Journal of Constructional Steel Research 101 (2014) 437–454

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Seismic performance of dual-steel moment resisting frames

Andre Tenchini a, Mario D'Aniello b, Carlos Rebelo a, Raffaele Landolfo b,⁎, Luis Simões da Silva a, Luciano Lima c
a
ISISE, Civil Engineering Department, University of Coimbra, Pinhal de Marrocos, 3030-201 Coimbra, Portugal
b
Department of Structures for Engineering and Architecture, University of “Federico II” of Naples, 80134 Naples, Italy
c
Faculty of Engineering, State University of Rio de Janeiro, Maracanã, 20550-013 Rio de Janeiro, Brazil

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

Article history: In seismic design of steel structures, the “dual-steel” concept concerns the combined use of High Strength Steel
Received 31 December 2013 (HSS) in non-dissipative members and Mild Carbon Steel (MCS) in dissipative zones, in order to control the glob-
Accepted 13 June 2014 al frame behavior into a ductile overall failure mode. In this paper, a comprehensive parametric study devoted to
Available online xxxx
investigate the seismic design and performance of Eurocode 8 compliant dual-steel Moment-Resisting Frames
(MRF) is presented and discussed. The overall seismic performance has been analysed through static and dynam-
Keywords:
Steel moment resisting frames
ic nonlinear analyses against three limit states: damage limitation (DL), severe damage (SL) and near collapse
High strength steel (NC). The investigated parameters cover both geometric and mechanical variables, as the type columns, span
Dual-steel frames length, number of storeys and spectral shape. The analyses showed that the use of HSS in Eurocode 8 compliant
Seismic performance based design MRFs is effective to provide overall ductile mechanism, but it may lead to inefficient and uneconomical structures
Nonlinear analysis characterized by limited plastic demand due to the large design overstrength. The comparison between dual steel
MRFs with those entirely made of MCS showed that: i) in order to fulfil the codified drift requirements and to
limit the stability coefficients, the same shapes for members should be used for both structures in the most of
cases; ii) a similar performance can be recognized in both dual steel and single grade steel structures under
both damage limitation and significant damage limit state; and iii) dual steel frames guarantee a better control
of plastic mechanism than single grade steel frames at near collapse limit state.
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction
In recent years, significant development in steel processing oc-
curred. Indeed, the improvements in industrial processes by the combi-
nation of rolling practices and cooling rates allowed obtaining high
strength steel (HSS) with very attractive properties. Owing to the high
performance, the use of HSS has a number of benefits in terms of eco-
nomic, architectural, environmental and safety aspects where the in-
crease of strength allows a size reduction of the structural members
enabling potential benefits also in terms of environment impact by sav-
ing energy and reduction of gas emissions [1]. Nowadays, in Japan and
USA, HSSs are widely used for bridge and building construction in seis-
mic zones. On the contrary, in Europe, there is still a limited number of
applications on buildings, and even less in seismic regions, although
some examples exist of use of HSS in bridge structures [2].
Seismic applications potentially represent the rational field to ex-
ploit the high performance of HSS. Indeed, according to modern codes
the seismic design of steel or composite buildings are based on the con-
cept of dissipative structures, where specific zones of the structures
should be able to develop plastic deformation, mainly on ductile mem-
ber, in order to dissipate the seismic energy. On the contrary, the non-
⁎ Corresponding author.
E-mail address: landolfo@unina.it (R. Landolfo).
dissipative zones and members should behave elastically under seismic
action in order to avoid the brittle collapse of the building. For this rea-
son, these zones should be designed to resist the full plastic strength of
the dissipative members. Consequently, the large overstrength de-
mands to non-dissipative zones lead to high material consumption,
and sometimes, huge size of members to fulfil this design requirement.
The combined use of HSS for non-dissipative members and of mild
carbon steel (MCS) for dissipative members may allow an easier appli-
cation of capacity design criteria. The expected design improvement
would be obtained in terms of smaller member sizes than those obtain-
ed when using MCS only. Structures designed using the combination of
HSS and MCS are termed “dual-steel” structures.
Recent studies [3–5] have highlighted the advantages of dual-steel
concept, especially for what concerns the control of seismic response
of multi-storey buildings to achieve overall ductile mechanism. In
particular, Dubina et al. [5] showed the potential benefits given by
using HSS in full strength moment-resisting steel beam-to-column
joints, in order to guarantee the formation of plastic hinge in the beam
and preserving both the connection and the column in moment
resisting frames (MRFs).
On the other hand, notwithstanding the fulfilment of codified rules
such those given by EN1998-1 [8], the design of MRFs with HSS columns
may lead to obtain structures with lateral stiffness smaller than those
designed with a single steel grade. Hence, some problems may arise

http://dx.doi.org/10.1016/j.jcsr.2014.06.007
0143-974X/© 2014 Elsevier Ltd. All rights reserved.
438 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454

2. Framework of the study

Nomenclature
2.1. Investigated parameters
agR reference peak ground acceleration
k coefficient depending on seismicity (generally assumed
Two sets of twenty-four MRFs were designed in accordance with
equal to 3)
EN1998-1 [8]. The first set of frames covers structures designed using
q behavior factor
S460 (HSS) for columns and S355 (MCS) for beams. The second set con-
s beam flexural overstrength
sists of frames made of S355 grade for both beams and columns. The
A acceleration from IDA
comparison between these two sets allows examining the potential
Ad design acceleration
benefit of dual steel frames as respect to those made of single steel
Ay acceleration corresponding to the yielding of the frame
grade. For both ensembles of structures the investigated design param-
Aθ acceleration corresponding to the maximum permitted
eters are shown as follows:
interstorey drift ratio
Ac acceleration corresponding to the column plastification
• number of storeys: 4-storey and 8-storey frames, as shown in Fig. 1,
AR acceleration corresponding to the maximum permitted
where the height of first floor is 4.0 m and all other are 3.5 m;
local rotation
• span length: 5.0 m and 7.5 m.
Au acceleration related to the failure
• composite steel–concrete column typologies (see Fig. 2): fully
H height of building
encased (FE), partially encased (PE) and concrete filled tube (CFT);
L bay length
• corner period of the design spectra: two types of soil conditions have
Μu peak bending moment experienced by the beam
been examined. The former representative of soil type C according to
Μp plastic bending moment on the beam
EN1998-1 [8] (hereinafter identified as “stiff soil”) and the latter
Mmax maximum bending moment on the beam
representative of very soft soil conditions with corner period of
Mj,bi,Ed bending moment in the beam at the intersection of the
1.6 s, which are representative of specific soil condition in Bucharest
member centerlines into the joints on the side “i”.
(hereinafter identified as “soft soil”);
TL return period
TLR reference return period for which the reference seismic
In order to identify each structure, a label code has been given to the
action may be computed
frames as follows:
Vy base shear from of the yield strength of the structure
V1y base shear at the formation of the first plastic hinge MRF ðStoreyÞ  ðSpanÞ  ðSoilÞ  ðColumnÞ
Vd design base shear
αu multiplier of horizontal seismic design action at forma-
where:
tion of global plastic hinge in the system
α1 multiplier of horizontal seismic design action at forma-
• (Storey) = 1 for eight storeys; 2 for four storeys.
tion of first plastic hinge in the system
• (Span) = 1 for 5.0 m span; 2 for 7.5 m span.
β transformation factor according to EN1993:1–8
• (Soil) = 1 for stiff soil; 2 for soft soil.
δ1y displacements corresponding to the formation of the
• (Column) = 1 for fully encased sections; 2 for partially encased
first plastic hinge
sections; 3 for concrete filled tube sections.
δmax roof displacement corresponding to the achievement of
an interstorey drift ratio equal to 3%
The frames integrate the structural system of buildings (Fig. 2) that
γI importance factor
comply with the regularity conditions in plan and elevations, according
γov material overstrength factor
to EN1998-1 [8]. As shown in Fig. 3 the building is intended as braced in
γmaterial density of material total mount
the transverse direction. Floors consist of orthogonal steel beams
μ ductility factor
(primary and secondary) and steel–concrete composite columns. Pri-
ν reduction effect accounting the lower return period of
mary beams are considered braced to avoid lateral–torsional buckling
the seismic action
in order to allow for plastic deformation in bending. All beam–column
θy beam yield rotation
joints were assumed as full strength and full rigid. The columns were
Ω overstrength factor
considered to be fixed at the base and continuous through the height.
In plan, columns are spaced at L = 7.5 m or L = 5.0 m in both directions,
as depicted in Fig. 2.

