You are on page 1of 19

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2008; 37:467–485


Published online 9 November 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.770

Non-linear seismic analysis and vulnerability evaluation of


a masonry building by means of the SAP2000 V.10 code

Laurent Pasticier1, § , Claudio Amadio2, ‡ and Massimo Fragiacomo3, ∗, †, ‡


1 Via Schiaparelli 8, 34143 Trieste, Italy
2 Department of Civil & Environmental Engineering, University of Trieste, Piazzale Europa 1, 34121 Trieste, Italy
3 Department of Architecture & Planning, University of Sassari, Piazza Duomo 6, 07041 Alghero, Italy

SUMMARY
The aim of the paper is to explore the possibilities offered by SAP2000® v.10, a software package with
user-friendly interface widely used by practising engineers, for seismic analyses of masonry buildings.
The reliability of the code was first investigated by carrying out static push-over (SPO) analyses of two
walls, already analysed by other researchers using advanced programs. The equivalent frame modelling
was employed in all analyses carried out. The code was then used to investigate the seismic performance
of an existing two-storey building typical of the north-east of Italy, with the walls being made of roughly
squared stones. An SPO analysis was performed first on the most significant wall, followed by a number
of time-history analyses aimed to evaluate the dynamic push-over curves. Finally, the seismic fragility
curves were derived, considering the seismic input as a random variable. Copyright q 2007 John Wiley
& Sons, Ltd.

Received 14 November 2006; Revised 12 July 2007; Accepted 5 October 2007

KEY WORDS: equivalent frame; fragility curves; incremental dynamic analysis; masonry building;
push-over analysis; SAP2000

1. INTRODUCTION

The assessment of the seismic vulnerability of existing buildings is one of the priorities of many
European countries. A large number of old masonry buildings, often characterized by degradation,
and in some cases with significant historical value, are in fact located in earthquake-prone areas
with different levels of seismic hazard. A proper evaluation of the seismic risk in existing buildings

∗ Correspondence to: Massimo Fragiacomo, Department of Architecture & Planning, University of Sassari, Piazza
Duomo 6, 07041 Alghero, Italy.

E-mail: fragiacomo@uniss.it

Associate Professor.
§ Consulting Engineer.

Copyright q 2007 John Wiley & Sons, Ltd.


468 L. PASTICIER, C. AMADIO AND M. FRAGIACOMO

Figure 1. Failure mechanisms of a masonry pier: (a) rocking; (b) sliding shear;
and (c) diagonal shear cracking.

is a necessary step in order to recognize the most critical areas and assess the priorities of the
retrofit work. In order to achieve this result, a proper modelling of the masonry structure is needed.
Several models, with different theoretical approaches [1], have been developed to date. The finite
element models [2, 3], based on proper constitutive laws for the masonry components [4], allow an
accurate determination of the critical points in the structure including the failure mechanisms, but
are time consuming and require the use of expensive and complex software. Other simpler models
are based on ‘macromodel’ modelling, where the masonry building is divided into a number of
one- or two-dimensional ‘macroelements’ [5–14]. In most of the models based on two-dimensional
elements, the hypothesis of material with no tensile strength is assumed [15], which usually results
in a quite complex iterative process. Among the models using one-dimensional elements, the
POR method is well known and extensively used. Such a method assumes, in its original version
improved later [16, 17], that the structural collapse occurs because of a storey mechanism. The
failure is assumed to take place only in the piers, and no allowance for the possible damage of the
spandrel beams is made.
An improvement of the POR method is provided in the so-called ‘equivalent frame’ method,
which allows the user to carry out a global analysis of the building. In such a method, a higher
number of possible failure mechanisms occurring inside each macroelement, such as shear with
diagonal cracking, shear with sliding, and rocking (Figure 1), can be considered. In accordance
with the use of this approach, which was used for example in the SAM code [10–12] and in
comparative analyses between macromodels and finite element models [18], the spandrels and
the piers are regarded as elastic, their intersections are modelled as fully rigid, and the possible
mechanical non-linearity is concentrated in some well-defined cross-sections inside the elastic
parts. The use of this approach is allowed by the FEMA 356 [19], the new Italian Seismic Code
[20], and the latest draft of the European Code (Eurocode 8) [21]. Both Italian and European
Codes encourage the use of non-linear static push-over (SPO) analysis and require a control of the
spandrels, which is not possible using the POR method. The use of suitable programs is therefore
needed for the design of masonry buildings according to those regulations.
The aim of this paper is to explore the possibilities offered by a widespread, user-friendly and
well-known package for structural analysis such as SAP2000® v.10 [22] for seismic design of
masonry buildings. This code was already used for seismic analysis of masonry buildings using
both one- and two-dimensional elements [18, 23]. The reliability and limitations of the code were
first investigated by performing an SPO analysis of two multi-storey walls, already analysed by
other authors using advanced programs. Some non-linear static and dynamic analyses were then

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
NON-LINEAR SEISMIC ANALYSIS AND VULNERABILITY EVALUATION 469

carried out on a facade wall of a typical Italian masonry building using the ‘equivalent frame’
method. The fragility curves were then drawn by assuming the seismic input as a random variable.

2. THE PROPOSED MODELLING

The proposed modelling of the masonry building is based on the use of the equivalent frame
method. The SAP2000® v.10 package allows the user to account for the non-linear mechanical
behaviour of the material by introducing the following elements with lumped plasticity in the
equivalent frame:
• plastic hinges;
• non-linear links.
The plastic hinges were used in SPO analyses since they allow the user to accurately follow the
structural performance beyond the elastic limit at each step of the incremental analysis. The non-
linear links were instead used in time-history analyses since they allow the user to accurately define
the cyclic behaviour of the elements including a proper degradation rule [24]. The mechanical
properties of these non-linear elements were defined based on the possible failure mechanisms of
masonry macroelements shown in Figure 1 [25–28]. The adopted modelling is described in the
following sections.

