You are on page 1of 37

Bull Earthquake Eng (2018) 16:259–294

https://doi.org/10.1007/s10518-017-0196-5

ORIGINAL RESEARCH PAPER

Precast industrial buildings in Southern Europe: loss


of support at frictional beam-to-column connections
under seismic actions

Cristoforo Demartino1,2 • Ivo Vanzi3,4 • Giorgio Monti1,2 •

Concetta Sulpizio3,4

Received: 22 November 2016 / Accepted: 18 July 2017 / Published online: 25 July 2017
Ó Springer Science+Business Media B.V. 2017

Abstract This paper presents the evaluation of the loss-of-support conditions in frictional
beam-to-column connections of industrial precast concrete buildings under seismic actions.
This type of connection is widespread throughout Southern Europe in non-seismically
designed industrial precast buildings. First, geometric properties of industrial precast
buildings and of the frictional beam-to-column connections, together with reference values
for the friction coefficient, are reviewed. Then, earthquake time histories taken from the
European Strong-Motion sets and recordings of the two major shocks of the 2012 Emilia-
Romagna events are presented and discussed showing the importance of the vertical
component. Two dynamic models of increasing complexity are used to ascertain loss-of-
support conditions under seismic action. The first model is an elastic one, representing a
single frame of the industrial buildings. Results are obtained according to: (1) 2D analyses,
disregarding the time correlation between the response peaks along the horizontal and
vertical directions, (2) 2D analyses taking into account time correlation, and (3) 3D
analyses to evaluate also directionality effects. The second model is a 2D non-linear planar
frame developed within the OpenSees framework. Results show that simplified (linear)
models are a good proxy to more refined (non-linear) ones. However, one must resort to
non-linear models if differential displacements between beam and column are of interest.
The non-linear numerical investigations show that friction coefficient, horizontal and

& Cristoforo Demartino


cristoforo.demartino@me.com
1
College of Civil Engineering, Nanjing Tech University, Nanjing 211816, People’s Republic of
China
2
Department of Structural Engineering and Geotechnics, Sapienza University of Rome, via A.
Gramsci 53, 00197 Rome, Italy
3
Department of Engineering and Geology, University G. d’Annunzio of Chieti-Pescara, viale
Pindaro, 42, 65127 Pescara, Italy
4
College of Civil Engineering, Fuzhou University, Fuzhou 350108, Fujian, People’s Republic of
China

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
260 Bull Earthquake Eng (2018) 16:259–294

vertical periods and damping, and column reinforcement ratio are the key variables in
estimating the loss-of-support conditions.

Keywords Precast industrial buildings  Beam-to-column connections  Loss


of support  Friction  Vertical ground motion

1 Introduction

Damages to industrial buildings under extreme natural events (e.g., earthquakes, floods,
winds, etc.) may be extensive. Apart from the utterly important issue of safeguarding the
life of workers, two types of notable losses may be identified, the first one being the
economic loss associated with the interruption of the production, while the second one is
the environmental loss associated with dispersion of hazardous materials usually therein
stored and/or employed. For instance, the Fukushima Daiichi nuclear power plant accident
in Japan triggered by a magnitude 9 earthquake and resulting tsunami on March 11, 2011,
caused a month-long discharge of radioactive materials into the atmosphere (Chino et al.
2011) and highlighted the related economic and environmental risks. This and other
remarkable events occurred in industrial plants highlighted the urgent need to investigate
strategies for risk evaluation and mitigation for one-storey precast buildings that are the
most common industrial building typology in Europe.
One-storey industrial precast buildings are an assemblage of precast elements that,
suitably connected, form a 3D framework capable of resisting very efficiently to vertical
and lateral loads. Under the seismic action, when correctly designed, they can attain very
good seismic performances (e.g. Negro et al. 2013; Brunesi et al. 2015; Biondini et al.
2013; Negro and Lamperti Tornaghi 2017). However, when non-seismically designed, they
prove highly vulnerable. Structural deficiencies were observed during recent earthquakes
in China (2008 and 2010), New Zealand (2011), Japan (2011), Turkey (2011) and Italy
(2009 and 2012). Vulnerabilities pertain to both structural and non-structural elements:
foundations, precast columns, connections, roof system and surrounding infill panels.
The main difference between a cast-in-place concrete structure and a precast one is the
presence of joints, which strongly affect the overall response. In many countries (USA,
New Zealand, Japan, Australia, etc.), rigid connections are preferred for beam-to-column
joints, while in Europe (Italy, Greece, Spain, Portugal, Slovenia etc.) and elsewhere
(Turkey, Armenia, etc.), simple dry-pinned connections are traditionally used in frame type
buildings (Psycharis and Mouzakis 2012a). This type of connection is the main reason of
the exhibited large vulnerability of these buildings in Southern Europe. It is worth noting
that nowadays simple frictional beam-to-column connections are not used in seismic
countries anymore since modern Codes forbid their use. Accordingly, in the following, we
will refer to simple frictional beam-to-column connections employed in existing buildings.
One of the recent events, showing all the deficiencies of precast industrial buildings,
was the seismic sequence in Emilia-Romagna (Administrative Region of Northern Italy) in
May-July 2012. Details of what happened following the earthquakes are given in the
following. The main shocks were of moderate intensity (around ML ¼ 6) although they
caused several collapses of precast industrial buildings, due to their high density in Emilia-
Romagna (Braga et al. 2014). Distinct categories of earthquake-induced structural damage
have been identified (ReLUIS et al. 2012): (1) flexural hinges at the base of precast
columns, (2) large sliding of the beam-to-column connections, and (3) large damage to
infill elements and to shelves. Damages to roof systems and to foundations were also

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 261

found. The earthquake caused 17 casualties, 13,000 homeless and severe damage to the
economy of the region. The event was deemed as the costliest events outside the U.S. in
2012 (Impact Forecasting, the center of excellence for the development of models of Aon
Benfield Catastrophe, in ‘‘Annual Global Climate and Catastrophe Report - 2012’’). Italy’s
state-financing body Cassa Depositi e Prestiti (CDP) allocated 12 billions of Euros to
subsidize the rebuilding process (Benfield 2013); insured losses were approximately 1.5
billion Euros. Overall, the total cost was larger than 1% of the Italian Gross National
Product (GNP).
Considering the Emilia-Romagna case, the observed large seismic risk can also be
related to the evolution of the requirements for precast structures of the Italian Building
Codes (e.g. Marzo et al. 2012; Mezzi et al. 2014; Magliulo et al. 2014) and of the seismic
classification. Specific indications for precast structures were given only in 1970, with no
special provision before this date. However, most industrial buildings were realized after
the 1970s (Bellotti et al. 2014), but the use of beam-to-column connections with shear
transfer mechanism based only on friction was allowed until 1987 (MIT 1987) in seismic
zones and until 2003 in other zones (PCM 2003). After 2003, according to Eurocode 8
(CEN 2004), the use of friction connection has been forbidden. On the hazard side, Emilia-
Romagna was declared seismic only in 2004, as low seismicity area, as it is also today
(PCM 2006). Bournas et al. (2014) observed that the 2003 seismic zoning was waived for
18 months and become compulsory in 2005. Recently, Dolce and Di Bucci (2015)
observed that the three similar magnitude earthquakes of Umbria and Marche (1997),
Abruzzo (2009) and Emilia (2012) caused very different damage to industrial buildings.
The former two caused little damage thanks to the low industrial density (Umbria) or to
compulsory seismic design since 1915 (Abruzzo). The latter caused extensive damage
because of high industrial density and inadequate seismic design. Liberatore et al. (2013)
reported similar observations. Finally, the Friuli earthquake caused little damage because
of low energy content in the medium-to-high periods (Fajfar et al. 1978).
Frictional beam-column connections are probably the main vulnerability source for
precast concrete industrial buildings (e.g. ReLUIS et al. 2012; Parisi et al. 2012; Marzo
et al. 2012; Savoia et al. 2012; Bournas et al. 2014; Clementi et al. 2016). Several authors
have analyzed beam-to-column connections with respect to the loss-of-support limit state.
Toniolo (2002), after the Kocaeli (1999) earthquake in Turkey, noticed that beams loss of
support and overturning were the main causes. Magliulo et al. (2008) performed 3D non-
linear dynamic analyses on two typical 1950–1970 industrial structures. Vertical accel-
eration was taken into account. They assumed a friction coefficient equal to 0.6 [i.e.,
neoprene-concrete contact surfaces; this value is lower than the value subsequently found
by the same research group (Magliulo et al. 2011)] and found this value was exceeded in
the corner columns, thus indicating a 3D structural behavior.
Mata et al. (2006) and Plumier and Calvi (2007) observed that the poor seismic per-
formance of some precast structures is due to their low level of structural damping, P-d
effects and low ductility of the structural joints. They proposed the use of passive dissi-
pating devices for improving their behavior. Martinelli and Mulas (2010) proposed
insertion of friction and mechanical devices in the beam-to-column connections. ReLUIS
et al. (2012) indicate different strategies for the beam loss of support; among them,
increasing the seating length and insertion of mechanical devices. Belleri et al. (2015)
investigated retrofit solutions foreseeing mechanical devices in frictional connections.
A large number of authors tried to explain at different levels the damage observed in
beam-to-column connections. Liberatore et al. (2013) performed 2D non-linear dynamic
analyses on a typical Emilia-Romagna industrial building. 0.3% and 1% column

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
262 Bull Earthquake Eng (2018) 16:259–294

longitudinal reinforcement ratios were considered. They modeled the beam-to-column


connections as either pin or friction connections; in the latter case, they assumed a friction
coefficient of 0.4. They used two time histories recorded in Mirandola during the 2012
Emilia-Romagna sequence. They found directionality of damage (i.e., dependence on the
horizontal component adopted in the analysis), relevance of the earthquake vertical
component of excitation and significant columns inelastic rotation, dependent on the
longitudinal reinforcement. Magliulo et al. (2014) evaluated the vulnerability of the beam-
to-column connections using the pseudo-acceleration spectrum of the horizontal time
histories evaluated at the fundamental horizontal periods of the buildings. Friction coef-
ficient was assumed equal to 0.13 (Magliulo et al. 2011) and safety factors well below 1
were found for a wide range of periods. Belleri et al. (2014) added the vertical component
of seismic input to the model of Magliulo et al. (2014), considering 50% of the mass of the
roof system (roof and beam) as vertical participating modal mass. They adopted a friction
coefficient between 0.2 and 0.25. The vertical component of earthquake excitation
decreased the safety factor by about 10%. A similar approach was also used by Demartino
et al. (2017). Recently, Casotto et al. (2015) presented a seismic fragility model for Italian
RC precast buildings, to be used in earthquake loss estimation and seismic risk assessment.
They performed 2D and 3D non-linear dynamic analyses and the friction coefficient (in the
range of 0.2–0.3) was considered as a random variable. Fragility curves were computed
with Monte Carlo simulations. Similarly, Babič and Dolšek (2016) derived fragility
functions for 12 classes of Italian precast buildings. Both peak ground and spectral
accelerations were used as intensity measures.
Although several authors evaluated frictional beam-to-column connections sliding
conditions, little attention has been paid to dynamic assessment of the most influential
variables, which is one of the goals of this paper. Moreover, the second scope is to
highlight the risk of loss of support at frictional beam-to-column connections under seismic
actions using the typical characteristics of buildings representing the Southern European
situation (with special reference to Italy). In fact, although nowadays this type of con-
nection is prohibited in Europe by Eurocodes, especially in Southern Europe a large
number of existing precast industrial buildings employed this technology leading to a large
risk. This risk in Southern Europe is enhanced by the presence of different sites with a
large density of vulnerable industrial buildings and by seismic hazard with characteristics
similar or worse than the Emilia-Romagna seismic events.
The geometrical and mechanical characteristics of Italian precast concrete industrial
buildings are reported (Sect. 2) and a review of the frictional beam-to-column connection
typologies and of reference values for the friction coefficient are given (Sect. 2.2). The
governing parameters of the transverse response and their statistic distribution are high-
lighted in Sect. 2.3. The earthquake data set used are the Emilia-Romagna earthquakes
plus a careful selection from the European Strong Motion database. The earthquake data
set is then presented (Sect. 3). Two dynamic models of increasing complexity are used to
ascertain loss-of-support conditions under seismic action. The first model (Sect. 4) is a
3-DoFs elastic one, representing a single frame of the industrial buildings. Results are
obtained (Sect. 4.2) both by disregarding the time correlation between the response peaks
along the horizontal and vertical directions and by taking it into account and, in Sect. 4.3,
by evaluating the directionality effects. Moreover, this model is applied using the elastic
spectra defined in the Italian Building Code (Sect. 4.4). The second model is a 2D non-
linear planar frame developed within the OpenSees framework (Sect. 5). Both geometric
non-linearity due to P-D effects, and material nonlinearity, arising from the non-linear
behavior of the columns (modeled with a fiber section) and for the connection frictional

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 263

behavior (modeled with the Coulomb’s law) are modeled (Sect. 5.1). A typical Italian
industrial precast building whose characteristics are described in Sect. 5.2 is used as case
study to analyze the output of the FEM model and results are reported in Sect. 5.3. Finally,
some conclusions and prospects are drawn (Sect. 6).