such as the nonfulfillment of damage limitation requirements and over-

all stability problems due to P-Delta effects. These considerations clearly
show the need to investigate the effectiveness of dual-steel concept in
capacity design of MRFs. To this aim, in this paper a wide parametric nu-
merical study has been carried out to evaluate the benefits of dual-steel
concept on the performance based seismic design [6,7] of MRF compli-
ant to EN1998-1 [8]. Furthermore, the overall seismic performance has
been analysed through static and dynamic nonlinear analyses against
three limit states as defined in EN1998-3 [9]: damage limitation (DL),
severe damage (SL) and near collapse (NC). After describing the results
of numerical analyses, the discussion looks for insights at the following
issues: (i) quantification of performance parameters for each limit
states, (ii) characterization of the behavior factors at each limit states, a) 4-Storey b) 8-Storey
(iii) comparison between dual steel MRFs and single grade steel MRFs,
and (iv) assessment of economic efficiency. Fig. 1. Structural scheme of study cases.
 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454
(FE) (PE) (CFT)
Fig. 2. Examined typologies of composite columns.
2.2. Performance levels
According to EN 1998, the seismic hazard is expressed in terms of
the value of the reference peak ground acceleration agR on bedrock
(namely ground type A as defined in the code). For each seismic zone
agR corresponds to the reference probability of exceedance in 50 years
of the seismic action for the no collapse requirement. To this reference
ground motion an importance factor γI = 1.0 is assigned and the design
ground acceleration agd is expressed as agd = γI agR for ground of type A.
The value of the importance factor γI multiplying the reference seismic
action to achieve the same probability of exceedance in TL years as in the
TLR years for which the reference seismic action may be computed as:
 −1=k
T LR
γI ¼
TL
where the exponent k depends on seismicity, but being generally
assumed equal to 3.
The seismic performance of study cases is evaluated for three perfor-
mance levels, which are associated to different annual rate of exceed-
ance: damage limitation (DL), severe damage (SD) and near collapse
(NC).
According to EN1998-3 [9] at the DL state the building is subjected to
the frequent earthquake with 95-year-return period, the structure shall
have no occurrence of damage and the associated limitations of use.
EN1998-3 [9] presented a reduction factor ν to take into account the de-
termination of the frequent earthquake from the design earthquake. In
the examined cases ν is equal to 0.5 and the corresponding structural
performance should provide interstorey drift ratios lesser than 0.75% [8].
The SD state corresponds to design condition where the structure
shall have no local or global collapse under the design seismic action
with 475-year-return period. At this performance level, the structure
is strongly damaged but has some residual lateral strength and stiffness
and vertical elements are capable of sustaining vertical loads, thus
providing the strength to sustain moderate after-shocks. To verify the
effectiveness of the structural performance, a limit value for residual
drift equal to 0.4% has been assumed.
bracing system
L single grade frames at the same design parameters. This outcome was
L
studied MRFs
L In order to assess the seismic behavior of frames, both nonlinear
L L L The models were developed using the force-based (FB) distributed
Fig. 3. Plan configuration for all the frames.
439
At the NC state the structures are expected to be heavily damaged,
with negligible residual lateral strength and stiffness, although vertical
elements are still capable of sustaining vertical loads. Large permanent
drifts are present. The structures are near collapse and are not able to re-
sist to moderate earthquake after-shocks. In this study is assumed to
verify this performance level against a seismic action with 2475-year-
return period. An interstorey drift ratio of 3.0% has been assumed to
characterize this damage scenario as early proposed by Grecea et al. [7].
Table 1 summarizes the performance levels used for each of the
three limit states where acceleration ratio Ad corresponds to peak
ground acceleration used in design.
3. Seismic design
The frames were designed in accordance with EN1998-1 [8],
EN1993-1 [10] and EN1994-1 [11]. In the design calculations, first
order elastic analyses have been carried out with amplification of rele-
vant action effects to account for P-Delta effects. In addition, the effects
of initial sway imperfection have been taken into account by systems of
equivalent horizontal forces as indicated by EN1993:1-1 [8].
As damage limitation requirement, the maximum interstorey drift
ratio was fixed equal to 0.75%, thus assuming ductile non-structural
elements.
Dead and live loads equal to 4.0 kN/m2 and 3.0 kN/m2, respectively,
were considered. The peak ground acceleration was assumed equal to
0.24 g for stiff soil and 0.16 g for soft soil. This assumption was mainly
due to investigate frames representative of realistic building structures,
ð1Þ thus achieving two objectives: 1) to have force demand large enough to
emphasize the need of using HSS for non-dissipative elements; and 2)
to limit the drift demand, because increasing the lateral stiffness leads
the use of HSS meaningless. Hence, a larger PGA allowed to pursue the
first objective in case of stiff soil condition, while a smaller PGA was nec-
essary to achieve the second objective in case of soft soil condition.
The reference behavior factor was assumed to be q = 4.0, which cor-
responds to the recommended value for “Ductility Class Medium”
(DCM) concept in EC8. According to EN1998-1 philosophy the design
for DCM concept ensures the formation of a stable and trustworthy dis-
sipative mechanism in predefined critical areas, by using a behavior fac-
tor q N1.5 and by detailing the critical zones in order to guarantee good
ductility and hysteretic behavior without the occurrence of a fragile rup-
ture. It should be noted that although higher values for the behavior fac-
tor q are proposed by EC8, in order to exploit the potentially large
ductility and energy dissipation capabilities of MRFs, the use of lower
behavior factors is generally the more rational design choice as shown
by [12,13] particularly to design frames in low-to-moderate seismicity
regions. Fig. 4 shows both the elastic and the design spectra for the
frames located on stiff or soft soil.
The design forces have been calculated by means of standard modal
response spectrum analyses, according to EN1998-1 [8], where all
modes of vibration that contribute significantly to the global response
were taken into account. The designed cross sections for beam and col-
umns and the relevant first and second natural periods per structure are
reported in Table 2. It should be noted that the design process resulted
in identical member sizes for the most of cases with either dual steel and
mainly due to the need to fulfil the requirements for the limitations of
both interstorey drift ratios and stability coefficients.
4. Numerical models
static and dynamic analyses have been performed using the software
SeismoStruct [14].
inelasticity elements [15,16]. These elements account for distributed in-
elasticity through integration of material response over the cross
440 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454

Table 1 EN 1992:1-1 [21], while different values of material overstrength factor

Performance levels. (γov) for each steel grade. In particular, γov equal to 1.25 was assumed
Limit state Return A/Ad Failure criteria for S355, while 1.10 for S460 [22].
period The numerical integration method used is based on the Gauss–
(years) Lobatto distribution [23,24], which includes, at a minimum, monitoring
Damage limitation (DL) 95 0.50 0.75% interstorey drift ratio points at each end of the element. Such feature allows each structural
Severe damage (SD) 475 1.00 0.40% residual interstorey drift ratio member to be modelled with a single FE element, thus requiring no
Near collapse (NC) 2475 1.72 3.0% interstorey drift ratio
meshing for each element. In the present study, 5 Gauss–Lobatto inte-
gration points have been used.
Second order effects have been accounted in all analyses presented
section and integration of the section response along the length of the in this paper, by assuming large displacements/rotations and large inde-
element. The cross-section behavior is reproduced by means of the pendent deformations relative to the chord of the frame element
fiber approach, assigning a uniaxial stress–strain relationship at each through the employment of the co‐rotational formulation given by [25].
fiber. The validity of the modelling assumptions has been verified against
The stress–strain relationship for concrete fibers in the column ele- the experimental results carried out by D'Aniello et al. [26] on steel
ments was determined using the model proposed by Martinez-Rueda beams, and by Wakawayashi et al. [27] on a single storey steel MRF.
and Elnashai [17]. The effects of confinement provided by steel profile The comparison between experimental and numerical response curves
and/or reinforcement has been determined according to Mander et al. are depicted in Fig. 5, where it can be noted the satisfactory accuracy of
[18] and Susantha et al. [19], for fully/partially encased and concrete the implemented numerical models. It should be noted that the
filled tube, respectively. In the case of steel members, the model pro- effectiveness and the accuracy of the numerical models is proved prior
posed by Menegotto and Pinto [20] for the stress–strain curve was degradation phenomena may occur in the beams. According to experi-
chosen. mental and analytical studies carried out by D'Aniello et al. [26] and
The average values of both concrete compression strength and steel more recently by Beg et al. [28] within the framework of European
yield stress have been used. The former has been assumed according to RFCS project HSS-SERF, the value of total beam rotation without

a) Elastic spectra b) Design spectra

Fig. 4. Elastic and design spectra for both stiff and soft soil.