2.1. Modelling of the non-linear behaviour for the SPO analysis


The standard force–displacement curve that can be implemented in the SAP2000 plastic hinges is
depicted in Figure 2(a) [24]. The masonry piers were modelled as elastoplastic (as also suggested
in [20]) with final brittle failure (Figure 2(b)) [11] by introducing two ‘rocking hinges’ at the end
of the deformable parts and one ‘shear hinge’ at mid-height (Figure 3(a)). A rigid-perfectly plastic
behaviour with final brittle failure was assumed for all these plastic hinges (Figure 2(c)).
The strength in terms of ultimate moment Mu is defined by Equation (1). As far as the shear
strength is concerned, according to the experimental test outcomes [28], it was decided to consider
two strength criteria. The first criterion (Equation (2)) is recommended in [20] for existing buildings.
This criterion, which refers to shear failure with diagonal cracking, was originally proposed by
Turnšek and Cacovic [29] and later modified by Turnšek and Sheppard [30]. The second criterion
(Equation (3)) refers to shear failure with sliding and is recommended in [20] for new buildings.
Although formulated differently, such a criterion is also recommended by the Eurocode [21, 31]:
 
0 D 2 t 0
Mu = 1− (1)
2 k fd

f 1.5 f v0d Dt 0
Vu = 1+ (2)
 1.5 f v0d
3 0
f v0d +
2 m
Vus = Dt (3)
3H0
1+ f v0d
D0

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
470 L. PASTICIER, C. AMADIO AND M. FRAGIACOMO

Figure 2. (a) Standard shape of the force vs displacement curve in SAP2000® v.10 for the
plastic hinge element [24]; (b) and (c): behaviour assumed, respectively, for the entire pier and
the correspondent plastic hinge; (d) and (e): behaviour assumed, respectively, for the entire
spandrel beam and the correspondent plastic hinge.

elastic part fully rigid part elastic part fully rigid part

Spandrel
Joint

Pier
Analysed
wall

"shear "rocking "shear hinge"


hinge" for hinge" for for the
0 1m 2m
(a) the piers the piers spandrels (b) "shear hinge" for the piers (c) Ground Floor

Figure 3. Analysed building: (a) plastic hinges’ location in the equivalent frame model of the wall facade
used for static analyses; (b) non-linear links’ location in the equivalent frame model of the wall facade
used for dynamic analyses; and (c) ground floor plan with the analysed facade wall (measures in metres).

where 0 is the mean vertical stress, D the pier width, t the pier thickness, k the coefficient taking
into account the vertical stress distribution at the compressed toe (a common assumption is an
equivalent rectangular stress block with k = 0.85), f d the design compression strength, f v0d the
design shear strength with no axial force;  (friction coefficient) = 0.4,  the coefficient related
to the pier geometrical ratio, H0 the effective pier height (distance of the cross-section in which
the strength criterion is applied from the point of zero bending moment), and m the safety factor
(assumed to be equal to 2). For the rocking hinges the strength is given by Equation (1), and
the ultimate rotation u corresponds to an ultimate lateral deflection u equal to 0.8% of the
deformable height of the pier, minus the elastic lateral deflection, as recommended in [20]. For
the shear hinge, the strength is given by the minimum value resulting from Equations (2) and (3).
The ultimate shear displacement u was assumed to be equal to 0.4% of the deformable height

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
NON-LINEAR SEISMIC ANALYSIS AND VULNERABILITY EVALUATION 471

of the pier, minus the elastic lateral deflection, as recommended in [20]. The failure was assumed
to have occurred in the pier when the first between the ultimate rotation u in the plastic hinge
and the ultimate shear displacement u in the shear hinge was attained. Since in SAP2000 it is
not possible to automatically control the total deflection of an entire macroelement if more than
one of its plastic hinges exceed the elastic limit, such a quantity was manually checked on every
macroelement at the end of each load step.
As far as the modelling of the spandrel beams is concerned, assuming the presence of a lintel
properly restrained at both supports, only one ‘shear hinge’ was introduced at mid-span (Figure
3(a)), with the shear strength Vu given by

Vu = ht f v0d (4)

where h is the spandrel depth, t the spandrel thickness, and f v0d the design shear strength with
no axial force. A brittle–elastic behaviour with residual strength after cracking equal to 14 th of the
maximum strength was assumed for the entire element, with no limit in deflection (Figure 2(d)
and (e)) [11].

2.2. Modelling of the non-linear behaviour for the time-history analyses


The ‘multilinear-plastic pivot’ non-linear link was used in the time-history analyses. This link
allows the user to reproduce the cyclic behaviour of the entire macroelement by defining the shape
of the hysteresis loop and the degradation of both strength and stiffness (Figure 4(a)) through a
proper choice of the mechanical parameters.
Owing to the complexity of the non-linear time-history analyses, in order to reduce the compu-
tational burden, only one ‘shear link’ was introduced at mid-height of the piers (Figure 3(b)). This
choice was suggested both by the outcomes of the SPO analyses presented in the next paragraph,
where the dominant failure mechanism was found to be shear in the pier, and by the geomet-
rical ratio (and therefore strength) of the spandrels relative to the piers. For the sake of safety, a
hysteretic behaviour characterized by shear failure with diagonal cracking was assumed. This type
of failure is, in fact, more fragile and, therefore, critical in terms of displacement demand than
sliding shear.

100
P4 P3 80
2 Fy2 Fy1=Fy2 A
base shear [kN]

B C V=30% 60
A 40
Fy1 1 Fy1= 2 Fy2 B 20
=0.45Fy1 0
1Fy1
-20 Magenes
-40 et al.
2Fy2 2Fy2= 1Fy1
-60 SAP2000
2Fy2 B' -80 v.10
A' =0.45Fy2 -100
V=30%
C' B' -8 -6 -4 -2 0 2 4 6 8
P2 P1
2Fy2
A' Fy2=Fy1 top displacement [mm]
(a) (b) (c)

Figure 4. (a) Standard hysteresis loop for the ‘multilinear-plastic pivot’ non-linear link in SAP2000® v10
[24]; (b) reference curve assumed for the non-linear link with the corresponding values of the parameters
for stiffness and strength degradation control; and (c) comparison between the experimental curve detected
with the quasi-static cyclic test performed on a pier by Anthoine et al. [26] and the numerical curve
detected with SAP2000® v.10 for the same pier using the multilinear-plastic pivot non-linear link with
the parameters indicated in (b).

