Thin-Walled Structures 61 (2012) 248–257

Contents lists available at SciVerse ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Performance-based design of sheathed CFS buildings in seismic area

Luigi Fiorino, Ornella Iuorio, Vincenzo Macillo, Raffaele Landolfo n
University of Naples Federico II, Naples, Italy

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

Available online 17 April 2012 The widely recognized structural performance provided by cold-formed steel (CFS) systems together
Keywords: with high levels of prefabrication, safety, durability and sustainability are spreading the CFS construc-
Behaviour factor tion systems all over the world. At the same time, the actual lack in European seismic design codes
Cold-formed steel requires the development of specific design procedures for the application in seismic area. The seismic
Performance-based behaviour of CFS structures is characterized by the lateral response of shear walls. In particular, when
Residential buildings the sheathing-design methodology is adopted, then the seismic behaviour is strongly influenced by the
Sheathing-braced nonlinear response of sheathing-to-frame connections. In this paper a performance-based design
methodology for the seismic design of sheathed CFS walls is proposed. The presented methodology is
based on the results of an extensive parametric nonlinear dynamic study carried out on one story
buildings by means of incremental dynamic analysis. Moreover, a multistep design procedure that
allows all the main structural components to be defined in three steps is developed. In particular, a
nonlinear dynamic (ND) nomograph for the assessment of the external screw spacing is presented in
this work in order provide a new tool that upgrade the design procedure under investigation by the
Authors in the last years. Finally the application of the proposed design methodology is verified through
a case study.
& 2012 Published by Elsevier Ltd.

1. Introduction by sheathing, steel frame and fasteners represents the main seismic-
force resisting system and the walls act as in-plane diaphragms.
Nowadays, the construction market offers a wide choice of When the design is carried out according to the ‘‘sheathing-braced’’
structural typologies and the selection of technologies and pro- approach, the advantages of choosing CFS solution are mainly due to
ducts to use for buildings is strongly influenced by economic costs the higher structural performance, with consequent material saving,
and execution times. Besides, the construction system has to which usually corresponds to ease of assembly, reduction of execu-
satisfy other fundamental requirements as safety, durability and tion times and lower costs [2].
eco-efficiency. From this perspective, the constructions made Different approaches are available to estimate the lateral
with cold-formed steel (CFS) structure combine successfully high response of sheathed CFS shear walls: experimental, analytical and
structural performance with elevated quality level offering extre- numerical methodologies. The experimental approach is based on
mely competitive solutions, which satisfy the needs of very full scale tests carried out on typical walls and it is frequently used.
demanding customers [1]. In fact, nominal shear strength design values provided by building
The design of a CFS structure under horizontal loads can be codes [3] in tabulated form are based on experimental test results.
carried out using two different approaches: ‘‘all-steel’’ design and Due to the required large number of test, it is clear that this
‘‘sheathing-braced’’ design. The first one considers only steel mem- approach is the most expensive one and, in addition, it can be used
bers as load carrying elements and the introduction of X or K bracing only when the wall characteristics (geometry and materials) are
systems in the lateral resisting walls is required. The latter approach within the range of experimental results.
takes into account the resisting effect given by sheathing panels, In order to overcome the limitations of the experimental
usually wood or gypsum based. In this case, the system composed approach, finite element methods can be used to evaluate the
shear response of sheathed CFS shear walls. Numerical models are
usually calibrated on available experimental results and they can
n
Correspondence to: Department of Constructions and Mathematical Methods be used to simulate the structural response of walls having
in Architecture University of Naples ‘‘Federico II’’ via Forno Vecchio, 36. 80134 characteristics different from tested walls.
Naples, Italy. Tel.: þ39 081 2538052; fax: þ39 081 2538989.
E-mail addresses: lfi[email protected] (L. Fiorino),
As far as analytical approaches are concerned, approaches
[email protected] (O. Iuorio), [email protected] (V. Macillo), specifically developed for sheathed CFS walls are not present in
[email protected] (R. Landolfo). literature. On the contrary, a large number of methods developed

0263-8231/$ - see front matter & 2012 Published by Elsevier Ltd.

http://dx.doi.org/10.1016/j.tws.2012.03.022
 L. Fiorino et al. / Thin-Walled Structures 61 (2012) 248–257 249

