Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
DOI 10.1007/s10518-013-9498-4
M. Badalassi A. Braconi
S. Caprili W. Salvatore
Received: 23 February 2013 / Accepted: 23 July 2013 / Published online: 15 August 2013
Springer Science+Business Media Dordrecht 2013
Abstract Among the resisting systems suitable for the design of ductile steel structures,
Eurocode 8 proposes MRFs and EBFs. The formers are considered more efficient in terms
of ductility, but they suffer a strong weakness in the lateral stiffness, with following cumber-
some design procedures to avoid excessive lateral displacements maintaining a quite high
ductile behaviour under seismic actions. Often, the design process leads to not optimized
structural members, oversized with respect to the minimum seismic requirements due to lat-
eral deformation limitations. EBFs combine high lateral stiffness, due to bracing elements,
and high dissipative capacities, provided by the plastic hinges developed in links. Eurocode
8 proposes a design procedure for EBF structures in which iterative checks are required to
design links with a defined level resistance dependent on all the other links strength. The
present paper investigates the seismic behaviour of EBFs using Incremental Dynamic Analy-
ses (IDA) to explore their mechanical response under increasing seismic action. IDAs are
executed considering the influence of variability of steel mechanical properties on the behav-
iour of EBFs, using seven artificial accelerograms according to Eurocode 8. The aims of IDAs
are the probabilistic assessment of the response of the system with respect to the variability
of the material properties, the analysis of structural safety and the ability of the structures to
internally redistribute plastic phenomena during the earthquake. Structural safety conditions
will be defined according to a multi-level performance approach. The paper presents also
some final suggestions for possible improvements and design simplifications.
A. Braconi
Riva FIRE S.p.A., Viale Certosa 249, 20151 Milano, Italy
e-mail: aurelio.braconi@rivagroup.com
123
2250 Bull Earthquake Eng (2013) 11:22492285
1 Introduction
Modern codes on seismic design of steel buildings, as Eurocode 8 (UNI EN 1998-1:2005) and
FEMA 350 (FEMA 2000), allow the adoption of ductile systems able to develop high plastic
deformations (e.g. plastic hinges) for dissipating the seismic energy stored in the building
during the earthquake.
The greater is the number of the plastic hinges developed, the greater is the attainable
ductility and therefore the greater is the dissipative capacity of the structure; at the same
time, the demand in terms of plastic rotational capacity on structural elements is reduced,
since plastic deformations are spread all over the structure (Fig. 1a). The possibility of
exploiting plastic resources, in fact, allows to consider lower values of the design seismic
action, generally expressed in terms of peak acceleration experienced by the structure, and to
consequently avoid the oversizing of the structural members in which plastic deformations
are expected to develop.
Such concepts are introduced in Eurocode 8 (EN 1998-1:2005) through the adoption of
the behaviour factor (or q factor) and the capacity design (CD) approach. The former is
a coefficient that reduces the design seismic actions according to the structural response,
the inelastic resources and the sensibility to the second-order effects of the buildings. The
latter assures the development of plastic phenomena only in selected regions of the structural
members (critical zones), protecting, at the same time, all the other parts where plastic
deformations shall be very limited or completely absent. The final detailing of the structural
joints and connections, together with CD, assures the development of an efficient dissipative
mechanism, without significant decreases of strength and stiffness.
Obviously, the location of the critical zones in the bearing elements depends on the struc-
tural typology and on the associated collapse modalities: as an example, in the Fig. 1b, the
most common global collapse mechanisms for moment resisting frames (MRF), concentri-
cally braced frames (CBF) and eccentrically braced frames (EBF) are presented.
The traditional approach used in the design of ductile steel structures, according to actual
European standards (UNI EN 1998-1:2005), foresees the execution of safety assessments in
terms of strength for all the structural elements (including also connections): the design values
of internal forces, evaluated using the seismic design combination, shall be lower than the
corresponding design strength, opportunely obtained considering the nominal yielding stress
of materials. Additional controls are required for what concerns the stiffness and ductility
performance of the structure: in order to verify that elements are provided by an adequate level
of ductility, specific detailing and sizing rules shall be adopted, especially in correspondence
of the selected dissipative zones.
Aiming to obtain the configuration of plastic hinges assumed during the design phase, spe-
cific requirements about materials resistance shall be satisfied and, moreover, a CD approach
shall be lay down among structural elements: these requirements govern the distribution of
plastic hinges, whose development is influenced by the actual values of the mechanical prop-
erties of materials.
On the other hand, the actual European production codes (for example EN10025-1
6:2004) do not provide adequate limitations for the mechanical properties of steel products,
resulting in inconsistencies between the design standards and the technical delivery condi-
tions of steel products. These differences can lead to great discrepancies between the effective
123
Bull Earthquake Eng (2013) 11:22492285 2251
plastic,1
plastic,2
plastic
hinge
(b)
resistance of structural members and the nominal one (i.e. material Over-StrengthOS), gen-
erally adopted in the design process; in particular, the higher will be OS phenomena, the higher
will be the earthquake intensity at which plastic hinges will develop affecting the expected
formation of the dissipative mechanism. Therefore, in order to avoid possible modifications
in the location of plastic hinges as planned during the design, Eurocode 8 introduces the use
of adequate safety factors and additional checks for the mechanical properties of materials,
to be adopted in the CD approach. As an example, in the dissipative zones, the ratio between
real and nominal yielding stress of the steel shall be upper limited by the an over-strength
coefficient (ov ), assumed equal to 1.25.
Analyzing the eccentrically braced solutions, it is evident that the ability to dissipate seis-
mic energy strictly depends on the plastic deformations essentially located in links, designed
123
2252 Bull Earthquake Eng (2013) 11:22492285
to plasticize before beams, braces and columns, otherwise sized using the forces generated
by yielded and hardened links (Rossi and Lombardo 2007) in order to remain in the elastic
field. In particular, the OS assigned to non dissipative elements depends on the mechanical
and geometrical characteristics of the links.
In CD framework, the OS is generally expressed trough the factor , defined as the ratio
between the plastic design resistance and the effective action on the dissipative elements:
shear force for short links and bending moment for long links. According to Eurocode 8 (EN
1998-1:2005), the distribution of the OS factor () shall be quite uniform in EBFs, with a
difference between minimum and maximum OS below the 25 %. This provision intends to
uniform the distribution of plastic phenomena in correspondence of links in all the storeys
to develop a global dissipative mechanism.
On the other hand, some studies in the current literature (Popov and Engelhardt 1988)
evidenced undesired collapse mechanisms in EBFs characterized by a high number of storeys,
despite the development of small plastic deformations of the links and the respect of the design
criteria. This event can be partially caused by the difference between nominal design values
and real effective values of the mechanical properties of steel products (Bruneau et al. 1998;
Bosco and Rossi 2009): this difference, generally already taken into account during the design
process through the introduction of the material OS coefficient ov , can lead to the alteration
of the failure modalities assumed in the design, causing premature local collapse phenomena
of the structure and consequently avoiding the global dissipation of seismic energy.
The present work aims to investigate the influence of the variability of material properties
on the dissipative behaviour of EBFs and was developed in the framework of a European
Research Project funded by the Research Fund for Coal and Steel called OPUS (Optimizing
the seismic Performance of steel and steel-composite concrete strUctures by Standardizing
material quality control, RFCS 2011).
The aim of the research consisted in the characterization of the structural behaviour of
different steel buildings designed according to Eurocode 8 (EN 1998-1:2005), taking into
account the variability of the mechanical properties. In particular, the structural response
was investigated through the execution of non linear Incremental Dynamic Analyses (IDA),
considering the variability of mechanical properties of materials and using 7 different seismic
inputs. The results obtained from IDAs using actual values of mechanical properties of
steel, assumed as representative of the European steel production, were presented in terms
of activation probability for each relevant collapse criteria, analyzing the variation of the
structural safety level as function of the demand imposed by the earthquake and of the
material properties.