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
472 L. PASTICIER, C. AMADIO AND M. FRAGIACOMO

The link parameters 1 , 2 , 1 , and 2 , which control the stiffness degradation during the
unloading procedure, were chosen so as to reproduce the experimental results obtained by Magenes
and Coworkers [26] on a single brick wall (Figure 4(b) and (c)). The parameters were all assumed to
be equal to 0.45. The same values were also adopted for the other stone masonry walls investigated
in this paper. The maximum shear strength and plastic displacements were assumed to be the same
as those used for the SPO analyses.

3. VALIDATION OF THE MODEL FOR SPO ANALYSES

In order to verify the reliability of the proposed modelling, two walls (designated as A and B
in Figure 5) of stone masonry buildings previously analysed in the ‘Catania Project’ [32] were
modelled with SAP2000 v.10. The ‘Catania Project’ was an extensive nationwide research project
focused on seismic performance of existing masonry buildings. In such a project, some laboratory
and in-situ tests were performed to characterize the mechanical properties of the masonry. In
addition, numerical modelling of the structural response was undertaken by a number of Italian
universities, each of them using a different advanced software package. The University of Pavia
used the SAM code, which is considered as an important reference for this work. Such a code, which
is based on the equivalent frame modelling, was previously validated on a number of experimental
tests providing satisfactory results [10, 25]. The pier walls are modelled using Equations (1) and
(3), whereas the shear strength with diagonal cracking is evaluated using a more suitable criterion
for regular brick masonry walls [28], which is different from Equation (2). For the spandrel
beams, Equations (1) and (4) are adopted. The Basilicata research group used a no tensile strength
macroelement model with crushing and shear failures [6, 7], while the Genoa research group used
a finite elements model with layer failures [3].
The mechanical properties used in the analyses were E (Young’s modulus) = 1500 N/mm2 ,
G (shear modulus) = 250 N/mm2 ,  (unit weight) = 1900 kg/m3 , f d (design compression strength)
= 2.4 N/mm2 , f v0d (design shear strength with no vertical stress) = 0.2 N/mm2 , and  (friction
coefficient) = 0.5. The lateral loads representing the seismic action were applied by assuming the
inverted triangular distribution. The weights of each floor and the ratios between the seismic force
on the floor and the base seismic shear are reported in Table I.
There is an important difference between the SAM and the SAP2000 v.10 programs that has
f
to be highlighted. The SAM program updates the strengths Mu , Vu and Vus of the plastic hinges
during the non-linear analysis if the quantities H0 and 0 (Equations (1)–(3)) change due to the
effect of the lateral loads. While the quantity H0 almost remains constant, the quantity 0 can
markedly change during the analysis when the lateral load rises. The SAP2000 v.10 code, however,
does not allow for the automatic update of the strengths during the analysis. In order to assess the
significance of this limitation, the two walls A and B were analysed by considering two different
0 distributions (No. 1 and No. 2) for the evaluation of the strengths of the plastic hinges. In
the distribution No. 1, the hinge strengths were calculated using, for 0 , the values read on the
structure at the step 0 of the analysis, considering only gravity loads and no lateral loads. In
the distribution No. 2, the hinge strengths were calculated using, for 0 , the values obtained by
applying the gravity load and increasing the lateral loads up to the attainment of the elastic limit of
the frame. Four independent SPO analyses were then carried out: two on the wall A with the two
stress distributions No. 1 and No. 2, denoted as SPO 1 and SPO 2, and the corresponding two on
the wall B, also denoted as SPO 1 and SPO 2. Details on the axial force, shear force, and bending

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
NON-LINEAR SEISMIC ANALYSIS AND VULNERABILITY EVALUATION 473

Figure 5. Validation: elevation of wall A (a) and wall B (b), with the equivalent frame modelling used in
SAM and the failure mechanisms detected by the same code [5, 32] (measures in metres).

Table I. Validation: seismic weights and distribution of the lateral forces at the different floors [3].
Wall A Wall B
Seismic weight Seismic force/base Seismic weight Seismic force/base
Floor Wi (kN) shear ratio Fi / Fi Wi (kN) shear ratio Fi / Fi
2nd 548.2 0.38 290.0 0.41
1st 1096.3 0.40 480.0 0.37
Ground 1277.0 0.22 626.0 0.22

moment distributions in the frames due to gravity only (No. 1) and both gravity and lateral load
(No. 2) are reported in Pasticier [33].

3.1. Numerical comparisons


The outcomes of the numerical comparisons are displayed in Figure 6(a) for wall A and Figure
6(b) for wall B. The failure mechanisms as detected by SAP 2000 in the SPO 1 and SPO 2
analyses are displayed in Figure 7 for wall A. Similar to the SAM method, both SPO 1 and SPO
2 analyses detected a storey mechanism at the second floor of wall A with the same value of
ultimate strength. Such an ultimate strength was higher than the strength obtained by the SAM
code, but lower than those obtained by the Basilicata and Genoa research groups. The same top
displacement was obtained in both SPO 1 and SPO 2 analyses. Such a value was close to that
detected by the Genoa research group and fairly different from those detected by the SAM and
Basilicata research group.

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
474 L. PASTICIER, C. AMADIO AND M. FRAGIACOMO

1800 800

1500 X
600
base shear [kN]

base shear [kN]


1200

900 SPO 1 400


SPO 2 SPO 1
600 SAM SPO 2
Genoa R.G. 200 SAM
300 Basilicata R.G. Genoa R.G.
Basilicata R.G.
0 0
0 10 20 30 40 0 10 20 30 40
(a) top displacement [mm] (b) top displacement [mm]

Figure 6. Validation: comparison between the push-over curve obtained with SAP2000® v.10 and those
obtained with the other codes for the walls A (a) and B (b).

