¼ 2  €x2 ðtÞ ð2Þ
ðXexp  Xan Þ2
F¼ þ ð1  MACÞ2 ð4Þ
X2exp

is minimized [15]. In (4) Xexp and Xan are the experimental and the
analytical frequencies, respectively, while MAC is the modal assur-
ance criterion [16] defined as

ðfUTexp g  fUan gÞ2

MAC ¼ ð5Þ
ðfUTexp g  fUexp gÞ  ðfUTan g  fUan gÞ

The MAC index measures the level of correlation between the

Fig. 8. PSD curves of the accelerations recorded at the TS1 station in the X, Y and Z
directions.
experimental modal shape fUexp g and the analytical modal shape
fUan g .
The analyses are performed considering four different cases: (a)
the tower is fixed at the base and completely free along the height,
without any interaction with the adjacent constructions (free (F)
model); (b) the tower is horizontally constrained in correspon-
dence of the top of the adjacent constructions (horizontally con-
strained (HC) model); (c) the tower stiffness is fictitiously
214 A. D’Ambrisi et al. / Engineering Structures 36 (2012) 210–219
increased up to the top of the adjacent constructions to simulate
their constraint effect (incremented stiffness (IS) model); (d) the
tower is considered fully restrained up to the top of the adjacent
constructions (fully restrained (FR) model).
The results show a good agreement between the analytical and
the experimental frequencies in all considered cases. On the con-
trary, the analytical modal shapes evidence significant differences
with respect to the experimental results for the different consid-
ered models. These differences can be expressed through the root
mean square percentage error (RMSEP) [17] between the analytical
modal displacements and the experimental response at two con-
trol points corresponding to the two intermediate recording sta-
tions TS1 and TS2 (Fig. 5). As evidenced by Fig. 9, the best fitting
is obtained with the IS model. These results confirm how the effect
of the adjacent constructions is very significant and not negligible
in the structural analysis of the tower.
The structural identification led to a value of E1 = 1000 MPa and
of E2 = 15–30 MPa. The values obtained for E2 are not significant
from a physical point of view, but they evidence the degradation
of the masonry in the upper part of the tower and the bad quality
of its connection with the lower part. In the hypothesis that the
damaged masonry in the upper part of the tower be repaired, a sin-
gle value of E = 1000 MPa has been assigned in the following to the
whole structure.
The elastic modulus identified from ambient vibrations is a dy-
namic modulus; a ratio of 1.2 has been assumed between the dy-
namic and the static modulus. This assumption led to a value of
833 MPa for the static modulus. A reasonable value of the com-
pressive strength has been estimated from the identified elastic
modulus. Considering a parabolic-rectangular stress–strain rela-
tion for the masonry and taking into account the stress state due
to vertical loads, at which the test was performed, a relation be-
tween E and fc has been derived. The relation is characterized by
a value of the correlation factor E/fc equal to 750. This value falls
within the interval of values typically proposed in literature for
masonry [18]. Consequently, a value of 1.1 MPa has been estimated
for the masonry compressive strength. This value is noticeably low
and can be regarded as a lower limit for this kind of masonry. To
investigate the influence of the assumed value of the masonry
strength on the results of the seismic assessment, a sensitivity
analysis is performed adopting, besides the estimated value of
1.1 MPa, also the values of 2.2 and 3.3 MPa. The 2.2 MPa value is
in accordance with typical values reported in literature for histor-
ical masonry towers [19,8], while the 3.3 MPa value approximates
the average of the reference values range provided by the Italian
code [12] for the compressive strength of this kind of masonry
(2.40–4.00 MPa).
Fig. 9. Root mean square percentage error between the analytical modal displace-
ments and the experimental response at the control points.
3. Seismic analysis
3.1. Linear elastic analysis
A preliminary linear elastic analysis has been performed on a
FEM model of the tower with the computer program SAP2000
[20]. The constraint effect of the adjacent constructions on the
tower has been simulated utilizing springs in the FEM model.
The stiffness of the springs in the X direction has been assumed
equal to the shear stiffness of the walls in this direction, while
the stiffness of the springs in the Y direction has been assumed
equal to that evaluated with reference to shear-type models of
the frames in this direction. A response spectrum analysis has been
performed referring to a peak ground acceleration characterized by
a return period of 475 years according to the current Italian code
[12], that for existing buildings suggests a behaviour factor
q ¼ 2  au =a1 , where au =a1 is the overstrength factor. In the case
of lack of specific studies a value of au =a1 ¼ 1:5 leading to a behav-
iour factor q ¼ 3 can be assumed. Other indications about the value
of the behaviour factor are provided by the guidelines [10] that
suggests a value q = 3 for towers, reduced to 2.25 for irregular
structures in elevation or when the tower is adjacent to other
buildings, as in the present case.
To investigate the dissipative capacity of the tower and support
the assumption of such values of the behaviour factor, a pushover
analysis on a simplified numerical model made up of linear ele-
ments with plastic hinges at the ends has been performed. The
hinges have been introduced at the base of the tower, at the height
of 11 m where there is no more interaction with the adjacent con-
structions and in correspondence of the changes of thickness of the
masonry walls. The moment–curvature relation of these hinges