Table 2
Sectional properties of columns and beams.

Frames Columns Beams Periods

1st to 4th floor 5th to 8th floor 1st to 2th floor 3th floor 4th floor 5th to 6th floor 7th to 8th floor First Second

MRF_1.1.1.1 HEB 280 HEB 260 IPE 360 IPE 360 IPE 300 IPE 300 IPE 240 1.91 0.69
MRF_1.1.1.2 HEB 300 HEB 280 IPE 360 IPE 360 IPE 300 IPE 300 IPE 240 1.92 0.70
MRF_1.1.1.3 SHS 300 × 16 SHS 300 × 12 IPE 360 IPE 360 IPE 300 IPE 300 IPE 240 1.84 0.63
MRF_1.1.2.1 HEB 340 HEB 300 IPE 400 IPE 400 IPE 360 IPE 360 IPE 240 1.53 0.58
MRF_1.1.2.2 HEB 360 HEB 320 IPE 400 IPE 400 IPE 360 IPE 360 IPE 240 1.53 0.58
MRF_1.1.2.3 SHS 500 × 20 SHS 450 × 16 IPE 400 IPE 400 IPE 360 IPE 360 IPE 240 1.41 0.51
MRF_1.2.1.1 HEB 600 HEB 500 IPE 450 IPE 450 IPE 400 IPE 400 IPE 360 1.89 0.60
MRF_1.2.1.2 HEB 650 HEB 550 IPE 450 IPE 450 IPE 400 IPE 400 IPE 360 1.86 0.59
MRF_1.2.1.3 SHS 550 × 25 SHS500 × 25 IPE 450 IPE 450 IPE 400 IPE 400 IPE 360 1.89 0.60
MRF_1.2.2.1 HEB 650 HEB 550 IPE 550 IPE 550 IPE 500 IPE 500 IPE 360 1.43 0.51
MRF_1.2.2.2 HEB 700 HEB 600 IPE 550 IPE 550 IPE 500 IPE 500 IPE 360 1.42 0.50
MRF_1.2.2.3 SHS 600 × 25 SHS550 × 25 IPE 550 IPE 550 IPE 500 IPE 500 IPE 360 1.44 0.50
MRF_2.1.1.1 HEB 280 – IPE 300 IPE 270 IPE 270 – – 1.08 0.32
MRF_2.1.1.2 HEB 300 – IPE 300 IPE 270 IPE 270 – – 1.10 0.33
MRF_2.1.1.3 SHS 300 × 16 – IPE 300 IPE 270 IPE 270 – – 1.11 0.34
MRF_2.1.2.1 HEB 320 – IPE 300 IPE 270 IPE 270 – – 1.02 0.29
MRF_2.1.2.2 HEB 360 – IPE 300 IPE 270 IPE 270 – – 1.02 0.29
MRF_2.1.2.3 SHS 400 × 12 – IPE 300 IPE 270 IPE 270 – – 1.02 0.28
MRF_2.2.1.1 HEB 400 – IPE 400 IPE 360 IPE 360 – – 1.10 0.31
MRF_2.2.1.2 HEB 450 – IPE 400 IPE 360 IPE 360 – – 1.08 0.31
MRF_2.2.1.3 SHS 400 × 20 – IPE 400 IPE 360 IPE 360 – – 1.12 0.33
MRF_2.2.2.1 HEB 500 – IPE 400 IPE 360 IPE 360 – – 1.00 0.26
MRF_2.2.2.2 HEB 550 – IPE 400 IPE 360 IPE 360 – – 0.99 0.26
MRF_2.2.2.3 SHS 500 × 20 – IPE 400 IPE 360 IPE 360 – – 1.01 0.27
 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454 441

a) b)
150 250
Experimental response
200

Bending Moment (kNm)

Calibrated model
150

Base shear (kN)

100
50
50
0
-50
-50 -100
-150
Experimental response
-200
Numerical response
-150 -250
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 -4.50% -3.00% -1.50% 0.00% 1.50% 3.00% 4.50%
rotation (rad) drift ratio (%)

Fig. 5. Comparison between numerical and experimental curves: cyclic tests on beams by D'Aniello et al. [26]; cyclic tests on single storey MRF by Wakawayashi et al. [27].

degradation of maximal strength is larger than the stringent limit value because it corresponds to the failure criteria assumed for NC limit state,
of 0.035 rad from EN 1998-1. Moreover, if full strength and full rigid as reported in Table 1.
welded beam-to-column joints are designed in order to guarantee Fig. 7 shows the capacity curves from pushover analyses for both
that plastic hinges develop away from the beam-to-column connec- patterns of lateral load, obtained normalizing the base shear V with
tions, the value of total joint rotation is larger than the limit value of the design base shear Vd. As it can be noted all frames possess an actual
0.04 rad for structures classified as special moment resisting frames as capacity considerably higher than that assumed in design. This effect is
in AISC 341-10 [29]. Therefore, it can be assumed that the models likely due to the design procedure which is governed by the require-
used in the present study may be considered valid within the range of ments to satisfy both interstorey drift limits and to control stability co-
± 0.04 rad. Moreover, since the performance limit for Near Collapse efficient. The comparison of response curves allows highlighting the
state has been assumed equal to 3%, the numerical results beyond that influence of the investigated parameters. In particular, for what con-
limit have been kept out of interest for the present study. cerns the influence of number of storeys it can be noted the 4-storey
Concerning dynamic analyses a 2% Rayleigh tangent stiffness frames experience larger V/Vd ratios than the 8-storey frames. This im-
damping has been used at both first and second mode. Indeed, different- plies that the smaller is the number of storeys and the larger is the de-
ly from initial damping formulation, it does not produce artificial sign overstrength.
overdamping at high ductility demand as shown by [30–32]. In the examined cases (namely within the range of 5–7.5 m) the in-
fluence of span length is found to be less significant. Frames with longer
5. Nonlinear analyses spans showed slightly larger overstrength. This result is ascribable to
the design process. Indeed, longer beams imply larger beam depth,
5.1. Pushover thus leading to stiffer and stronger columns to satisfy capacity design
rules.
The pushover analyses were carried out applying two types of lateral The frames under 1st mode pushover showed an almost uniform
load distribution, the first proportional to the first mode and the second interstorey drift distribution till the formation of the first plastic hinge,
proportional to the masses along frame height. while pushover analyses under uniform load pattern showed a pro-
The response parameters monitored by the performed pushover nounced concentration of drifts at lower storeys also in elastic range.
analyses are illustrated in Fig. 6. In particular, Vy refers to the yield At ultimate state for both 1st mode and uniform pushover lower storeys
strength of the structure, V1y is the base shear at the formation of the exhibit significant inter-storey drifts compared to upper levels, but soft
first plastic hinge and Vd corresponds to the design base shear; δ1y and storey mechanism occurred only for the uniform load pattern at the first
δmax are the roof displacements corresponding to the formation of the storey.
first plastic hinge and to the first occurrence of an interstorey drift The inelastic distribution and the sequence of plastic hinges are
ratio equal to 3%, respectively. The latter parameter has been monitored pointed out by the numbering close to the symbol of plastic hinge (i.e.
bold circle) shown in Fig. 8, where it can be observed both the effective-
3% interstorey drift ratio
roof displacement at the

ness of capacity design criteria and the influence of the building aspect
the first plastic event
roof displacement at

ratio (namely the ratio between the building height and the width) on
first occurrence of a

the damage evolution up to the attainment of the defined collapse

limit. The former is evidenced by the weak-beam/strong-column be-
havior, being all plastic hinges located on ends of beam and column
base. The influence of the latter can be recognized by the different distri-
bution of damage along the building height. Indeed, the shorter is the
Vy ultimate strength bay and the larger is the plastic beam rotation at the same total rotation
demand as compared to a beam belonging to longer span. In addition,
the structures with the shorter span (i.e. 5.0 m) fully developed the
plastic mechanism with plastic hinge formed at all beam ends and at
V1y shear force at the first plastic event
column bases, as well. On the contrary, the frames with the longer
design base shear span (i.e. 7.5 m) do not achieve the complete formation of plastic mech-
Vd
anism, even though the damage redistribution was fully satisfactory.
This difference is due to the lateral stiffness and to the profile of lateral
δ 1y δmax displacements exhibited by both ensembles of frames. Indeed, the
structures with 5.0 m spans showed a cantilever dominant behavior,
Fig. 6. Parameters monitored in pushover analyses. thus leading to large rotation demand on the top. Conversely, those
442 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454