2 Precast concrete industrial buildings

This Section introduces the most important characteristics, from the structural point of
view, of precast concrete industrial buildings. A general description is first given
(Sect. 2.1). Then, Sect. 2.2 shows the behavior of the most important structural part, i.e.
the beam-to-column connection. The behavior of such connections is, in fact, mainly
affected by the value adopted for the friction coefficient (typically concrete-to-concrete or
neoprene-to-concrete surfaces), which does not appear to be adequately estimated for this
type of buildings. The final section individuates, from a qualitative point of view, the most
important geometrical and mechanical properties influencing the overall response. Refer-
ence values, for a typical case representing the Southern European situation (especially
Italy), are given.

2.1 General description

Precast concrete frames are widely used for both low-rise and multi-storey offices and for
elevated car parks (Elliott 2002). Since industrial facilities may be standardized for space
distribution and layout, precast concrete frame buildings are widely used. One-storey
buildings are the most common structural configuration (Dassori and Assobeton 2001).
This type of buildings are mostly located in Southern Europe, e.g., in Italy (Dassori and
Assobeton 2001) and in Turkey (Sezen et al. 2000; Posada and Wood 2002), and also in
China (Zhu 2015).

Fig. 1 Typical structure of single-story precast concrete industrial buildings

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
264 Bull Earthquake Eng (2018) 16:259–294

The typical structure of one-storey precast concrete industrial buildings analyzed in this
study is shown in Fig. 1. A precast frame is composed of two foundation systems, two
columns and one beam (Fig. 1), all precast. Fixed-base cantilever columns (inserted in
precast socket footings) are usually assumed in any model. Precast beams and roof tiles are
modeled as simply supported. Each part of the typical building is detailed within this
section while frictional beam-to-column connections are presented in Sect. 2.2.
The column-foundation connection used in precast concrete structures is built by
embedding a column portion (embedded length) into a pocket within the foundation
(Canha et al. 2009). The space between the pocket and the column is filled with cast-in-
place concrete. Vertical loads are transmitted to the foundation by forces acting on the
bottom and the lateral faces; roughness of the contact surfaces is the key parameter. The
pocket is formed in the foundation itself (Elliott 2002). Columns embedment depth larger
than 1.5 times the column cross-section is required to obtain fixed-base restraint (Osanai
et al. 1996). Flexibility of the soil-foundation interface requires separate assessment if
needed (Belletti et al. 2016).
The standard cross-section for precast concrete columns for single-storey sheds is either
rectangular or square and the most common size is 500  500 mm, with possible variation
in 50 mm steps (Dassori and Assobeton 2001). Minimum cross-section of 250  300 mm
have been observed, and are usually related to beam-to-column connector size. The pre-
casting method allows geometric reinforcement ratio up to 10%, although values of 1–2%
are usually observed (Liberatore et al. 2013). As for beams, cross-sections may have
various shapes (I, Y, T) in order to lighten the structural system. Depths may vary along the
length, according to the bending moment, and prestressing is used for long spans. Beams
may be classified as (Elliott 2002): internal (inside the building) and external (between two
corner columns, often L-shaped and loaded also in torsion). Concrete in precast columns
and beams is usually of high quality, due to the care that was taken in a controlled
environment. Bar diameters are from 8 up to 12 mm for stirrups and from 16 up to 40 mm
for flexural bars.
Roof system and cladding panels are not reviewed in this Section since they do not fall
within the scope of this study. The interested reader is referred to Bonfanti et al. (2008) for
more details about the characteristics of Italian single-story precast concrete industrial
buildings and to Elliott (2002), Bachmann et al. (2011), Elliott and Jolly (2013) for precast
concrete structures.

2.2 Frictional beam-to-column connections and reference values


of the friction coefficient

In frictional beam-to-column connections, the shear transfer is based on friction only, as


already observed in Sect. 1 for non-seismically designed precast industrial buildings. In
this case, the frame behavior depends on the relative intensity of horizontal and vertical
forces acting on the joint. The behavior is a hinge (allowing a limited rotation between the
column and beam) under small horizontal loads (such as wind actions), and the static
scheme is two columns clamped at the base with a pin-ended beam. For higher horizontal
loads (e.g., strong seismic actions), horizontal connection resistance (sum of friction and
possible dowel) may be overcome, and horizontal relative displacements between the top
of the column and the beam end may occur. Considering the typical Southern European
precast industrial buildings, the dowel stiffness and resistance as a restraint may be dis-
regarded. Although pinned joints (with dowels) were widely investigated, the analysis of
this type of connections would not fall within the scope of this study since referring to

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 265

seismically designed buildings and will not be discussed in the following. The interested
reader is referred to the extensive literature available (e.g. Psycharis and Mouzakis
2012a, b; Zoubek et al. 2013, 2015; Clementi et al. 2016).
Frictional beam-to-column connections are realized using different contact surfaces,
e.g., concrete-to-concrete or neoprene-to-concrete. The connection to the column head can
be realized in several ways, e.g., directly on the plain head, with a corbel, or by shaping the
head of the columns as a fork or tenon. Beam-to-column connection shapes define kine-
matics: plain heads release both horizontal displacements; all other cases restrain at least
one. Connections are further detailed in Dassori and Assobeton (2001), Elliott and Jolly
(2013), Bellotti et al. (2014).
Friction in the connection has a complex behavior since it is based on shear transfer
across discontinuities. Experiments indicate dependence upon many variables, including
material coupling, sliding displacement, speed and acceleration, temperature, normal
stress, humidity and surface preparation (Berger 2002; Andersson et al. 2007; Foerster
et al. 1989; Magliulo et al. 2011; Zoubek et al. 2013). The friction model most commonly
used is the Coulomb one, which can be formulated as (Andersson et al. 2007):

FF ðtÞ _ ¼ 0 and FF  ls FN ðtÞ
if xðtÞ
FF ðtÞ ¼ ð1Þ
ld FN ðtÞ _ 6¼ 0
if xðtÞ
where FF ðtÞ and FN ðtÞ are respectively the tangent and normal force, xðtÞ
_ is the velocity, ls
and ld are the static and dynamic friction coefficients. For the connection case, Eq. 1
defines two distinct behaviors (slip and non-slip modes), activated according to the
earthquake excitation level.
ls is higher than ld , although, after the first cycle, ls decreases and can be assumed
equal to ld (Foerster et al. 1989). Accordingly, the pre-cracked interface does not expe-
rience the difference between the static friction load and sliding resistance, which are the
same (Foerster et al. 1989). In the following, the difference between static and dynamic
friction coefficient will be neglected.
Large variability is found in the literature for the friction coefficient, l. Some studies
concern aggregate interlock (for shear transfer mechanism in beams) (e.g. Taylor 1970;
Paulay and Loeber 1974; Santos and Júlio 2012). The frictional properties of support joints
were first examined by Mollesten and Packalen (1966); l values between 0.6 and 0.8 were
found for smooth concrete surfaces, up to 1.4 for roughened concrete. For concrete
members cast at different times, the shear-friction theory (Birkeland and Birkeland 1966)
was proposed for longitudinal strength assessment. Friction coefficients were experimen-
tally assessed for variable smoothnesses, ranging from 1.7 (monolithic concrete) to 1.4
(artificially roughened joints), to 0.8–1. Birkeland and Birkeland (1966) determined the
friction coefficient varying for several situations, namely: (1) l ¼ 1:7, for monolithic
concrete, (2) l ¼ 1:4, for artificially roughened construction joints and (3) l ¼ 0.8–1
(ordinary joints). Shaikh (1978) proposed the friction coefficient (1) l ¼ 1:4, for concrete-
to-concrete cast monolithically, (2) l ¼ 1:0, for a concrete to hardened concrete interface
with a roughness amplitude of 6.4 mm, (3) l ¼ 0:6, for a concrete-to-steel interface and
(iv) l ¼ 0:4, for a concrete-to-concrete smooth interface. Randl (1997) found values within
the 0.5–1 interval, depending on the surface roughness. Papanicolaou and Triantafillou
(2002) investigated the shear transfer capacity of interfaces between pumice aggregate
concrete and high-performance concrete. They found l ¼ 0.33–0.46, depending on the
surface roughness. Santos and Júlio (2010) found l within the 1.1–1.4 range, depending on
the mean valley depth (Muralikrishnan and Raja 2008). Mohamad et al. (2015) found l

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
266 Bull Earthquake Eng (2018) 16:259–294

values within the 0.6–2.1 range, depending on the mean peak height (Muralikrishnan and
Raja 2008).
For the case of elastomeric bearing pad inserted between column and beam, concrete-to-
elastomeric material friction must be considered. Magliulo et al. (2011) tested both neo-
prene and generic rubber pads. Normal stress was kept low, at 2  103 MPa. The friction
coefficient mean value and c.o.v. were estimated equal to 0.492 and 12%, respectively.
Higher normal stress, between 1.7 and 5:3 MPa showed reduction to about 0.1. They
proposed the following relationships for concrete-to-neoprene friction:
8
< 0:49 rn  0:14 MPa
l¼ 0:055 MPa ð2Þ
: 0:1 þ 0:14 MPa\rn  5 MPa
rn
Comparison with literature and standards (CNR 1999; Schrage 1981; PCI 1985; UNI 2005)
highlighted good agreement for the low normal stress case.
As a final note of synthesis, literature indicates concrete-to-concrete friction coefficients
ranging from 0.5 to 1.2, mainly depending on surface roughness and normal stress. Neo-
prene-to-concrete friction coefficients range from 0.1 to 0.5, mainly depending on normal
tension.

2.3 Governing parameters of the transverse response and their statistic


distribution

The dynamic transverse response of one-storey industrial precast concrete buildings is


largely dependent on strength and stiffness of the roof system. The latter may be rigid or
flexible. The case of Italian buildings is retained as an example in this paper; for the
majority of the Italian building stock, the roof system is flexible and has limited capability
to redistribute forces to different columns.
In order to perform a parametric study on the results, which will be illustrated in the
next Sections, the most important geometric quantities have been retained as variables.
These include frames spacing I, column height H, beam length L (Fig. 1). Beams and
columns are parametrized via their equivalent (in terms of strength and stiffness) rectan-
gular shapes. The parameters retained are the cross-section dimensions bb , hb , bc and bc
(subscripts b and c refer to beam and column, respectively). Masses applied on structural
elements are estimated via tributary areas. Reinforcement ratio (i.e., the ratio of steel to
concrete areas) is considered only for columns. This is an important parameter as it
influences resistance, stiffness, and ductility of the columns.
A further parameter to be defined is structural viscous damping, that must be defined
separately for beams (where elastic behavior is expected) and columns (with possible
inelastic behavior). For beams, no literature study has been found, except the tests by
Franchetti et al. (2009). These are relative to a 7.3 m long precast prestressed reinforced
concrete beam, with 300  600 mm rectangular cross section. Reinforcement ratios varied
from 0.13% (ordinary reinforcement) to 0.35% (tendons). Franchetti et al. (2009) found very
low structural damping, in the range of 0.3–0.8% for the first three modes. These results
disregard the energy loss due to the roof system. In this study, relatively small values of the
damping ratio for the vertical mode (beam and roof system) have been assumed.
The above variables have been defined based on the authors’ experience and databases,
and on available literature data. After the Emilia (2012) earthquake, the authors surveyed
about one hundred industrial buildings, collected available data and built up a first

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 267

database. The variables mean values have been found consistent with the results from
Bellotti et al. (2014), who used data retrieved from an Italian and two regional (Emilia-
Romagna and Tuscany) databases. Mean values of the geometric variables are reported in
Table 1. A correlation study on the authors’ dataset has further shown large positive
correlation between bc and hc , i.e., columns are generally square.

