14th World Conference on Seismic Isolation,

Energy Dissipation and Active Vibration Control of Structures

September 9-11 2015 San Diego, Ca USA

NUMERICAL INVESTIGATION OF RE-CENTRING CAPABILITY OF

FRICTION PENDULUM ISOLATORS
Emanuele GANDELLI
Politecnico di Milano
Piazza Leonardo da Vinci 32, 20133, Milano, Italy
emanuele.gandelli@polimi.it

Paolo DUBINI
Politecnico di Milano
Piazza Leonardo da Vinci 32, 20133, Milano, Italy
paolo.dubini@polimi.it

Virginio QUAGLINI
Politecnico di Milano
Piazza Leonardo da Vinci 32, 20133, Milano, Italy
virginio.quaglini@polimi.it

Maria Giuseppina LIMONGELLI

Politecnico di Milano
Piazza Leonardo da Vinci 32, 20133, Milano, Italy
mariapina.limongelli@polimi.it

ABSTRACT - Self-centring capability after the seismic shaking is one of the fundamental functions
required to seismic isolation systems. Low re-centring capacity may lead to serious damage and even
structural collapse due to excessive cumulative displacements during mainshock-aftershock seismic
sequences. Near fault quakes, frequently characterized by large pulse, may have a strong impact on the
behaviour of the isolation systems leading in some cases, to instability phenomena.
In this study some hundreds of nonlinear time-history analyses of SDOF systems were conducted
within an extensive parametric study aimed to investigate the re-centring capability of Friction Pendulum
System (FPS) isolators and its sensitivity to ground motions characteristics. A wide range of devices and
earthquakes, characterized by different values of the isolator design-parameters (restoring stiffness and
frictional damping) and characteristics of the ground motion (“pulse-level” and frequency content), has
been considered. The latters have been quantified in terms of the “predominant period” of the ground
motion and through the introduction of a “kinetic Pulse Index” defined in terms of the rate of transmission
of the kinetic energy. Moreover, the results of a shake table investigation are reported in order to assess the
consistency of the proposed numerical formulation with the experimental response of structures
implementing FPS isolators.
The preliminary results, discussed herein in terms of maximum and residual displacements of the
isolators, show that both the mechanical properties of the isolator and the characteristics of the ground
motion, but also their relationships, may have an important influence on the re-centring capability of the
devices.

Keywords: friction pendulum isolator, re-centring capability, parametric study, shake table tests

1
1 INTRODUCTION

Re-centring related to a seismic isolation device represents the capability of the same to
recover its original configuration after a seismic event. The parameter used to quantify this aspect is
the residual displacement of the device at the end of the seismic shaking (often measured as a
percentage of the maximum one). Re-centring capability must be carefully taken into account
particularly for devices with low restoring forces since, in certain cases, the lack of re-centring
capability may lead to serious damages and even structural collapse due to excessive cumulative
displacements. The evaluation of the re-centring capability of isolated structures is a controversial
matter as witnessed by the fact that current standards, AASHTO (2010), EN1998-2 (2005), and EN
15129 (2009), do not provide homogenous readings [Cardone et al, 2012, Medeot et al., 2013].
According to Katsaras (2008) the response of a hysteretic isolation system (both rubber and
sliding isolators) in terms of restoring capability may be subdivided into two stages: the first one is
the strong-motion stage, in which the system absorbs energy from the seismic shaking, and the
second one is the coda stage, in which seismic energy input is insignificant with respect to that
dissipated by the isolation system. Furthermore the coda stage is mainly governed by the isolator
properties whereas the strong-motion stage is also strongly affected by ground motion details.
Several seismologists have suggested that base-isolated buildings are vulnerable to large pulse-like
ground motions generated at near-fault locations [Heaton et al., 1995, Hall et al., 1995]. Such
ground motions are characterized by a clear predominant period, one or more displacement pulses
with peak velocities up to 1 m/s, and durations of few seconds. These pulses have a large impact on
the isolation systems with natural periods in the same range since they usually produce large
displacements which, especially for elastomeric bearings, may cause instability. In this regard a
proposal of an optimal design of both elastomeric and sliding isolators was first formulated by
Jangid et al. (2005 and 2007).
Some recent numerical studies [Katsaras et al., 2008, Cardone, 2012] have shown that the
main parameter influencing the restoring capability of the isolation system is the ratio dmax/drm,
where dmax is the maximum seismic displacement and drm is the maximum residual displacement
under which the system can be in static equilibrium (see the following section). It must be noted
that the maximum earthquake displacement dmax includes the effect of the excitation, whereas drm is
a characteristic parameter of the isolation system. It was shown that, for a given system, the re-
centring capability is higher for ground motions inducing larger displacements.
Dicleli et al. (2005) studied the combined effect of isolator, ground motion, and substructure
characteristics on the performance of seismic-isolated bridges. In particular it was found that the
response of this kind of structures strongly depends on the ratio between the peak ground
acceleration and the peak ground velocity (apeak/vpeak) of the earthquake.
Cardone et al. (2015) investigated the re-centring capability of FPS isolators by means of an
extensive parametric study involving thousands of nonlinear time-history analyses of SDOF
systems. A regression analysis was performed in order to investigate the dependence of the residual
displacement on the governing parameters of FPS isolators and verify the possible influence of
near-fault events.
The same aspects are further deepened in this study: some preliminary results of a parametric
study on the re-centring capability of friction pendulum isolators are herein presented. A wide range
of devices and earthquakes, characterized by different values of the isolator design-parameters
(restoring stiffness and frictional damping) and characteristics of the ground motion (“pulse-level”
and frequency content), has been considered. The latters have been quantified in terms of the
“predominant period” of the ground motion and through the introduction of a “kinetic Pulse Index”
defined in terms of the rate of transmission of the kinetic energy.
.