(Fig. 10) is characterized by a yielding moment My corresponding
to the decompression moment that produces zero stress at the ten-
sile edge of the section and by an ultimate moment Mu correspond-
ing to the masonry crashing in compression. The obtained
pushover curve (Fig. 11) shows a value of 1.65 for au =a1 that leads
to a strength reduction factor of 3.3. The response spectrum anal-
ysis has been then performed assuming the value q = 3 suggested
by the current Italian code [12].
The performed linear elastic analysis allowed to preliminarily
investigate the stress levels in the tower under the design seismic
action. In particular the stress level has been defined at the most
critical sections of the tower: at the base, at the height of 12 m
Fig. 10. Moment–curvature relation of the plastic hinges.
 A. D’Ambrisi et al. / Engineering Structures 36 (2012) 210–219
Fig. 11. Pushover curve obtained with the simplified model.
where the constraint effects of the adjacent constructions are neg-
ligible, and at the height of 19.1 m in correspondence of the first
reduction of thickness of the walls. The calculated stress levels
are generally compatible with the masonry compressive strength,
even if at the base of the tower the principal compressive stress
reaches the value of 1.06 MPa, that is very close to the masonry
compressive strength. These results have been obtained consider-
ing a behaviour factor q = 3, that is in the hypothesis that the tower
dissipates energy when structural damages occur. The actual
capacity of the tower has been then defined by performing a non-
linear analysis on a refined numerical model.
3.2. Nonlinear static analysis
The nonlinear seismic analyses of the masonry tower have been
performed on a refined three-dimensional model made of 3D-solid
elements using the computer code DIANA [21]. For the compres-
sive regime the Drucker-Prager failure surface with a compressive
strength of 1.1 MPa, a cohesion c ¼ 0:45MPa and a friction angle
/ ¼ 11 has been considered, while for the tensile regime a con-
stant tension cut-off with smeared cracking, a tensile strength of
0.1 MPa and a linear softening until the maximum tensile strain
of 1‰ have been considered. A pushover analyses has been first
performed, both in the X and the Y direction, applying along the
tower forces proportional to the masses for the displacements of
the normalized deformed shape in the considered direction. The
pushover analysis has been carried out up to the collapse, in corre-
spondence of which the structure shows an extended cracking and
crushing. Fig. 12a and b show the principal stress–strain relation in
the section at the height of 3 m at the compression and the tension
side, respectively; the reported diagrams evidence that the princi-
pal stresses reach the strength limit of the masonry. They are also
representative of the constitutive behaviour of the implemented
material, that shows a hardening in compression (Fig. 12a) and a
linear softening in tension, approximately reaching the value of
the limit strain (1‰) (Fig. 12b). Fig. 13a reports the vertical stress
in the east façade of the tower at the collapse load step, while
Fig. 13b and c show the crushing and cracking patterns in the crit-
ical zones at the base of this façade for the same load step.
The performed analysis has allowed to define the pushover
curve of the tower. The curve has been then compared with the
one obtained from the simplified initial analysis on the model
including plastic hinges (Fig. 14). The comparison shows a good
correlation between the two curves in spite of the different models
adopted. Each pushover curve has been then processed to obtain
the curve of the equivalent SDOF system finally transformed in a
bilinear curve (Fig. 15).
215
Fig. 12. Principal stress–strain relation at the (a) compression and (b) tension side
of the horizontal section at the height of 3 m.
The bilinear capacity curve evidences a value of the behaviour
factor q, defined as the ratio between the maximum displacement
dmax and the yielding displacement dy, equal to 1.38. The pushover
curve of the simplified numerical model leads instead to a value of
the behaviour factor q equal to 1.74 (Fig. 11). These values of q re-
sult lower than those suggested by the Italian code [12] and by the
guidelines [10], equal to 3 and 2.8, respectively, underlining the
uncertainty in its definition and its conventional nature.
To evaluate the influence of the adopted value of the masonry
strength on the results of the seismic assessment, a sensitivity
analysis has been performed considering two other values of the
compressive strength, 2.2 and 3.3 MPa, according to the consider-
ations reported in paragraph 2.3. Fig. 16 reports the capacity curves
for the three considered values of the compressive strength, while
the corresponding bilinear curves of the equivalent SDOF system
are reported in Fig. 17. Fig. 16 shows that the increase of the com-
pressive strength determines a non proportional increase of the
structure strength and a reduction of the ultimate displacement.
The assumed strength values of 1.1, 2.2 and 3.3 MPa, yield to val-
ues of the ultimate base shear equal to 1204, 1428 and 1481 kN,
and to values of the ultimate displacement equal to 211, 196 and
195 mm, respectively. From Fig. 17 it is evident a reduction of
the structure ductility with the increase of the compressive
strength, indeed the behaviour factor q is equal to 1.38 for
fc = 1.1 MPa, to 1.10 for fc = 2.2 MPa, and to 1.05 for fc = 3.3 MPa.
3.3. Nonlinear dynamic analysis
Successively an incremental nonlinear dynamic analysis has
been carried out applying at the tower base the El Centro recorded
accelerogram of the 1940 Imperial Valley earthquake. The analysis
has been performed varying the peak acceleration from a value of
216 A. D’Ambrisi et al. / Engineering Structures 36 (2012) 210–219