Fig. 7. Normalized pushover response curves.

frames with 7.5 m spans exhibited a shear-type dominant response,
that is characterized by large rotation demand at lower storeys.
5.1.1. Overall overstrength factor
The overall overstrength factor Ω is defined as the ratio between the
base shear corresponding to the overall yield strength of the frame and
the design base shear. This ratio can be decomposed in two terms:
Vy V y V 1y
Ω¼ ¼ 
V d V 1y V d
where the first term, Ω1 = Vy/V1y, corresponds to αu/α1 defined in the
EN1998-1 [8]. This value depends on the frame configuration, formation
of the collapse mechanism, redistribution capacity and gravity loading
[12]. The second term Ω2 = V1y/Vd is related to aspects of the design
procedure such as differences between actual and nominal material
strength, member oversizing due to choices of commercial cross-
section and design governed by deformation and/or non-seismic
loading. Therefore, this term may be associated to the degree of special-
ization of frame for seismic resistance [33].
The pushover curves obtained under uniform load distribution
give the lower values with a mean value of Ω1 within the range of
ð2Þ
1.33–1.38 (see Fig. 9a), thus slightly larger than 1.30 recommended by
EN1998-1 [8].
Concerning the column type, no substantial differences can be rec-
ognized in terms of lateral capacity. Only the cases with FE columns
 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454
MRF_1.2.2.1 MRF_1.1.2.1
MRF_2.2.2.1 MRF_2.1.2.1
a) 7.5 m span b) 5 m span
Fig. 8. 1st mode pushover: damage distribution.
exhibited slightly smaller base shear ratio. Indeed, although designed to
satisfy hierarchy criteria, FE columns are characterized by slightly small-
er flexural stiffness than the relevant PE and CFT sections. As a conse-
quence, in such cases the distribution of damage tends to be less
uniform than in the frames with stiffer columns. Hence, the damage
tends to be concentrated at the lower storeys, thus resulting in a smaller
overstrength by plastic redistribution. The other parameters have no ap-
preciable influence on the Ω1 factors.
For what concerns the second term of Eq. (2), the pushover curves
obtained with 1st mode load distribution show the lower Ω2 factors.
Fig. 9b clearly shows that significantly large values have been obtained.
Indeed, the mean of (V1y/Vd) ranges from 3.14 to 4.27, thus very close to
the design behavior factor (q = 4). This result is ascribable to the need
to satisfy the code drift requirements, which compels to oversize the
structural members to provide adequate lateral stiffness. This issue
often arises in the design of ductile seismic resistant MRFs. However,
in case of dual-steel frames this design procedure is even more neces-
sary. Indeed, although the use of HSS allowed to guarantee the hierarchy
criteria, the higher is steel grade of non-dissipative members the smaller
is the corresponding size of cross sections, thus potentially resulting in
very flexible frames. Analysing the influence of design parameters, no
appreciable difference can be recognized for Ω2 factors. However,
some differences for tall frames (i.e. 8-storey) can be observed for differ-
ent soil types. Indeed, in the cases with stiff soil the design base shear is
about 20% smaller than in the cases with soft soil (see Fig. 4b), while the
design displacement demand are comparable. Hence, in order to satisfy
the drift limitations the selected cross sections were proportionately
larger for frames located on stiff soils, thus resulting in the larger nor-
malized base shear ratios.
This effect can be also recognized in the overall Ω factors. Indeed, as
depicted in Fig. 9c the mean Ω factors are smaller for frames designed
443
for soft soil than those for stiff soil. However, it should be noted that
4-storey frames exhibit larger Ω than 8-storey ones, namely the former
from 5.43 to 5.97 and the latter from 4.25 to 4.97. This pronounced dif-
ference is due to the fact that the design forces for shorter frames are
quite small while the displacement demand is proportionally larger,
thus clarifying the reason for so large values of overstrength.
5.1.2. Overall ductility factor
The overall ductility factor (μ) has been computed for all frames in
order to evaluate the plastic displacement capacity that dual steel
frames can provide till the achievement of an interstorey drift ratio
equal to 3%, that has been assumed as limit of performance at NC limit
state. The μ factor is defined as follows:
δmax
μ¼ ð3Þ
δ1y
where δmax is the roof displacement when an interstorey drift ratio
equal to 3% is achieved into the frame and δ1y is the roof displacement
associated to formation of the first plastic hinge (see Fig. 6).
Fig. 9d shows the overall ductility factors obtained from both 1st
mode and uniform pattern pushovers. As it can be noted both 4- and
8-storey frames exhibit similar values for (μ) with an average equal to
2.03 for the former and 2.11 for the latter.
This limited value of ductility is mainly due to relatively high values
of interstorey drift ratio corresponding to the formation of the first plas-
tic event, which ranges from 1.2% to 1.6% of the storey height with an
average value equal to 1.42%. Two combined reasons produced this
behavior, which are: i) the large flexibility of the frames and (ii) the no-
ticeably high design overstrength. Anyway, some slightly differences
can be recognized comparing frames designed for stiff soil with those
for soft soil. Indeed, the former cases showed lower values of the overall
ductility factor (about 1.97 for 4-storey and 1.96 for 8-storey buildings,
respectively) than the second set of frames (about 2.16 for 4-storey
and 2.36 for 8-storey buildings, respectively). This result is a direct
consequence of the larger design base shear for frames under soft soil
conditions, which leads having stronger and stiffer structures that are
characterized by the smaller yield drift ratios.
For what concerns the influence of span length, the numerical re-
sults show that low-rise frames with the shorter span experience the
larger lateral displacement capacity. On the contrary, in taller frames
those with the longer span show the larger ductility factor. Regarding
the column type, no appreciably differences can be recognized compar-
ing the ductility factors obtained for the examined frames.
5.2. Incremental dynamic analyses
Two sets of both artificial and natural accelerograms have been used
to perform nonlinear dynamic analyses. These records have been select-
ed to match the elastic spectra given by EN1998-1 [8] according to the
procedure described in [34]. Fig. 10 depicts the comparison between
the codified response spectra and those given by each record and the
relevant mean spectrum, while the properties of the selected records
are reported in Table 3.
Incremental dynamic analyses (IDAs) were carried out scaling the
PGA up to 12 times the design value, as follows:
• from 0.2 PGA to 1.2 PGA with a scaling step of 0.2
• from1.2 PGA to 12.0 PGA with a scaling step of 0.4
In addition, nonlinear dynamic analyses were performed at 0.5 PGA
in order to assess the damage limitation state.
Nonlinear incremental dynamic analyses were performed to charac-
terize both the inelastic behavior and the performance against the three
reference limit states defined by EN1998-3 [9]. Fig. 11 depicts the capac-
ity curves in terms of normalized base shear (i.e. base shear divided by
444 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454

a)

b)

c)

d)
4-Storey 8-Storey

Fig. 9. Performance parameters from Pushover: a) Overstrength factor (Vy/V1y); b) Overstrength factor (V1y/Vd); c) Overstrength factor (Ω); Overall ductility factor (μ).

a) Stiff soil b) Soft soil

Fig. 10. Response spectra of selected accelerograms vs. EC8 response spectra.
 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454 445

design base shear) and peak roof displacement obtained from incre-

Scaling factor
mental dynamic analyses. As expected, the maxima V/Vd ratios from
IDAs are within the range defined by the 1st mode and uniform push-