for the analysis of sheathed wood shear walls is available. Because 10 mm (outside-to-outside web depth  outside-to-outside flange
the global response of steel-framed and wood-framed walls size  outside-to-outside lip size) lipped channel studs, spaced at
sheathed with panels under lateral loads is qualitatively similar, 600 mm and sheathed with GWB panels on both sides (GþG) or
then the application of existing analytical methods for wood- GWB on one side and OSB panels on the other side (Gþ O). The
framed walls is reasonable also in the case of steel-framed walls. thicknesses of GWB and OSB panels are 12.5 mm and 9.0 mm,
The possibility to predict only strength and deflection, without respectively. For connecting sheathing panels and frame
giving a reliable evaluation of the whole load vs. deflection 4.2  25 mm (diameter  length) flat head self drilling screws have
response curve, represents the main limitation of existing analy- been considered for OSB and 3.5  25 mm bugle head for GWB. The
tical approaches, especially when a nonlinear static procedure is fasteners have spacing equal to 300 mm in the field, while different
selected for seismic analysis of the structure. Also this limitation dimensions for the external screw spacing have been investigated.
can be overcome by means of developing numerical models. Hold-down devices type Simpson Strong-Tie [8] and chemical ancho-
From a seismic point of view, the cyclic behaviour of sheathed rage (type HIT-RE 500 with HIS-N 8.8 by Hilti [9] have been
CFS shear walls is characterized by a remarkable nonlinear considered at each end of the wall, while mechanical shear anchors
response and a strong pinched forces versus displacement hyster- (type HST M8 by Hilti) have been hypothesized. Sheathing panel
esis. Therefore, the seismic design according to force-based design typology, wall geometry (height h and length l) and external screw
linear methods, traditionally used in current building codes, spacing (s) have been varied as summarized in Table 1. For each wall
requires specific study dedicated to the evaluation of the seismic configuration obtained by combining the parameters given in Table 1,
performance for such seismic-force resisting systems. The main the stud thickness and hold-down device typology have been selected
objective is to evaluate the seismic ductility and to define the in such a way to promote the sheathing fasteners collapse. The
capacity design rules, which are quantitatively obtained through monotonic response of the walls has been obtained by considering
seismic performance parameters, like the behaviour and over- materials and components properties as defined in Table 2.
strength factors. For the evaluation of wall cyclic response, the model proposed in
As an attempt to give a contribution to design of seismic-force Della Corte et al. [10] and upgraded in Fiorino et al. [11], which takes
resisting systems made of sheathed CFS shear walls, a research into account of strength degradation, has been used. This approach
project devoted to develop a specific performance-based methodo- allows to model the cycle lateral response of sheathed CFS shear
logy was undertaken at the University of Naples Federico II. The walls in terms of horizontal force (H) vs. lateral deflection (d)
methodology is based on the results of a large number of nonlinear through the definition of three limit curves together with the
dynamic parametric analyses (Section 2) and it consents the design definition of transition and strength degradation laws. The three
of sheathed CFS walls to be performed by either linear or nonlinear limit curves (monotonic curve) represent the monotonic response,
dynamic analysis. In particular, the methodology consists in a Multi- the upper bound cyclic curve and the lower bound cyclic curve to all
Step (MS) design procedure (Section 4), that allows the seismic possible H–d values obtained in the cyclic response. For each curve,
design of walls on the basis of a Linear Dynamic (LD) or Nonlinear a 6 independent parameters Richard–Abbott type law [12] is
Dynamic (ND) seismic analysis by means of the definition of specific assumed and considering also the transition and degradation laws,
design nomographs. According to this procedure, if the LD analysis is the model depends on a total number of 22 independent para-
selected for the seismic design, the evaluation of nonlinear response meters. The calibration of these parameters (Fig. 1), which describe
is performed through the definition of behaviour factors according the whole cyclic response, has been carried out on the basis of
to a performance-based approach, called Multi-performance (MP) available results of experimental monotonic and cyclic full scale
approach (Section 5). In contrast, if the ND analysis is chosen to be tests [5], in such a way that the numerical cyclic response would be
applied, then the nonlinear behaviour is directly taken into account as much as possible similar to the experimental cyclic response in
by means of nonlinear Incremental Dynamic Analysis (IDA) results. terms of H–d curve and dissipated energy. Further details about
assumptions and applicability of the model for cyclic prediction can
be found in Fiorino et al. [11].
2. Incremental dynamic analyses For the parametric analysis, one-story buildings have been
considered as case studies. They refer to a stick-built construc-
In order to perform nonlinear dynamic analyses on sheathed tions in which both floors and walls are realized with CFS framing
CFS wall, it is necessary to define models which simulate the sheathed with structural panels. In particular, in order to obtain a
monotonic and cyclic wall lateral response. In particular, the large range of solutions, a schematic plan has been considered
monotonic lateral response of sheathed CFS shear walls is with wall length (L) variable between 3 and 7 m and lengths of
evaluated by means of the analytical-numerical model proposed
in Landolfo et al. [4]. The model allows the evaluation of each wall
Table 1
deflection contribution, from which the total wall deflection can
Variables assumed in the parametric study.
be obtained by adding the deformation due to bending (dF),
sheathing fasteners (dF  S), sheathing panels (dS), and anchorages Sheathing panel typology GWBþ GWB (G þ G),GWBþ OSB (Gþ O)
(dF  F):
Wall height (h) (mm) 2400, 2700, 3000
d ¼ dF þdFS þdS þ dFF ð1Þ Wall length (l) (mm) 1200, 2400, 9600
External screw spacing (s) (mm) 50, 75, 100, 150
The model has been calibrated on the basis of monotonic tests
on full scale specimens [5] carried out on CFS shear walls
sheathed with OSB and gypsum board (GWB) and tests performed
Table 2
on sheathing connections [6], that are nominally identical to
Materials and components properties assumed in
those used for the walls. Further details about hypothesis and the parametric study.
applicability of the model for monotonic prediction can be found
in Fiorino et al. [7]. OSB shear modulus 1400 MPa
In order to study the behaviour of different wall configurations, GWB shear modulus 750 MPa
Steel Young modulus 200,000 MPa
the model has been applied to simulate the monotonic response of 72 Hold-down stiffness 27–31 kN/mm
different wall configurations. All the walls are made of 100  50 
250 L. Fiorino et al. / Thin-Walled Structures 61 (2012) 248–257