3 Seismic input

This Section is devoted to presenting the seismic input adopted for the numerical simu-
lations that follow. In particular, two earthquakes databases are used. The first one is the
collection of the time histories of the two main events of the Emilia-Romagna 2012
earthquake sequence. The second one is a selection of the records with characteristics
similar to the Emilia-Romagna earthquakes, taken from the European Strong-Motion
Database (Ambraseys et al. 2004). The choice of sets of earthquake records representing
the Emilia-Romagna earthquakes has been made to simulate the condition of a typical case
representing the Southern European situation, characterized by a moderate intensity. First,
details on the Emilia sequence are given (Sect. 3.1). In Sect. 3.2, the adopted set is pre-
sented. It contains the Emilia-Romagna earthquakes plus a careful selection from the
European Strong Motion database.

3.1 Characteristics of the 2012 Emilia-Romagna seismic sequence

Two main events were recorded an ML 5:9 on May 20th near Finale Emilia, at 6.3 km
depth an ML 5:8 on May 29th near Mirandola, at a depth of 10.2 km (Scognamiglio et al.
2012). The major events were close in time (9 days), space (11.91 km) and had the same
focal mechanisms, i.e., thrust-type, (Ganas et al. 2012). Historical hazard analyses,
excerpted from the Italian Macro-seismic Database CPTI04 (Stucchi et al. 2007) show the
area is earthquakes prone. Within 30 km from Mirandola, starting in 91 A.C., 19 earth-
quakes with MMS  5 were found, the strongest one ðMMS ¼ 5:85Þ in 1501 in the
Appennino Modenese. The disaggregation of the peak ground acceleration performed
according to the procedure reported in Spallarossa and Barani (2007) showed that, for both
Mirandola and San Felice Panaro, it is influenced chiefly by ground motions with mag-
nitude M  5 at a distance of about 10 km. This is compatible with the Emilia-Romagna
earthquakes main events of 2012.
The Po plain is one of the largest sedimentary basins in the world, with an area of about
50,000 km2 and a sediment thickness varying from few tens of meters to about 8 km (Luzi
et al. 2013). The subsoil class is mainly C, according to the Italian Building Code [NTC
(MIT 2008)] and Eurocode 8 (CEN 2004). For reference, codes subsoil classes are shown

Table 1 Mean geometric characteristics of precast concrete industrial buildings


Source L ½m H ½m I ½m bc ½m hc ½m

Authors’ survey 15 6.2 9.4 0.45 (mode 0.5) 0.46 (mode 0.5)
Bellotti et al. (2014) 16.2 6.5 7.0 n/a n/a

L, H, I are defined in Fig. 1; bc and hc are the base and height of the column cross-section, respectively

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
268 Bull Earthquake Eng (2018) 16:259–294

in Table 2 as a function of the shear wave velocity averaged on 30 m thickness, vs;30 ,


together with the soil factor, S, (i.e., factor accounting for the soil conditions in the code
response spectra definitions). Emilia-Romagna subsoil classification has been estimated
with the geological map of Italy of the Italian Geological Service (Vanzi et al. 2015) and
the shear wave velocity estimation reported in Michelini et al. (2008). The result of the
estimation is shown in Fig. 2 for the Emilia-Romagna area. According to the Italian
Building Code [NTC (MIT 2008)] the design peak ground acceleration, with 975 years
return period, is equal to about 4 m=s2 , soil amplification effects included.
Strong motion data during the Emilia-Romagna sequence were mainly recorded by the
Italian Accelerometric Network (Rete Accelerometrica Nazionale, RAN), operated by the
Department of Civil Protection (DPC 2016). RAN includes at present 528 digital stations;
71 of them are located within 100 km from the epicenters of the two main events. The 71
stations are equally spaced in this area with a slightly larger density in the south part. In the
following, only records from these stations will be considered and presented in Sect. 3.2.
It is worth highlighting that, nowadays, the Italian Building Code [NTC (MIT 2008)]
and Eurocode 8 (CEN 2004) do not consider near-field effects in the seismic design of
structures located in the near-field domain (Grimaz and Malisan 2014). Figure 3 shows a
comparison of the design elastic response spectra with the measured response spectra.
Design elastic response spectra refer to the Italian Building Code [NTC (MIT 2008)] and
the Eurocode 8 (CEN 2004) type 1 and 2 evaluated with a return period of 975 years, for
soil class C (Table 2), for structural damping of n ¼ 5%. Measured response spectra refer
to the horizontal and vertical components for the event of the 20-5-2012 at Mirandola
(Fig. 3a) and of the 29-5-2012 at San Felice sul Panaro (Fig. 3b). Both stations were
located near the epicenter at a distance of 17 and 4 km, respectively. The return period of
the design elastic response spectra was chosen in order to better fit the horizontal com-
ponents of the recorded accelerograms. Compared to the site response analysis results, the
Italian Building Code [NTC (MIT 2008)] and Eurocode 8 (CEN 2004) type 1 and 2 elastic
spectra gives a good estimate of the horizontal response. In particular, Eurocode 8 (CEN
2004) type 2 spectrum gives the best estimate. Conversely, the vertical component is
strongly underestimated at low periods, while a sufficiently good estimate is given at high
periods. In this case, the best estimate is given by Eurocode 8 (CEN 2004) type 2 spectrum.

Table 2 Seismic ground classification according to the Italian Building Code [NTC (MIT 2008)] or the
Eurocode 8 (CEN 2004) as a function of the shear wave velocity averaged on 30 m thickness, vs;30
Soil class A B C D

vs;30 ½m=s  800 360 800 180 360 \180


S-[NTC (MIT 2008)]a 1 1  1:4  0:4  k  1:2 1  1:7  0:6  k  1:5 0:9  2:4  1:5  k  1:8
S-Type 1 [EC 8 (CEN 1 1.2 1.15 1.35
2004)]
S-Type 2 [EC 8 (CEN 1 1.35 1.5 1.8
2004)]
Comparison of the soil factor, S, reported in the Italian Building Code [NTC (MIT 2008)] and in the
Eurocode 8 (CEN 2004). (Fo is the factor accounting for the maximum spectral acceleration according to
[NTC (MIT 2008)], ag is the peak ground acceleration, g is the gravity acceleration)
a ag
k ¼ Fo  g

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 269

Fig. 2 Map of the soil class Soil class - DM 2008 or EC8


46
according to Italian Building
Code [NTC (MIT 2008)] or
D
Eurocode 8 (CEN 2004) 45.5
estimated using the Global vs;30
Map Server of the USGS. The
red and the blue circles indicate 45
C

Latitude [°]
an area of 30 and 100 km from
Mirandola, respectively. The red 44.5
dots indicate the epicenters of the
two main events B
44

43.5
A

43
10 10.5 11 11.5 12 12.5 13
Longitude [°]

(a)

(b)

Fig. 3 Comparison of the design elastic response spectra of the Italian Building Code [NTC (MIT 2008)]
and the Eurocode 8 (CEN 2004) type 1 and 2 evaluated with a return period of 975 years for n ¼ 5%, soil
class C (Table 2) with the measured response spectrum for the horizontal and vertical components for the
event of the 20-5-2012 at Mirandola (a) and for the event of the 29-5-2012 at San Felice sul Panaro (b). The
subscripts NS and EW indicates the north-south and east–west components, v the vertical component and
h an arbitrary horizontal component

This difference is due to the derivation of the code vertical spectra on the base of
accelerations derived from ordinary Probabilistic Seismic Hazard Analyses (PSHA) under
the hypothesis of far-field conditions. The vertical ground motion presents most of its

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
270 Bull Earthquake Eng (2018) 16:259–294

energy in a narrow low-periods band because it is mainly associated with the arrival of
vertically propagating compressive P-waves (characterized by short wavelength), whilst
secondary, shear S-waves are the main cause of horizontal components (Collier and
Elnashai 2001). Moreover, the effect of P-waves is stronger at low epicentral distances as
the higher frequency makes the energy attenuation (i.e., dissipation) with distance faster.
Moreover, near-fault ground motions often contain large long-period (2–5 s) wave pulses
(Mollaioli et al. 2006).
Vertical ground motions are often considered in the seismic design of critical structures
such as nuclear power plants and dams, but not for standard structures; however, different
studies suggest that the effect of the vertical ground-motion component can also be sig-
nificant for the seismic response of ordinary highway bridges on sites located within about
15 km of major faults (Kunnath et al. 2008; Gülerce and Abrahamson 2010) and moreover
can also increase the axial force on columns, especially in the inner ones as they have
generally a greater load (Broderick et al. 1994; Papazoglou and Elnashai 1996). In the
following, the importance of the vertical component in the evaluation of the loss-of-support
conditions of frictional beam-to-column connections will be demonstrated.

3.2 Emilia-Romagna earthquake records and European Strong Motion


Database

In the following, the Emilia-Romagna main events earthquake records having an epicentral
distance less than 100 km will be compared with selected records from the European
Strong Motion Database (Ambraseys et al. 2004). The latter were selected in order to be
similar to the Emilia-Romagna earthquakes with the following characteristics: epicentral
distance less than 100 km, magnitude in the range of 5.5–6 and soil class C. A total of 62
3-components accelerograms were selected.
In Fig. 4 the peak ground acceleration, ag , for the three mutual directions as a function
of the epicentral distance is compared between the Emilia-Romagna earthquake records
and the selected records from the European Strong Motion Database and with a common
Ground-Motion Prediction Equation (GMPE) available in the literature. The mean in the
horizontal and vertical directions shows the typical decrease of ag with the distance. The
mean is the average of the data with a bin of 10 km and the horizontal direction is the mean
of the two horizontal components (as defined in GMPEs Barani et al. 2015). In order to
validate the two sets of records, these were compared with the GMPE of Ambraseys et al.
(1996) that are representative of European strong motion in terms of the attenuation of
peak ground acceleration. In particular, the following conditions were employed: soft soil
and MS ¼ 5:75 [i.e., approximately the mean value of the range of MW selected in the
European Strong Motion Database converted using the relationship of Yenier et al.
(2008)]. The comparison reported in Fig. 4 shows the good agreement between the mean of
the horizontal direction and the GMPE of Ambraseys et al. (1996) for both sets of
accelerograms. Moreover, it can be observed that the peak ground acceleration is in
agreement with the Italian Building Code [NTC (MIT 2008)] provisions for a seismic
event with return period of 975 years with roughly predicted values considering the soil
effect of about 4 m=s2 .
Similar considerations can be done for the maximum spectral acceleration values and
associated periods. These are not reported for the sake of brevity.
In Fig. 5 the ratio between the peak ground acceleration in the vertical and horizontal
directions is reported, namely the V/H ratio, as a function of the epicentral distance. The mean is

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 271

Fig. 4 Peak ground acceleration, ag , for the three mutual directions as a function of the epicentral distance
found using the Emilia-Romagna earthquake records (left) and the selected records from the European
Strong Motion Database (right). The grey area indicates the peak ground acceleration reported in OPCM
3431 (PCM 2005) for the area hit by the Emilia-Romagna earthquake. Comparison with the ag GMPE of
Ambraseys et al. (1996)

the average of the data with a bin of 10 km and the mean of V/H is computed as the mean of the
two horizontal components. It is visible that generally V/H is not very sensitive to the epicentral
distance with the exception of the Emilia-Romagna earthquake records at distance smaller than
10 km, although only a limited number of earthquake records in this range of epicentral distance
(i.e., \10 km) was found and its statistic reliability can be considered low. Again in order to
validate the data found, these were compared with the GMPE of Bozorgnia and Campbell
(2004) with the following conditions employed: soft soil and MW ¼ 5:5. Also in this case, the
comparison reported in Fig. 5 shows a good agreement with the exceptions reported above. It
is worth noting that the vertical component has a strong influence on the frictional behavior of
the beam-to-column connections as it will be shown in the following.
The comparison reported above validates the considered two sets of data and shows a
good agreement (in a statistical sense) between them. According to these observations, in
the following, the two sets will be joined and used in all the subsequent analyses; this is
done with the aim of obtaining a larger set of data with a better statistical definition. The
histogram of the seismic events as a function of the epicentral distance obtained combining
the Emilia-Romagna earthquake records and the selected records from the European
Strong Motion Database is shown in Fig. 6.
In the following, all the results will be reported using the epicentral distance as primary
variable. The choice of the epicentral distance as primary variable is due to provide
analytical evidence of the damaging effect of vertical earthquake ground motion since
higher values of V/H ratios occurs for near-fault records (see Fig. 5). Moreover, in this
way, this information can be used to evaluate the territorial extension of the damage given
the seismic event.