Figure 7. Validation: deformed shape of the equivalent frame at the attainment of the ultimate deformation
in the first plastic hinge in the SPO 1 (a) and SPO 2 (b) analyses for wall A, with RO, rocking; SL,
sliding shear; DC, diagonal cracking shear, and underlining, attainment of the ultimate deformation.

The results of both SPO 1 and SPO 2 analyses were almost the same as those of SAM also for
wall B, with a storey mechanism occurring at the second floor. The top displacement was almost
the same as that detected by the Genoa research group, but different from that detected by the
Basilicata research group. The ultimate shear strength observed was different in both cases.

3.2. Discussion of the results


The proposed model in SAP 2000 led to strength evaluation close to that given by the SAM program
for both walls A and B. The ultimate displacement was similar for wall B but more different for
wall A. Such a difference is due to the change from compression to tension taking place during the
analysis in some piers (the circled ones in Figure 7). Such behaviour is not automatically detected
by SAP2000 v.10, which assumes pier strength independent of the axial force. Differently, the

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
NON-LINEAR SEISMIC ANALYSIS AND VULNERABILITY EVALUATION 475

SAM method updates the pier strength at each step of analysis. Since no tensile strength was
assumed for the masonry piers, the analysis with the SAM code ended well before the SAP 2000
model, leading to a far lower ultimate displacement capacity. The top displacement attained in the
SAP 2000 model when the axial force in the piers turned from compression to tension corresponds
to point X in Figure 6(a). This new value matches quite well with the one obtained as a result of
the analysis carried out with the SAM program.
The results of SPO 1 and SPO 2 analyses in terms of ultimate strength and displacement were
almost the same for both walls. In the cases of shear walls of usual geometry, in fact, the variation
of vertical stresses in the walls due to overturning moment caused by lateral load is fairly low. The
main limit of the SAP2000 v.10 model, which is the impossibility to update the strengths of the
piers based on the variation of axial force, seems therefore not to be so crucial in the SPO analysis
of ordinary masonry buildings. However, in terms of failure mechanisms, only the outcomes from
the SPO 2 analysis are close to the results obtained using the SAM program (see Figures 5 and 7).

4. THE ANALYSED BUILDING

The plan of the ground floor is displayed in Figure 3(c) for the analysed building. This is a
stone masonry house typical of the north-east of Italy. The vertical structure is a single layer non-
retrofitted masonry made of roughly squared sandstone, while the floors are composites of concrete
slabs and timber beams. In order to reduce the computational burden of the dynamic analyses
needed for the vulnerability assessment, only the facade wall was analysed using the proposed
SAP2000 v.10 model. The design values assumed for the mechanical properties are based on the
mean values measured in situ on a number of similar buildings located in the same area as the
analysed building: f d = 0.8 N/mm2 , f v0d = 0.032 N/mm2 , E = 1600 N/mm2 , and G = 640 N/mm2 .
Only the in-plane seismic performance of the wall was investigated, assuming that the wall was
effectively connected to the floors.

5. SPO ANALYSIS

The SPO curves were obtained for the wall using the same procedure described in Section 3.
Such curves were needed to identify the limit states considered in the evaluation of the fragility
curves. As recommended by recent codes of practice and regulations [20], the horizontal seismic
loads were applied adopting two different distributions (Table II): (i) proportional to the product
of the masses by the floor heights (inverted triangular distribution) and (ii) proportional to the
floor masses (uniform distribution). Two SPO analyses, denoted as ‘SPO 1’ and ‘SPO 2’ and
corresponding to the two different stress distributions in the wall described in Section 3, were
carried out on the two models for each load distribution, so that in total four analyses were carried
out. Both piers and spandrels were modelled as described in Section 2.1.

5.1. Discussion of the results


In the SPO 1 analysis with the inverted triangular distribution, the collapse occurred after a storey
mechanism was initiated at the first floor, while with the uniform distribution the mechanism
occurred at the ground floor (Figures 8(a), 9(a), and (b)). In the first case, all the piers of the weak

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
476 L. PASTICIER, C. AMADIO AND M. FRAGIACOMO

Table II. Analysed building—seismic weights and distribution of the lateral forces at the different floors.
Seismic force/base shear ratio at each floor Fi / Fi
Floor Seismic weight Wi (kN) Inverted triangular distribution Uniform distribution
1st 278.7 0.67 0.47
Ground 281.5 0.33 0.53

180 180
B
150 150
2

storey shear [kN]


base shear [kN]

A
120 3 120
1
2' 3'
90 90
SPO 1 triangular distribution 1'
60 SPO 2 triangular distribution 60
SPO 1 uniform distribution
30 SPO 2 uniform distribution 30

0 0
0 2 4 6 8 10 0 2 4 6 8 10
(a) top displacement [mm] (b) storey displacement [mm]

Figure 8. Analysed building: (a) SPO global curves obtained using SAP 2000 v.10, including the
points corresponding to the attainment of the global limit states (1, 2, 3) on the most conservative
curve (SPO 1 with inverted triangular distribution); (b) SPO curve of the weakest floor with the
points corresponding to the attainment of the local limit states (1 , 2 , 3 ), with: 1 , limited damage;
2 , significant damage; 3 , near collapse.

storey collapsed due to a sliding shear mechanism. In the second case, only one pier failed due
to diagonal cracking, leading to a strength loss higher than 20% of the maximum strength, which
defined the ultimate global displacement. Also in the SPO 2 analyses, the structure collapsed due
to a storey mechanism characterized by a type of failure similar to that occurred in the SPO 1
analyses.
By comparing the four SPO curves, it is clear that the minimum base shear strength was always
obtained when the triangular distribution of seismic forces is applied (Figure 8(a)). By using this
type of distribution, the ultimate strengths obtained in the two corresponding analyses were almost
the same. In terms of displacement capacity, all the analyses led to nearly the same value of
8.5 mm. This good agreement among the different analyses is very important since the damage
index representing the limit states will be defined in terms of displacement (interstorey drift, ISD).