full height (resisting) wall segment (l) in the range l ¼0.4L through
l¼0.7L (Fig. 2). Unit weights ranging from 0.4 to 1.5 kN/m2 and
from 0.3 to 1.2 kN/m2 have been considered for floors and walls,
respectively. Moreover, the building has been considered without
and with attic (Fig. 3). In the first case, a variable live load of
2.0 kN/m2 has been considered and, in the latter case, a snow
variable load ranging from 0.60 to 1.20 kN/m2 has been added. The
seismic weights have been defined according to the following
relationship:
SGKi þ Sj ð0:3UQ Kj Þ ð2Þ
where Gki and Qki are the characteristic values of permanent and
variable actions, respectively. With these conditions, a seismic weight
per unit wall length ranging between 2 and 38 kN/m has been
obtained. Therefore, 7 seismic weights per unit wall length have
Fig. 1. Calibration of wall cyclic response.
been considered (10, 15, 20, 25, 30, 35 and 40 kN/m) and have been
applied to the 72 wall configurations defined previously in the
parametric study.
The seismic inputs have been selected in such a way that they
could cover all the soil typologies classified by Eurocode 8 [13].
In particular, Eurocode 8 provides five different soil types A, B, C,
D, and E, but in this study only three spectra are adopted grouping
the soil types B, C and E under one spectrum type. Therefore, 21
earthquake records have been selected from the ESD (European
Strong-motion Database, www.isesd.cv.ic.ac.uk). For each soil
type 7 accelerograms have been considered so that the shape of
the average elastic response spectrum is close as much as possible
to the shape of the corresponding Eurocode 8 elastic acceleration
spectrum [14]. The selected earthquakes include records from
different European and Mediterranean regions. For these earth-
quakes the Richter magnitude ranges from 5.8 to 7.6. In order to
match the design spectra the natural accelerograms have been
scaled to the peak ground acceleration (ag).
Each wall defined in the parametric study has been schematized
as single degree of freedom structure, in which the hysteretic
behaviour under horizontal loads is described by the model pre-
viously presented. In order to account the second order effects, a
Fig. 2. Schematic building for a large range of solutions. vertical load equal to the 100% of the mass has been considered.

Fig. 3. Unit weights defined for the parametric analysis.

 L. Fiorino et al. / Thin-Walled Structures 61 (2012) 248–257 251

Moreover, the viscous damping ratio has been set equal to 5%, be associated a numerical value of the interstory drift. In quali-
according to the experimental results obtained by Kawai et al. [15], tative terms, following the definitions given in FEMA P-750 [18]
Gad et al. [16] and Dubina [17]. The adopted procedure is at immediate occupancy (IO) level, the structure has to be able to
the well known Incremental Dynamic Analysis (IDA). Outputs of guarantee that the structural damage is negligible and non-
the analysis are the IDA curves, which are presented as adimentio- structural elements remains serviceable. The building has to be
nalized elastic spectral acceleration (Sae/g) vs. maximum required safe to be occupied and only little repair could need. At life safety
interstory drift angle (d/ h). In particular, for obtaining a single IDA (LS) level, the structure can be subjected to significant structural
curve, which represents the response for a specific wall condition, and non-structural damage without remarkable risk to life. Non-
each accelerogram has been scaled in the range from 0.05 to 1.95 by structural elements need some repair because they may lose
considering 50 values. Therefore, a total number of 529,200 single functionality and some structural repair could be necessary.
nonlinear dynamic analyses have been performed. Fig. 4 shows Finally, in the case of collapse prevention (CP) level, the structure
typical IDA curves obtained in the present study for a fixed wall is near to reach complete damage and it is unsafe to be occupied.
configuration and different accelerograms. Structural repair is generally not economically feasible.
Several research programs have been devoted to define quanti-
tatively the interstory drift limits for sheathed CFS structures. On the
3. Definition of interstory drift limits basis of test results on steel sheet-to-steel sheet screw connections,
for CFS walls sheathed with trapezoidal CFS sheets, Dubina [17]
For the application of a performance-based methodology to suggested different interstory drift limits, corresponding to different
the case of sheathed CFS shear walls is necessary to define the performance levels: 0.3, 1.5 and 2.5% for fully operational, partially
interstory drift (d/h) limits for such type of seismic-force resisting operational and ultimate limit states, respectively.
systems. In fact, from an operative point of view, for each Blais [19] suggested as interstory drift limits for ultimate condi-
qualitative definition of the seismic performance level, it should tion the range from 1.91 to 2.23%. This range was defined starting
from the results of a large number of experimental tests carried out
on CFS shear walls sheathed with OSB panels. Taking into account
the results of an experimental campaign carried out on typical
Japanese CFS walls sheathed with plywood and plasterboard panels,
Kawai et al. [15] proposed a seismic design methodology based on
direct evaluation of seismic capacity in terms of interstory drift, in
which the 2% represents the target (limit) value for sever earthquake
in order to ensure the building safety.
In this study the numerical quantification of the interstory
drift limits is associated to the definition of three different limit
displacements on the generic response curve of a sheathed CFS
shear wall (Fig. 5): the peak (dp), the ultimate (du) and the
yielding (dy) displacement [11]. Therefore, the capacities of the
generic wall associated to the different performance levels has
been assumed as follows: yield displacement for IO (dC,IO ¼dy),
peak displacement for LS (dC,LS ¼dp) and ultimate displacement
for CP (dC,CP ¼du); then a seismic design which allows an adequate
damage control for all selected earthquake hazard levels should
be reached. In fact, for IO performance level, the assumed
displacement capacities correspond to the interstory drift limits
(d/h) ranging from 0.13 to 0.97%, which are less than those (1.5%)
Fig. 4. Typical IDA curves. given by Dubina [17]. For LS and CP performance levels the

Fig. 5. Interstory drift limits for an experimental test [5].