In such a way it was possible to assess the structural performance considering the vari-
ability of material properties (i.e. q factor estimation, Braconi et al. 2013) and the influence
of imposing an upper limit to the yielding stress (i.e. fictitious additional quality check for
EN10025 produced steels) on structural safety as well as to evaluate the sensitiveness of CD
approach to the ov factor, introduced for taking into account steel OS.
2 Proposed methodology
In general, the failure of a structure during an earthquake occurs when the capacity (C) is
exceeded in one or more elements by the demand (D): being both C and D time-dependant
and mutually inter-dependant, the failure of the whole structure is related to the sequence of
the single collapses in structural elements.
123
Bull Earthquake Eng (2013) 11:22492285 2253
In the widest framework of OPUS project, fifteen different buildings in steel and steel
composite concrete structure (Somja et al. 2013) were designed, in order to cover the most
common geometrical and functional structural typologies adopted in the European Countries:
MRFs, CBFs and EBFs for offices, industrial buildings and car parks were analyzed.
123
2254 Bull Earthquake Eng (2013) 11:22492285
Secondary
(a) Main beams beams (b)
6000
1000 800 800 800 800 800 1000
600
Secondary
6300
beams
600
Main beams
600
Y Y
(c)
Sheat Studs
Secondary beam Coupled Beam for vertical load
Fig. 2 General plan of buildings a office buildings 34, b car park 16 and c beam duplication for decoupling
vertical and seismic loads
The present paper deals with the seismic behaviour of EBF steel structures; three different
buildings, in the following called building 3, 4 and 16, were designed according to the criteria
imposed by Eurocodes (EN 1990-1, EN 1991-1, EN 1993-1 and EN 1998-1:2005). External
EBF frames were designed in order to sustain the total seismic horizontal forces: stating
the symmetry of elements disposition and mass distribution, the design of the buildings
was calibrated on single eccentrically braced frames, referring to the two main directions of
each structure. The reliability of the design so executed was checked comparing the results
with the ones coming from dynamic modal analyses on 3D global models of the buildings,
opportunely elaborated using SAP2000 software.
Different functional destinations were adopted for the EBF case studies: office buildings
for case studies 3 and 4 and car park for building 16.
Buildings 3 and 16, located in high seismicity region, were designed using short shear
links, while building 4, located in medium-low seismicity area, was provided by long bending
links.
Buildings 3 and 4 were characterized by a similar geometry, presenting 5 storeys with an
interstorey height of 3.50 m and a span length variable between 6 and 7 m (Figs. 3 and 4);
building 16 was characterized by a span length between 8 and 10.5 m and was designed with
only two storeys of interstorey height equal to 4.0 m (Fig. 5). The location of links and their
length were varied in relation to their typology (shear or bending) and to the floor position.
A duplication of secondary beams was applied in buildings 3 and 4 in order to avoid the
amount of vertical loads and the connection of elements in correspondence of the dissipative
links (see Fig. 2). Pinned connections were used at the ends of non dissipative elements,
123
Bull Earthquake Eng (2013) 11:22492285 2255
2100
(a) 700 700 700
HEB120 e=450mm
350
HEB140 e=450mm
HEB180 e=700mm
350
HEB200 e=700mm
350
2400
(b) 600 600 600 600
HEB100 e=250mm
350
HEB140 e=350mm
350 350
HEB160 e=450mm
1750
HEB200 e=600mm
350
HEB200 e=600mm
350
such as braces and columns, and between beams and columns for eccentric inverted V-brace
frames (frame 3xz, 16xz and 16yz, Figs. 3a and 5a, b); welded connections were adopted
for the beam to column joints in eccentric D-brace frames (Figs. 3b and 4a, b). The general
geometrical properties of EBF buildings are presented in Table 1.
For all case study buildings a floor system characterized by a concrete slab on prefabricated
trussed slab with a global thickness of 23 cm was used; in the design of buildings 3 and 4
steel grade S355 was adopted (nominal yielding strength equal to 355 MPa) while building
16 was designed using steel grade S275 (nominal yielding strength equal to 275 MPa).
Buildings 3 and 16 were designed considering a PGA equal to 0.25 g and soil category B,
while for building 4 a PGA equal to 0.10 g and soil category C were adopted; the response
spectra adopted in the design were compatible with both Eurocode 8 and actual Italian
standard for constructions (D. M. 14/01/2008). Table 2 summarizes the values of vertical and
horizontal loads adopted in the design.
According to Eurocode 8 and in relation to the location of the buildings, behaviour factors
equal to 6 and 4 were adopted respectively for high ductility class (HDC, case studies 3 and
16) and low ductility class buildings (LDC, case study 4). All the EBF buildings described so
123
2256 Bull Earthquake Eng (2013) 11:22492285
2100
(a) 700 700 700
IPE160 e=1000mm
IPE240 e=1000mm
1750
IPE270 e=1000mm
IPE270 e=1000mm
HEB240 HEB240
Braces Beams without
HEB200 link IPE500
2400
(b)
600 600 600 600
IPE160 e=1000mm
350 350 350 350 350
IPE220 e=1000mm
IPE240 e=1000mm
1750
IPE270 e=1000mm
IPE270 e=1000mm
6000
(a) 1000 800 800 800 800 800 1000
HEB320 e=600mm
400 400
800
HEB360 e=600mm
6300
(b) 1050 1050 1050 1050 1050 1050
HEB300 e=700mm
400 400
800
HEB320 e=700mm
Fig. 5 Building 16 (short links), geometry and elements: a xz frame and b yz frame
far were designed to resist vertical and horizontal forces according to actual standards, both
for seismic and static combinations, without suffering global or local failures. The design
was optimized to obtain a uniform distribution of plastic phenomena of links in all the floors:
an accurate distribution of the OS factors i was pursued, obtaining variations smaller than
123
Bull Earthquake Eng (2013) 11:22492285 2257
Building Type Typology Live load Snow Wind Soil type Seismic Seismic mass (ton)
(kN/m2 ) load load action
(kN/m2 ) (kN/m2 ) (g)
Floor Roof
25 % among the floors, according to Eurocode 8 (EN 1998-1:2005). The obtained values for
OS factors are summarized in Table 3.
For the design of links, the actions coming from the linear dynamic analysis with behaviour
factor were used; the design strength of links was consequently adopted for the overstrength-
ening of other elements, such as beams, columns and braces, according to the CD approach.
Buckling phenomena of compressed elements and the satisfaction of the interstorey drift
limitation were relevant aspects in the design of structural elements.
123
2258 Bull Earthquake Eng (2013) 11:22492285
Link profile Link length (mm) Link profile Link length (mm)
Typically, HEB sections were used for columns and braces in all the buildings; otherwise,
HEB or IPE sections were adopted for links: HDC buildings with short shear links presented
HEB sections for dissipative elements (the link length e varied between 250 and 700 mm),
while LDC building adopted IPE sections for long bending links (link length e equal to
1,000 mm), as presented in Table 4.
In the following, a specific identification code will be used to quickly refer to each structure,
indicating the number of the building (3, 4 or 16), the structural typology (EBF) and the
direction of the considered frame (X for plane xz or Y for plane yz, resulting, as an example,
in frame 3EBFX, frame 16EBFY and so on).
In order to evaluate the influence of material properties variability on the effective seismic
behaviour of EBF structures, non linear Incremental Dynamic Analyses (IDA) were executed
on plane frame models corresponding to the two main directions of the case study buildings
previously described.