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
272 Bull Earthquake Eng (2018) 16:259–294

Fig. 5 V/H ratio for the peak ground acceleration, ag , as a function of the epicentral distance found using
the Emilia-Romagna earthquake records (left) and the selected records from the European Strong Motion
Database (right). Comparison with the V / H GMPE of Bozorgnia and Campbell (2004) for ag , magnitude of
5.5 and firm soil

Fig. 6 Histogram of the seismic events as a function of the epicentral distance obtained combining the
Emilia-Romagna earthquake records and the selected records from the European Strong Motion Database

4 2 and 3-DoFs elastic model of the transverse response

In this Section, a portal-like 2 and 3 Degrees-of-Freedom (DoFs) elastic dynamic model


representing the transverse response of one frame of a single-story precast concrete
industrial building is presented (Sect. 4.1). The 3-DoFs dynamic model is used to deter-
mine: (1) the lower and upper bound (considering perfect correlation of the maximum of

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 273

the accelerograms) and (2) the linear solution (considering the real correlation of the peaks
of the accelerograms) of the minimum friction coefficient required to avoid sliding con-
ditions, i.e., differential displacements in the beam-to-column connection (Sect. 4.2). Then,
the effects of the input directionality are evaluated (Sect. 4.3). Finally, the lower and upper
bound of the minimum required friction coefficient are evaluated using the elastic response
spectra given in the Italian Building Code (MIT 2008) for different values of the return
periods and soil conditions (Sect. 4.4).

4.1 Dynamic system and equations of motion

Typically, precast industrial buildings in Southern Europe are realized with connected
elements leading to a statically-determined structure (Sect. 2) whose transverse portal-like
response can be modeled as 3 DoFs oscillators, as represented in Fig. 7. The first DoF
refers to the vertical response of the beam, while the second and third DoFs refer to the
horizontal response of the beam. They are used in order to evaluate the horizontal response
of the frame in two directions, each perpendicular to each other. The two horizontal DoFs
should be considered as ‘‘virtual’’ DoFs since they are used to simultaneously evaluate the
response in two directions (or by a combination of them of all the possible directions)
according to the arbitrary orientation of the frames in real buildings. This corresponds to

Fig. 7 One-story precast industrial building and simplified 3-DoFs elastic mechanical model. The
subscripts v and i ¼ fNS; EW g indicate that the variable refers to the vertical and horizontal (North–South
and East–West) directions, respectively

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
274 Bull Earthquake Eng (2018) 16:259–294

figuring that the frame is first located in one direction and then in the perpendicular one. In
particular, the three DoFs refer to the three acceleration components v, NS, EW (vertical,
North-South and East-West) of the ground motion time histories (Sect. 3).
The model herein proposed considers the following assumptions: (1) the column mass is
neglected; (2) the mass of the beam and of the roof system is considered lumped at the
beam midspan; (3) the column and beam cross-sections are considered constant with
equivalent characteristics to the actual one; (iv) the beam is modeled as a simply-supported
beam considering the beam-to-column connections as hinges; (v) the materials are
described by elastic constitute laws; (vi) linear geometric response; (vii) the axial
deformability of the columns and of the beam is neglected. Moreover, the difference
between static and dynamic friction coefficient is neglected.
Under these assumptions, the Equations of Motion (EoMs) of the 3-DoFs elastic
mechanical model are:
0:5  M x€v ðtÞ þ Cv x_v ðtÞ þ Kv xv ðtÞ ¼ 0:5  Mav ðtÞ  Mg
ð3Þ
xh;i ðtÞ þ Ch;i x_h ðtÞ þ Kh;i xh;i ðtÞ ¼ Mah;i ðtÞ þ Ff ;i
M€ i ¼ fNS; EW g

where t is the time, M is the lumped mass of the beam and of the roof system, C is the
damping coefficient, K is the stiffness, a(t) is the ground-motion time history and Ff ¼
l  Fn is the friction force with l friction coefficient and Fn normal force exerted by each
surface on the other in the beam-to-column connection. The subscripts v and i ¼ fNS; EW g
indicate that the variable refers to the vertical and horizontal (North-South and East-West)
directions, respectively. The mass multiplier 0.5 in the vertical direction transforms the
total mass into the modal mass of a simply supported beam. Here and in the following, the
dynamic characteristics in the two horizontal DoFs will be considered the same, i.e.,
Kh;NS ¼ Kh;EW and Ch;NS ¼ Ch;EW .
The natural periods in the vertical and horizontal directions can be evaluated as:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffi
0:5  M M
Tv ¼ 2p  Th ¼ 2p  ð4Þ
Kv Kh

Given the vertical natural period, Tv , the length of the beam (Fig. 7) can be evaluated
considering a constant-section simply-supported beam:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rffiffiffiffiffiffiffiffiffiffiffiffi
Ec  Ib ð5Þ
L ¼ 0:5  Tv  p 
m
where Ec is the elastic modulus of the concrete (i.e., the cross-section material), Ib is the
beam cross-section area moment of inertia and m is the mass per unit length of the beam
and of the roof system, i.e., m ¼ M=L. The column height can be estimated using the
horizontal natural period, Th , considering a cantilever beam with the beam and roof system
mass located at the free-end (i.e., neglecting the columns mass) with stiffness equivalent to
the two columns:
"   #13
6  Ec  Ic Th 2
H¼  ð6Þ
mL 2p

where Ic is the sum of the area moment of inertia of the two column cross-sections and L is
the length of the beam (Fig. 7).

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 275

4.2 Lower and upper bound and linear solution of the minimum required
friction coefficient

The total acceleration, i.e., x


v ðt; Tv ; nv Þ and x
h;i ðt; Th ; nh Þ, in each direction is obtained by
integrating Eq. 3, mainly depending on the natural periods and structural damping. In
particular, the non-sliding condition at time t in the DoF i, occurs when the capacity of the
connection, Ff ;i , is greater than the demand, which is the absolute value of the force exerted
parallel to the contacting surfaces, Fp;i :
 
Ff ;i ¼ li Fn;i ðtÞ  Fp;i ðtÞ ð7Þ

The absolute value accounts for the symmetric domain of the shear capacity of the
connection, i.e., same shear resistance for positive or negative values of the demand.
Considering a 3-component earthquake (vertical and two horizontal), Fn;i ðtÞ is equal to
M multiplied by the acceleration of gravity, g, (static part or dead load) and half (modal
mass of a simply supported beam) of the total vertical acceleration and Fp;i ðtÞ is equal to
M multiplied the total horizontal acceleration. Eliminating the mass, this leads to:
   

li g þ 0:5  x€
v ðt; Tv ; nv Þ  x€
h;i ðt; Th ; nh Þ ð8Þ

The positive vertical direction is downward, i.e., the same of the gravity vector, while
the horizontal one is arbitrary for the absolute value. For all the duration of the earthquake
(i.e., for all values of t), the minimum required friction coefficient, lmin;i , preventing beam-
column differential displacements from occurring is:
8   9
< x€
h;i ðt; Th ; nh Þ =
lmin;i ðTv ; Th ; nv ; nh Þ ¼ max ð9Þ
t :g þ 0:5  x €
v ðt; Tv ; nv Þ;

This solution accounts for the non-perfect correlation of the maxima of the vertical and
horizontal components. As a matter of fact, in the near-field domain, the peak of the
vertical component occurs slightly before the horizontal one; the time-lag among the two
peaks generally increases with distance (Collier and Elnashai 2001; Shrestha 2009).
Moreover, horizontal and vertical peak ground motion can be coincident when the distance
from the source is less than 5 km (Collier and Elnashai 2001).
Neglecting the correlation between the peaks of the horizontal and vertical ground
motion, Eq. 9 can be rewritten using the absolute maximum acceleration in an earthquake
on a damped, harmonic oscillator moving in one physical dimension (i.e., acceleration
response spectrum):
 
 Sh;i ðTh ; nh Þ
ð10Þ
lmin;i ðTv ; Th ; nv ; nh Þ ¼
ðg 0:5  Sv ðTv ; nv ÞÞ
where SðT; nÞ is the acceleration response spectrum. The plus-minus sign indicates that the
maximum response spectrum acceleration can be either positive or negative, although for
the horizontal component the absolute value vanishes its effect. In Eq. 10, it is considered
that the peaks of the horizontal and vertical actions occur simultaneously. A similar
approach was previously proposed by Magliulo et al. (2014), Belleri et al. (2014).
The plus-minus defines either the worst-case (-, i.e., negative maximum of the spectral
vertical acceleration with maximum of the horizontal one) or the best-case (þ, i.e., positive

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
276 Bull Earthquake Eng (2018) 16:259–294

maximum of the spectral vertical acceleration with maximum of the horizontal one) sce-
narios namely the lower bound of the minimum required friction coefficient preventing
beam-to-column differential displacements. As real earthquakes induce horizontal and
vertical accelerations whose maxima are not perfectly correlated, the solution of Eq. 9 is in
the range of the worst-case (-) to the best-case (þ) scenarios of Eq. 10; this is true for each
earthquake record. The two conditions are the case of addition and subtraction of the
weight with the vertical earthquake force, i.e., positive and negative maximum vertical
acceleration. In particular, the solution of Eq. 9 is in the middle of the worst-case (-) to the
best-case (þ) scenarios for zero vertical loads and as a function of the horizontal and
vertical peak ground motion correlation is nearest to the one or other case.
In order to account for the worst directional scenario, the direction corresponding to the
maximum required friction coefficient is used. This can be evaluated by computing the
magnitude of the vectorial sum of the two components in the two horizontal directions:
rhffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
i2 h i2ffi
xh ðtÞ ¼

x€ ðt; Th ; nh Þ þ x€
ðt; Th ; nh Þ ð11Þ
h;NS h;EW

Accordingly, Eq. 9 can be written as:



jxh ðtÞj
lmin ðTv ; Th ; nv ; nh Þ ¼ max ð12Þ
t g þ 0:5  x€
v ðt; Tv ; nv Þ
and Eq. 10 accounting for the worst directional scenario becomes:
jSh ðTh ; nh Þj
l
min ðTv ; Th ; nv ; nh Þ ¼ ð13Þ
ðg 0:5  Sv ðTv ; nv ÞÞ

In the following, the set of earthquake records presented in Sect. 3 will be used to
evaluate the minimum required friction coefficient. In order to find the lowest bound of the
minimum required friction coefficient (i.e., worst possible scenario), a set of precast
industrial buildings with fundamental periods corresponding to the periods of the maxi-
mum spectral acceleration of each ground-motion time history is selected. Figure 8 depicts
the periods corresponding to the maximum spectral acceleration Tmax;v and Tmax;h in the
vertical and horizontal directions, respectively, of the earthquakes presented in Sect. 3.
L and H of the one-story precast concrete industrial building (Fig. 7) were calculated
using Eqs. 5 and 6 assuming a spacing of the frames of 10 m (rounded value of the spacing
reported in Table 1 by the authors’ survey), weight for unit area of the roof system
qr ¼ 200 kg=m2 , density and elastic modulus of the concrete qc ¼ 2500 kg=m3 and
Ec ¼ 33000 MPa, respectively, column section: Bc ¼ 0:5 m and Hc ¼ 0:5 m, beam section:
Bb ¼ 0:5 m and Hb ¼ 1 m; these are mean parameters of the typical Italian precast concrete
buildings (Sect. 2.3).
The computed values of L and H are shown in Fig. 9. In particular, L is almost entirely
in the range of typical values of the geometric size (gray filled rectangles) while H assumes
slightly smaller values compared to the range of typical values of the geometric size (red
filled rectangles). The computed H does not consider the presence of cladding panels that
can increase the stiffness, thus reducing the horizontal fundamental period (Biondini et al.
2013; Magliulo et al. 2015). Through a benchmark design study among Italy, Greece,
Slovenia and Turkey, Olgiati et al. (2011) found that the horizontal fundamental period of
a typical single-storey industrial precast concrete building ranges between 0.8 and 1.4 s.
Although larger than Tmax;h (Fig. 8), this range corresponds to spectral accelerations that

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 277

Fig. 8 Periods corresponding to the maximum spectral acceleration Tmax;v and Tmax;h in the vertical and
horizontal directions, respectively, of the earthquakes set as a function of the epicentral distance

are still significant. It should be highlighted that the values of L and H reported in Fig. 9
were evaluated considering a full cross-section area moment of inertia; some code such as
Eurocode 8 (CEN 2004) suggests to use a 50% reduced cross-section area moment of
inertia. If the latter assumption is taken, a reduction of L and H equal to 84% is observed
compared with the values reported in Fig. 9.
lþ 
min , lmin (Eq. 13) and lmin (Eq. 12) evaluated using the worst scenario set of precast
concrete industrial buildings (i.e., Fig. 9) are shown in Fig. 10a as a function of the
epicentral distance while those evaluated using a reference case with Tv ¼ 0:18 s and
Th ¼ 1 s [typical values of industrial precast concrete buildings (e.g. Olgiati et al. 2011)]
are shown in Fig. 10b. In both Figures, the mean of lmin (dashed red line) is the average
with a bin of 10 km. The gray and light red shaded areas show the typical values of the
friction coefficients for neoprene-to-concrete and concrete-to-concrete contact surfaces
(Sect. 2.2), respectively. As previously explained, lmin (blue circle sign) is included in the