252 L. Fiorino et al. / Thin-Walled Structures 61 (2012) 248–257

assumed limits of d/h range from 0.61 to 1.87% and from 0.73 to spacing) and some design parameters of wall components
2.44%, respectively, which are generally more conservative values (i.e., type of the sheathing panels; type and edge distance of the
respects to those suggested by different Authors: 2.5% for Dubina sheathing fasteners; stud size with the exception of the thickness;
[17], 2.0% for Kawai et al. [15], and from 1.91 to 2.23% for Blais hold-down anchor and shear anchor typologies), which derives
[19]. The possibility of damage containment based on the from vertical loads design together with architectural and tech-
assumed interstory drift limits is also supported by available nological considerations; (2) the assessment of the external screw
experimental experiences, which confirm a damage of sheathing spacing (s), which is the only design parameter that directly
connections negligible for d/h less than 1.0%, which became derives from seismic analysis results; (3) the evaluation of stud
tolerable for d/h in the range 1.0% through 2.5% (Fig. 5). Finally, thickness, hold-down anchor diameter, and shear anchor spacing,
the difference in terms of interstory drift limits between walls which is carried out on the basis of capacity design criteria.
sheathed with steel sheets and panels is due to an initial stiffer The approach is conceived for single-storey constructions (SDOF
behaviour and a larger ultimate displacement of steel-to-steel systems) having symmetric structural configuration (lack of torsional
connections with respect to steel-to-panel connections. effects), in which all sheathed CFS shear walls have the same
geometry and structural components. In fact, under these assump-
tions the behaviour of the whole building under seismic actions can
4. Multi-Step design procedure based on nonlinear dynamic be directly obtained from the load-deflection response curve of the
analysis results single wall. As a result, the dynamic lateral behaviour of the structure
can be well represented by the dynamic lateral behaviour of the
A Multi-Step (MS) seismic design procedure for sheathed CFS walls, which can be described by a SDOF system. The procedure may
housing that allows to define the screws spacing and all the be extended to the case of multi-storey constructions (MDOF
other shear walls components was presented in Fiorino et al. [7]. systems) through the use of an equivalent SDOF system.
The procedure refers to sheathed CFS walls, without openings, in For the assessment of the external screw spacing (step 2), which
which the wall components are designed in such a way to has to be derived by seismic analyses, the proposed procedure
promote the sheathing fastener failure. The design procedure introduces two nomographs defined on the basis of linear dynamic
can be summarized in three consecutive steps (Fig. 6): (1) the analysis and nonlinear static analysis that allow the user to design
selection of the wall geometry (wall height and length, stud following both seismic procedure. In particular, in the case of Linear

Fig. 6. Wall design parameter for the proposed methodology.

 L. Fiorino et al. / Thin-Walled Structures 61 (2012) 248–257 253

d C dD and overstrength are taken into account by reducing the seismic

force, that usually corresponds to a seismic hazard level equal to 10%
si -1, wj, probability of exceedance in 50 years (10%/50). As a result, these
IDA curves soil type k approaches allow significant structural damage under 10%/50 earth-
quakes, by neglecting the seismic performance evaluation in the case
si ,wj, of seismic events having lower probability of exceedance.
dC = fC (s) soil type k As attempt to overcome the limitations of this traditional
approach, in this paper a methodology named Multi-performance
si +1,wj, (MP) design approach is proposed, in which different seismic
soil type k performance levels are required for different seismic hazard
levels, in such a way to obtain a ‘‘multi-performance’’ seismic
design. The peculiarity of the proposed approach is to provide the
s ag possibility to achieve an ‘‘enhanced objective’’ [18], consisting of
si-1 s si+1 the following goals (Table 3): (1) immediate occupancy (IO)
performance level for earthquakes having 50% probability of
Fig. 7. Nomograph for ND procedure. exceedance in 50 years (50%/50) i.e., ground motions with mean
return period of about 75 years or so called ‘‘frequent’’ earth-
Dynamic (LD) procedure, a force-based design approach is usually quakes; (2) life safety (LS) performance level for earthquakes
used, in which the inelastic behaviour and the structural over- having 10% probability of exceedance in 50 years (10%/50) i.e.,
strength are taken into account by behaviour factors; while for ground motions with mean return period of about 500 years or so
Nonlinear Static (NS) procedure, the inelastic behaviour and the called ‘‘rare’’ earthquakes; (3) collapse prevention (CP) perfor-
structural overstrength are directly considered. Finally, in order to mance level for earthquake with 2% probability of exceedance in
complete the wall design, an additional nomograph is defined 50 years (2%/50) i.e., ground motions with mean return period of
to evaluate all the other design parameter (step 3) according to about 2500 years or so called ‘‘very rare’’ earthquakes.
capacity design criteria. More details about the hypothesis, limit of Based on the results of performed dynamic analyses, the non-
applicability and the definition of design nomographs (LD and NS) linear capacity of sheathed CFS shear walls, in terms of behaviour
can be found in Fiorino et al. [7]. factor, has been assessed. The numerical results have been inter-
As an attempt to upgrade this methodology, a design nomo- preted by considering the interstory drift limits (yielding dy, peak dp
graph for the seismic design of single-storey sheathed CFS frame and ultimate du displacements) defined in Section 3.
structures developed on the basis of nonlinear dynamic analysis For each IDA curve the seismic intensity measures Sa,y, Sa,p and
results is presented. In particular, the procedure for the evalua- S,ua corresponding to the limit displacements dy, dp and du,
tion of external screw spacing (step 2) by means of Nonlinear respectively, have been evaluated and these spectral accelerations
Dynamic (ND) analyses is described hereafter. have been used to define three different behaviour factors, as
As it is well known, when the ND procedure is selected for follows (Fig. 8):
seismic analysis, the ductility and the overstrength of the struc-
ture are directly considered and the comparison between seismic q1 ¼ Sa,p =Sa,y , q2 ¼ Sa,u =Sa,p , q3 ¼ Sa,u =Sa,y , ð5Þ
capacity and demand can be obtained in terms of displacements
(displacement-based approach): Table 3
Performance objective matrix: traditional vs. MP design approach.
dC ZdD ð3Þ
Performance levels
where dC and dD are the seismic displacement capacity and demand,
respectively. In particular, the displacement demand can be IO LS CP
achieved by means of IDA, in which the results can be given in
terms of displacement demand (dD) vs. peak ground acceleration Earthquake hazard level 50%50 
10%50 
(ag) curve, which corresponds to the response of a fixed wall
2%50 O
condition (wall geometry and materials, seismic weight and soil
type). Instead, the displacement capacity corresponds to a given  goal achieved by traditional design approach.
limit displacement (e.g., ultimate displacement du) on wall response O goal achieved by MP design approach.