As many past works evidenced (Rides and Popov 1993; Mastrandrea and Piluso 2009)
the modelling of link elements needed to be very accurate to obtain numerical outcomes
consistent with the EBF response prediction; both short and long links, despite the different
mechanism involved for the dissipation of seismic energy, generally develop flexural forces
combined with shear ones: the model of link shall be consequently able to reproduce both
the two effects.
Many numerical models were proposed in the current scientific literature to represent the
behaviour of link elements: for example, one component models with concentrated plastic
hinges at the ends of the element (Gilberson 1969) or two component models constituted
by beams working in parallel (Rides and Popov 1993) were presented and deeply analyzed.
123
Bull Earthquake Eng (2013) 11:22492285 2259
V
xh
V
Shear V
V
+
M
xL
0
Fiber element N, M
N
(a) (b)
Fig. 6 a General scheme of fibre elements and b model of imperfections of braces and columns
(a) (b)
V
Fu Fu
Epl
Kpl
Fy Fy
Es
Ks
y u Dy
Fig. 7 Constitutive law adopted for a flexural behaviour, b shear behaviour of dissipative elements
Nevertheless, only more recent models were able to directly take into account the shear
behaviour of the dissipative elements (Rides and Popov 1993).
In the present work, bi-dimensional models of the two main frames of the buildings
were elaborated using the numerical software OpenSees (Mazzoni et al. 2007). A different
technique was used for the modelling of dissipative (links) and not dissipative elements
(beams and columns), in order to take into consideration the differences about the behaviour
towards seismic action due to the aforementioned CD approach (Fig. 5).
All the elements were modelled as fibre section elements (Fig. 6a); an opportune mesh
of sections, with fibres dimensions usually equal to 1.0 1.0 cm2 , was adopted in order to
better approximate the effective structural behaviour of elements.
The Menegotto-Pinto law (1973), directly implemented in OpenSees through Steel02
material model, was used for the representation of the flexural behaviour of elements (beams,
columns, links and braces): an elasto-plastic with kinematic hardening stress-strain relation-
ship was adopted (Fig. 7a).
For what concerns the shear behaviour of elements, a distinction was made between
elements designed to remain in the elastic field (beams, columns and braces) and dissipative
components (links); the formers were modelled assuming a linear elastic shear-displacement
law, while dissipative elements, both in the case of shear and bending links, were characterized
by an elastic-plastic with hardening relationship (Fig. 7b). The relevant parameters defining
the shear-displacement relationships are presented by Eq. (1), in which KShear,Link represents
the elastic shear stiffness, Av,beam the area of the web section of the considered link element,
e the link length, Gs the elastic shear modulus and Fy,Link , Fu,Link are, respectively, the
yielding and tensile strength of the steel material and Dy,Link and Du.Link the corresponding
displacements, and finally, Link represents the ultimate plastic rotation of links, opportunely
123
2260 Bull Earthquake Eng (2013) 11:22492285
evaluated taking into account standards prescriptions and current literature, assumed equal
to 110 and 80 mrad respectively for shear and bending links.
G s Av,beam
K Shear,Link =
e
fy fu
Fy,Link = Av,beam , Fu,Link = Av,beam
3 3
Fy,Link
D y,Link = , Du,Link = e Link (1)
K Shear,Link
A single inelastic fibre element was used to represent columns, beams without links and short
shear links; on the other hand, two elements were used to model each long bending link and
four elements were employed for each brace. Buckling phenomena of braces were directly
taken into account providing by an initial imperfection equal to 1/500 of the brace length the
middle point of the brace, as represented in Fig. 6b; a similar modelling approach was also
adopted to include in the analysis P effects of column elements (Fig. 6b).
The calibration of the parameter adopted in the models was executed comparing the
outcomes from cyclic loading histories on single components (for example the braces) with
literature results (Tremblay 2002).
In order to evaluate the global ductile behaviour of the structures under seismic action, all
the possible collapse mechanisms for EBFs were individuated and analyzed, considering the
prescriptions provided by actual standards (EN 1993-1:2005, FEMA 2000) and what already
presented in the current literature. The reaching of the different collapse criteria represented
the limit condition at which IDA simulations were generally stopped: for higher levels of
seismic action collapse took place and the behaviour of the structure was, consequently,
not significant. The collapse limitations were adopted in relation to different limit states,
assessing the structural performance both at ultimate and serviceability limit conditions.
One of the most significant collapse criteria for EBFs was, obviously, the failure of link
elements due to the over passing of the plastic rotation limit. The plastic rotation was evaluated
as the ratio between the relative vertical displacement () and the link length (e), as presented
by Eq. (2):
v1 v2
= = L I N K (2)
e e
being v1 and v2 the vertical displacements in correspondence of the two ends of the link.For
shear short link was evaluated as the relative vertical displacement between the two ends of
the link (Fig. 8a); for long bending links, whose behaviour can be generally well approximated
to the one of a beam element in flexure, the relative displacement referred to the mid length
of the element, assuming that the point of contraflexure occurred in correspondence of the
mid length of the element, as briefly schematized in the Fig. 8b (FEMA 356 2000).
The limits assumed according to the standards are briefly presented in Table 5.
For the evaluation of the plastic rotation limit for links, the prescriptions of both Eurocode
8 (EN 1998-1:2005) and FEMA 356 (2000) were considered. According to Eurocode 8, the
plastic rotation limit shall be assumed equal to 80 and 20 mrad respectively for short shear
links and long bending links; on the other hand, looking at the prescriptions imposed by
FEMA 356, the limits imposed for the ultimate rotation of short shear links are equal to 110
and 140 mrad respectively in the case of Life Safety (LS) or Collapse Prevention (CP) limit
123
Bull Earthquake Eng (2013) 11:22492285 2261
e
(b)
e
e 0.5
0.5
Ultimate plastic rotation EC8, FEMA 356 110 mrad (shear), 20 mrad (bending)
Global buckling EN 1993-1 according to Eq. (3)
Interstorey drift ratio EN 1998-1 1.5 % Interstorey height
state. Moreover, in the case of long bending links, FEMA 356 suggests the adoption of the
same values used for beam elements, resulting in values varying between 2 and 8 times the
chord rotation at yielding (y ), in relation to the geometrical characteristics of the sections
profiles. As a consequence of what already presented, an intermediate condition between
FEMA 356 and Eurocode 8 was assumed in the present work, resulting in values of the
ultimate plastic rotations respectively equal to 110 and 20 mrad for shear and bending links.
The limit axial load for the buckling of steel members in compression (columns and
braces) was evaluated according to Eurocode 3 (UNI EN 1993-1:2005) using expression (3),
and was consequently strongly influenced by the mechanical properties of materials (yielding
strength fy ):
A fy
Nb,Rd = (3)
M1
123
2262 Bull Earthquake Eng (2013) 11:22492285
The limitation imposed to the interstorey drift was also assessed; the respect of the lateral
stiffness requirements strongly influenced the sizing of steel members such as braces, espe-
cially in building 4, designed for low-ductility class. Once again, the prescriptions imposed
by FEMA 356 were considered and little modified: the limitations proposed by the afore-
mentioned code are, in fact, respectively equal to 2.0 % in the case of CP limit state and
1.5 % (transient) and 0.50 % (permanent) in the case of LS limit state. In the present work,
an interstorey drift limit equal to 1.5 % was assumed as ultimate value for EBF structures, as
briefly summarized in the Table 5, being the respect of the interstorey drift a very significant
aspect in the design of case study buildings.