1.053
0.957

1.249
1.671
1.567
1.401
1.372
1.100
over curves. In addition, consistently within the results given by push-

1
1
1
1
1
1
Significant duration over analyses, the 4-storey frames are characterized by V/Vd ratios
larger than those for 8-storey frames. This result confirms the consider-
ations about the significant influence of design criteria highlighted by
between 5 and
95% of AI (s)
pushover analyses. In line with that, it can be noted that the smaller V/
Vd ratios are obtained for 8-storey frames under soft soil conditions, be-
11.12
12.85
13.25
13.93
13.22
13.41
35.27
6.85

37.92
50.03
26.49
7.43
28.45

19.90
cause those frames are characterized by the larger seismic demand.
In the next sections, the performance evaluation of dual steel frames
are described and discussed at the light of outcomes from IDAs.
Arias Intensity (AI)

6. Performance evaluation
0.130
0.319
0.150
0.218
0.217
0.227
0.121
0.372

0.184
0.154
0.181
0.182
0.266
0.193
The performance indicators monitored for all limit states are the fol-
lowing: i) peak interstorey drift ratios; ii) residual interstorey drift ra-
tios; iii) peak storey accelerations; iv) beam ductility demand; and v)
Arias Intensity (AI)
Scaled to PGA = 1

beam flexural overstrength. The results are presented hereinafter.

6.1. Peak interstorey drift ratios

0.107
0.223
0.223
0.155
0.173
0.127
0.284
0.205

0.424
0.339
0.426
0.108
0.353
0.4

Fig. 12 depicts the median value of interstorey drift ratio (IDR) de-
mand along the building height for the three limit states, while Table 4
Epicenter distance (km)

reports the values obtained for the 16th, 50th and 84th percentiles.
As a general remark, the examined design parameters do not high-
light appreciable influence on the IDR demand. However, it is interest-
ing to note that the numerical results show an almost uniform
distribution of IDR along the building height, expect for the first storey.
80
161

13
65

127
228
72
26

The IDR demand at each limit state was fairly lower than the perfor-
mance limits (see Table 1). It is interesting to note that although the
Foundation category

seismic design of the examined frames has been influenced by drift lim-
itation, the median values of IDR at DL state are significantly lower than
the limit of 0.75%. At DL state all frames behave in elastic field, being the
Alluvium

yield IDR larger than 1% in most of the cases. Thus, consistently with the
Stiff soil
Stiff soil

Stiff soil
Soft soil

results from pushover analyses, a very limited inelastic demand can be

Rock

Rock
Rock

observed at both SD and NC limit state. In particular, the median IDR

for the SD also in the range of 1%–1.2% for 4- and 8-storey frames,
magnitude Ms
Surface wave

thus significantly lower than the limit of 3.0%. Such results show that
dual-steel solution may lead to inefficient and uneconomical structures.
However, such considerations have been also drafted for EN1998-1 [8]
7.32
6.87

7.33

6.87

6.75
7.05

6.87

6.87

compliant MRFs made of one steel grade [33]. The reason of that may
be found in the severe requirements for both drift limitations and stabil-
Focal depth

ity criteria.
(km)

19
16

10
86

16
16

16

6.2. Residual interstorey drift ratios

Erzincan-Meteorologij Mudurlugu (Turkey)

The residual interstorey drift ratios (RIDR) have been monitored at

each limit state, because they provide useful data on the damage distri-
bution and on the post-quake reparability of the frames.
As discussed in the previous section, at DL state all frames behave
Bucharest Building Res Inst.

elastically and no linear events were observed, except for some cracking
Measurement location

Torre del Greco (Italy)

Roccamonfina (Italy)

in the composite columns. Hence, the RIDRs are close to zero at this limit
Artificial by SIMQKE
Artificial by SIMQKE
Artificial by SIMQKE
Artificial by SIMQKE
Artificial by SIMQKE
Artificial by SIMQKE

state.
Tricarico (Italy)
Kashmar (Iran)

Bisaccia (Italy)
Rudsar (Iran)

Small RIDRs have been recognized at SD and NC limit states, as

(Romania)

depicted in Fig. 13 and Table 5 the maximum values are lower than
the limit of 0.40%. In particular, the maximum median RIDRs are about
Basic data of the selected records.

0.05% and 0.19% at SD and NC limit state, respectively. It is interesting

to observe that, owing to the smaller design overstrength, all 8-storey
1992/03/13
1980/11/23
1977/03/04

1978/09/16

1980/11/23
1990/06/20
1980/11/23

1980/11/23

frames showed the larger demand and consequently residual drift ratios
larger than those experienced by 4-storey frames.
Date

Anyway, the extent of the residual drifts is relatively small, thus

allowing easily the repair after the earthquake. Nowadays, this issue
000155XA

000535XA
000184XA

000294XA

000303XA
000296YA

000297YA

000479YA
2S-R5-sc1
2S-R6-sc1
2S-R7-sc1
1S_2SR1
1S_2SR6
1S_2SR7

has a primary importance at the light of the recent earthquakes (e.g.

Record
Table 3

L'Aquila 2009, Emilia 2012, Christchurch 2010 and 2011, etc.). Indeed,
at the Authors' experience, the repairing costs may overcome in some
446 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454

a) Stiff soil b) Soft soil

Fig. 11. Capacity curves from IDAs.
4-Storey frames
8-Storey frames

a) Stiff soil b) Soft soil

Fig. 12. Interstorey drift ratio demand for the three limit states.

cases the constructional cost of the new building. Hence, although the Fig. 14 shows the distribution of the median values of PSA along the
performance of the examined frames is less efficient, rough economic building height for each limit state and Table 6 reports the correspond-
considerations may suggest that the examined frames might potentially ing percentiles. The results are shown varying the number of storeys
result in cheaper structures by accounting also the repairing costs in
case of a refined benefit–cost analysis. This matter is relatively recent Table 4
in European design practice, being the current codes (e.g. EN1998-1, Dispersion of interstorey drift ratio demand (values in %).
EN1998-3, NTC 2008, etc.) strictly oriented to design ductile structures
Percentiles at 4-Storey 8-Storey
experiencing plastic deformations even under moderate seismic action each limit state
(as also those at DL limit state). More detailed investigation on this Stiff soil Soft soil Stiff soil Soft soil

topic is beyond the scope of this study but it is clearly a research field DL 16th 0.38 0.36 0.52 0.49
that deserves further analysis. 50th 0.54 0.44 0.63 0.66
84th 0.83 0.54 0.74 0.82
SD 16th 0.83 0.71 0.89 0.98
6.3. Peak storey accelerations 50th 1.07 0.91 1.19 1.16
84th 1.43 1.29 1.48 1.64
Peak storey accelerations (PSAs) are usually related to the non- NC 16th 1.39 1.08 1.19 1.35
structural damage, thus allowing to quantify the potential economic 50th 1.72 1.58 1.62 1.65
84th 2.37 1.78 1.98 2.10
loss depending on the type of facilities and non-structural elements.
 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454 447

4-Storey frames
8-Storey frames

a) Stiff soil b) Soft soil

Fig. 13. Residual interstorey drift ratios.