curve can be expressed as a function of external screw spacing (s):

Sa
dC ¼ f C ðsÞ ð4Þ
Sa,u
Therefore, for a fixed wall condition, with exception of the Sa,p
external screw spacing (which is the design parameter), if the selected IDA curve
function fC(s) and the relevant IDA curves are known, then for each
value of external screw spacing (s), both displacement demand (dD)
and capacity (dC) can be evaluated and compared until Eq. (3) is Sa,y
satisfied. The design procedure can be represented graphically by d
means of the schematic ND nomograph shown in Fig. 7.
dy dp du

monotonic curve
5. Behaviour factor evaluation according to the
multi-performance design approach bilinear curve

Nowadays, the seismic design according to current building codes H

is generally performed by means of force-based approaches, in which
in the case of the ultimate limit state the structural inelastic response Fig. 8. Definition of behaviour factors.
254 L. Fiorino et al. / Thin-Walled Structures 61 (2012) 248–257

Table 4 Table 5
ag values for selected seismic hazard levels and intensities. Behaviour factors for the MP design approach.

Seismic hazard level Seismic intensity zone ag(g) Wall configuration q1 q2 q3

50%/50 ‘‘frequent’’ earthquake Low (LO) 0.06 Gþ G Average 2.23 1.35 3.05
Medium (ME) 0.10 St. Dev. 0.44 0.17 0.83
High (HI) 0.14 C.o.V. 0.20 0.13 0.27
10%/50 ‘‘rare’’ earthquake LO 0.15 Gþ O Average 2.35 1.23 2.88
ME 0.25 St. Dev. 0.40 0.12 0.57
HI 0.35 C.o.V. 0.17 0.10 0.20
2%/50 ‘‘very rare’’ earthquake LO 0.23 All types(G þG and GþO) Average 2.29 1.29 2.96
HI 0.38 St. Dev. 0.43 0.16 0.71
ME 0.53 C.o.V. 0.19 0.12 0.24

ag ag
ag (2%/50) ag (10%/50)
0.06 g for LO
ag (50%/50) = 0.10 g for ME
ag (10%/50) 0.14 g for HI
0.15 g for LO
0.15 g for LO ag (10%/50) = 0.25 g for ME
ag (10%/50) = 0.25 g for ME 0.35 g for HI
0.35 g for HI
IDA (demand) curves
0.23 g for LO IDA (demand) curves
dD d
ag (50%/50) ag (2%/50) = 0.38 g for ME
0.53 g for HI 0.5du dC = du
dD,IO dD,LS dD,CP d Selected curve IDA curve
dC,IO=dy 0.5dC,LS dC,LS=d p dC,CP =du Unselected IDA curve

Selected curve IDA curve capacity curve

Unselected IDA curve

H
H
capacity curve Fig. 10. Selection of IDA curves for traditional approach.

Fig. 9. Selection of IDA curves for the MP design approach.

ranges from 0.10 to 0.27, which corresponds to moderately
scattered results.
in which q1 takes into account the overstrength, q2 takes into According to the proposed MP design approach and by con-
account the ductility and q3 ¼q1 q2 takes into account both sidering the obtained results, a behaviour factor q¼q1 ¼2.0 is
overstrength and ductility. proposed in the case of ‘‘rare’’ (10%/50) earthquakes, while its
In order to obtain an assessment of the behaviour factors q1, q2 value should be assumed equal to q¼q3 ¼3.0 for ‘‘very rare’’
and q3 on the basis of significant IDA results, only the IDA curves (2%/50) earthquakes.
representing realistic design conditions have been selected. As complementary task, in order to compare these results with
The selection has been performed by comparing the displacement those obtained by applying a traditional approach, one required
demand (dD) obtained for a given earthquake hazard level and performance objective only has been selected, in which the
different seismic intensity zones with the relevant displacement displacement demand (dD) has been obtained under 10%/50
capacity (dC). As result of this comparison, a performance coeffi- earthquakes, with ag equal to 0.15, 0.25, and 0.35 g for low (LO),
cient (p¼dD/ dC) has been defined and only the IDA curves for medium (ME) and high (HI) seismic intensity zone, respectively,
which p satisfies specified target conditions (e.g., 0.5 rpr1.0) and by adopting a displacement capacity corresponding to the
have been chosen. ultimate displacement (dC ¼du). For each IDA curve for which the
For the case of the proposed MP design approach, for each performance coefficient p ¼dD/ dC is in the range from 0.5 to 1.0
selected seismic hazard level, three different seismic intensities (Fig. 10), the behaviour factors q1, q2 and q3 have been evaluated.
have been assumed in such a way that ag assumes the values The obtained average, standard deviation and coefficient of
given in Table 4. Hence, in order to choose realistic cases, only IDA variation values of behaviour factors for G þG, GþO and all walls
curves satisfying the following criterion (Fig. 9) have been (GþG and GþO together) are shown in Table 6. The results show
considered: that q1 is about 2.4 for all types of wall configurations, q2 is about
ðpIO ¼ dD,IO =dC,IO r 1Þ \ ð0:5 rpLS ¼ dD,LS =dC,LS r 1:0Þ \ ðpCP 1.3, even if in case of walls sheathed with G þO the value
decreases to about 1.2, and q3 is about 3.0 for all walls, while
¼ dD,CP =dC,CP r1Þ ð6Þ
for G þG and G þO configurations is about 3.2 and 2.9, respec-
The obtained average, standard deviation and coefficient of tively. The results in terms of dispersion are very similar to those
variation values of the behaviour factors q1, q2 and q3 are shown observed in case of the MP approach, with a coefficient of
in Table 5. From the these results, it can be observed that q1 is variation in the range from 0.10 to 0.32.
about 2.2, 2.4 and 2.3 for walls sheathed with GþG, Gþ O and all On the basis of results shown in Table 6, a behaviour factor
walls, respectively; q2 is about 1.3 considering all types of wall q¼q3 ¼3.0 is proposed for a traditional seismic design under
configurations, while it is about 1.4 and 1.2 for G þG and GþO 10%/50 earthquakes.
configurations, respectively; q3 is about 3.1, 2.9 and 3.0 for walls Considering all behaviour factors defined on the basis of
sheathed with G þG, Gþ O and all wall typologies. As far as the nonlinear dynamic analysis results (both MP approach and tradi-
dispersion of the data is concerned, the coefficient of variation tional approaches), they range between 2.2 and 2.4 for q1
 L. Fiorino et al. / Thin-Walled Structures 61 (2012) 248–257 255