The seismic hazard of a region represents its natural exposure to earthquakes, whose char-
acteristics are defined by the variables mainly correlated to regional seismic features. In
particular, the maximum peak ground acceleration is related to the probability of exceedance
and to the mean return period of the seismic event. According to EN1998-1 (2005) guide-
lines, it is possible to assume that the annual rate of exceedance of the reference peak ground
acceleration (agR ) may be taken to vary with agR as presented by:
H ag R = k0 agkR (4)
TL
TR = (5)
ln (1 PN C R )
that gives a return period of 475 years for the design PGA. According to PGA levels assumed
during seismic design, 475 years of return period corresponds, respectively, to 0.25 g in high
seismicity areas and 0.10 g in low seismicity areas and fixes k0 parameter in Eq. (4) equal to
3.29 105 and 2.10 106 for high and low seismicity areas. Resulting hazard functions
are presented in Fig. 9.
123
Bull Earthquake Eng (2013) 11:22492285 2263
1.0E-03
1.0E-04
1.0E-05
0.0 0.5 1.0 1.5
a gR -Peak Ground Acceleration -[g]
TL = 1 year
1.0E-02 TL = 50 years
1.0E-03
1.0E-04
1.0E-05
1.0E-06
0.0 0.5 1.0 1.5
a gR -Peak Ground Acceleration -[g]
Fig. 9 Hazard function according to EN1998-1 prescriptions: a high seismic hazard; b low seismic hazard
Two types of seismic intensities were considered, according to what adopted in the design
phase: for buildings designed in high seismicity areas (i.e. buildings n 3 and 16), the PGA
level was equal to 0.25 g and the type 1 spectrum for soil category B was used, while for
building designed in low seismicity areas (i.e. building n 4) the PGA was equal to 0.10 g
and the type 2 spectrum for soil category C was applied for the simulations (Fig. 10a). The
filter function was defined by a trapezoidal shape, where the time intervals for the initial and
ending ramps were 5 s and the strong motion duration was equal to 10 s or 5 s respectively
for high and low seismicity (Fig. 10b).
The relevant eigen-periods were assumed to be in a range between 0.1 and 3.0 s. The
chosen sampling interval of t = 0.01 s allowed a sufficient accurate calculation for Eigen-
frequencies up to 20 Hz (5 points for each period).
The assessment of the accelerograms by determining the velocity and displacement time
histories showed that the displacements were running out (Fig. 11): hence, a baseline cor-
rection was applied to obtain a sufficient small displacement at the end of the record. The
adequacy of the accelerograms was checked by the determination of their elastic response
spectra (Fig. 12). For periods lower than TB the spectrum value Sa was slightly too high
123
2264 Bull Earthquake Eng (2013) 11:22492285
(a) 0.8
high - 5 %
low - 5 %
0.6
Sa [g] 0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
T [s]
(b)
f (t)
t
t2 - strong motion
Fig. 10 Target spectra (a) and filter function (b) for the generation of artificial time histories
(Fig. 13); however, the target spectrum was sufficiently met and the requirements defined in
EN1998-1 (2005) were obtained.
The COV of the spectral values for the 7 accelerograms was finally between 0.04 and 0.12,
as presented in the Fig. 14. It shall be noted that the energy density of artificial accelerograms
is much higher than the one of natural accelerograms, as all frequencies of interest are
included.
IDA simulations were executed assuming as Intensity Measure (IM) the PGA, due to the
fact that the 7 time histories, for each Hazard curve, were generated from the EN1998
spectra. Consequently, scaling the PGA of each single time-history influences the values
of its corresponding response spectrum with the same ratio, making the PGA the only
free variable. This fact legitimated the adoption of the PGA as IM for carrying out IDAs
(Table 6).
The PGA levels adopted for the scaling of the earthquakes were previously determined
studying structural response of case study buildings. For each plain structure (see Figs. 3, 4
and 5), different excitation levels were individuated according to the activation of the col-
lapse modalities for increasing levels of PGA, see Table 7. As an example, this preliminary
assessment revealed that collapse of columns and collapse of braces in 3EBFX were modal-
123
Bull Earthquake Eng (2013) 11:22492285 2265
Fig. 11 Baseline correction for an artificial accelerogram (high seismicity): a displacement; b velocity
ities that cannot be activated for PGA levels lower than 1.50g. Hence, these two modes were
considered as not significant for the study and neglected.
These high PGA activation levels are a consequence of the design procedure: the columns
were sized under full vertical static actions at Ultimate Limit State, while braces in 3EBFX
were over-sized due to the extreme sensitivity of structural configuration to second order
effects.
For each case study, collapse criteria were analyzed for each considered PGA level, executing
IDAs adopting alternatively the 7 artificially generated accelerograms.
Monte Carlo Method was applied to each analysis generating 500 samples of mechanical
variables and running IDA for each of them; each set of mechanical variables includes a
specific value for yielding strength (Re ), a specific value for tensile strength (Rm ) and a spe-
cific values for ultimate elongation (A), opportunely generated according to the probabilistic
model based on real production data. In particular, to be adherent to the real assembling
of steel structures, all beams and braces were considered as probabilistically not dependant
(generating independent sets of mechanical variables) while columns of two subsequent
floors were considered as characterized by the same probabilistic variables, according to the
generation scheme presented in Fig. 15c.
123
2266 Bull Earthquake Eng (2013) 11:22492285
(a) 1.0
0.8
0.6
S a [g]
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
T [s]
(b) 1.0
0.8
0.6
Sa [g]
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
T [s]
Fig. 12 Target spectrum and elastic response spectra of 7 artificial accelerograms: low seismicity (a) and high
seismicity (b)
For each set of 500 nonlinear analyses, the results (related to a single collapse criterion, a
PGA level and an accelerogram) were suitably standardized referring to the values being,
123
Bull Earthquake Eng (2013) 11:22492285 2267
(a) 1.0
0.8
0.6
Sa [g]
0.4 target spectrum
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
T [s]
(b) 1.0
target spectrum
0.8
0.6
Sa [g]
mean value
0.4
0.2
90 % target spectrum
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
T [s]
Fig. 13 Target spectrum and mean value of the elastic response spectra of 7 artificial accelerograms: low
seismicity (a) and high seismicity (b)
for the specified collapse criterion, DMi the damage measure assumed by the EDP in the i-th
analysis and DMu its limit value corresponding to collapse.
The so obtained new set of data was statistically analyzed evaluating the basic parameters
(maximum, minimum and mean values and standard deviation) and executing the 2 test
to check the hypothesis of Normal or Log-Normal distributions. When the 2 test was not
negative a Normal or Log-Normal distribution was assumed; alternatively the statistical
cumulative density function was built, completed in correspondence of tails by suitable
exponential functions (Braconi et al. 2010).
The probability of failure related to each set of 500 data (related to a single collapse
criterion, a PGA level and an accelerogram) was so simply evaluated using its cumulative
density function. Clearly, for each collapse criterion and each PGA level, 7 values of Pf , and
so 7 fragility curves were obtained, one for each accelerogram. The average of 7 fragility
curves was assumed as the fragility curve related to that specific collapse criterion (see
Fig. 16).
Fragility of case studies referred to a collapse mode was finally integrated with European
Seismic Hazard function, as described in (UNI EN 1998-1:2005), furnishing annual proba-
bility of failure for relevant collapse criteria of all case studies, shortly presented in Tables
from 9af.