Fig. 9 Length of the beam, L, and height of the columns, H, giving the fundamental period equal to that of
the maximum spectral acceleration, Tmax , of the earthquakes set as a function of the epicentral distance. The
gray and light red shaded areas show the typical values of L and H, respectively for Italian precast concrete
buildings

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
278 Bull Earthquake Eng (2018) 16:259–294

þ
range of l min (black plus sign) to lmin (black x-mark sign) whose domain is shown in
Fig. 10b with a vertical solid black line. Neglecting the vertical component of the seismic
þ
input, lmin is in the middle between l min and lmin . Accordingly, especially at low epicentral
distances, it is essential to consider the vertical component to correctly predict the mini-
mum required friction coefficient.
Generally, as the seismic input energy decreases with the epicentral distance, the
minimum required friction coefficient reduces with it. Considering Fig. 10a, in only 4
earthquake time histories at an epicentral distance smaller than 20 km negative values of
lmin were found; these are indicated using a red downward-pointing triangle. Negative
values of lmin (perfectly correlated maxima of the vertical and horizontal action) indicate
uplift conditions, i.e., negative values of normal force, Fn ðtÞ, namely traction on the beam-
to-column connection. In these cases, a theoretical infinite value of the friction coefficient
is required. It can be seen that for distances lower than about 30 km, the average friction
coefficient is higher than 0.3, and relative displacement may occur. However, this

(a)

(b)

Fig. 10 lþ 
min , lmin (Eq. 13) and lmin (Eq. 12) and mean value of lmin as a function of the epicentral distance.
The vertical black lines indicate the range between lþ 
min (x-mark sign) and lmin (plus sign). The gray and
light red shaded areas show the typical values of the friction coefficients for neoprene-to-concrete and
concrete-concrete contact surfaces, respectively. a Tv ¼ Tmax;v ; Th;i ¼ Tmax;h;i . b Tv ¼ 0:18 s; Th;i ¼ 1 s;
nv ¼ 1%; nh;i ¼ 3%

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 279

condition should be considered as the possible worst scenario, although as previously


demonstrated it corresponds to a possible condition.
Differently, considering a specific industrial precast concrete building with common
characteristics (Fig. 10b), no uplift conditions were detected and for distances lower than
about 15 km, the friction coefficient is higher than 0.2. In this case, only for neoprene-
concrete contact surfaces, relative displacements in the connection may occur. In 5 cases at
all distances lower than about 15 km, a minimum required friction coefficient compatible
with the concrete-to-concrete contact surfaces was found. These were disregarded in the
computation of the mean of lmin .
In order to check if by changing the main variable (i.e., epicentral distance) a reduction
in the scatter of the results at a lower epicentral distance could be obtained, the maximum
peak ground acceleration is used as intensity measure to plot the same results reported in
Fig. 10b. In particular, lþ 
min , lmin (Eq. 13) and lmin (Eq. 12) evaluated using a reference
case with Tv ¼ 0:18 s and Th ¼ 1 s are shown in Fig. 11 as a function of the maximum
peak ground acceleration, ag , for the two horizontal directions. The x-axis is shown
reversed so to better compare Figs. 10b and 11. It can be seen that by using the peak
ground acceleration as intensity measure, the scatter in the results is not reduced for high
values of ag (corresponding to low epicentral distances, i.e., Fig. 4). Similar considerations
can be done for the worst scenario case (Fig. 10a).

4.3 Ground-motion directionality

In the last few years, research efforts have focused on the effect of different definitions of
horizontal ground motion (e.g., the larger horizontal component, any average between the
two horizontal components) on the median values predicted by GMPEs (Barani et al.
2015). This subject is often referred to as ground-motion directionality. The directivity
effect depends on the direction of the rupture front: if the rupture arises towards the site, it
is called forward directivity and, if it is in the opposite direction, is called backward

Fig. 11 lþ 
min , lmin (Eq. 13) and lmin (Eq. 12) and mean value of lmin as a function of the maximum peak
ground acceleration, ag , for the two horizontal directions. The vertical black lines indicate the range
between lþ 
min (x-mark sign) and lmin (plus sign). The gray and light red shaded areas show the typical values
of the friction coefficients for neoprene-to-concrete and concrete-concrete contact surfaces, respectively.
(Tv ¼ 0:18 s; Th;i ¼ 1 s; nv ¼ 1%; nh;i ¼ 3%)

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
280 Bull Earthquake Eng (2018) 16:259–294

directivity (Grimaz and Malisan 2014). The choice of one horizontal ground-motion
definition or another may result in significant changes in the hazard definition (Barani et al.
2015). As a matter of fact, seismic input has the maximum value along the fault-normal
direction and the minimum for the fault-parallel direction. Generally speaking, the effects
connected to forward directivity can be recognized at distances less than 50 km from the
fault depending on the earthquake magnitude and on the geometry of the site in relation to
the fault (Somerville 2000). The Emilia-Romagna earthquakes were characterized by
ground-motion directionality (Gallipoli et al. 2014) leading to significant effects on the
damage of precast industrial buildings (Liberatore et al. 2013; Casotto et al. 2015).
In order to study the ground-motion directionality effect, first, the North-South and
East-West directions separately are employed to calculate lmin;NS and lmin;EW ; then, these
are compared with seismic input giving in the direction of interest the maximum normal
force in the beam-to-column connection lmin (Eq. 12). The results obtained using the worst
scenario set of precast concrete industrial buildings are shown in Fig. 12a, while those
evaluated for the reference case with Tv ¼ 0:18 s and Th ¼ 1 s are shown in Fig. 12b. In
Fig. 12, dashed lines show the mean values of lmin;NS , lmin;EW and lmin with a bin of
10 km. The vertical gray lines indicate the range between lmin;NS (upward-pointing tri-
angle) and lmin;EW (right-pointing triangle). It should be noted that lmin is the same
reported in Fig. 10.
Looking at epicentral distances larger than 20 km, it can be observed that generally
lmin;NS , lmin;EW and lmin and their mean values approximately coincide. On the other hand,
at epicentral distances smaller than 20 km, strong differences can be found as a function of
the distance indicating ground-motion directivity. In fact, the directivity is associated with
a different behavior in the two directions. Moreover, in some cases, it is observed that lmin
can be larger than lmin;NS , lmin;EW , thus indicating that the dominating direction of the
earthquake is not located in the North-South or East-West directions. In particular, it is
observed that in terms of mean (i.e., dashed lines) the worst case occurs in the North-South
direction (green line) compared with the East-West direction (blue line).

4.4 Code based approach

In order to compare results obtained using real accelerograms (Sect. 4.2) with the code
seismic action, lþ 
min and lmin are computed using the elastic response spectra; this was
evaluated according to the Italian Building Code [NTC (MIT 2008)] for Mirandola for
nv ¼ 1%, nh;i ¼ 3% and for two different soil classes, types A and D. These two soil
classes are representative of the minimum and maximum values of the soil factor given by
the Italian Building Code [NTC (MIT 2008)], see Table 2. lþ 
min and lmin are reported in
Fig. 13, as a function of the return period, Tr . Results are reported as the maximum of lþ min
and l min for any Tv and Th;i and for Tv ¼ 0:18 s and Th;i ¼ 1 s (reference case previously
adopted).
lþ 
min and lmin increase with increasing Tr as the seismic action increases. Moreover, it is
observed that the difference between lþ 
min and lmin increases with increasing Tr as the
vertical component increases. Moreover, the strong influence of the soil class on lþ min and
lmin with soft soil leading to larger minimum required friction coefficients. Looking at the
worst scenario case, l min is compatible with concrete-to-concrete friction coefficients for
Tr  180 years for soft soil (type D) and for Tr  560 years for stiff soil (type A). On the
other hand, for the reference case, l min is compatible with concrete-to-concrete friction

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 281

(a)

(b)

Fig. 12 lmin;NS , lmin;EW and lmin and their mean values as a function of the epicentral distance. The vertical
gray lines indicate the range between lmin;NS (upward-pointing triangle) and lmin;EW (right-pointing
triangle). The gray and light red shaded areas show the typical values of the friction coefficients for
neoprene-to-concrete and concrete-to-concrete contact surfaces, respectively. a Tv ¼ Tmax;v ; Th;i ¼ Tmax;h;i ;
nv ¼ 1%; nh;i ¼ 3%. b Tv ¼ 0:18 s; Th;i ¼ 1 s; nv ¼ 1%; nh;i ¼ 3%

coefficients only for soft soil (type D) for Tr  560 years. However, for the reference case,
lþ 
min and lmin are compatible with the neoprene-to-concrete friction coefficients also for
relatively low values of the return periods for both soil class types.
In conclusion, although the vertical component is underestimated by the code elastic
response spectra (Fig. 3a), the code-based approach gives a good estimate of the mean of
the minimum required friction coefficient at distance lower than 10 km (Fig. 10b) for a
return period of approximately 975 years. However, very near-field records, at distance
lower than 5 km in Fig. 10b are characterized by larger values of the minimum required
friction coefficient compared with the code-based approach for approximately 975 years.
As previously stated (Sect. 3.1), this is due to the inability to consider near-field events by
the code elastic response spectra.

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
282 Bull Earthquake Eng (2018) 16:259–294

Fig. 13 lþ 
min (in blue) and lmin (in red) evaluated according to Eq. 13 evaluated using the elastic response
spectra in according to Italian Building Code [NTC (MIT 2008)] as a function of the return period, Tr , for
Mirandola. The gray and light red shaded areas show the typical values of the friction coefficients for
neoprene-to-concrete and concrete-to-concrete contact surfaces, respectively (Solid thick line maximum
values for any Th and Tv ; nv ¼ 1%; nh;i ¼ 3%; soil class type A. Solid thin line maximum values for any Th
and Tv ; nv ¼ 1%; nh;i ¼ 3%; soil class type D. Dashed thick line Tv ¼ 0:18 s; Th;i ¼ 1 s; nv ¼ 1%; nh;i ¼ 3%;
soil class type A. Dashed thin line Tv ¼ 0:18 s; Th;i ¼ 1 s; nv ¼ 1%; nh;i ¼ 3%; soil class type A)

5 FEM analyses

In this Section, a FEM portal-like model representing the transverse response of one frame
of a single story precast concrete industrial building is presented (Sect. 5.1). Then, the
geometrical and mechanical characteristics of the precast concrete frame analyzed are
described (Sect. 5.2); these represent a typical Italian industrial precast building (Table 1).
Finally, Sect. 5.3 presents results of the analyses performed considering two longitudinal
reinforcement ratios explaining when the simplified elastic model fails to predict the
minimum required friction coefficients.

5.1 FEM model of the transverse response

A 2D FEM model of one frame of the precast industrial building was built in OpenSees
(Version 2.4.6) framework (McKenna et al. 2013) invoking each analysis using MatlabÓ .
The proposed model is shown in Fig. 14. The structure is modeled as one frame composed
of precast concrete elements made with slender cantilevering columns and simply sup-
ported beam (Sect. 2). The model is made of 7 nodes: 2 fully constrained nodes placed at
the bases of the columns, 4 overlapped nodes located in the connection between the beam
and the columns and 1 node in the middle of the beam; the latter is required in order to
guarantee the vertical dynamic behavior of the beam.
The beam and the columns were modeled using Force-Based Beam-Column Element
and only the column elements account for the P-D effect. The beam section was defined
using an Elastic Section and the columns sections using the Fiber Section. The column
cross-section is subdivided into 80 concrete layers: 1 in the neutral axis direction and 80 in
the perpendicular direction. The reinforcement bars are modeled as single fibers. The
concrete was modeled using Concrete02 Material—Linear Tension Softening and the steel
using Steel01 Material (McKenna et al. 2013).