(overstrength), between 1.2 and 1.4 for q2 (ductility), and from The geometry and materials of the walls are selected according
2.9 to 3.2 for q3 (both overstrength and ductility). The values of to the assumptions made in Section 2, while the external spacing
overstrength related factor (q1) achieved in this study are very of sheathing-to-frame connection represents the design para-
similar to those (q1 ¼2.2C2.6) proposed by Fülöp and Dubina meter. The unit dead loads of the main elements are 0.90 kN/m2
[20], while they are larger than the corresponding result (q1 for the walls, 1.00 kN/m2 for the floor and 0.85 kN/m2 for the roof,
¼Ro ¼1.7) given by Boudreault et al. [21]. For the values of while the live load is assumed equal to 2.00 kN/m2. The resulting
the ductility related factor (q2), they are slightly smaller than unit seismic weight evaluated by Eq. (2) is 2.95 kN/m2.
the results (q2 ¼ 1.4C1.6) obtained by Fülöp and Dubina [20] The house is located in a medium intensity seismic zone,
and quite smaller than the factor (q2 ¼Rd ¼2.5) proposed by therefore following the proposed MP approach the assumed ag
Boudreault et al. [21]. As consequence, the results obtained in values are 0.10, 0.25 and 0.38 g for 50%/50, 10%/50 and 2%/50
terms of global behaviour (q3), which represents the product of q1 earthquakes, respectively. In the case in which a traditional
and q2, are smaller than those achieved by Fülöp and Dubina [20] approach is selected for seismic design, then a value of ag
(q3 ¼3.6C3.7) and Boudreault et al. [21] (q3 ¼Ro  Rd ¼4.3). equal to 0.25 g should be used for 10%/50 earthquakes. The assumed
Finally, considering the global behaviour factor (q3), the compar- soil conditions are type D, according to Eurocode 8 [13] classification.
ison with the prescriptions of applicable building codes shows All the assumptions that allow the seismic behaviour of the
that the value proposed in this paper is in the very large range building to be described by the SDOF represented by a sheathed
(q3 ¼ R ¼ 2C6.5) of those given by ASCE/SEI 7-10 [22] and FEMA CFS shear wall without torsional effects have been made. In
P-750 [18]. particular, the seismic analysis has been carried out by adopting
the following hypotheses: the ground motion acts separately in x
or y direction; the floor is assumed as a rigid diaphragms; the
6. Application of the proposed methodologies walls have symmetric configuration; the mass distribution is
uniform; the ‘‘segment’’ method is used for describing the shear
6.1. The case study behaviour of the sheathed CFS walls, assuming that the sum of the
length of the resistant wall segments in both x and y direction is
In the case study presented hereafter, a seismic design for a 9.6 m.
small structure is carried out by using the proposed MS design
procedure. The analyzed building is a typical one-family one- 6.2. The seismic design with handmade calculations
story dwelling with attic. The plan dimensions are 12  8 m, while
the height is 7.0 m including a pitched roof with 100% slope. The design consists mainly in the selection of the adequate
The structure is a stick-built construction in which floor, roof and external screw spacing that can be obtained by a linear seismic
walls are sheathed CFS frames. analysis, in which the design seismic force acting on a wall with
The sheathed CFS shear walls are placed symmetrically along unit length may be calculated by the following relationship:
the house perimeter (Fig. 11) and represent the seismic-force
resistant system. The walls are sheathed with G þO panels Se ðTÞw
HD ¼ ð7Þ
and they have dimensions 2400  2700 mm (length  height). q

where the seismic weight per unit length (w) results equal to
Table 6
29.5 kN/m and the behaviour factor (q) is assumed according to
Behaviour factors for the traditional approach.
those proposed in Section 5.
Wall configuration q1 q2 q3 The normalized (respect to gravity acceleration) elastic spec-
tral acceleration (Se(T)) has been assumed as the maximum value
Gþ G Average 2.37 1.32 3.16 obtained for natural vibration periods ranging between 0.21 and
St. Dev. 0.53 0.20 1.02
C.o.V. 0.22 0.15 0.32
0.29 s, which have been evaluated on the basis of all possible
Gþ O Average 2.40 1.20 2.87 lateral stiffness values for the selected wall configuration (lateral
St. Dev. 0.41 0.12 0.58 stiffness per unit length ranging from 1.4 to 2.7 kN/mm/m,
C.o.V. 0.17 0.10 0.20 depending on the external screw spacing). As result, Se(T) equal
All types (Gþ G and Gþ O) Average 2.38 1.27 3.04
to 0.34, 0.84 and 1.28 have been assumed for frequent, rare and
St. Dev. 0.49 0.19 0.88
C.o.V. 0.20 0.15 0.29 very rare earthquake, respectively. Therefore, the values of the
design forces (H D ) for the MP approach are 0.34  29.5/1 ¼

Fig. 11. Plan and elevation view of the dwelling.