2
2 THE FRICTION PENDULUM SYSTEM

The Friction Pendulum System (FPS), also known as Curved Surface Slider (CSS), is today a
well-established anti-seismic hardware for base isolation of buildings and structures. Among the
main advantages offered by CSS isolators are the compact design and smaller dimensions with
respect to seismic rubber isolators with equal load and displacement capacity, the period of
vibration independent on the mass of the isolated superstructure, and the absence of torsional effects
for asymmetric buildings.
The principal elements of an FPS device are two or more concave steel sliding surfaces in
contact through friction pads of self-lubricant material. The principles of operation of the Friction
Pendulum is the same of a typical pendulum: the relative motion along the steel sliding surfaces
lengthens the natural period of the structure decreasing the seismic forces transmitted to the
structure itself. The combined effects of the curvature of the sliding surfaces and of the weight of
the superstructure provides a certain re-centring capability and the seismic energy is dissipated by
means of frictional forces at the sliding interfaces. The frictional damping depends on the
characteristics of the material used for the sliding pads. The nonlinear force-displacement behaviour
of the CSS isolator can be described according to the hysteretic model illustrated in Fig.1. The
lateral force F developed by the isolator at a displacement d is given by the formula:

 d 
F = Q ⋅ sign (d& ) + k 2 ⋅ d = N ⋅  µ eq ⋅ sign (d& ) + (1)
 Req 


where Q = µ eq ⋅ N is the characteristic strength, k 2 = N Req is the restoring stiffness, N is the weight
of the supported structure, and Req and µ eq are respectively the equivalent radius and the equivalent friction
coefficient of the curved surfaces [Zayas et al., 1987, 1990].
When the linear modelling is allowed by the considered design standard the effective
stiffness keff, period Teff, and damping ξeff of the device can be determined through the following
equations:

 1 µ eq 
k eff = N ⋅  +  (2)
 Req d max 
 

Req
Teff = 2π ⋅ (3)
 µ eq ⋅ Req 
g ⋅ 1 + 
 d max 

2 1
ξ eff = ⋅ (4)
π d
+1
µ eq ⋅ Req

where dmax is the amplitude of the cycle.

3
 Figure 1 - Typical hysteretic loop of a friction pendulum isolator

It is well known that the friction coefficient depends on the instant velocity at the sliding
surface according to the following exponential law [Constantinou et al., 1999]:

µ = µ HV − ( µ HV − µ LV ) ⋅ exp( −α ⋅ v ) (5)

where the parameters µ LV and µ HV are usually acknowledged to represent the coefficient of
friction at very low and very high velocity respectively, and α is a transition parameter.
During severe earthquakes, devices with low stiffness k2 may undergo significant
displacements and may exhibit inadequate re-centring capability particularly in the case that high
friction materials are used for the pads. In this regard the maximum residual displacement drm,
identified by the first equilibrium position attained when the system is unloaded under quasi-static
conditions from its design capacity dmax, can be calculated as follows:

d rm = µ LV ⋅ Req (6)

The acceleration threshold over which a seismic event activates the device, that is the sliding
along the curved surfaces occurs, is the critical acceleration a cr ; this, assuming a basic SDOF model
(only isolator displacements, superstructure supposed infinitely rigid), can be approximated as
follows:

a cr = µ ST ⋅ g (7)

while for more refined MDOF models (isolators and superstructure displacements) can be
calculated with the following formula [Calvi et al., 2010]:

( N BS + N SS ) ⋅ µ ST
acr = (8)
M BS + β ⋅ M SS

where µ ST is the static friction coefficient at the motion breakaway, N BS and NUS are
respectively the weights (related masses M BS and M SS ) of the base slab at the isolation level and
of the supported superstructure, β is an amplification factor related to the response spectrum of the
superstructure without base isolation.
It is worth noting that Eq. 5 is not suitable to catch the transition from the static to the kinetic
friction coefficients at the motion breakaway leading to possible underestimation of the critical
acceleration required for the isolator activation.