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 283

Fig. 14 2D model of a one-story single-bay precast concrete industrial building built in OpenSees
framework (McKenna et al. 2013)

The connection between beam and columns is assigned to frictional type connectors
modeled using Flat Slider Bearing Element with the Coulomb Friction model characterized
by a friction coefficient equal to l taken equally for the left and right sides. The hinge
behavior was obtained assuming an elastic material with zero stiffness in the moment
direction in the Flat Slider Bearing Element. The effect of the dowel in the connection is
disregarded as a result of its small resistance; this is in agreement with the field obser-
vations of collapses of precast concrete industrial buildings not designed for seismic
actions during the Emilia-Romagna earthquakes (ReLUIS et al. 2012). The Coulomb
friction model of the connection is shown in Fig. 15. According to the Coulomb’s Law of
Friction, l is independent of the sliding velocity and the friction force is proportional to the
normal force exerted by each surface on the other in the beam-to-column connection. In
this way, the difference between the static and dynamic friction coefficient is disregarded
according to the findings of Foerster et al. (1989).
The structural damping is expressed in terms of Rayleigh damping ratio (Rayleigh
1877):
a0 a1 xi
ni ¼ þ ð14Þ
2xi 2
where i indicated the i-th mode and xi is are the angular natural frequencies, which
corresponded to the initial tangent stiffness evaluated after the application of the vertical

Fig. 15 Coulomb friction model


of the connection: Ff ðtÞ is the
friction force, Fn ðtÞ is the normal
force exerted by each surface on
the other in the beam-to-column
connection and lis the friction
coefficient

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
284 Bull Earthquake Eng (2018) 16:259–294

static loads. The coefficients a0 and a1 were determined from specified damping ratios nh
and nv of the horizontal and vertical modes (i.e., first and second modes), respectively.

5.2 Case study: typical Italian industrial precast building

The beam length is L ¼ 10 m and the column height is H ¼ 6:2 m. L was chosen smaller
with respect to the values reported in Table 1 in order to obtain a building with smaller Tv
(Eq. 5) better falling in the range of maximum vertical spectral accelerations (Sect. 3). The
mass of the roof system is computed considering the specific weight of the beam equal to
2500 kg=m3 and the weight per unit area of the roof system equal to 200 kg=m2 with a
spacing of the frames equal to 9.4 m. The columns have a square section of 0:5  0:5 m
and the beam has a rectangular section of 1  0:5 m.
The properties of concrete and steel are selected following the prescriptions of the
Italian building code (MIT 2008) on the basis of design values. The concrete adopted in the
columns is of strength class C45 / 55 and is characterized by compressive strength fc equal
to 45 MPa, the strain corresponding to peak stress is ey ¼ 0:002, the ultimate strain eu ¼
0:035 and tensile strength fct ¼ 3:8 MPa. The concrete adopted in the beam is described by
an elastic material with Ec ¼ 3:3  105 MPa, i.e., concrete elastic modulus for a strength
class C45 / 55. The choice of high class of concrete is due to the good quality of the
materials usually employed in precast elements. Longitudinal reinforcing steel bars are of
type B450C and are characterized by yield stress fy ¼ 450 MPa and elastic stiffness
Es ¼ 2:1  106 MPa.
The amount of reinforcement

steel is expressed in terms of longitudinal reinforcement
ratio, i.e., qp ¼ As = Bp  Hp with As reinforcement steel area and Bp and Hp base and
height of the column cross-section. The reinforcement bars were symmetrically located on
two layers with a cover of 20 mm. Two values of the longitudinal reinforcement ratio were
analyzed, i.e., qp ¼ 0:3% (4 U16) and qp ¼ 1:5% (12 U20). The first value corresponds to a
low amount of longitudinal reinforcement steel in the columns [value found with field
inspections by Liberatore et al. (2013)] while the second to a large amount approximately
corresponding to the code maximum allowable value (Sect. 2). In both cases, longitudinal
reinforcement bars were placed in two rows symmetric with respect to the cross-section
geometric center.
The obtained natural periods of the structure are Tv ¼ 0:1 s for both values of qp and
Th ¼ 0:51 s and 0:46 s for qp ¼ 0:3% and 1:5%, respectively. The damping ratios of the
horizontal and vertical modes were assumed equal to nh ¼ 3% and nv ¼ 1%, as previously
done in the application of the simplified 3-DoFs model (Sect. 4). The friction coefficient is
varied in the range of 0.01–0.5 with a step of 0.05.
A comparison of the vertical and horizontal fundamental periods, Tv and Th between the
OpenSees model and the simplified 3-DoFs model (Sect. 4) was performed in order to
validate the model. In the OpenSees model, the fundamental periods were evaluated using
the initial tangent stiffness evaluated after the application of the vertical static loads, as
reported in Eq. 14. The prediction of the vertical fundamental period perfectly corresponds
between the OpenSees model and the simplified 3-DoFs model because in both cases an
elastic simply supported beam was adopted; accordingly, Tv predicted using the OpenSees
model does not depend on qp . On the other hand, some differences were found in Th . In this
case, the amount of longitudinal reinforcement steel increases the column stiffness thus
reducing Th . Accordingly, Th predicted using the OpenSees model is lower than the one

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 285

predicted using simplified 3-DoFs model and a maximum difference of approximately 15%
was found for the largest value of qp analyzed.

5.3 Results

The analyses were performed using as seismic input the set of earthquake records (Emilia-
Romagna earthquake records and European Strong Motion Database selection) presented
in Sect. 3. The strong motion time histories are applied at the two fully restrained nodes
consisting of two components: one horizontal (alternately North-South and East-West
directions) and one vertical. Non-synchronous earthquake effects are not accounted. Of the
two analyses performed alternately using the North-South and East-West horizontal
components, that one giving the maximum response is taken. The top-column displace-
ments and beam-to-column connection relative displacements were evaluated during the
analyses and only their absolute maximum values with respect to both left and right sides
were used in the following.
In Figs. 16 and 17 the top-column displacement and the beam-to-column connection
relative displacement are shown for qp ¼ 1:5% and qp ¼ 0:3%, respectively. Each Fig-
ure shows the epicentral distance on the x-axis and the assumed friction coefficient on the
y-axis. The cell color indicates the displacement computed from the analyses. The latter
variable is the mean of the absolute maxima across earthquake simulations. For example,
the yellow cell in the relative displacement Figure, at l 2 ½0:05  0:1 and epicentral
distance [0–10 km] indicates that the mean of the absolute values of the relative dis-
placement for these analyses is equal to approximately 50 mm. The red dashed line
indicates the Mean lmin predicted using the simplified 3-DoFs model (Sect. 4). The small
differences in Mean lmin between the two cases depend on the different natural periods
depending on qp .

Fig. 16 Contour plot of the mean of the absolute maximum values of the top-column displacement and of
the beam-to-column connection relative displacement in mm for qp ¼ 1:5% as a function of the epicentral
distance and of the friction coefficient. The red dashed line indicates Mean lmin (same meaning of Fig. 10).
(nv ¼ 1%; nh;i ¼ 3%)

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
286 Bull Earthquake Eng (2018) 16:259–294

Fig. 17 Contour plot of the mean of the absolute maximum values of the top-column displacement and of
the beam-to-column connection relative displacement in mm for qp ¼ 0:3% as a function of the epicentral
distance and of the friction coefficient. The red dashed line indicates Mean lmin (same meaning of Fig. 10).
(nv ¼ 1%; nh;i ¼ 3%)

From the Figures, it is visible that increasing the epicentral distance the top-column
displacement and the beam-to-column connection relative displacement reduces when the
intensity of the seismic input decreases. On the other hand, an increase of the friction
coefficient generally leads to an increase of the top-column displacement and a reduction
of the beam-to-column connection relative displacement. This occurs until some value at
which no relative displacement in the connection occurs and the hinge-connection con-
ditions are reached. Exceeded this limit, any increase of the friction coefficients does not
have influences on the structural response.
Looking at the qp ¼ 1:5% case (Fig. 16), it is visible that the two models (FEM and
simplified 3-DoFs) are in good agreement since the boundaries of the region where relative
displacements can occur are well described by the 3-DoFs simplified model. At approxi-
mately l ¼ 0:4, the relative displacement in the connection goes to zero also at low
epicentral distances and the connection behaves like a perfect hinge. An increase of the
top-column displacement for epicentral distance smaller than 10 km was observed; this is
related to the large values of the beam-to-column connection relative displacement leading
to enhanced P-D effect.
For qp ¼ 0:3%, the top-column displacement values are larger and the beam-to-column
connection relative displacement values smaller. This is due to the lower stiffness and
strength of the columns and to the activation of the plastic behavior. In this case, the two
models (i.e., FEM and simplified 3-DoFs) are not in good agreement. The explanation is
given by the different mechanical behavior accounted by the two models. The 3-DoFs
simplified model accounts for only the elastic behavior of the columns, while the FEM
model accounts for all the non-linear geometric and mechanic behavior. In fact, the
activation of the non-linear behavior in the columns makes the elastic model not capable to
correctly predict the structural response. This also justifies the good agreement for

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 287

qp ¼ 1:5%, as in this case due to larger values of the column strength less evident non-
linear behavior occurs.
In both cases (qp ¼ 1:5% in Fig. 16 and qp ¼ 0:3% in Fig. 17), it can be observed that
simplified models (linear 3-DoFs) are a good proxy to more refined ones (non-linear FEM).
In particular, for both values of the longitudinal reinforcement ratio analyzed, results
coming from the simplified models are on the safe side with safer predictions for qp ¼
0:3% where non-linear effects of the columns dominate. The latter observations allow for
the validation of the simplified 3-DoFs and its use to estimate the minimum required
friction coefficient.
Finally, in order to explain the differences between the two cases analyzed (i.e., qp ¼
0:3% and qp ¼ 1:5%), a static pushover analysis was performed to determine the elastic
capacity and the non-linear response of the portal-like structure. The Flat Slider Bearing
Element was replaced with a rigid element, leaving the hinge, neglecting in this way the
frictional behavior; this is equivalent to assuming an infinite friction coefficient in the flat
slider bearing elements. After vertical dead loads application, one-directional monotonic
horizontal displacement-controlled static loading, Fh , was applied on the beam in the range
of 0–200 mm (drift ratio from 0 to 0.32%). Results are shown in Fig. 18, expressed as the
minimum required friction coefficient in the beam-to-column connections to avoid sliding:
Fh
l ¼ ð15Þ
Mg

The latter coefficient assumes the same meaning as those reported in Eq. 9, although the
seismic input vertical component is disregarded. It should be highlighted that the latter is
an approximation, since, as previously demonstrated, the loads induced by the earthquake
vertical component has a strong influence. The columns elastic limit is approximately 60
and 75 mm for qp ¼ 0:3% and qp ¼ 1:5%, respectively. The maximum minimum required
friction coefficient is reached for l ¼ 0:11 and l ¼ 0:45 for qp ¼ 0:3% and qp ¼ 1:5%,
respectively. These results are in agreement with the l-value at which the beam-to-column
connection relative displacement goes to zero and hinge-connection conditions are reached
and the differences are related to the vertical component effect. As a matter of fact, being
the system a series system (i.e., column and connection) the system strength is equal to that

Fig. 18 Push over curve in terms of l (Eq. 15) for qp ¼ 0:3% and qp ¼ 1:5%

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
288 Bull Earthquake Eng (2018) 16:259–294

of the element having the smallest one. Accordingly, if the column strength is smaller than
the friction capacity, no beam-to-column connection relative displacement occurs and
damage in the columns occurs. Differently, if the column strength is larger than the friction
capacity, beam-to-column connection relative displacement occurs and no damage in the
columns occurs. However, it should be highlighted that for large displacements non-linear
geometric effects can affect the described behavior. An increase of the column strength
leads to an increase of the required friction coefficient demand and an increase of the
friction coefficient can lead to an increase of force and/or ductility demand in the columns.
A correct design strategy of the retrofit should consider both aspects.
Finally, in order to better appreciate the distribution of the absolute maximum dis-
placements, it was computed the Empirical Cumulative Distribution Function (ECDF) of
the top-column displacement and of the beam-to-column connection relative displacement
for qp ¼ 1:5 and 0.3%, for epicentral distances in the range of 0–20 km and for different
values of the friction coefficient in the range of 0.15–0.3 (Fig. 19). It can be seen that for
qp ¼ 1:5% the variation of l leads to large strong variations of the ECDF, while for
qp ¼ 0:3% the effect is limited as the damage in the columns reduces the forces on the
beam-to-column connections. Considering the qp ¼ 1:5% case, it is observed that lower
values of the friction coefficients lead to large values of the displacements except for the
largest one, where the non-linear effects dominate. Moreover, it can be observed that
approximately from 50 to 70% of the cases, relative displacements in the beam-to-column
connection can occur.
As a final note, it is important to highlight that the relative displacement predicted using
the FEM model are only a simplified approximation of the real behavior, as the Coulomb
Friction model can strongly differ from the real friction behavior, especially under cyclic
conditions. However, these results can be used for the rough estimation of the beam-to-
column connection relative displacement demand in order to check if loss-of-support
conditions can occur (Fig. 19).