and the soil conditions, because these design parameters are those actu-
ally influencing the PSAs. It is interesting to note that the median PSA
are similar for both 4- and 8-storey buildings located on stiff soil, with
values of about 1.6, 3 and 4 times the design acceleration Ad at DL, SD
and NC respectively. As predictable due to the larger input acceleration
at the fundamental period, the buildings located on soft soil show larger
values for PSAs, ranging from 2, 4 to 5 times Ad for DL, SD and NC,
respectively.
The significant dynamic magnification of storey accelerations in all
examined cases clearly highlights that very severe non-structural dam-
age can be expected, which corresponds to a very limited structural
damage. Indeed, the most of architectural components (i.e. cladding
systems, ceiling and lighting systems, interior partition walls, etc.), me-
chanical and electrical equipment and systems (i.e. heating, ventilation
and cooling systems, fire protection systems, and emergency power
systems, etc.), and building contents and inventory are prone to high
amplitude of storey accelerations, as shown by recent earthquakes
[35]. Hence, it should be observed that such a kind of structural systems
may be unsuitable for critical facilities, like hospitals where the indirect
losses due to damaged equipment and lost inventory represent more
than 80% of the total construction cost [35].
6.4. Beam ductility demand
The ductility demand for dissipative beams of MRFs is expressed in
terms of total chord rotation. The beams should have adequate rotation
capacity to assure that a determined portion of the input seismic energy
is dissipated by plastic behavior. Therefore, steel beams need to develop
a ductile behavior with high rotation capacity [26,36,37].
According to EN1998-3 [9], the acceptable plastic rotation limits for
class 1 beams should be no lesser than 1θy, 6θy and 8θy, respectively for
the DL, SD and NC, being θy the beam yield rotation.
Fig. 15 illustrates the median ductility demand along the building
height. As it can be observed the most of the beams is in elastic range
at SD limit state, while very limited plastic rotation demand can be rec-
ognized at NC limit state. Once more, these results can be explained by
the design procedure that leaded to oversize the structural element to
fulfil the requirement of damage limitation.
The small beam ductility demand suggests that for such a kind of
structures the use of class 2 beams can be more convenient, because
the large reserve of ductility of class 1 profiles cannot be exploited. In
addition, at the same overall performance cheaper structures would
be designed.
6.5. Beam flexural overstrength
The beam flexural overstrength (s) is the non-dimensional measure
of the ultimate bending capacity of steel beams, due to the amount of
strain hardening which can be exhibited prior beam failure [26,38,39].
The maximum flexural overstrength (s) may be defined in terms of
bending moments as follows:
Mu
s¼ ð4Þ
Mp

where Μu is the peak bending moment experienced by the beam, and

Μp is the beam plastic bending moment.
Table 5 This factor plays a key role in the application of hierarchy criteria in
Dispersion of residual interstorey drift ratio demand (values in %).
seismic design. EN1998-1 [8] accounts only for the possible
Percentiles at 4-Storey 8-Storey overstrength due to random material variability by considering an
each limit state
Stiff soil Soft soil Stiff soil Soft soil overstrength factor equal to 1.1 γov, for all types of members. Converse-
ly, the amount of beam strain-hardening is neglected, thus potentially
SD 16th 0.00 0.00 0.01 0.01
50th 0.02 0.01 0.05 0.02 underestimating the actual ultimate strength of members, especially
84th 0.08 0.03 0.20 0.05 for those belonging to class 1 according to EN1993:1-1 [10].
NC 16th 0.04 0.01 0.04 0.05 In order to verify this issue, the ratios (s*) between the maximum
50th 0.14 0.05 0.19 0.16 bending moment Mmax and Mp have been monitored. Fig. 16 shows
84th 0.25 0.16 0.35 0.36
the median distribution of s* along the building height. As it can be
448 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454

4-Storey frames

PSA/A d PSA/A d
8-Storey frames

PSA/A d PSA/A d

a) Stiff soil b) Soft soil

Fig. 14. Storey acceleration for DL, SD and NC limit state.

noted the most of beams behave elastically thus the median s* is lesser 7. Behavior factors
than unity. Some cases slightly larger than 1 occur for 8-storey frames
with soft soil at SD and NC limit states. Such results confirm that the de- According to [40,41] the actual behavior factor is given as the ratio
sign approach of EN1998-1 [8] is consistent with the expected perfor- between the peak ground acceleration leading to accepted failure for
mance of the frames, namely limited ductility demand resulting in the selected performance level (Au) and the peak ground acceleration
small hardening developed by plastic hinge. corresponding to the yielding of the frame (Ay).
Notwithstanding the moderate beam plastic engagement, the analy- In the present study the values of Ay and Au have been derived from
ses show that the seismic action may induce bi-triangular distribution of IDAs. The amplitude corresponding to the acceleration Au is the mini-
bending moments close to the plastic capacity at both beam ends. This mum value corresponding to all possible theoretical states of collapse:
implies that significant shear forces are expected in the panel zone of
the columns, especially in the inner columns where the shear force act- Au ¼ minðAθ ; Ac ; AR Þ ð5Þ
ing on the web panel is almost twice that in the corner columns.
The analyses show that the transformation factor (β) can vary with- where Aθ is the acceleration corresponding to the maximum permitted
in the range of 1.29 to 2.01 at SD and 1.67 to 2.14 at NC, being β defined interstorey drift ratio (either transient and residual, as reported in
according to EN 1993:1–8 as the ratio between the beam moments Mj,b1, Table 1); Ac corresponds to the column plastification; AR corresponds
Ed and Mj,b2,Ed at the intersection of the member centerlines into the to the maximum permitted local rotation (as given by EN1998-3 [9]).
joints. It is interesting to note that according to EN 1993:1–8 β should It is worth noting that the Aθ limit was the limiting level in the most of
be no larger than 2, which corresponds to the contemporary formation cases.
of plastic hinge in the two beams belonging to the joints. However, in Fig. 17 shows the median behavior factors obtained and the 16th and
actual cases the amount of strain hardening may lead to overcome the 84th percentiles for each soil type at both SD and NC limit states. The be-
plastic moment in one side of the joints. This effect should be accounted havior factors obtained for NC limit state are obviously larger than those
for designing full strength joints according to the provisions of EC8, for SD limit state, owing to the more demanding collapse criterion.
where column panel zone should be designed in order to avoid signifi- As a general remark, it should be noted that the median behavior
cant yielding within this component. In the reference frames, in order to factors (namely 3.44 and 3.30 for stiff and soft soil) obtained from
satisfy this code requirement, each web panel of interior column was IDAs at SD are slightly lower than that used at the design stage (namely
strengthened with one doubler plate thinner than the column web in q = 4). This result is consistent with recent research findings [42] that
all cases. highlight the need to a rational recalibration of the values of behavior
factors reported in EN1998-1 [8].
Table 6 Analysing the role of the examined design parameters it is not pos-
Peak PSA/Ad ratios. sible to define a general trend. Anyway, it is interesting to note that
the frames located on soft soil present behavior factors slightly lower
Percentiles at 4-Storey 8-Storey
each limit state
than those designed for stiff soil. Such result is mainly due to the larger
Stiff soil Soft soil Stiff soil Soft soil displacement demand of records for soft soil that are characterized by
DL 16th 1.34 1.45 1.40 2.07 the more severe accelerations at the fundamental periods of the exam-
50th 1.64 1.63 1.61 2.79 ined frames.
84th 1.94 2.10 1.92 3.48
Aside from the soil conditions, Fig. 17 shows that the span length is
SD 16th 2.69 2.76 2.54 3.75
50th 3.24 3.33 2.98 4.30 the geometrical parameter influencing the value of behavior factor.
84th 3.71 4.27 3.65 5.14 For 4-storey frames those with the shorter span are characterized by
NC 16th 3.77 4.66 3.51 4.90 the larger behavior factors, while the contrary for 8-storey structures.
50th 4.71 5.07 4.05 5.71 This result can be explained by considering that for all cases the drift cri-
84th 5.43 5.70 4.88 6.50
terion was the main condition limiting the seismic performance.
 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454 449

4-Storey frames
8-Storey frames

a) Stiff soil b) Soft soil

Fig. 15. Ductility demand ratios.