123
2268 Bull Earthquake Eng (2013) 11:22492285
(a) 0.15
COV (S a )[-]
0.10
0.05
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
T [s]
(b) 0.15
COV (S a )[-]
0.10
0.05
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
T [s]
Fig. 14 COV of the elastic response spectra of 7 artificial accelerograms: low seismicity (a) and high seismicity
(b)
The execution of all IDA simulations created a bulk of results in which the modification of
structural performance can be related with the scattering of mechanical properties of steel
products. This relevant database was analyzed in order to assess the following points:
123
Bull Earthquake Eng (2013) 11:22492285 2269
Table 6 Parameters of target spectra and filter function for low and high seismicity
Table 7 PGA levels determined according to relevant collapse modes, for (a) building 3, (b) building 4 and
(c) building 16
Link (g) Col. (g) Brace (g) Drift (g) Link (g) Col. (g) Brace (g) Drift (g)
(a) building 3
1 0.60 2.00 2.00 0.40 0.45 2.00 0.75 0.50
2 0.50 2.00 2.00 0.55 0.50 2.00 0.70 0.55
3 0.50 2.00 2.00 0.60 0.50 2.00 0.65 0.55
4 0.45 2.00 2.00 0.45 0.45 2.00 0.65 0.50
5 0.55 2.00 2.00 0.40 0.40 2.00 0.65 0.40
6 0.45 2.00 2.00 0.50 0.55 2.00 0.70 0.60
7 0.50 2.00 2.00 0.60 0.55 2.00 0.70 0.55
Acc. Frame 4X Frame 4Y
Link (g) Col. (g) Brace (g) Drift (g) Link (g) Col. (g) Brace (g) Drift (g)
(b) building 4
1 0.40 2.00 0.60 0.90 0.40 2.00 2.00 1.20
2 0.50 2.00 0.80 1.40 0.50 2.00 2.00 0.90
3 0.50 2.00 0.60 1.00 0.50 2.00 1.90 0.95
4 0.45 2.00 0.50 0.95 0.40 2.00 1.60 0.70
5 0.50 2.00 0.50 1.10 0.50 2.00 2.00 1.15
6 0.45 2.00 0.50 1.20 0.40 2.00 1.90 0.80
7 0.50 2.00 0.55 1.20 0.50 2.00 2.00 1.20
Acc. Frame 16X Frame 16Y
Link (g) Col. (g) Brace (g) Drift (g) Link (g) Col. (g) Brace (g) Drift (g)
(c) building 16
1 0.60 1.70 0.55 0.70 0.70 2.00 0.50 0.65
2 0.60 1.70 0.55 0.70 0.70 2.00 0.40 0.80
3 0.60 2.00 0.55 0.70 0.75 2.00 0.35 0.80
4 0.50 1.50 0.50 0.55 0.60 2.00 0.35 0.70
5 0.60 1.60 0.50 0.65 0.65 2.00 0.35 0.75
6 0.55 1.70 0.55 0.65 0.60 2.00 0.35 0.75
7 0.70 2.00 0.60 0.80 0.70 2.00 0.35 0.80
The model codes handle seismic risk by specifying earthquake design loading conditions
and then requiring strength levels adequate to resist those loads. The earthquake condi-
tion under which the design requirements apply, is the designlevel earthquake (DBE) (EN
123
2270 Bull Earthquake Eng (2013) 11:22492285
(a) (c)
B5
Br9
C5 Br10 C6
B4
Br7 Br8
C3 B3 C4
Br5 Br6
B2
Br3 Br4
C1 B1 C2
Fig. 15 IDA results in terms of Br1 forcea material and seismic input variability; b material variability;
c distribution of independent variables inside 3EBFX
123
Bull Earthquake Eng (2013) 11:22492285 2271
fy ft u
Again, this reliability index refers to failure of one building component, where failure is
typically defined in the context of seismic loading as fracture, rather than yielding. Overstress
of a single component is considered life-threatening damage, but is not equivalent to the
probability of casualties.
More recently, other studies and researches were carried out in order to define adequate
threshold as acceptance level for seismic reliability problems (Porter et al. 1998) fixing,
for ordinary buildings, a Pfail threshold equal to 103 for building elements and 104 for
emergence response facilities (for strategic buildings or for critical equipments contained in
ordinary buildingsi.e. sprinkler).
Other researchers confirmed the difficulties to give useful and, at the same time, abstract
values of the Pfail without reference to the context of the reliability estimation calculations
(Legerer 1990; Veneziano 1976; Veneziano et al. 1983; Ditlevsen and Madsen 1996). As
reported in the previous considerations and on the basis of many calculations presented in
the literature for design formulas and for safety factor calibration, an appropriate value of
the Pfail Pfail,Nominal over the lifetime of the structure for failure under extreme loading
conditions (but no human errors, etc.) can be assumed between 103 and 104 (Melchers
2002), corresponding to a factor in the range 3.03.5. In the analysis carried out on the base
of IDA results, the Pfail judgement was made considering as acceptance threshold the interval
between 103 and 104 (yearly failure probability associated to seismic action return period
of 475 years). It is important to remind that this analysis was made on structures accurately
designed, in which human errors or wrong understanding of involved probabilistic aspects
were avoided.
The IDA results were employed to estimate the annual probability of exceeding the limit
values (), identifying selected collapse modes for the structural cast study buildings 3, 4
and 16. The estimation was executed adopting the approach proposed by PEER:
(E D P) = G ( E D P| M V ) |dG ( M V | D M)| |dG ( D M| I M)| |d (I M)| (6)
where EDP is the engineering demand parameter (i.e. outputs from IDA), MV is the material
variable (i.e. steel mechanical properties), DM is the damage measurement and IM is the
intensity measurement of the seismic input.
The integration reported in the expression (6) was carried out employing fragility curves
obtained following the procedure presented in Sect. 6.1 and the seismic hazard function
123
2272 Bull Earthquake Eng (2013) 11:22492285
(a)
Probability
Upper tail
Numerical fitting
PG (x)=1
Lower tail
Numerical fitting
G (X)=1 G (X i )
(b) 1.00
0.90
-Demand > Capacity
Probablity of failure
0.80
0.70 Quake 1
Quake 2
0.60 Quake 3
0.50 Quake 4
0.40 Quake 5
Quake 6
0.30 Quake 7
0.20 Mean
0.10 Normal CDF
0.00
0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Peak ground acceleration -[g]
Fig. 16 a Numerical CDF directly derived from IDA results (when 2 failed); b fragility of 3EBFX for
ultimate plastic rotation of the link B1
calibrated according to Sect. 5.1. The exceedance probability of limit values identifying
collapse modes so evaluated are reported in the Fig. 17af.
As visible, the safety level estimated for all collapse modes are in-line with the probability
threshold defined according to the conceptual process presented in Sect. 7.1. For sake of
clarity, collapse modalities identified by B refer to the exhaustion of plastic rotation in seismic
links, those identified by Br refer to the buckling of steel braces, C are related to the column
buckling and Drift are the ones referred to the reaching of maximum interstorey drift. The
numbers refer to the structural element ID (for B, Br and C) or to the storey number in the
case of Drift.
The lowest levels of failure probability (or associated seismic risk) are the ones related to
columns and braces, denoting that EN1998 design rules worked properly in the protection
of not dissipative members.
Ductile failure modes also presented failure probabilities higher than the ones presented
by brittle modes as buckling, as expected from structures characterized by ductilemedium
or highbehaviour; anyway, those failure probability levels are in the range of safety defined
in Sect. 7.1.