256 L. Fiorino et al. / Thin-Walled Structures 61 (2012) 248–257

9.97 kN/m, 0.84  29.5/2¼12.5 kN/m and 1.28  29.5/3¼12.6 kN/m wall lateral strength ðHC Þ, which represents the seismic capacity,
for frequent, rare and very rare earthquake, respectively. is evaluated by Eq. (8) and it is given as a function of external
The design wall lateral strength may be estimated by the screw spacing (s). The seismic demand ðHD Þ is calculated through
following relationship: Eq. (7), in which the elastic spectral acceleration (Se(T)) is
obtained by the elastic spectra for type D soil provided by
0:625H y,av
HC ¼ ð8Þ Eurocode 8 [13], by adopting the behaviour factors (q) proposed
gm
in Section 5 and a viscous damping ratio (x) equal to 5%. The
in which H y,av is the conventional yielding strength on the elastic spectral acceleration is a function of the natural vibration
idealized bilinear schematization of monotonic average response period calculated by the following well known relationship:
curve, 0.625 is the factor which converts the average value of wall rffiffiffiffiffiffiffiffiffi
strength into the characteristic value and gm is a partial safety w=g
T ¼ 2p ð9Þ
factor assumed equal to 1.25 [23]. The design values of wall k
lateral strength for different external screw spacing (s) are shown
in Table 7. in which g is the gravity acceleration and k is the wall lateral
Therefore, an external screw spacing equal to 75 mm is adequate stiffness. In the nomograph the wall stiffness is given as a
for the examined case study (H D,max /H C ¼12.6/16.3¼0.77). If the function of external screw spacing (s).
seismic design is performed with a traditional approach, then the According to the MP design approach, the normalized wall
relevant value of Se(T) is 0.84 (for ‘‘rare’’ earthquake) and the design lateral strength is evaluated from top left graph. For s¼75 mm
force (H D ) obtained from Eq. (7) results equal to 0.84  29.5/3¼ and w¼ 30 kN/m (the real unit seismic weight is 25.9 kN/m), the
8.3 kN/m. Therefore, a screw spacing of 100 mm should be used seismic capacity ðHC =wÞ is equal to 0.54 (Point A), while the wall
(H D /H C ¼8.3/12.2¼0.68). This result confirms that in order to limit stiffness, calculated on bottom left graph is equal to 2.15 kN/m/mm
the wall damage, it is necessary to increase the wall strength by (Point B). The point C, obtained as intersection of horizontal line
reducing the screw spacing. drawn from point B and the curve relevant to w¼30 kN/m, defines
the vibration period (T¼0.24 s). For each performance level, the
6.3. LD procedure acting seismic force is estimated by the curves relevant to medium
(ME) seismic intensity zone on the top right graph (IO: ag ¼0.10 g, LS:
In the LD nomograph illustrated in Fig. 12, both wall strength ag ¼ 0.25 g, NC: ag ¼0.38 g) and the obtained values are: HD =w ¼ 0:34
ðHC Þ and design seismic action ðHD Þ for wall unit length are for IO (point D), HD =w ¼ ¼ 0:42 for LS (point E), HD =w ¼ ¼ 0:43 for
normalized respect to the seismic weight (w). In particular, the CP (point F). All the points D, E and F are lower than the point A, then
the external spacing is adequate and the performance coefficients are
Table 7 p(IO)¼0.63, p(LS)¼ 0.78 and p(CP)¼0.80.
Design lateral strength for a G þ O 2400  2700 mm wall. For the sake of completeness, the LD nomograph is applied
also in the case of traditional design approach. Therefore, for
External screw spacing (s) (mm) Design lateral strength (H C ) (kN/m)
s¼100 mm and w¼30 kN/m the normalized wall strength and
150 8.2 stiffness are represented by point G ðHC =w ¼ 0:40Þ and point H
100 12.2 (k¼ 1.75 kN/m/mm), respectively. For this stiffness value, on the
75 16.3 curve relevant to w¼30 kN/m of bottom right graph the natural
50 24.5 vibration period is equal to 0.26 s (Point I). For this value of period
on the top right graph and considering the curve relevant to
current approach and medium (ME) seismic intensity zone
(ag ¼0.25 g), the seismic demand ðHD =wÞ is equal to 0.28 (Point J).
The Point G is upper then point J, then the external spacing is
adequate ðp ¼ HC =HD ¼ 0:70Þ.

6.4. ND procedure

In the ND nomograph (Fig. 13), each demand curve is derived

by considering the average of the IDA curves obtained for a given
wall condition. As mentioned before, the curves provide the
displacement demand as function of seismic intensity (ag/g).
In addition, yield (dy), peak (dp) and ultimate displacement (du),
as defined in Section 3, are assumed for seismic capacity and are
given as a function of external screw spacing (s).
For the application of nomograph in the case of the MP design
approach, the displacement demand for each level is evaluated on
the demand graph by considering the curve 10 (s ¼75 mm,
w¼30 kN/m). In particular, for ag ¼0.10 g (IO) the displacement
demand is 7 mm (point A), for ag ¼0.25 g (LS) the displacement
demand is 24 mm (point B), while for ag ¼0.38 g (CP) the
displacement demand is 40 mm (point C). On the capacity graph,
for s equal to 75 mm, the displacement capacity is: dy ¼16 mm for
IO (point D), dp ¼45 mm for LS (point E) and du ¼49 mm for CP
(point F). The points D, E and F are higher than points A, B and C,
respectively, then the external spacing is adequate and the perfor-
Fig. 12. Application of LD dynamic nomograph. mance coefficients are p(IO)¼0.44, p(LS)¼0.53 and p(CP)¼0.82.
 L. Fiorino et al. / Thin-Walled Structures 61 (2012) 248–257 257

approaches, which usually are widely based on 10%/50 earth-

quakes. In particular, a behaviour factor equal to q¼ q1 ¼2 is
proposed for 10%/50 hazard level, which is related to live safety
structural performance, while a behaviour factor of q¼ q3 ¼3
should be used for 2%/50 hazard level, which corresponds to
collapse prevention structural performance. As complementary
task, also the evaluation of the behaviour factor according to
traditional approaches has been carried out and a behaviour
factor equal to q¼q3 ¼3 has been found.