Fig. 13. (a) Vertical stresses in the east façade at the collapse load step; (b) crushing and (c) cracking patterns in the critical zones at the base of the east façade at the collapse
load step.

Fig. 14. Comparison between the capacity curves of the nonlinear material model Fig. 15. Capacity curves and bilinear curves of the equivalent SDOF system for the
and of the plastic hinges model. nonlinear material model and the plastic hinges model.

0.05g up to the value determining the collapse of the structure.
Fig. 18a and b report the obtained top displacement and base shear
time histories, respectively, for the different peak ground accelera-
tions, that evidence how the base shear oscillates at a different fre-
quency with respect to the top displacement. In particular the
qualitative comparison between the two curves (Fig. 18c) shows
that, with the exclusion of the initial transitory, the base shear
oscillates at two frequencies: a principal frequency corresponding
to the first fundamental frequency of the structure (the same at
which the top displacement oscillates), and a secondary frequency,
corresponding to the frequency of the second vibration mode of
the structure. Therefore while the top displacement is dominated
by the first modal shape, the base shear is also conditioned by
the effects of the higher modes. The base shear-top displacement
curves are very irregular (Fig. 19) since the sign inversions of the
base shear are not followed by the top displacement; this effect be-
comes less relevant after the initial transitory. All the base shear-
top displacement curves have as a backbone the static pushover
curve (Fig. 19), evidencing a good correlation between the results
obtained with the nonlinear static analysis and those obtained
 A. D’Ambrisi et al. / Engineering Structures 36 (2012) 210–219 217

Fig. 16. Capacity curves for the three considered values of the compressive
strength.