The large amount of data taken from IDA analyses allowed also to evaluate, with a simplified
approach, the influence of the upper yielding stress limitation on the failure probability
associated to relevant collapse modes. In particular, it was planned to analyze IDA results
123
Bull Earthquake Eng (2013) 11:22492285 2273
(a)
5 Br1 Br2
3EBFX
3EBFX
3EBFX
3 4 3EBFX
3EBFX
2
3EBFX
1 B1 B2 B3 B4 B5
3EBFX
3EBFX
3EBFX
3EBFX
3EBFX
3EBFY
3EBFY
3EBFY
Relevant collapse criteria
3EBFY
3EBFY
Drift Drift Drift Drift
3EBFY
3EBFY
4
3EBFY
3
3EBFY
B1 B4 B5 B8 B9 B12 B13 B16 B17 B20 2
3EBFY
3EBFY
3EBFY
3EBFY
3EBFY
3EBFY
3EBFY
3EBFY
3EBFY
3EBFY
4EBFX
4EBFX
B7 B9 B10 B12 B13 B15 1 2 3 4 5 Br1 Br2 C1 C2
4EBFX
4EBFX
Relevant collapse criteria
4EBFX
4EBFX
Drfit Drfit Drfit Drfit Drfit
4EBFX
4EBFX
4EBFX
4EBFX
4EBFX
4EBFX
4EBFX
4EBFX
4EBFX
4EBFX
4EBFX
B1 B3 B4 B6
4EBFX
4EBFX
4EBFX
4EBFX
123
2274 Bull Earthquake Eng (2013) 11:22492285
(d)
Drift2
4EBFY
Drfit1
4EBFY
Br1
4EBFY
B16
4EBFY
B13
4EBFY
B12
4EBFY
B9
4EBFY
B8
4EBFY
B5
4EBFY
B4
4EBFY
B1
4EBFY
(e) Drfit2
Limit
Drift1
C4
Buckling
Relevant collapse criteria
C3
C2
C1
Br6
Br5
Buckling
Br4
Br3
Br2
Br1
B6
B5
PH Rot.
B4
B3
B2
B1
1.0E-06 1.0E-05 1.0E-04 1.0E-03 1.0E-02 1.0E-01 1.0E+00
Exceedance probability of fixed threshold
(f) C7
C6
Buckling
C5
C3
Relevant collapse criteria
C2
C1
Br6
Buckling
Br4
Br3
Br2
Br1
B12
B11
B8
PH Rot.
B7
B6
B5
B2
B1
1.0E-08 1.0E-07 1.0E-06 1.0E-05 1.0E-04 1.0E-03 1.0E-02 1.0E-01 1.0E+00
Exceedance probability of fixed threshold
Fig. 17 continued
123
Bull Earthquake Eng (2013) 11:22492285 2275
580 580
530 530
480 480
430 430
350 400 450 500 350 400 450 500
Yielding stress -[MPa] Yielding stress -[MPa]
680 680
(b) (d)
630 630
580 580
530 530
480 480
430 430
350 400 450 500 350 400 450 500
Yielding stress -[MPa] Yielding stress -[MPa]
Tensile strength -[MPa]
680
(e)
630
580
530
480
430
350 400 450 500
Yielding stress -[MPa]
Fig. 18 Mechanical properties of link B1 in 3EBFX considered for failure probability estimation assuming
different fy limitation on all dissipative members (links)a 500 samples EN10025 full generation; b reduced
number imposing fy,act /fy,nom = 1.375; c reduced number imposing fy,act /fy,nom = 1.35; d reduced number
imposing fy,act /fy,nom = 1.30; e reduced number imposing fy,act /fy,nom = 1.25
using the general procedure presented in Sect. 6.1 but with preselected mechanical properties
sample sets.
More in the details, it was decided to estimate the dependency of the Pfail associated to
relevant collapse modes from the upper limit imposed on actual yielding strength (fy ) in the
dissipative zones (that means to assume different OVs imposed as an additional quality check
on produced steels). For such purpose, IDAs characterized by a minimum OV in seismic
links higher than a fixed upper limit were neglected in the statistical evaluation of results,
and consequently in the final estimation of Pfail . In Fig. 18af the modification of mechanical
properties population for Link1 in 3EBFX, due to the variation of the imposed maximum
OV in the links, is presented.
Figure 19af show the risk associated to failure modes as function of the maximum OV :
in particular, in figures a, c and e the ductile failure modes (exhaustion of plastic rotation in
links imposed adopting FEMA limits) and in figures b, d and f the not-ductile failure modes
123
2276 Bull Earthquake Eng (2013) 11:22492285
Variation of
failure mode (exhaustion of 4%
plastic rotation capacity in 3%
Links); (b, d, f) not-dissipative 2%
failure mode (buckling of braces) 1%
0%
-1%
-2%
- 1.375 1.350 1.300 1.250
Limitation on f y -(f y,max /f y,nom)
(b) 2%
1%
annual Pfail (Risk) 0%
Variation of
-1%
-2%
-3%
-4% 3EBFX Br1 3EBFX Br2
-5% 3EBFY Br1 3EBFY Br2
-6%
- 1.375 1.350 1.300 1.250
Limitation on f y -(f y,max /f y,nom)
(c) 30%
4EBFX B1 4EBFX B3 4EBFX B4
25%
annual Pfail (Risk)
4EBFX B6 4EBFY B1
20%
Variation of
15%
10%
5%
0%
-5%
- 1.375 1.350 1.300 1.250
Limitation on f y -(f y,max /f y,nom)
(buckling of protected members) are presented. The upper limitations on yielding strength
in dissipative members produced following effects:
upward trend of Pfail for ductile failure modes when OV lowers: real fy values close the
nominal ones produce premature plastic phenomena in seismic links leading to a major
exploitation of their post-elastic resources;
downward trend of Pfail for not-dissipative failure modes when OV lowers: stricter OV
values produce additional protection of bracing elements;
downward trend of Pfail for not-dissipative failure modes slows down after further decre-
ments of OV suggesting an asymptotic tendency, see Fig. 19b and d.
The OV assures a balanced protection, i.e. controlling forces acting into protected members
without increasing too much plastic resources demand into dissipative members, only when
OV is not lower than a certain threshold. The general trend summarized in Fig. 19af suggests
that an effective limitation of fy (fy,max /fy,nom ) should be not lower than 1.35 satisfying two
123
Bull Earthquake Eng (2013) 11:22492285 2277
Variation of
-10%
-15%
-20%
4EBFX Br1
-25% 4EBFX Br2
-30% 4EBFY Br1
4EBFY Br2
-35%
-40%
- 1.375 1.350 1.300 1.250
Limitation on f y -(f y,max /f y,nom)
(e) 25%
16EBFX B1
16EBFX B3
annual Pfail (Risk) 20% 16EBFY B2
16EBFY B6
Variation of
16EBFX B2
15%
16EBFY B1
16EBFY B5
10%
5%
0%
- 1.375 1.350
Limitation on f y -(f y,max /f y,nom)
20%
(f) 15%
16EBFX Br1
16EBFY Br1
10%
annual Pfail (Risk)
16EBFY Br3
5% 16EBFX Br2
Variation of
0% 16EBFY Br2
16EBFY Br4
-5%
-10%
-15%
-20%
-25%
-30%
-35%
- 1.375 1.350
Limitation on f y -(f y,max /f y,nom)
Database of results coming from IDAs was also employed for analyzing the efficiency of
CD formula in the prediction of real forces acting in the protected members. The CD in
EN1998-1-1 is expressed by the formula
E d = E G + 1.1 O V E E
dissi pative
R P L ,i
= 1.5 min dissi pative
(7)
E E+G,i
123
2278 Bull Earthquake Eng (2013) 11:22492285
1.E+00
(a) 1.E-05
1.E-10
(b) 1.E-02
1.E-06
1.E-10
1.E-14
forces < IDA forces
Probability of C.D.
1.E-18
1.E-22
1.E-26 Level - 0.40g
1.E-30 Level - 0.45g
1.E-34
1.E-38 Level - 0.50g
1.E-42
Level - 0.55g
1.E-46
1.E-50 Level - 0.60g
1.E-54
1.E-58 Level - 0.65g
1.E-62
1.E-66
1.E-70
1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.45 1.50
Over-strength factor - OV
(c) 1.E-01
1.E-09
1.E-17
1.E-25
forces < IDA forces
Probability of C.D.