Therefore, since in low-rise frames the overall displacement shape at
the peak is close to shear type profile, the lateral stiffness increases re-
ducing the beam length. Thus, a larger acceleration is necessary to over-
come the drift capacity and a larger behavior factor is obtained. In case
of 8-storey frames the lateral displacement profiles differ with the
beam length. In particular, the cases with the shorter spans tend to an
overall cantilever response, while a shear type shape corresponds to
the frames with the longer bays. This different behavior is associated
to different drift demand. The cantilever-type structures showed the
4-Storey frames
8-Storey frames
larger demand especially at the upper storeys, thus leading to the small-
er Aθ limits and the smaller behavior factors.
8. Dual steel vs. single steel moment-resisting frames
The analyses carried out on the single steel MRFs showed a seismic
performance very close to that of dual steel frames in terms of overall dis-
placements. Indeed, the IDR demand of single steel frames is almost equal
to that experienced by the relevant dual steel structures at DL, while some

a) Stiff soil b) Soft soil

Fig. 16. Overstrength demand ratios.
450 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454

SD limit State
NC limit State

a) Stiff soil b) Soft soil

Fig. 17. Behavior factors for the SD and NC limit state.

slight differences can be recognized at both SD and NC limit states, espe-
cially for 8-storey frames as depicted in Fig. 18, where the median IDR de-
mand along the building height is plotted for both dual steel (i.e. tagged as
“HSS”) and single steel (i.e. tagged as “MCS”) frames. The comparison at
SD shows that the maxima IDRs for single steel frames are 2.86 and 5.08
times the corresponding values of dual steel 4- and 8-storey structures,
respectively. At NC those variations are larger, ranging from 3.85 to
30.28 for 4- and 8-storey frames, respectively. Such results are mainly
due to the different plastic engagement exhibited by single grade frames
as compared to dual steel ones. Indeed, for the former cases moderate
SD limit State
NC limit State
plastic deformations can occur in interior columns at building mid-
height. The differences enlarge at NC where the plastic deformations in
the columns increase. As a consequence, larger RIDRs are observed for
single steel frames than for dual steel frames, as illustrated in Fig. 19. In
addition, concerning the shear demand in the column web panel it is
observed that for single steel frames the additional plates necessary to
strengthen the joints should be thicker than the column web thickness,
thus theoretically ineffective according to EN1993:1–8. On the contrary,
dual steel frames do not show this problem. This comparison suggests
that dual steel solution allows controlling better the plastic mechanism,

4-Storey 8-Storey

Fig. 18. Dual steel vs. single steel frames: interstorey drift ratio demand.
 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454 451

SD limit State
NC limit State

4-Storey 8-Storey

Fig. 19. Dual steel vs. single steel frames: residual interstorey drift ratios.

thus avoiding the formation of plastic deformations into the column and
minimizing the repairing costs even though the initial constructional
prices may be larger than those for the single steel solution. This outcome
is not easily predictable, since limited seismic demand was expected for
both solutions because of the significant design overstrength resulting
from EN1998-1 provisions.
9. Material consumption
In order to evaluate the influence of the design parameters on the
constructional costs, the material consumption has been calculated for
all the examined cases. In particular, the steel consumption has been
computed in terms of total weight, while concrete consumption in
terms of total cast volume. The amounts of steel grade S355 (MCS)
used for beams, steel grade S460 (HSS) used for columns and reinforce-
ment steel (RS) are presented separately. The total amount for each
type of material has been divided by the tributary area of the frame,
3 L2 (see Fig. 3) and by the number of storeys n, as follows:
Total Amount
γ material ¼
3L2  n
The average value obtained in Eq. (6) is designated as “material
density” hereinafter. Table 7 reports the material density showing the
dependency of this index on the examined design parameters, while
the interrelations with those parameters are better described from
Figs. 20 to 22.
As a general comment, the structures designed for soft soil are char-
acterized by a consumption of materials larger than those designed for
stiff soil. This result is even more interesting by considering that the de-
sign PGA for frames on soft soil is smaller (namely equal to 0.16 g) than
that used for cases on stiff soil (namely equal to 0.24 g). However, it is
not surprising because the soft soil spectrum is characterized by design
accelerations larger than those for stiff soil spectrum at the fundamental
periods of the reference buildings.
The influence of column cross section typology is shown in Fig. 20.
As it can be noted the structures with CFT columns are characterized
by the larger consumption of steel especially for the columns that are
made of HSS. In addition, considering that the configuration of beam-
to-column joints that should be used to have a full strength connections
[28] are more complex and expensive than those used for wide flange
columns, the structures made of CFT are the less effective from an eco-
nomical point of view. For what concerns the cases with FE and PE col-
umns are characterized by similar material consumptions.
Fig. 21 shows the steel densities highlighting the influence of span
length. As it can be noted, for both 4- and 8-storey frames the cases
with 5.0 m span show generally higher density of both MCS and HSS
ð6Þ
than those with 7.5 m span. This result is ascribable to the need to provide
adequate lateral stiffness, thus compelling the designer to select heavy
profiles.
Fig. 21 shows the variability of steel density highlighting the influ-
ence of number of storeys. As expected, the plots highlight that the larg-
er steel density generally characterizes the taller frames expect for the
cases with 4-storeys and span length equal to 5.0 m, where the HSS

Table 7
Average weight (kg) of steel for each structural element.

Span 8-Storey 4-Storey

Stiff soil Soft soil Stiff soil Soft soil

Column Beam Column Beam Column Beam Column Beam

5.0 m 12,729 8084 17,533 9711 6979 3524 8004 3524

7.5 m 29,752 12,283 33,015 16,082 10,846 5553 13,263 5553
Δγ = γ7.5/γ5 2.34 1.52 1.88 1.66 1.55 1.58 1.66 1.58
452 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454

Fig. 20. Average of amount of steel and concrete.

density is the larger. As previously discussed, this outcome derives from
the design process to satisfy both drift and stability limitations.
10. Conclusions
The seismic performance-based evaluation has been carried out con-
sidering three limit states according to EN1998-3 [9], namely damage
limitation (DL), significant damage (SD) and near collapse (NC).
Based on the analysis of the results, the following conclusions can be
drawn:

A parametric study based on both nonlinear static and dynamic anal-
yses has been presented in order to investigate the seismic performance
and economic assessment of dual-steel moment-resisting frames
(MRFs) designed according to EN1998-1 [8]. High strength steel S460
has been used for columns and mild carbon steel S355 for beams.
– The use of HSS showed to be efficient to guarantee the weak-beam/
strong-column behavior.
– In the most of the cases, pushover analyses showed overall
V V V
overstrength factors Ω ¼ V dy ¼ V 1yy  V1yd larger than the design behav-
ior factor (q = 4). This result is ascribable to the codified design