5.2. Choice of the damage index and limit states


The assumed damage index was the ISD. The Draft No. 7 of the Eurocode 8, Part 3 [21] defines the
limit states corresponding to the achievement of the global displacement capacity of the structure
on the push-over curve. Three points were located on such a curve in terms of top displacement:
(i) yield point (Limit State of Limited Damage), (ii) 34 th of the ultimate top displacement capacity

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
NON-LINEAR SEISMIC ANALYSIS AND VULNERABILITY EVALUATION 477

Figure 9. Analysed building: deformed shape of the equivalent frame at the attainment of the ultimate
deformation in the first plastic hinge in SPO 1 analysis with inverted triangular distribution ((a)—point
A in Figure 8(a)) and with uniform distribution ((b)—point B in Figure 8(a)), with: RO, rocking; SL,
sliding shear; DC, diagonal cracking shear, and underlining, attainment of the ultimate deformation.

(Limit State of Significant Damage), and (iii) point corresponding to at least 20% reduction in the
peak strength (Limit State of Near Collapse).
Among the four SPO analyses carried out, the one characterized by the smallest ISD at collapse
was selected (SPO 1 with inverted triangular distribution). The three limit states were then located
on the global push-over curve as discussed above (points 1, 2, and 3 in Figure 8(a)). From these
values of total displacement, the corresponding limits in terms of ISD were evaluated, and the
local push-over curve for the weakest floor (the first floor) was drawn (points 1 , 2 , and 3 in
Figure 8(b)). For the building under study, the obtained ISD/storey height ratios were
Limit State of Limited Damage: ISD/ h = 0.007%;
Limit State of Significant Damage: ISD/ h = 0.2%;
Limit State of Near Collapse: ISD/ h = 0.3%.
Since the Limit State of Limited Damage was conservatively defined by the yielding of the very
first hinge in the most unfavourable of the four analyses, the corresponding displacement value is
quite low. Also the other two values are fairly low, mainly due to the inherent brittle behaviour of
the analysed stone masonry wall.

6. INCREMENTAL DYNAMIC ANALYSIS

An incremental dynamic analysis (IDA) consists of a series of non-linear time-history analyses,


each one carried out using the same seismic record but a different scale factor for the seismic
intensity [34, 35]. The peak ground acceleration (PGA) is properly scaled in each analysis, in order
to cover the entire range of the structural response, from the yield point to the collapse. A seismic
event can differ from another one in the frequency content, the energy content, the duration, the
number of passages through zero of the acceleration, etc., causing therefore different effects on
the same structure [35]. Fourteen different recorded earthquake ground motions were used in the

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
478 L. PASTICIER, C. AMADIO AND M. FRAGIACOMO

Table III. Characteristics of the selected earthquake ground motions.


Station of PGA IA PDH td Duration
Earthquake Country detection Date Component (g) (cm/s) (cm s) (s) (s)
Imperial-Valley U.S.A. El Centro 15/05/1940 SOOE 0.348 148 2.2 13 20
Mexico City Mexico Sct 27 19/09/1985 N–S 1.79 244 121.5 32 180
Bucharest Romania Incerc 04/03/1977 N–S 1.930 75 6.0 14 45
Friuli Italy Tolmezzo 06/05/1976 E–W 0.315 117 1.5 5 35
Friuli Italy Buia 15/09/1976 N–S 0.109 15 0.7 8 26
Tabas Iran Boshroych 16/09/1978 N79E 1.004 28 0.3 20 35
Campano Lucano Italy Irpinia, Calitri 23/11/1980 E–W 0.175 134 7.3 47 86
Umbro- Italy Assisi 26/09/1997 E–W 1.083 22 1.3 32 55
Marchigiano
Kocaeli Turkey Yesilkoy 17/08/1999 N–S 0.089 19 1.4 37 106
Caldiran Turkey Maku 24/11/1976 S–E 0.956 9 0.1 19 28
Gazli Uzbekistan Gazli 17/05/1976 E–W 0.720 495 1.3 7 13
Montenegro Montenegro Bar-S.O. 15/04/1979 E–W 0.363 303 4.4 19 48
Umbro- Italy Colfiorito 26/09/1997 E–W 2.968 54 0.6 5 44
Marchigiano
Thessalonika Greece Thessaloniki- 20/06/1978 E–W 1.431 6 0.0 22 28
City Hotel

140 El Centro
Mexico City
Bucharest
120 Friuli (Tolmezzo)
Friuli (Buia)
Tabas
100
Irpinia
Umbro-Marchigiano (Assisi)
Kocaeli
80
Sd [cm]

Caldiran
Gazli
60 Montenegro
Umbro-Marchigiano (Colfiorito)
Thessalonika
40

20

0
0 0.5 1 1.5 2 2.5 3 3.5 4
natural period [s]

Figure 10. Displacement spectra of the selected earthquake ground motions.

analyses. The record properties such as the PGA, arias intensity (IA ), destructive potential (PDH )
[36], time of significant damage (td ), and duration are displayed in Table III, and the corresponding
elastic displacement spectra are shown in Figure 10.

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
NON-LINEAR SEISMIC ANALYSIS AND VULNERABILITY EVALUATION 479

240

200 El Centro
Mexico City
Bucharest
160 Friuli (Tolmezzo)
base shear [kN]

Friuli (Buia)
Ta bas
120 Irpinia
Umbro-Marchigiano (Assisi)
Kocaeli
Caldiran
80
Gazli
Montenegro
Umbro-Marchigiano (Colfiorito)
40
Thessalonika
SPO 1 Triangular distribution
SPO 1 Uniform distribution
0
0 2 4 6 8 10 12
top displacement [mm]

Figure 11. Analysed building: comparison among the higher SPO curve, the lower SPO curve, and the
curves obtained with the IDAs.

As previously discussed in Section 2.2, the hysteresis loop used for all the links in the IDAs
was chosen by assuming the only failure mechanism of shear with diagonal cracking, using the
same strength values assumed for the ‘shear hinges’ in the SPO 1.