6 Conclusions

Emilia-Romagna 2012 seismic events highlighted the large economic and environmental
risks associated with industrial plants. In particular, precast concrete industrial buildings
whose beam-to-column connection are of frictional type showed all their structural defi-
ciencies under seismic action. Large sliding and loss of support conditions were observed.
The occurrence of this phenomena was mainly due to a combination of different factors
such as near fault effects characterized by high vertical peak accelerations, lower friction
coefficient, natural frequencies of the structures falling in the range of high spectral
accelerations and lower structural damping.
This condition of precast industrial buildings is common in many countries, especially
in Southern Europe, where frictional beam-to-column connections have been largely
employed until a few years ago. Accordingly, Emilia-Romagna 2012 seismic events,
characterized by relatively low seismic intensity, i.e., ML around 5.8 in the two main
events, can be considered as premonitory events of diffused risk conditions in Southern
Europe. As a matter of fact, there exist a large number of precast industrial buildings with
beam-to-column connection vulnerability (i.e., with the shear transfer mechanism mainly
based on friction) and located in sites with seismic hazard with characteristics similar or
worse than Emilia-Romagna 2012 seismic events. Fortunately, nowadays frictional beam-

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 289

Fig. 19 ECDF of the absolute maximum values of the top-column displacement and of the beam-to-column
connection relative displacement for qp ¼ 1:5 and 0.3%, for epicentral distances in the range of 0–20 km
and for four values of the friction coefficient: 0.15, 0.2, 0.25 and 0.3. Note that when only one line is shown,
this indicates that the four lines are overlapped

to-column connections are not used in seismic countries since modern Codes have for-
bidden their use.
In this paper, the evaluation of the loss-of-support conditions in frictional beam-to-
column connections of industrial precast concrete buildings was presented, employing
different numerical models of increasing complexity. A summary of the characteristics of
the typical precast industrial buildings pointed out the variables defining the dynamic
response of these buildings. Then, the analysis of the Emilia-Romagna strong-motion time
histories highlighted the seismic input characteristics leading to so large a damage. In
particular, the vertical component played a fundamental role, especially at low epicentral
distances. The comparison with GMPEs available in literature and with a selection from
the European Strong Motion Database showed that the intensity of the latter was com-
parable to previously recorded earthquakes having similar characteristics. The two sets of
seismic input (Emilia-Romagna strong-motion time histories and selection from the
European Strong Motion Database) were joined and used in the numerical analyses.
Two dynamic models representing a single frame of the industrial buildings were used.
First, a 3-DoFs elastic model was employed where the time correlation between the
response peaks along the horizontal and vertical directions was either disregarded or
accounted for. Moreover, this model was used to evaluate directionality effects and
employing the elastic spectra defined in the Italian Building Code. Analyses were per-
formed using the worst scenario set of precast concrete industrial buildings (i.e., charac-
terized by fundamental periods coincident with those corresponding to the maximum
spectral acceleration) and using a reference case with Tv ¼ 0:18 s and Th ¼ 1 s [typical
values of industrial precast concrete buildings (e.g. Olgiati et al. 2011)]. In both cases, it is

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
290 Bull Earthquake Eng (2018) 16:259–294

shown that for epicentral distances lower than 30 km, relative displacements in the con-
nection can occur for neoprene-to-concrete surfaces while for the first case (worst sce-
nario), for epicentral distances lower than 15 km, these can occur also for concrete-to-
concrete surfaces. The ground-motion directional analyses showed that at epicentral dis-
tances lower than 20 km strong directional effects occur. Moreover, the application of the
elastic spectra defined in the Italian Building code showed a good agreement with the
prediction of the mean of the minimum required friction coefficient to avoid sliding.
However, near-field conditions are not well predicted: this is due to the differences in the
elastic spectra of the vertical component. Then, a 2D non-linear planar frame developed
within the OpenSees framework accounting for both geometric and material non-linearities
was presented. The analyses showed that the elastic model well reproduces the sliding
condition when the column longitudinal reinforcement ratio is high and it fails when it is
low. A simple push-over analysis was adopted to explain that the system is a parallel one in
which the column and the beam-to-column connection are a series system with resulting
strength equal to the minimum one; in such system, an increase of the column strength
leads to an increase of the minimum required friction coefficient and an increase of the
friction coefficient leads to an increase of the force and/or ductility demand in the columns.
A correct retrofitting design strategy should consider both aspects.
However, some limitations must be highlighted. The minimum required friction coef-
ficient was mainly evaluated in terms of its mean, disregarding the evaluation of the
standard deviation. Moreover, attention was paid to the evaluation of the minimum
required coefficient in order to avoid sliding conditions, estimating in a simplified way
(i.e., employing the Coulomb Friction model) the maximum relative displacement in the
beam-to-column connections. This evaluation requires the definition of cyclic constitutive
law defining the connection response. Future work should, therefore, include the accurate
evaluation of the entire distribution of the minimum required friction coefficient and the
definition of a more accurate cyclic constitutive law for this type of connection, validated
by experimental results. Finally, an interesting application of the simplified models pre-
sented in this study can be the evaluation of the loss-of-support risk map derived at a
regional scale for the Southern Europe area, combining the seismic input information with
spatial and geometrical information. Further work, currently in progress, is required to
investigate these aspects.

Acknowledgements ReLUIS 2014–2018 project, research line 2.1, is acknowledged for the financial
support given to the present research.

References
Ambraseys NN, Simpson KA, Bommer JJ (1996) Prediction of horizontal response spectra in Europe.
Earthq Eng Struct Dyn 25(4):371–400
Ambraseys N, Smit P, Douglas J, Margaris B, Sigbjörnsson R, Olafsson S, Suhadolc P, Costa G (2004)
Internet site for European strong-motion data. Bollettino di Geofisica Teorica ed Applicata
45(3):113–129
Andersson S, Söderberg A, Björklund S (2007) Friction models for sliding dry, boundary and mixed
lubricated contacts. Tribol Int 40(4):580–587
Babič A, Dolšek M (2016) Seismic fragility functions of industrial precast building classes. Eng Struct
118:357–370
Bachmann H, Steinle A, Thrift P (2011) Precast concrete structures. Ernst & Sohn, Berlin
Barani S, Albarello D, Spallarossa D, Massa M (2015) On the influence of horizontal ground-shaking
definition on probabilistic seismic-hazard analysis. Bull Seismol Soc Am 105(5):2704–2712

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 291

Belleri A, Brunesi E, Nascimbene R, Pagani M, Riva P (2014) Seismic performance of precast industrial
facilities following major earthquakes in the Italian territory. J Perform Constr Facil 29(5):04014135-
1–04014135-10
Belleri A, Torquati M, Riva P, Nascimbene R (2015) Vulnerability assessment and retrofit solutions of
precast industrial structures. Earthq Struct 8(3):801–820
Belletti B, Gasperi A, Spagnoli A, Valentino R (2016) Role of soil-structure interaction on the response of
precast RC structures under seismic loading: case study. In: Practice periodical on structural design and
construction, 04016014
Bellotti D, Casotto C, Crowley H, Deyanova MG, Germagnoli F, Fianchisti G, Lucarelli E, Riva S,
Nascimbene R (2014) Capannoni monopiano prefabbricati: distribuzione proba-bilistica dei sistemi e
sottosistemi strutturali dagli anni sessanta ad oggi. Progettazione Sismica 5(3):41–70
Benfield A (2013) Annual global climate and catastrophe report: 2012. Accessed from, p 411
Berger EJ (2002) Friction modeling for dynamic system simulation. Appl Mech Rev 55(6):535–577
Biondini F, Dal Lago B, Toniolo G (2013) Role of wall panel connections on the seismic performance of
precast structures. Bull Earthq Eng 11(4):1061–1081
Birkeland PW, Birkeland HW (1966) Connections in precast concrete construction. J Proc 63:345–368
Bonfanti C, Carabellese A, Toniolo G (2008) Strutture prefabbricate: catalogo delle tipologie esistenti
Bournas DA, Negro P, Taucer FF (2014) Performance of industrial buildings during the Emilia earthquakes
in Northern Italy and recommendations for their strengthening. Bull Earthq Eng 12(5):2383–2404
Bozorgnia Y, Campbell KW (2004) The vertical-to-horizontal response spectral ratio and tentative proce-
dures for developing simplified V/H and vertical design spectra. J Earthq Eng 8(02):175–207
Braga F, Gigliotti R, Monti G, Morelli F, Nuti C, Salvatore W, Vanzi I (2014) Speedup of post earthquake
community recovery: the case of precast industrial buildings after the Emilia 2012 earthquake. Bull
Earthq Eng 12(5):2405–2418
Broderick BM, Elnashai AS, Ambraseys NN, Barr J, Goodfellow R, Higazy M (1994) The Northridge
(California) earthquake of 17 October 1994; observations, strong-motion and correlative response
analyses. In: Engineering seismology and earthquake engineering report no. ESEE vol 4, p 94
Brunesi E, Nascimbene R, Bolognini D, Bellotti D (2015) Experimental investigation of the cyclic response
of reinforced precast concrete framed structures. PCI J 60(2):57–79
Canha RMF, Ebeling EB, de Cresce El ALH, El Debs MK (2009) Analysing the base of precast column in
socket foundations with smooth interfaces. Mater Struct 42(6):725–737
Casotto C, Silva V, Crowley H, Nascimbene R, Pinho R (2015) Seismic fragility of Italian RC precast
industrial structures. Eng Struct 94:122–136
CEN (2004) Eurocode 8: Design of Structures for earthquake resistance. European Committee for Stan-
dardization, Brussels, p 1
Chino M, Nakayama H, Nagai H, Terada H, Katata G, Yamazawa H (2011) Preliminary estimation of
release amounts of 131I and 137Cs accidentally discharged from the Fukushima Daiichi nuclear power
plant into the atmosphere. J Nucl Sci Technol 48(7):1129–1134
Clementi F, Scalbi A, Lenci S (2016) Seismic performance of precast reinforced concrete buildings with
dowel pin connections. J Build Eng 7:224–238
CNR (1999) Apparecchi di appoggio per le costruzioni. Istruzioni per l’impiego (CNR 10018). Consiglio
Nazionale delle Ricerche
Collier CJ, Elnashai AS (2001) A procedure for combining vertical and horizontal seismic action effects.
J Earthq Eng 5(04):521–539
Dassori E, Assobeton (2001) La prefabbricazione in calcestruzzo: guida all’utilizzo nella progettazione. BE-
MA
Demartino C, Monti G, Vanzi I (2017) Seismic loss-of-support conditions of frictional beam-to-column
connections. Str Eng Mech 61(4):527–538
Dolce M, Di Bucci D (2015) Comparing recent Italian earthquakes. Bull Earthq Eng 15(2):497–533
DPC (2016) Ran - Rete Accelerometrica Nazionale
Elliott K (2002) Precast concrete structures. Elsevier, New York
Elliott KS, Jolly C (2013) Multi-storey precast concrete framed structures. Wiley, New York
Fajfar P, Banovec J, Saje F (1978) Behaviour of prefabricated industrial building in Breginj during the Friuli
earthquake. In: 6th European conference on earthquake engineering, vol 2, pp 493–500
Foerster HR, Rizkalla SH, Heuvel JS (1989) Behavior and design of shear connections for loadbearing wall
panels. PCI J 34(1):102–119
Franchetti P, Modena C, Feng MQ (2009) Nonlinear damping identification in precast prestressed reinforced
concrete beams. Comput Aided Civ Infrastruct Eng 24(8):577–592

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
292 Bull Earthquake Eng (2018) 16:259–294