a) 8-Storey frames

b) 4-Storey frames
Fig. 21. Steel consumption vs. span length.
 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454
a) Frames with 5.0m span
b) Frames with 7.5m span
Fig. 22. Steel consumption vs. storey number.
procedure, which leads to increase the member size to satisfy the
drift limitations. Indeed, using HSS for columns small sections need
to satisfy hierarchy criteria, thus having flexible structures.
V
– The average ratio V 1yy obtained from pushover curves confirmed the
value of 1.30 recommended by EN1998-1 [8].
– Nonlinear dynamic analyses showed that the frames have a seismic
demand (namely, transient and residual drift ratios, beam ductility)
fairly below the proposed limit for DL, SD and NC states. In particular,
at SD limit state the most of frames behave in elastic field. This result
is mainly due to the design oversizing.
– The median peak storey accelerations range between 2 and 3 times
the design PGA. Therefore, significant amplification effects can
occur and should be accounted for preserving the integrity of facili-
ties and non-structural elements.
– The behavior factors obtained from incremental dynamic analyses
for SD limit state are smaller than the code value with an average
value equal to 3.4. The average behavior factor at NC limit state is
equal to 3.7, hence still smaller than the code value. These results
suggest the need to calibrate the behavior factors given by
EN1998-1 [8] as also highlighted in [42].
– The comparison between dual steel and single steel frames showed
that the former solution is more effective to control the plastic
mechanism.
Regarding to material consumptions, the following remarks can be
drawn:
– The taller is the building height the larger is the material consump-
tion especially in terms of steel density;
– The frames with concrete filled tubular columns are characterized by
the larger amount of high strength steel (N50%) and concrete
(N30%) than the cases with full encased and partial encased
columns, that are characterized by similar material consumptions.
– All frames located on soft soil are characterized by larger consump-
tion of material than the corresponding designed for stiff soil. This
result is worth of noting especially considering that the design PGA
453
for frames on soft soil is smaller (namely equal to 0.16 g) than that
used for cases on stiff soil (namely equal to 0.24 g). Hence, this
aspect should be carefully accounted for the economic planning at
design stage in all cases the cost of the building is the leading design
parameter.
Acknowledgments
The Authors thank the financial support granted by the Research
Fund for Coal and Steel (RFCS) and all the partners involved in the
research project HSS-SERF (High Strength Steel in Seismic Resistant
Buildings Frames—grant no. RFSR-CT-2009-00024).
The first author wishes to thank the financial support from Erasmus
Mundus External Cooperation Window (no. 141285-EM-1-2008-1-PT-
ERA MUNDUS-ECW) —ISAC, the Brazilian PhD program—“Ciências
Sem Fronteiras”.
In addition, the Authors would like to thank the two anonymous
reviewers for their helpful comments and suggestions to improve the
quality of the paper.
References
[1] Samuelson A, Schroter F. High performance steel in Europe—production processes,
mechanical and chemical properties, fabrication properties. IABSE—Structural
Engineering Document Nº 8—Use and application of high performance steels
(HPS) for steel structures; 2005.
[2] Mikia C, Homma K, Tominaga T. High strength and high performance steels and
their use in bridge structures. J Constr Steel Res 2002;58(1):3–20.
[3] Dubina D, Dinu F, Zaharia R, Ungureanu V, Grecea D. Opportunity and effectiveness
of using high strength steel in seismic resistant building frames. International
Conference in Metal Structures; 2006.
[4] Dubina D, Stratan A, Dinu F. Dual high-strength steel eccentrically braced frames
with removable links. Earth Eng Struct Dyn; 2008 [published online].
[5] Dubina D. Dual-steel frames for multistory buildings in seismic areas. Internacional
Colloquium Stability and Ductility of Steel Structures; 2010.
[6] Ghoberah A. Performance-based design in earthquake engineering: state of develop-
ment. Eng Struct 2001;23:878–84.
[7] Grecea D, Dinu F, Dubina D. Performance criteria for MR steel frames in seismic
zones. J Constr Steel Res 2004;60:739–49.
454 A. Tenchini et al. / Journal of Constructional Steel Research 101 (2014) 437–454
[8] EN 1998-1-1. Eurocode 8: Design of structures for earthquake resistance—Part 1:
General rules, seismic actions and rules for buildings. CEN; 2004.
[9] EN 1998-3. Eurocode 8: Design of structures for earthquake resistance—Part 3:
Assessment and retrofitting of buildings. CEN; 2005.
[10] EN 1993-1. Eurocode 3: Design of steel structures—Part 1-1: General rules and rules
for buildings. CEN; 2005.
[11] EN 1994-1. Eurocode 4: Design of composite steel and concrete structures—Part 1-1:
General rules and rules for buildings. CEN; 2004.
[12] Elghazouli AY. Assessment of capacity design approaches for steel framed structures.
J Steel Struct 2005;5(5):465–75.
[13] Elghazouli AY, Castro JM, Izzuddin BA. Seismic performance of composite moment-
resisting frames. Eng Struct 2008;30:1802–19.
[14] SeismoStruct, Version 5.0.5. Seismosoft—Earthquake Engineering Software Solution
2011.
[15] Spacone E, Ciampi V, Filippou FC. Mixed formulation of nonlinear beam finite
element. Comput Struct 1996;58(I):71–83.
[16] Calabrese A, Almeida JP, Pinho R. Numerical issues in distributed inelasticity model-
ling of RC frame elements for seismic analysis. J Earthq Eng 2010;14(1):38–68.
[17] Martinez-Rueda JE, Elnashai AS. Confined concrete model under cyclic load. Mater
Struct 1997;30:139–47.
[18] Mander JB, Priestley MJN, Park R. Theorical stress–strain model for confined con-
crete. J Struct Eng 1988;8(114):1804–26.
[19] Susantha KAS, Ge H, Usami T. Unixial stress–strain relationship of concrete confined
by various shaped steel tubes. Eng Struct 2001;23(10):1331–47.
[20] Menegotto M, Pinto PE. Method of analysis for cyclically loaded R.C. plane frames in-
cluding changes in geometry and non-elastic behaviour of elements under com-
bined normal force and bending. Symposium on the Resistance and Ultimate
Deformability of Structures Acted on by Well Defined Repeated Loads; 1973.
[21] EN 1992-1-1. Eurocode 2: Design of concrete structures—Part 1-1: General rules and
rules for buildings. CEN; 2004.
[22] RFSR-CT-2007-00039. Optimizing the seismic performance of steel and steel–
concrete structures by standardizing material quality control (OPUS)978-92-79-
29037-4; 2013. http://dx.doi.org/10.2777/79330.
[23] Abramowitz M, Stegun IA. Handbook of mathematical functions. National Bureau of
Standards, Applied Math. Series; 1964.
[24] Szabó BA, Babuška I. Finite element analysis. John Wiley & Sons; 1991.
[25] Correia AA, Virtuoso FBE. Nonlinear analysis of space frames. In: Mota Soares, et al,
editors. Proceedings of the Third European Conference on Computational
Mechanics: Solids, Structures and Coupled Problems in Engineering; 2006.
[26] D'Aniello M, Landolfo R, Piluso V, Rizzano G. Ultimate behaviour of steel beams
under non-uniform bending. J Constr Steel Res 2012;78:144–58.
[27] Wakawayashi M, Matsui C, Minami K, Mitani I. Inelastic behaviour of full scale steel
frames. Kyoto University Research Information Repository, Disaster Prevention
Research Institute annuals; 1970.
[28] Beg Darko, Cermelj Blaž, Rejec Klemen, Lopatic Jože, Dubina Dan, Vulcu Cristian,
et al. Prequalification criteria for welded beam-to-column joints in dual-steel
frames. Deliverable 5 within High Strength Steel in Seismic Resistant Building Frames
(HSS-SERF) Grant agreement no. RFSR-CT-2009-00024. Doc. ID: hss-d-0005-wp5-
ul-v1; 2014.
[29] American Institute of Steel Construction, Inc. (AISC). Seismic provisions for structur-
al steel buildings. Standard ANSI/AISC 341-10, Chicago (IL, USA); 2010.
[30] Priestley MJN, Grant DN. Viscous damping in seismic design and analysis. J Earthq
Eng 2005;9(1):229–55.
[31] D'Aniello M, La Manna Ambrosino G, Portioli F, Landolfo R. Modelling aspects of the
seismic response of steel concentric braced frames. Steel Compos Struct
2013;15(5):539–66.
[32] D'Aniello M, La Manna Ambrosino G, Portioli F, Landolfo R. The influence of out-of-
straightness imperfection in physical-theory models of bracing members on seismic
performance assessment of concentric braced structures. Struct Des Tall Spec Build
2014. http://dx.doi.org/10.1002/tal.1160.
[33] Villani A, Castro JM, Elghazouli AY. Improved seismic design procedure for steel
moment frames. STESSA 2009: Behaviour of Steel Structures in Seismic Areas,
Philadelphia; 2009.
[34] Fulop L. Selection of earthquake records for the parametric analysis. Research report
VTT-R-03238-10, VTT, Espoo; 2010.
[35] Federal Emergency Management Agency (FEMA). Reducing the risks of nonstructur-
al earthquake damage—a practical guide. FEMA E-74, Prepared by ATC for FEMA,
January 2011, Washington, D.C.; 2011.
[36] Tortorelli S, D'Aniello M, Landolfo R. Lateral capacity of steel structures designed ac-
cording to EC8 under catastrophic seismic events. COST ACTION C26: Urban Habitat
Constructions under Catastrophic Events—Proceedings of the Final Conference;
2010.
[37] D'Aniello M, Güneyisi EM, Landolfo R, Mermerdaş K. Analytical prediction of avail-
able rotation capacity of cold-formed rectangular and square hollow section
beams. Thin-Walled Struct 2013. http://dx.doi.org/10.1016/j.tws.2013.09.015.
[38] Güneyisi EM, D'Aniello M, Landolfo R, Mermerdaş K. A novel formulation of the
flexural overstrength factor for steel beams. J Constr Steel Res 2013;90:60–71.
[39] Güneyisi EM, D'Aniello M, Landolfo R, Mermerdaş K. Prediction of the flexural
overstrength factor for steel beams using artificial neural network. Steel Compos
Struct 2014 [in press].
[40] Mazzolani FM, Piluso V. ECCS manual on design of steel structures in seismic zones.
Technical Committee 13—Seismic Design, No 76/1994. ECCS—European Convention
for Constructional Steelwork; 1994.
[41] Elnashai AS, Broderick BM. Seismic response of composite frames II. Calculation of
behaviour factors. Eng Struct 1996;18(9):707–23.
[42] Landolfo Raffaele, editor. Assessment of EC8 provisions for seismic design of
steel structures. Technical Committee 13—Seismic Design, No 131/2013. ECCS—
European Convention for Constructional Steelwork; 2013.