6.1. The followed procedure


For each earthquake ground motion, the proper PGA scale factors corresponding to the achievement
in the equivalent frame of the three limit states of limited damage, significant damage, and near
collapse were determined with the bisection method. A number of other PGA scale factors were
assumed in order to draw the whole incremental dynamic push-over curve. Such a curve joins
together the ‘base shear-top floor displacement’ points characterized by different PGA scale factors
for the same earthquake ground motion. The maximum base shear and the maximum top floor
displacement values are reported in the curve even if they do not occur simultaneously. The 14
obtained IDA curves are displayed in Figure 11 together with the two SPO curves characterized
by the largest and smallest strength values (SPO 1 with uniform distribution and SPO 1 with
triangular distribution, respectively).

6.2. Discussion of the results


Figure 11 clearly shows the strong dependence of the structural response on the seismic record used
as input. Figure 12(a) and (c) depicts the IDA curves for two of the 14 earthquake ground motions
(Kocaeli and Colfiorito, respectively) together with the three points 1 , 2 , and 3 corresponding to
the three limit states, and more points corresponding to different values of the PGA scale factor. It
can be observed that the same PGA of 0.30g would cause a very different top displacement and,

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
480 L. PASTICIER, C. AMADIO AND M. FRAGIACOMO

Kocaeli ground motion


200
2' 3'
0.28g
0.24g 0.30g 0.305g
160
0.22g
base shear [kN]

0.26g
0.18g
120
1'

80 Kocaeli
SPO 1 triangular distribution
40 SPO 1 uniform distribution

0
0 2 4 6 8 10
(a) top displacement [mm] (b)

Colfiorito ground motion


200 2'
0.37g 0.38g 3'
0.33g 0.35g
160 0.39g
0.30g
base shear [kN]

0.20g 0.32g
120 1'

80 Colfiorito
SPO 1 triangular distribution
40 SPO 1 uniform distribution

0
0 2 4 6 8 10
(c) top displacement [mm] (d)

Figure 12. Analysed building: (a) and (c): comparison between the IDA curves and the limit SPO
curves for the ground motions of Kocaeli and Colfiorito, respectively; (b) and (d): SAP2000 v.10
base shear vs top floor displacement curves for a PGA scale factor of 0.30 for the ground motion of
Kocaeli and Colfiorito, respectively.

hence, damage level in the two earthquake ground motions. The wall, in fact, would experience
high plastic deformations that would lead to the attainment of the limit state of significant damage
with the Kocaeli seismic record (Figure 12(b)). Conversely, the limit state of limited damage would
just be overcome, with little plastic deformation, with the Colfiorito record (Figure 12(d)).
Owing to the different shape of the seismic records, also the range of PGA values required to
lead the structure from the elastic (limited damage) to the near collapse limit markedly changes.
This is clearly shown in Figure 13(a), where it can be observed that such a range extends from
0.09g to 0.54g for the Thessalonika earthquake and from 0.11g to 0.39g for the Gazly earthquake,
while for the Bucharest and Mexico City ground motions it extends from 0.23g to 0.28g only.
This difference can be explained with some considerations on the shape of the acceleration
spectra of the seismic events (Figure 13(b)). The reduction in stiffness due to the plasticization
caused by the seismic actions leads to an increase in the natural period of the structure, which is

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
NON-LINEAR SEISMIC ANALYSIS AND VULNERABILITY EVALUATION 481

Thessalonika
Gazli
20
Umbro (Assisi)
Caldiran Mexico City
El Centro 15 Gazli
Tabas

S a [m/s2 ]
Friuli (Buia)
Friuli (Tolmezzo) 10
Kocaeli
Umbro (Colfiorito)
Campano
Montenegro 5
Bucharest
Mexico City
0
0.10g 0.15g 0.20g 0.25g 0.30g 0.35g 0.40g 0.45g 0.50g 0.55g 0 1 2 3 4
(a) acceleration (b) natural period [s]
Limited Significant Near Collapse
Damage Damage

Figure 13. (a) Summary of the PGA values necessary to reach the three limit states for the different seismic
events and (b) acceleration elastic spectra of the Mexico City and Gazli earthquake ground motions.

0.1 s in elastic phase. A seismic event with a higher spectral acceleration at low natural periods,
such as Gazly, can then become less destructive for higher periods, allowing the structure to
resist far beyond its elastic limit. The opposite would happen for shakings with lower spectral
acceleration for low natural periods, such as Mexico City. Another significant outcome is that the
IDA curves give higher base shears than those obtained by the SPO analysis. Possible reasons for
this outcome are [34] (i) the influence of the higher vibration modes (particularly the 2nd mode)
in the IDA curves, ignored in the SPO analysis and (ii) the use in the IDAs of recorded shakings,
which are very different from the artificial motions compatible with the design spectrum that would
generally lead to IDA curves inside the SPO curves.
In terms of collapse mechanisms, the results obtained with the IDAs are similar to those of the
SPO analyses with uniform distribution, since both analyses detect the weak storey at the ground
floor.

7. SEISMIC FRAGILITY CURVES

The seismic fragility of a structure is defined as the probability of reaching a defined limit state in
correspondence with a specific value of the chosen seismic intensity parameter. The fragility curves
were evaluated for the wall under study by considering the seismic record as the only uncertainty
parameter and, therefore, using the results of the IDAs. No allowance for the variability of the
mechanical properties of the masonry wall was made, since it is believed that the uncertainty
of the seismic record is far more important than the scatter in mechanical properties. Also, no
out-of-plane failure mechanisms were considered. In order to have an immediate comparison with
the PGA levels that identify the four seismic zones defined by the Italian seismic regulation [20],
the PGA was considered as the seismic intensity parameter. This choice was also suggested by
the too scattered distribution of PGA values that would have been obtained for the natural period
of the analysed wall under different earthquakes if other intensity parameters such as the spectral
displacement Sd or the spectral acceleration Sa had been used. Such a scatter of PGA values, in

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
482 L. PASTICIER, C. AMADIO AND M. FRAGIACOMO

100 100
cumulative probability [%]

cumulative probability [%]


80 80

60 60

Limited Damage
Limited Damage
40 Significant Damage 40 Significant Damage
Near Collapse Near Collapse
IV category

IV category

III category
III category

II category

II category

I category
I category

20 20

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
(a) scale factor for g (b) scale factor for g