1.E-33
1.E-41
Level - 0.40g
1.E-49
1.E-57 Level - 0.45g
1.E-65 Level - 0.50g
1.E-73
Level - 0.55g
1.E-81
1.E-89 Level - 1.05g
1.E-97 Level - 1.10g
1.E-105 Level - 1.15g
1.E-113
1.E-121
1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.45 1.50
Over-strength factor - OV
Fig. 20 Some results of statistical analysis carried out on G(X,OV ) functionExceedance probability of
IDA forces respect to predicted CD forcesa Brace 1 in 3EBFX-S355; b Brace 2 in 3EBFX-S355; c Brace
1 in 4EBFX-S355; d Brace 5 16EBFX-S355; e Brace 6 16EBFX-S275; f Brace 7 16EBFY-S275
where EG and EE are the effects of gravity loads and seismic action coming from elastic
analysis, OV is the material OS fixed equal to 1.25 and is the structural OS measuring
the discrepancy between plastic resistance of most stressed dissipative member and internal
123
Bull Earthquake Eng (2013) 11:22492285 2279
1.E+00
(d)
1.E-01
(e) 1.E+00
1.E-01
Level -0.35g
forces < IDA forces
Probability of C.D.
1.E-07
1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.45 1.50
Over-strength factor - OV
(f) 1.E+00
1.E-01
forces < IDA forces
Probability of C.D.
1.E-02
Level -0.35g
1.E-03 Level -0.40g
Level -0.45g
1.E-04 Level -0.60g
Level -0.65g
1.E-05 Level -0.70g
Level -0.75g
Level -0.80g
1.E-06 Level -0.85g
1.E-07
1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.45 1.50
Over-strength factor - OV
Fig. 20 continued
forces in seismic combination. The IDA results contained also the forces acting in protected
members; assuming those as representative of the real internal forces state during ground
shaking at increasing levels of seismic action, it was possible to compare results coming from
123
2280 Bull Earthquake Eng (2013) 11:22492285
0.8 1.45
0.7 1.40
0.6 1.35
1.30
0.5
1.25
0.4
1.20
0.3
1.15
0.2
1.10
0.1
0
0.25 0.38 0.50 0.63 0.75 0.88 1.00
Peak Ground Acceleration -[g]
(c) OV protection in C.D. framework
1
Real Force > C.D. Force
Exceedance Probability
0.9
0.8
0.7
1.50
0.6 1.45
0.5 1.40
1.35
0.4 1.30
0.3 1.25
1.20
0.2 1.15
0.1 1.10
0
0.15 0.45 0.75 1.05 1.35 1.65
Peak Ground Acceleration -[g]
(7) with IDA results; moreover, Eq. (7) was employed considering also different OV values
in order to test its effectiveness in the mitigation of material properties scattering.
The study was performed employing the following testing function
+
E G,i 1.1 O V E E,i
G i (X, O V ) = (8)
E I D A,r eal,i
that assumes values higher or lower than 1 when real forces from IDA are, respectively,
lower or higher than forces predicted by Eq. (7). The function (8) was statistically evaluated
inserting all IDA results and elastic forces of three case studies and considering different OV
values, varying between 1.5 and 1.1. The results of this analysis are reported in the Fig. 20af
where the probability that CD forces were lower than real IDA forces are plotted as function
of PGA level and of assumed OV ; each curve is referred to one structural member belonging
123
Bull Earthquake Eng (2013) 11:22492285 2281
0.9 1.50
1.45
0.8 1.40
0.7 1.35
1.30
0.6
1.25
0.5 1.20
0.4 1.15
1.10
0.3
0.2
0.1
0
0.00 0.50 1.00 1.50 2.00
Peak Ground Acceleration -[g]
to case studies. Statistical data, adopted for obtaining the curves depicted in Fig. 20, were
also analyzed to obtain fragility curves of the Eq. (7) efficiency. More in the details, results
presented in Fig. 20 were deeply processed in order to obtain, for each analyzed structural
member, a family of fragility curves (Fig. 21), interpolating PGA level with probability of
having real forces higher than those obtained from CD rules. These curves were to assess
the exceedance probability of real forces respect to those obtained through CD: probability
was evaluated in correspondence of design PGA and is reported in Table 9. The shape of the
fragility curves and the values reported in the Table 9 are a direct consequence of the design
process. Buildings 3 and 4 were conditioned by the verification of drift control and second
order effect sensitiveness limitation, following EN1998-1-1; according to this verifications,
structural members were oversized in order to laterally stiffen the frames. These design
checks coupled with CD formula properly controlled the influence of seismic input and
123
2282 Bull Earthquake Eng (2013) 11:22492285
Frame 3EBFX 3EBFX 3EBFX 3EBFX 3EBFY 3EBFY 3EBFY 3EBFY 3EBFY
(a)
ag,design (g) 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25
Element Brace1 Brace2 Col_1 Col_2 Brace1 Brace2 Col_1 Col_2 Col_4
Pfail <1E10 <1E10 <1E10 <1E10 <1E10 <1E10 <1E10 <1E10 <1E10
Frame 16EBFX 16EBFX 16EBFX 16EBFX 16EBFX 16EBFX 16EBFX
(b)
ag,design (g) 0.25 0.25 0.25 0.25 0.25 0.25 0.25
Element Brace1 Brace2 Brace3 Brace4 Brace5 Brace6 Col_1
Pfail 9.42E05 6.24E05 1.14E03 8.16E04 1.38E04 2.00E04 9.42E05
Frame 16EBFY 16EBFY 16EBFY 16EBFY 16EBFY 16EBFY 16EBFY
(c)
ag,design (g) 0.25 0.25 0.25 0.25 0.25 0.25 0.25
Element Brace1 Brace2 Brace3 Brace4 Brace7 Brace8 Col_1
Pfail 1.04E04 8.60E05 3.98E05 3.81E05 9.46E05 8.48E05 <1E10
Frame 4EBFX 4EBFX 4EBFX 4EBFX 4EBFX 4EBFX
(d)
ag,design (g) 0.10 0.10 0.10 0.10 0.10 0.10
Element Brace_1 Brace_2 Column_1 Column_2 Column_3 Column_4
Pfail <1E10 <1E10 <1E10 <1E10 <1E10 <1E10
Frame 4EBFY 4EBFY 4EBFY 4EBFY 4EBFY 4EBFY
(e)
ag,design (g) 0.10 0.10 0.10 0.10 0.10 0.10
Element Brace_1 Brace_2 Column_1 Column_2 Column_4 Column_5
Pfail <1E10 <1E10 <1E10 <1E10 <1E10 <1E10
123
Bull Earthquake Eng (2013) 11:22492285 2283
8 Conclusions
to assess the influence of material properties scattering on the seismic response of steel
eccentrically braced frames (EBFs);
to analyze the efficiency of CD rules employed by EN1998-1-1 for the finalization of
seismic design of EBFs;
to evaluate the influence of fixing an upper limit on yielding strength of steel products
on the estimated probability of failure of structural members in EBFs.
These three questions were analyzed and discussed employing a large amount of results
coming from the execution of thousands IDA simulations.
The influence of material properties scattering on the failure probability of structural
members of case studies was estimated as marginal. The Pfail associated to all relevant
collapse modes individuated in the buildings 3, 4 and 16 were often lower than the fixed
acceptance threshold, confirming the appropriate working of the capacity design adopted
by EN1998-1-1. Nevertheless, design procedure often forces to oversize the non-dissipative
members for the seismic induced demand to comply with the relevant capacity parame-
ters in order to satisfy structural checks; in such a way, the failure probabilities associated
to member collapse were lower than the minimum values estimable with the employed
procedure.