References

[1] Landolfo R. Advances in Italian cold-formed steel structures research.

In: Loughlan J, editor. Thin-Walled Structures. Recent Advances and Future
Trends in Thin-Walled Structures Technology. London: Canopus Publishing
Limited; 2004. p. 21–42.
[2] Landolfo R. Cold-formed steel structures in seismic area: research and
applications, in the Proceedings of VIII Congresso de Construc- a~ o Metálica e
Mista, Guimara~ es, Portugal, 2011, pp. 3–22.
[3] AISI S213-07/S1-09, North American Standard for Cold-Formed Steel
Framing—Lateral Design 2007 Edition with Supplement No. 1, American Iron
and Steel Institute (AISI), Washington, 2009.
[4] Landolfo R, Fiorino L, Iuorio O. A specific procedure for seismic design of cold-
formed steel housing. Advanced Steel Construction 2010;6:603–18.
[5] Landolfo R, Fiorino L, Della Corte G. Seismic behaviour of sheathed cold-formed
structures: physical tests. Journal of Structural Engineering 2006;132:
570–81.
[6] Fiorino L, Della Corte G, Landolfo R. Experimental tests on typical screw
Fig. 13. Application of ND dynamic nomograph. connections for cold-formed steel housing. Engineering Structures 2007;29:
1761–73.
[7] Fiorino L, Iuorio O, Landolfo R. Sheathed cold-formed steel housing: a seismic
For the traditional design approach, the seismic displacement design procedure. Thin Walled Structures 2009;47:919–30.
demand is estimated on the demand graph considering the curve [8] Simpson Strong-Tie Company, /www.strongtie.comS, 2007.
[9] Hilti, North America product technical guide, 2005.
11 (s¼ 100 mm, w¼30 kN/m). On this curve, for an ag ¼0.25 g, the [10] Della Corte G, Landolfo R, Fiorino L. Seismic behavior of sheathed cold formed
displacement demand dD is equal to 31 mm (point G). The point H structures: numerical study. Journal of Structural Engineering 2006;132:558–69.
on capacity graph represents the seismic displacement capacity [11] Fiorino L, Iuorio O, Landolfo R. Sheathed cold-formed steel housing: a seismic
design procedure. Engineering Structures 2011;34:538–47.
(dC ¼du ¼46 mm) corresponding to s¼100 mm. The point H is [12] Richard RM, Abbott BJ. Versatile elastic-plastic stress-strain formula.
upper than point G, then the external spacing is adequate (p ¼dD/ Journal of the Engineering Mechanics Division 1975;101:511–5.
dC ¼ 0.67). [13] CEN, EN 1998-1–Eurocode 8: Design of structures for earthquake resistance
— Part 1: General Rules, Seismic Actions and Rules for Buildings, European
Committee for Standardization, Bruxelles, 2004.
[14] Iervolino I, Maddaloni G, Cosenza E.. Unscaled real record sets compliant with
7. Conclusions Eurocode 8, First European Conference on Earthquake Engineering and
Seismology, Geneva, Switzerland, Paper n. 113, 2006.
In this paper a performance-based methodology for the seis- [15] Kawai Y, Kanno R, Uno N, Sakumoto Y., Seismic resistance and design of steel
mic design of sheathed CFS walls has been proposed. Based on the framed-houses. Nippon Steel Technical Report n. 79, 1999.
[16] Gad EF, Duffield CF, Hutchinson GL, Mansell DS, Stark G. Lateral performance
results of an extensive parametric dynamic analysis, a seismic of cold-formed steel-framed domestic structures. Engineering Structures
design procedure, which allows to define in three steps all the 1999;21:83–95.
shear wall components, has been presented. The proposed Multi- [17] Dubina D. Behaviour and performance of cold-formed steel-framed houses under
seismic action. Journal of Constructional Steel Research 2008;64:896–913.
Step (MS) procedure allows to define the external screw spacing, [18] FEMA P-750, NEHRP Recommended Provision for seismic provisions for new
which represents the main seismic design parameter, through buildings and other structures. Washington: Federal Emergency Manage-
design nomographs based on linear dynamic (LD) or nonlinear ment Agency; 2009.
[19] Blais C., Testing and Analysis of Light Gauge Steel Frame/9mm OSB Wood
dynamic (ND) analyses. In case of ND analysis, the results of the
Panel Shear Walls, Master’s Thesis, Department of Civil Engineering and
parametric study can be directly applied to design the sheathed Applied Mechanics, McGill University, Montreal, 2006.
CFS shear walls by using incremental dynamic analysis response [20] Fülöp LA, Dubina D. Performance of wall-stud cold-formed shear panels
curves according to a performance-based methodology. Instead, under monotonic and cyclic loading Part II: numerical modelling and
performance analysis. Thin Walled Structures 2004;42:339–49.
in case of LD analysis, the nonlinear response is taken into [21] Boudreault FA, Blais C, Rogers CA. Seismic force modification factors for light-
account by defining different behaviour factors according to a gauge steel-frame–wood structural panel shear walls. Canadian Journal of
Multi-performance (MP) approach. This approach proposes to Civil Engineering 2007;34:56–65.
[22] ASCE/SEI 7-10, Minimum design loads for buildings and other structures.
relate different seismic hazard levels to different seismic perfor-
Reston: American Society of Civil Engineers (ASCE); 2010.
mance levels, in such a way to achieve an ‘‘enhanced objective’’ by [23] Veljkovic M, Johansson B. Light steel framing for residential buildings.
allowing an effective damage control, differently from the current Thin Walled Structures 2006;44:1272–9.