Figure 14. Analysed building: fragility curves obtained, respectively, (a) as a result of the IDAs and (b)
by a linear regression of the IDA outcomes by assuming a lognormal distribution.

conjunction with the low deformation level of masonry walls, would lead to results with little or
no sense. For a same value of Sd or Sa , in fact, some earthquake ground motions would lead the
structure to the attainment of the elastic limit, while some others would lead far beyond the collapse
limit. Conversely, the use of the PGA as seismic intensity parameter, although characterized by
significant variation, led to consistent results (Figure 13(a)).
The cumulative probability curves corresponding to the three limit states are depicted in
Figure 14(a). They were drawn using the IDAs by evaluating, for each value of PGA, the percentage
of earthquake ground motions that reached the given limit state. Two ground motions, Bucharest
and Mexico City, were ignored since their peculiar features would lead to outcomes very different
from all the other shakings. The piecewise-linear curves in Figure 14(a) were then approximated
using a lognormal distribution (Figure 14(b)) [35, 37]. The procedure used is based on the repre-
sentation of the (PGA, normalized cumulative probability) points on a lognormal probabilistic
chart. By evaluating the linear regression curve, the parameters y and  y (lognormal mean and
lognormal standard deviation, respectively) were then derived and the fit cumulative distribution
curve was drawn.
From Figure 14 it can be observed that the significant damage and the near collapse curves
are very close to each other. This means that, apart from the seismic records of Bucharest and
Mexico City where the two limits would be almost the same, once the significant damage limit
state is reached, only small PGA increments are needed for reaching the near collapse limit state.
The fragility curves allow the designer to evaluate the seismic vulnerability of the analysed wall
by direct comparison with the design PGA values assumed by the Italian seismic regulation for
the four different seismic zones [20], all of those corresponding to a return period of 475 years.
The analysed wall does not suffer from damage for PGA values lower than 0.1g (fourth category
according to [20]), showing an elastic behaviour for all the 14 seismic records. Conversely, for a
PGA of 0.35g (first category) there is a 60% probability that the wall will collapse. The analysed
building therefore needs proper retrofit to reduce the seismic vulnerability when located in seismic

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
NON-LINEAR SEISMIC ANALYSIS AND VULNERABILITY EVALUATION 483

areas characterized by a design PGA of 0.35g. For a PGA of 0.25g (second category), the same
structure is significantly damaged but still exhibits some residual strength.

8. CONCLUDING REMARKS

The first aim of this work was to test the reliability of a widespread and simple software package,
such as SAP2000® , v.10 for performing SPO analyses on masonry buildings using the ‘equiv-
alent frame’ simplified modelling. The proposed modelling was validated on two walls of an
existing masonry building, already analysed with different advanced programs by other researchers.
A significant limitation of the SAP2000 v.10 modelling is the impossibility, during the SPO anal-
ysis, to take into account the possible influence on the structural global strength of the axial force
variation in the piers (stress 0 ). Based on the outcomes of some comparisons with numerical
results carried out using more advanced software, it was found that the proposed model in SAP
2000 v.10 can be used for push-over analyses of masonry walls of usual and regular geometry. In
this case the strengths of the piers can be evaluated using, for 0 , the values corresponding to the
step 0 of loading (only gravity load, no horizontal loads applied yet).
The second aim of this work was to investigate the seismic performance of a typical standard
masonry building located in the north-east of Italy. An SPO analysis and an incremental dynamic
analysis were carried out on the facade wall of the building. The IDA pointed out how sensitive the
structural response is to the type of earthquake ground motion assumed as input. Same PGAs in
different seismic records may lead to very different results in terms of displacement and strength
demands on the same wall. The range of PGA values that led the structure from the elastic to the
collapse limit markedly changed depending on the assumed seismic event. The seismic fragility
curves were then derived assuming the seismic event as the only uncertainty parameter. Based on
the obtained curves, there was a 60% probability for the analysed wall to reach the collapse in
seismic regions characterized by PGA = 0.35g, such as the first category areas according to the new
Italian seismic code. A proper seismic retrofit would therefore be required to reduce the seismic
vulnerability of the building under study when located in those areas. In second category areas
(PGA = 0.25g), the same structure was significantly damaged but still exhibited some residual
strength. Those conclusions are applicable to buildings with walls having similar geometrical and
mechanical properties to the analysed one, and where adequate connection to the floors is provided
so that no significant out-of-plane damage may occur.

ACKNOWLEDGEMENTS
The authors wish to thank Mr Davide Bolognini from the European Centre for Training and Research
in Earthquake Engineering (EUCENTRE) of Pavia (Italy) for the information provided on the use of
the SAM code. Associate Professor Natalino Gattesco from the Department of Architectural and Urban
Design, Faculty of Architecture, University of Trieste (Italy), is also acknowledged for his valuable advice.

REFERENCES
1. Lourenço PB. Computations on historic masonry structures. Progress in Structural Engineering and Materials
2002; 4(3):301–319.
2. Amadio C, Fragiacomo M. Seismic analysis of a historical stone-masonry industrial building by the Abaqus
code. European Earthquake Engineering 2003; 17(1):18–30.