Gallipoli MR, Chiauzzi L, Stabile TA, Mucciarelli M, Masi A, Lizza C, Vignola L (2014) The role of site
effects in the comparison between code provisions and the near field strong motion of the Emilia 2012
earthquakes. Bull Earthq Eng 12(5):2211–2230
Ganas A, Roumelioti Z, Chousianitis K (2012) Static stress transfer from the May 20, 2012, M 6.1 Emilia-
Romagna (northern Italy) earthquake using a co-seismic slip distribution model. Ann Geophys
55(4):655–662
Grimaz S, Malisan P (2014) Near field domain effects and their consideration in the international and Italian
seismic codes. Bollettino di Geofisica Teorica ed Applicata 55(4):717–738
Gülerce Z, Abrahamson NA (2010) Vector-valued probabilistic seismic hazard assessment for the effects of
vertical ground motions on the seismic response of highway bridges. Earthq Spectra 26(4):999–1016
Kunnath SK, Erduran E, Chai YH, Yashinsky M (2008) Effect of near-fault vertical ground motions on
seismic response of highway overcrossings. J Bridge Eng 13(3):282–290
Liberatore L, Sorrentino L, Liberatore D, Decanini LD (2013) Failure of industrial structures induced by the
Emilia (Italy) 2012 earthquakes. Eng Fail Anal 34:629–647
Luzi L, Pacor F, Ameri G, Puglia R, Burrato P, Massa M, Augliera P, Franceschina G, Lovati S, Castro R
(2013) Overview on the strong-motion data recorded during the May–June 2012 Emilia seismic
sequence. Seismol Res Lett 84(4):629–644
Magliulo G, Fabbrocino G, Manfredi G (2008) Seismic assessment of existing precast industrial buildings
using static and dynamic nonlinear analyses. Eng Struct 30(9):2580–2588
Magliulo G, Capozzi V, Fabbrocino G, Manfredi G (2011) Neoprene-concrete friction relationships for
seismic assessment of existing precast buildings. Eng Struct 33(2):532–538
Magliulo G, Ercolino M, Petrone C, Coppola O, Manfredi G (2014) The Emilia earthquake: seismic
performance of precast reinforced concrete buildings. Earthq Spectra 30(2):891–912
Magliulo G, Ercolino M, Manfredi G (2015) Influence of cladding panels on the first period of one-story
precast buildings. Bull Earthq Eng 13(5):1531–1555
Martinelli P, Mulas MG (2010) An innovative passive control technique for industrial precast frames. Eng
Struct 32(4):1123–1132
Marzo A, Marghella G, Indirli M (2012) The Emilia-Romagna earthquake: damages to precast/prestressed
reinforced concrete factories. Ing Sismica 2:132–147
Mata P, Barbat AH, Oller S, Boroschek R (2006) Evaluation of the seismic behavior of precast concrete
buildings with energy dissipating devices. In: First European conference on earthquake engineering
and seismology
McKenna F, Fenves GL, Scott MH, Jeremic B (2013) Open system for earthquake engineering simulation
(OpenSees) [program], Version 2.4.6, Pacific Earthquake Engineering Research Center. University of
California at Berkeley, United States
Mezzi M, Comodini F, Rossi L (2014) Precast industrial buildings in Italy-current building code and new
provisions since the 2012 earthquake. In: Seismic design of industrial facilities. Springer, pp 75–85
Michelini A, Faenza L, Lauciani V, Malagnini L (2008) ShakeMap implementation in Italy. Seismol Res
Lett 79(5):688–697
MIT (1987) Ministerial Decree: ‘‘Norme tecniche per la progettazione, esecuzione e collaudo delle cost-
ruzioni prefabbricate’’ (in Italian)
MIT (2008) Ministerial Decree: ’’NTC 2008 - Norme tecniche per le costruzioni’’ (in Italian)
Mohamad ME, Ibrahim IS, Abdullah R, Rahman AB Abd, Kueh ABH, Usman J (2015) Friction and
cohesion coefficients of composite concrete-to-concrete bond. Cem Concr Compos 56:1–14
Mollaioli F, Bruno S, Decanini LD, Panza GF (2006) Characterization of the dynamic response of structures
to damaging pulse-type near-fault ground motions. Meccanica 41(1):23–46
Mollesten NBI, Packalen JM (1966) Experimentell undersoknig av frictionen vid upplag for fortillverkade
balkar. Royal Institute of Technology, Department of Bridge Engineering, Master’s thesis 129,
Stockholm
Muralikrishnan B, Raja J (2008) Computational surface and roundness metrology. Springer, Berlin
Negro P, Bournas DA, Molina FJ (2013) Pseudodynamic tests on a full-scale 3-storey precast concrete
building: global response. Eng Struct 57:594–608
Negro P, Lamperti Tornaghi M (2017) Seismic response of precast structures with vertical cladding panels:
the SAFECLADDING experimental campaign. Eng Struct 132:205–228
Olgiati M, Negro P, Bournas D (2011) Literature survey and identification of needs–part two: general survey
and design procedures. In: Contribution of the Joint Research Centre. SAFECAST-Deliverable vol 1,
p 218417-2
Osanai Y, Watanabe F, Okamoto S (1996) Stress transfer mechanism of socket base connections with
precast concrete columns. ACI Struct J 93(3):266–276

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Bull Earthquake Eng (2018) 16:259–294 293

Papanicolaou CG, Triantafillou TC (2002) Shear transfer capacity along pumice aggregate concrete and
high-performance concrete interfaces. Mater Struct 35(4):237–245
Papazoglou AJ, Elnashai AS (1996) Analytical and field evidence of the damaging effect of vertical
earthquake ground motion. Earthq Eng Struct Dyn 25(10):1109–1138
Parisi F, De Luca F, Petruzzelli F, De Risi R, Chioccarelli E, Iervolino I (2012) Field inspection after the
May 20th and 29th 2012 Emilia-Romagna earthquakes. Rep., Italian Network of Earthquake Engi-
neering University Laboratories
Paulay T, Loeber PJ (1974) Shear transfer by aggregate interlock. Spec Publ 42:1–16
PCI (1985) PCI Design Handbook: Precast and Prestressed Concrete. PCI Industry Handbook Committee
and others
PCM (2003) OPCM-3274: Primi elementi in materia di criteri generali per la classificazione sismica del
territorio nazionale e di normative tecniche per le costruzioni in zona sismica (in Italian). Ordinanza
del Presidente del Consiglio dei Ministri 3274, 30 March 2003
PCM (2005) OPCM-3431: Ulteriori modifiche ed integrazioni all’ordinanza del Presidente del Consiglio dei
Ministri n. 3274 del 20 marzo 2003, recante ’’Primi elementi in materia di criteri generali per la
classificazione sismica del territorio nazionale e di normative tecniche per le costruzioni in zona
sismica (in Italian). Ordinanza del Presidente del Consiglio dei Ministri 3431, 3 May 2005
PCM (2006) OPCM-3519: Criteri generali per l’individuazione delle zone sismiche e per la formazione e
l’aggiornamento degli elenchi delle stesse zone (in Italian). Ordinanza del Presidente del Consiglio dei
Ministri 3519, 11 May 2006
Plumier A, Calvi GM (2007) Guidelines for seismic vulnerability reduction in the urban environment. IUSS
Press, Pavia
Posada M, Wood S (2002)Seismic performance of precast industrial buildings in Turkey. In: 7th US national
conference on earthquake engineering (7NCEE)
Psycharis IN, Mouzakis HP (2012a) Assessment of the seismic design of precast frames with pinned
connections from shaking table tests. Bull Earthq Eng 10(6):1795–1817
Psycharis IN, Mouzakis HP (2012b) Shear resistance of pinned connections of precast members to
monotonic and cyclic loading. Eng Struct 41:413–427
Randl N (1997) Investigations on transfer of forces between old and new concrete at different joint
roughness. Ph.D. thesis, University of Innsbruck, Austria
Rayleigh L (1877) Theory of sound (two volumes). Dover Publications, New York
ReLUIS, ASSOBETON, CNI, DPC (2012) Linee di indirizzo per interventi locali e globali su edifici
industriali monopiano non progettati con criteri antisismici. www.reluis.it
Santos PMD, Júlio EBS (2010) Recommended improvements to current shear-friction provisions of model
code. In: Proceedings of the 3rd fib international congress
Santos PMD, Júlio ENBS (2012) A state-of-the-art review on shear-friction. Eng Struct 45:435–448
Savoia M, Mazzotti C, Buratti N, Ferracuti B, Bovo M, Ligabue V, Vincenzi L (2012) Damages and
collapses in industrial precast buildings after the Emilia earthquake. Ing Sismica 29(2–3):120–131
Schrage I (1981) Anchoring of bearings by friction. In: Joint sealing and bearing systems for concrete
structures, world congress on joints and bearings, vol 1
Scognamiglio L, Margheriti L, Mele FM, Tinti E, Bono A, De Gori P, Lauciani V, Lucente FP, Mandiello
AG, Marcocci C, Mazza S, Pintore S, Quintiliani M (2012) The 2012 Pianura Padana Emiliana seimic
sequence: locations, moment tensors and magnitudes. Ann Geophys 55(4):549–559
Sezen H, Elwood KJ, Whittaker AS, Mosalam KM, Wallace JW, Stanton JF (2000) Structural engineering
reconnaissance of the August 17, 1999 earthquake: Kocaeli (Izmit). Pacific Earthquake Engineering
Research Center, Turkey
Shaikh AF (1978) Proposed revisions to shear-friction provisions. PCI J 23(2):12–21
Shrestha B (2009) Vertical ground motions and its effect on engineering structures: a state-of-the-art review.
In: Proceeding of international seminar on hazard management for sustainable development in
Kathmandu, Nepal, pp 29–30
Somerville P (2000) Seismic hazard evaluation. Bull N Z Soc Earthq Eng 33(3):371–386
Spallarossa D, Barani S (2007) Disaggregazione della pericolosità sismica in termini di MR-e, Progetto
DPC-INGV S1, Deliverable D14
Stucchi M, Camassi R, Rovida A, Locati M, Ercolani E, Meletti C, Migliavacca P, Bernardini F, Azzaro R
(2007) DBMI04, il database delle osservazioni macrosismiche dei terremoti italiani utilizzate per la
compilazione del catalogo parametrico CPTI04. Quaderni di Geofisica
Taylor HPJ (1970) Investigation of the forces carried across cracks in reinforced concrete beams in shear by
interlock of aggregate. Technical report
Toniolo G (2002) The seismic design of precast concrete structures in the New Eurocode 8. In: BIBM 17th
international congress of precast concrete industry, Istanbul, Turkey

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
294 Bull Earthquake Eng (2018) 16:259–294

UNI (2005) Appoggi strutturali – Parte3: Appoggi elastomerici. UNI-EN1337:3:2005


Vanzi I, Marano GC, Monti G, Nuti C (2015) A synthetic formulation for the Italian seismic hazard and
code implications for the seismic risk. Soil Dyn Earthq Eng 77:111–122
Yenier E, Erdoğan O, Akkar S (2008) Empirical relationships for magnitude and source-to-site distance
conversions using recently compiled Turkish strong-ground motion database. In: The 14th world
conference on earthquake engineering, pp 1–8
Zhu J, Tan P, Jin J (2015) Fragility analysis of existing precast industrial frames using CFRP reinforcement
Zoubek B, Isakovic T, Fahjan Y, Fischinger M (2013) Cyclic failure analysis of the beam-to-column dowel
connections in precast industrial buildings. Eng Struct 52:179–191
Zoubek B, Fischinger M, Isakovic T (2015) Estimation of the cyclic capacity of beam-to-column dowel
connections in precast industrial buildings. Bull Earthq Eng 13(7):2145–2168

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center
GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers
and authorised users (“Users”), for small-scale personal, non-commercial use provided that all
copyright, trade and service marks and other proprietary notices are maintained. By accessing,
sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of
use (“Terms”). For these purposes, Springer Nature considers academic use (by researchers and
students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and
conditions, a relevant site licence or a personal subscription. These Terms will prevail over any
conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription (to
the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of
the Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may
also use these personal data internally within ResearchGate and Springer Nature and as agreed share
it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not otherwise
disclose your personal data outside the ResearchGate or the Springer Nature group of companies
unless we have your permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial
use, it is important to note that Users may not:

1. use such content for the purpose of providing other users with access on a regular or large scale
basis or as a means to circumvent access control;
2. use such content where to do so would be considered a criminal or statutory offence in any
jurisdiction, or gives rise to civil liability, or is otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association
unless explicitly agreed to by Springer Nature in writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a
systematic database of Springer Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a
product or service that creates revenue, royalties, rent or income from our content or its inclusion as
part of a paid for service or for other commercial gain. Springer Nature journal content cannot be
used for inter-library loans and librarians may not upload Springer Nature journal content on a large
scale into their, or any other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not
obligated to publish any information or content on this website and may remove it or features or
functionality at our sole discretion, at any time with or without notice. Springer Nature may revoke
this licence to you at any time and remove access to any copies of the Springer Nature journal content
which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or
guarantees to Users, either express or implied with respect to the Springer nature journal content and
all parties disclaim and waive any implied warranties or warranties imposed by law, including
merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published
by Springer Nature that may be licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a
regular basis or in any other manner not expressly permitted by these Terms, please contact Springer
Nature at

onlineservice@springernature.com

You might also like