Fig. 17. Bilinear capacity curves of the equivalent SDOF system for the three
considered values of the compressive strength.

with the nonlinear dynamic analysis. However the nonlinear dy-

namic analysis provides levels of displacement capacity lower than
those resulting from the nonlinear static analysis. This gap evi-
dences how the nonlinear static analysis is unable to capture the
inertial effects associated with a dynamic input that can lead to
an early collapse of the structure. This is an important outcome be-
cause generally it is expected that more simplified models and
analysis should lead to more conservative results.
Fig. 18. (a) Top displacement time history, (b) base shear time history and (c)
relative qualitative comparison.
4. Assessment of the seismic vulnerability

The definition of a standard procedure for the evaluation of the

vulnerability of historical constructions is a difficult task, because
of the complexity and the large variety of the structural systems
and the types of masonry. Moreover there are many economical
and cultural implications that can not be neglected. Nevertheless
it is necessary to define guidelines for the vulnerability assessment
and the retrofitting interventions of these structures. For this pur-
pose in recent years the Italian guidelines for the seismic risk
assessment and mitigation of the cultural heritage [10] have been
published. They contain suggestions for the seismic vulnerability
assessment of masonry historical constructions and for the seismic
enhancement interventions conceived as an upgrading of the seis-
mic performance respecting the preservation requirements.