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
484 L. PASTICIER, C. AMADIO AND M. FRAGIACOMO

3. Gambarotta L, Lagomarsino S. Damage models for the seismic response of brick masonry shear walls. Part II: the
continuum model and its applications. Earthquake Engineering and Structural Dynamics 1997; 26(4):441–462.
4. Lourenço PB, Rots JG, Blaauwendraad J. Continuum model for masonry: parameter estimation and validation.
Journal of Structural Engineering 1998; 124(6):642–652.
5. Calderoni B, Marone P, Pagano M. Models for static verification of masonry buildings in earthquake-prone areas.
Ingegneria Sismica 1987; 3:19–27 (in Italian).
6. Braga F, Liberatore D. A finite element for the analysis of the response of masonry buildings. Proceedings of
the 5th North American Masonry Conference, Urbana, 1990; 201–212.
7. Braga F, Liberatore D, Spera G. A computer program for the seismic analysis of complex masonry buildings.
Fourth International Symposium on Computer Methods in Structural Masonry, Pratolino (Fi), Pande GN, Middleton
J, Kralj B (eds). E&FN Spon: London, 1997; 309–316.
8. Brencich A, Lagomarsino S. A macro-elements dynamic model for masonry shear walls. STRUMAS IV—4th
International Symposium on Computer Methods in Structural Masonry, Pratolino (Fi), Pande GN, Middleton J,
Kralj B (eds). E&FN Spon: London, 1997; 67–75.
9. Brencich A, Gambarotta L, Lagomarsino S. A macroelement approach to the three-dimensional seismic analysis
of masonry buildings. Eleventh European Conference on Earthquake Engineering, Paris, Bisch P, Labbé P, Pecker
A (eds). Balkema: Rotterdam, 1998.
10. Magenes G, Della Fontana A. Simplified non-linear seismic analysis of masonry buildings. Proceedings of the
British Masonry Society 1998; 8:190–195.
11. Magenes G, Bolognini D, Braggio C (eds). Simplified Methods for Non-linear Seismic Analysis of Masonry
Buildings. CNR—National Group for Seismic Protection: Rome, 2000; 99 (in Italian).
12. Magenes GA. Method for pushover analysis in seismic assessment of masonry buildings. Twelfth World Conference
on Earthquake Engineering, Auckland, New Zealand, 2000. CD-ROM.
13. Sucuoglu H, Erberik A. Performance evaluation of a three-storey unreinforced masonry building during the 1992
Erzincan earthquake. Earthquake Engineering and Structural Dynamics 1997; 26(3):319–336.
14. Azevedo J, Eeri M, Sincraian G, Lemos JV. Seismic behaviour of blocky masonry structures. Earthquake Spectra
2000; 16(2):337–365.
15. D’Asdia P, Viskovic A. Seismic analysis of masonry buildings. Ingegneria Sismica 1994; 1:32–42 (in Italian).
16. Tomazevic M. The computer program POR. Report ZRMK, 1978.
17. Tomazevic M, Weiss P. A rational, experimentally based method for the verification of earthquake resistance of
masonry buildings. Fourth U.S. National Conference on Earthquake Engineering, Palm Springs, 1990; 349–359.
18. Salonikios T, Karakostas C, Lekidis V, Anthoine A. Comparative inelastic pushover analysis of masonry frames.
Engineering Structures 2003; 25(12):1515–1523.
19. Federal Emergency Management Agency. FEMA 356. Prestandard and Commentary for the Seismic Rehabilitation
of Buildings. Federal Emergency Management Agency: Washington, DC, November 2000.
20. Decree of the cabinet president No. 3274. Annex 2: provisions for design, seismic evaluation and retrofit of
buildings. Appendix No. 72 to The Italian Official Gazette, vol. 105, 20 March 2003 (in Italian).
21. CEN. Eurocode 8: Design of Structures for Earthquake Resistance. Part 3: Assessment and Retrofitting of
Buildings. DRAFT No. 7. Stage 49. June 2004.
22. CSI (Computers and Structures Inc.). SAP2000 v10 Integrated Finite Element Analysis and Design of Structures.
CSI, Berkeley, 2004.
23. Cardoso R, Lopes M, Bento R. Seismic evaluation of old masonry buildings. Part I: method description and
application to a case-study. Engineering Structures 2005; 27(14):2024–2035.
24. CSI (Computers and Structures Inc.). SAP2000 v10 Analysis Reference Manual. CSI, Berkeley, 2004.
25. Magenes G, Kingsley G, Calvi GM. Static testing of a full-scale, two storey masonry building: test procedure
and measured experimental response. Experimental and numerical investigation on a brick masonry building
prototype—Numerical prediction of the experiment. CNR-GNDT, Report 3.0, 1995.
26. Anthoine A, Magonette G, Magenes G. Shear-compression testing and analysis of brick masonry walls. Tenth
European Conference on European Engineering. Balkema: Rotterdam, 1995; 1657–1962.
27. Magenes G, Calvi GM. Shaking table tests on brick masonry walls. Tenth European Conference on European
Engineering. Balkema: Rotterdam, 1995.
28. Magenes G, Calvi GM. In-plane seismic response of brick masonry walls. Earthquake Engineering and Structural
Dynamics 1997; 26(11):1091–1112.
29. Turnšek V, Cacovic F. Some experimental results on the strength of brick masonry walls. Proceedings of the
2nd International Brick Masonry Conference, Stoke-on-Trent, 1971; 149–156.

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe
NON-LINEAR SEISMIC ANALYSIS AND VULNERABILITY EVALUATION 485

30. Turnšek V, Sheppard P. The shear and flexural resistance of masonry walls. Proceedings of the International
Research Conference on Earthquake Engineering, Skopje, 1980; 517–573.
31. CEN. Eurocode 6: Design of Masonry Structures. Part 1-1: General Rules for Buildings—Rules for Unreinforced
and Reinforced Masonry, March 1998.
32. Liberatore D. Catania Project: Investigation on the Seismic Response of Two Masonry Buildings. CNR—National
Group for Seismic Protection: Rome, 2000; 275 (in Italian).
33. Pasticier L. Nonlinear analysis of a masonry building in a seismic prone region. Graduation Thesis, University
of Trieste, 2005 (in Italian).
34. Vamvatsikos D, Cornell CA. Incremental dynamic analysis. Earthquake Engineering and Structural Dynamics
2002; 31(3):491–514.
35. Altug Erberik M, Elnashai AS. Fragility analysis of flat-slab structures. Engineering Structures 2004; 26(7):
937–948.
36. Saragoni G. Response spectra and earthquake destructiveness. Fourth U.S. National Conference on Earthquake
Engineering, Palm Springs, 1990; 35–43.
37. Piluso V, Rizzano G, Tolone I. Reliability of criteria for seismic design of steel moment-resisting frames.
Ingegneria Sismica 2003; 2:37–52 (in Italian).

Copyright q 2007 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:467–485
DOI: 10.1002/eqe

You might also like