According to the presented results, the CD rule proposed by EN1998-1-1 for EBFs worked
properly, producing internal forces higher than the ones obtained from IDA simulations.
Moreover, the analysis executed on CD formula suggested that the assumption of OV as
material OS is too conservative and can lead to an excessive structural members over-
sizing. In fact, the adoption of OV equal to 1.25 in Eq. (7), as suggested by EN1998, with
steel products whose variability leads to ov,material higher than 1.35 or more, produced
safe structures and the CD formula gave safe-side evaluation of real internal forces regime.
This discrepancy suggested that the CD formula should adopt OV coefficient calibrated
on structural type and expected ductile behaviour, not directly employing the material OS
measured on the steel products.
Finally, the analysis performed in order to assess the influence of upper yielding strength
limitation at the production site for steel elements was carried out.
It was observed, after the repetitive application of statistical procedure, that the adoption
of fy,max limits in the dissipative zone produced two equal scale effects: the increasing of fail-
ure probability associated to ductile failure modes and the decreasing of failure probability
associated to not-dissipative failure modes. Moreover, the decreasing of Pfail associated to
not-dissipative collapse criteria presented an horizontal asymptote when fy,max limits became
stricter. On the contrary, the increasing of Pfail associated to dissipative failure modes pre-
sented, sometimes, an increasing exponential behaviour, suggesting that too strict limits on
fy,max in dissipative zones could lead to excessive plastic demand on dissipative members
for lower seismic hazard level.
According to these considerations and on the basis of Fig. 19af, it could appear as rational
the application of an upper limit on fy in dissipative zones not lower than 1.375 (like Eurocode
suggests): in this range it was observed a balanced variation of Pfail between ductile and not-
ductile failure modes. Stricter limits as 1.20 for example showed in discussed example a too
high increment of Pfail associated to ductile failure modes.
123
2284 Bull Earthquake Eng (2013) 11:22492285
References
Bosco M, Rossi PP (2009) Seismic behaviour of eccentrically braced frames. Eng Struct 31:664674
Braconi A, Badalassi M, Salvatore W (2010) Modeling of European steel qualities mechanical properties
scattering and its influence on Eurocode 8 design requirements. In: 14th ECEE ProceedingsEuropean
conference on earthquake engineering. Ohrid, Makedonia, August 31September 03 2010
Braconi A, Caprili S, Degee H, Guendel M, Hjaij M, Hoffmeister B, Karmanos S, Rinaldi V, Salvatore W
(2013) Design and assessment of steel and steel-concrete composite structures: efficacy of en1998 design
procedure. In: 4th ECCOMAS thematic conference on computational methods in structural dynamics and
earthquake engineering. Kos Island, Greece, 1114 June
Bruneau M, Uang C, Whittaker A (1998) Ductile design of steel structures. McGraw Hill, NY
CEN, Technical Commission 250 (2005) UNI-EN1991-1-1: Eurocode 1actions on structures. Part 1 1:Gen-
eral actionsdensities, self-weight, imposed loads for buildings
Cornell CA (1969) Structural safety specification based on second-moment reliability. In: Proceedings of
symposium of the international association of bridge and structural engineers. London, England
D. M. 14/01/2008. (2008) Norme Tecniche per le Costruzioni, (In Italian)
Ditlevsen O, Madsen HO (1996) Structural reliability methods. Wiley, Chichester
Ellingwood B, Galambos TV, MacGregor JG, Cornell CA (1980) Development of a probability based crite-
rion for American National Standard A58, Washington, DC: Special Publication 577, National Bureau of
Standards, pp 222
EN10025-16:2004. (2004) General technical delivery conditions for: non-alloy, normalized/normalized rolled
weldable fine grain, thermomechanical rolled weldable fine grain, improved atmospheric corrosion resis-
tance, flat products of high yield strength in the quenched and tempered condition. European Committee
for Standardization, Brussels
FEMA 350 (2000) ASCErecommended seismic design criteria for new steel moment-frame buildings
FEMA 356 (2000) ASCEprestandard and commentary for the seismic rehabilitation of buildings
Gilberson MF (1969) Two nonlinear beams with definitions of ductility. J Struct Eng 95(2):137157
Hasofer AM, Lind N (1974) An exact and invariant first-order reliability format. ASCE J Eng Mech
100(EM1):111121
Legerer F (1990) Code theorya new branch of engineering science. In: Lind NC (ed) Structural reliability
and codified design, SM Study No 3:113127, University of Waterloo, Ontario
Mastrandrea L, Piluso V (2009) Plastic design of eccentrically braced frames, I: momentshear interaction.
J Constr Steel Res 65:10071014
Mazzoni S, McKenna F, Scott MH et al. (2007) Opensees command language manual
Melchers RE (2002) Structural reliability analysis and prediction, 2nd edn. Wiley, Chichester
Menegotto M, Pinto P (1973) Method of analysis for cyclically loaded reinforced concrete plane frame
including changes of geometry and non elastic behavior of elements under combined normal force and
bending. In: IABSE Symposium on resistance and ultimate deformability of structures acted on by well
defined repeated loads, final report. Lisbon, Portugal
Pinto PE, Gainnini R, Franchin P (2004) Seismic reliability analysis of structures. IUSS Press, Pavia
Popov EP, Engelhardt MD (1988) Seismic eccentrically braced frames. J Constr Steel Res 10:321345
Porter K, Scawthorn C, Taylor C, Blais N (1998) Appropriate seismic reliability for critical equipment
systemsrecommendations based on regional analysis of financial andlife loss. NCEER project numbers
946201 and 956201, MCEER, State University of New York at Buffalo, USA
RFCS (2011) Steel RTD programmeoptimizing the seismic performance of steel and steel-concrete struc-
tures by standardizing material quality control. In: Braconi A (Ed) RFCS Contract: RFSR-2007-CT-00039,
Brussels (Belgium): European Commission
Rides JM, Popov EP (1993) Inelastic link element for EBF seismic analysis. J Struct Eng 120:441463
Rossi PP, Lombardo A (2007) Influence of the link overstrength factor on the seismic behaviour of eccentrically
braced frames. J Construct Steel Res 63:15291545
Somja H, Nofal S, Hjiaj M, Degee H (2013) Effect of the steel material variability on the seismic capacity
design of steel-concrete composite structures: a parametric study. Bulletin of Earthquake Engineering,
published online: 12 February 2013
Tamast G (1977) Bounds for probability in multivariate normal distribution. In: I.S.I. Proceedings pp 203204
Technical Commission 250 (2005) UNI-EN1990: Eurocode Basis of structural design. CEN, Brussels
Tremblay R (2002) Inelastic seismic response of steel bracing members. J Constr Steel Res 58:665701
UNI EN 19981:2005. (2005) Eurocode 8: design of structures for earthquake resistance. Part 1: general rules,
seismic action and rules for buildings. European Committee for Standardisation, Brussels
UNI EN 19931:2005. (2005) Eurocode 3design of steel structurespart 11: general rules and rules for
buildings
123
Bull Earthquake Eng (2013) 11:22492285 2285
Vanmarcke EH, Gordon GA, Heredia-Zavoni E (1999) SIMQKE-II, conditioned earthquake ground motion
simulator: users manual, version 2.1. Princeton University, USA
Veneziano D (1976) Basic principles and methods of structural safety. Bull DInformation, Comit Europen
du Bton, Paris 112:212288
Veneziano D, Casciati F, Faravelli L (1983) Methods of seismic fragility for complicated systems. In: Pro-
ceedings of the second conference on the safety of nuclear installations (CSNI), specialist meeting on
probabilistic methods in seismic risk assessment for NPP. Livermore, California
123