4.1. Italian guidelines

The Italian guidelines for the seismic risk assessment and miti-
gation of the cultural heritage [10] define three different limit Fig. 19. Base shear vs. top displacement curves and pushover curve.
218 A. D’Ambrisi et al. / Engineering Structures 36 (2012) 210–219
states: the ultimate limit state (ULS), the damage limit state (DLS)
and the artistic damage limit state (ALS). The ULS is referred to the
collapse of the structure, the DLS is referred to the serviceability
loss of the structure, while the ALS has to be considered when
the structure contains artistic heritage to be protected. In the pres-
ent case the tower has been verified with reference to the ULS.
Moreover the guidelines define three different approach levels
for the seismic vulnerability assessment of historical construc-
tions: LV1, LV2 and LV3. The LV1 and LV2 levels are respectively
used for territorial-scale seismic evaluation and for local interven-
tion on individual constructions. The LV3 level, considered in the
present study, is based on the use of numerical models that simu-
late the global structural behaviour of the construction, leading to
an accurate assessment of its seismic vulnerability.According to
the guidelines provisions the values of the material properties have
to be reduced through the confidence factor FC
X
4 (Fig. 20). The blue line in Fig. 20 represents the elastic ULS demand
FC ¼ 1 þ F Ck
k¼1
The four parameters FCk in (6) are partial confidence factors that
take into account the accuracy of the data concerning the following
aspects: geometric survey (FC1), material types and structural de-
tails (FC2), mechanical characteristics of the materials (FC3), subsoil
and foundation characteristics (FC4). In this case it is assumed
FC1 = FC2 = 0 because of the exhaustive survey of the geometrical
characteristics and of the structural details. The values of the
mechanical properties of the masonry were obtained through the
structural identification. Despite these values were not supported
by the results of destructive tests, not performed to preserve the
integrity of the monument, FC3 is assumed equal to zero in that
the assumed values of the mechanical parameters are lower than
the reference values suggested by the current Italian code [12].
The FC4 factor depends on the influence that the foundation system
and the geotechnical characteristics can have on the collapse phe-
nomena. Since the tower does not show signs of foundation settle-
ments or problems connected with the foundation system, FC4 is
assumed equal to zero. With this assumptions a global confidence
factor Fc = 1 is obtained from (6). The guidelines [10] define the ULS
earthquake as the event having a predefined exceedance probabil-
ity in 50 years. This probability is defined taking into account the
nominal lifetime and the class of use of the considered construc-
tion. For the tower of Soncino a nominal lifetime of 50 years and
a class of use ‘‘frequent’’ have been assumed, considering the pres-
ence of the adjacent public constructions. These assumptions lead
to an ULS seismic action characterized by a reference peak ground
acceleration on the bedrock ag = 0.119 g with a 10% exceedance
probability in 50 years, corresponding to a 475 years return period.
The seismic assessment is based on the evaluation of the seis-
mic safety index ISS, defined as the ratio between the ULS peak
ground acceleration aULS obtained from the numerical model, re-
duced by the subsoil amplification factor S, and the reference peak
ground acceleration on the bedrock ag
aULS
ISS ¼
S  ag
ISS quantifies the deficiency of the seismic capacity of the struc-
ture with respect to the seismic demand.
4.2. Verification of the ultimate limit state
The verification of the ULS is performed utilizing the pushover
curve obtained with the model implemented under the DIANA
computer code [21] (Fig. 15), considering the better accuracy of
this model. The capacity curve of the equivalent SDOF system,
transformed in a bilinear curve for the ULS verification, is reported
Fig. 20. Verification of the ULS in the ADRS plane.
in the acceleration–displacement response spectrum (ADRS) plane
ð6Þ spectrum characterized by the design peak ground acceleration
Sag = 0.214 g, being ag = 0.119 g and S = 1.8. The green line is the
reduced elastic demand spectrum consistent with the structural
capacity that provides a value of aULS = 0.188 g. These values yield
to a seismic safety index ISS = 0.87, that clearly evidences the
inability of the structure to resist the design seismic action. Appro-
priate measures to improve the seismic behaviour of the tower and
to ensure its preservation should be therefore adopted, according
to the provisions of the current Italian guidelines for the seismic
risk mitigation of the cultural heritage [10].
5. Conclusions
The performed study evidences the fundamental importance in
the seismic assessment of historical masonry towers of the dy-
namic identification with ambient vibration tests. This procedure
allows to estimate the material properties and to define the actual
constraint conditions of these kind of structures without perform-
ing destructive tests.
The analyses performed on analytical models characterized by
different levels of refinement show that different modeling ap-
proaches can lead to different values of the behavior factor q. This
evidences the very conventional nature of the behaviour factor q
for this kind of structures.
Moreover from the analyses it results that the q values sug-
gested by the Italian codes overestimate the actual q values by a
factor ranging between 1.3 and 2.2. Therefore in the case of ma-
sonry towers the code provisions overestimate both the actual over
strength and the displacement ductility.
The results obtained with the nonlinear dynamic analysis are in
a good agreement with those obtained with the nonlinear static
analysis, indeed the obtained base shear-top displacement curves
derived from dynamic analysis have as a backbone the static push-
over curve. However the nonlinear dynamic analysis provides lev-
els of displacement capacity lower than those resulting from the
nonlinear static analysis. In this case, therefore, the more simple
ð7Þ static analysis does not lead to more conservative results.
Moreover the dynamic analyses have evidenced that while the
top displacement time-history is dominated by the fundamental
oscillation mode, the base shear time-history is also significantly
influenced by the higher frequencies.
The performed study underlines the importance of the nonlin-
ear dynamic analysis with refined models in the evaluation of
the actual seismic capacity of a masonry structure, also when the
structure is characterized by a simple static configuration, as the
analyzed slender tower. The more accurate seismic assessment al-
lows to identify more suitable design strategies to increase the
seismic capacity of the considered structure.
 A. D’Ambrisi et al. / Engineering Structures 36 (2012) 210–219
Acknowledgments
The authors thank Alberto Dusi and Elena Manzoni of Numeria
s.r.l of Cremona (IT), Giovanni Rossi of the municipality of Soncino,
Alberto Marcellini, Alberto Tento and Rossella Daminelli of IDPA-
CNR of Milano (IT) for making available the data of the tower of
Soncino (Cremona, Italy) used in the present study.
References
[1] Abruzzese D, Miccoli L, Ferraioli M, Mandara A, Froncillo S. Dynamic
Investigations on Medieval Masonry Towers: Sesmic Resistance and
Strenghtening Techniques. Proceedings of the International Conference on
Protection of Historical Buildings (PROHITEC). Rome; 2009.
[2] Ivorra S, Pallarés FJ. Dynamic Investigations on a Masonry Bell Tower. Eng
Struct 2006;28:660–7.
[3] Gentile C, Saisi A. Ambient Vibration Testing of Historical Masonry Tower for
Structural Identification and Damage Assessment. Constr Build Mater
2007;21:1311–21.
[4] Bayraktar A, Türker T, Sevim B, Altunisßik AC, Yildirim F. Modal Parameter
Identification of Hagia Sophia Bell-Tower via Ambient Vibration Test. J
Nondest Eval 2009;28:37–47.
[5] Russo G, Bergamo O, Damiani L, Lugato D. Experimental analysis of the ‘‘Saint
Andrea’’ masonry bell tower in venice. a new method for the determination of
‘‘tower global Young’s Modulus E’’. Eng Struct 2010;32:353–60.
[6] Ceroni F, Pecce M, Manfredi G. Seismic assessment of the bell tower of Santa
Maria del Carmine: problems and solutions. J Earthquake Eng
2010;14(1):30–56.
[7] Bayraktar A, Sahin A, Mehmet Ozcan D, Yildirim F. Numerical damage
assessment of Hagia Sophia bell tower by nonlinear FE modeling. Appl Math
Model 2010;34:92–121.
219
[8] Carpinteri A, Invernizzi S, Lacidogna G. In situ damage assessment and
nonlinear modelling of a historical masonry tower. Eng Struct
2005;27:387–95.
[9] Ivorra S, Pallarés FJ, Adam JM. Experimental and numerical results from the
seismic study of a masonry bell tower. Adv Struct Eng 2009;12(2):287–93.
[10] DPCM. Valutazione e riduzione del rischio sismico del patrimonio culturale
con riferimento alle NTC 14 gennaio 2008. DPCM 9 febbraio 2011, G.U.R.I.
February 26th 2011, Roma, Italy.
[11] Galantino F. Storia di Soncino con note documentarie. Milano; 1869.
[12] NTC. Norme tecniche per le costruzioni. D.M. Ministero Infrastrutture e
Trasporti 14 gennaio 2008, G.U.R.I. February 4th 2008, Roma, Italy.
[13] Dusi A, Manzoni E, Mezzi M, Rossi G. Estimation of the seismic response of civil
engineering structures using ambient excitations: a case study. 10th Int. Conf.
Inspection, Appraisal, Repairs & Maintenance of Structures. Hong Kong; 2006.
[14] Dusi A, Manzoni E, Marcellini A, Tento A, Daminelli R, Mezzi M. Seismic
assessment of structures by ambient vibrations: an application to a Medieval
Tower. 8th PCEE. Singapore; 2007.
[15] Friswell MI, Mottershead JE. Finite element model updating in structural
dynamics. Dordrecht, The Netherlands: Kluwer Academic Publishers; 1995.
[16] Allemang RJ, Brown DL. A correlation coefficient for modal vector analysis.
Proceedings of the 1st International Modal Analysis Conference. Orlando USA;
1982.
[17] DeGroot MH. Probability and statistics. Addison-Wesley; 1980.
[18] Hendry AW. Structural Brickwork. London: Macmillan; 1981.
[19] Binda L, Tiraboschi C. Flat-jack test as a slightly destructive technique for the
diagnosis of brick and stone masonry structures. Int J Res Build Monu
1999:449–72.
[20] SAP2000 Advanced 14.0.0, Structural analysis program. Analysis reference
manual. Computer and Structures, Inc. Berkley, California, USA; 2009.
[21] DIANA finite elements analysis. User’s manual – Release 9.3 – Material Library.
TNO Building and Construction Research, Department of Computational
Mechanics, A.A. Delft, NL; 2008.