4
3 CHARACTERISTICS OF THE GROUND MOTION

Several experimental and numerical studies [Dicleli et al., 2005, Cardone, 2012) have shown
the strong relationship between the re-centring behaviour of seismic isolation devices and the
content of strong velocity pulse in the input ground motion. Specifically earthquakes containing
strong velocity pulse are likely to impose high demand in terms of displacements thus affecting the
re-centring after a strong ground shaking. A ground motion is classified as “pulse-like” if the
velocity time-history contains a pulse which is a large portion of the ground motion [Baker, 2007].
A quantitative criterion for the identification of “pulse-like” quakes, based on signal processing
through wavelet analysis, was first proposed by Baker in 2007 and later enhanced by Shahi et al. in
2011. The Baker’s approach is based on the quantification of the relative importance of the largest
velocity pulse extracted from the ground motion with respect to the remaining part of the signal.
In this paper a different approach is adopted, based on the rate of transmission of the kinetic
energy to the structure. Specifically a kinetic pulse index PI k is defined in terms of the ratio
between the time interval Dv,T across which a significant amount of kinetic energy is transmitted to
the structure with respect to the significant duration of the quake Dv,B:

D v ,T
PI k = 1 − (9)
Dv, B

being Dv,T and Dv,B respectively the Trifunac [Trifunac et al., 1975] and bracketed [Bolt,
1969] durations of the ground motion, both calculated in terms of velocities.
The higher PIk, the shorter the time interval of transmission of the kinetic energy. The
Trifunac duration Dv,T = t0.95 I − t0.05 I in terms of velocity is calculated as the time interval between
E E

5% and 95% of the energy integral IE [Anderson, 2004]:

∞

I E = ∫ v g2 dt (10)
0

The bracketed duration Dv,B is the total time between the first and the last exceeding of a given
threshold during the strong motion. Herein the threshold has been fixed at 1% of the absolute peak
velocity.
The basic idea behind this approach for the detection of “pulse-like” earthquakes is that the
shorter the time interval during which a significant amount of kinetic energy is transmitted to the
structure, the higher the demand imposed to the structure in terms of peak velocity. Moreover, also
the predominant period of the pulse Tp can have a strong effect on the structural response depending
on its relationship with the effective period of the structure. Herein the value of Tp has been
estimated as the period Tsv corresponding to the peak of the un-damped ground motion velocity
response spectrum.

4 PARAMETRIC STUDY

The re-centring capability of friction pendulum isolators was investigated by means of an

extensive parametric study in terms of the values of the maximum dmax and residual dres
displacement under a given earthquake. The variables considered in the investigation were both the
governing properties of the devices (restoring stiffness and damping) and the characteristics of the
ground motion (“pulse-level” and frequency content).
A wide range of devices were taken into account: twenty-five isolator prototypes were
defined varying both the equivalent radius Req and the friction coefficients μLV, μHV, and α over
5
typical ranges for the design of seismically isolated structures. A vertical load N=100 tons, a typical
value for a medium-height residential building, was assumed acting on the device to model the
weight of the superstructure. Table 1 reports the main design parameters and the un-damped period
of vibration Tis = 2π Req g of the considered twenty-five isolator prototypes.

Table 1 - Characteristic parameters of the considered curved surface slider prototypes

Req Friction µ LV µ HV α Tis
Prototypes
(mm) id (-) (-) (s/mm) (s)
FPS 1.1 f1 0.01 0.025 0.0055 3.0
FPS 1.2 f2 0.02 0.050 0.0055 3.0
FPS 1.3 2200 f3 0.03 0.075 0.0055 3.0
FPS 1.4 f4 0.04 0.100 0.0055 3.0
FPS 1.5 f5 0.05 0.125 0.0055 3.0
FPS 2.1 f1 0.01 0.025 0.0055 3.5
FPS 2.2 f2 0.02 0.050 0.0055 3.5
FPS 2.3 3000 f3 0.03 0.075 0.0055 3.5
FPS 2.4 f4 0.04 0.100 0.0055 3.5
FPS 2.5 f5 0.05 0.125 0.0055 3.5
FPS 3.1 f1 0.01 0.025 0.0055 3.8
FPS 3.2 f2 0.02 0.050 0.0055 3.8
FPS 3.3 3500 f3 0.03 0.075 0.0055 3.8
FPS 3.4 f4 0.04 0.100 0.0055 3.8
FPS 3.5 f5 0.05 0.125 0.0055 3.8
FPS 4.1 f1 0.01 0.025 0.0055 4.0
FPS 4.2 f2 0.02 0.050 0.0055 4.0
FPS 4.3 4000 f3 0.03 0.075 0.0055 4.0
FPS 4.4 f4 0.04 0.100 0.0055 4.0
FPS 4.5 f5 0.05 0.125 0.0055 4.0
FPS 5.1 f1 0.01 0.025 0.0055 5.0
FPS 5.2 f2 0.02 0.050 0.0055 5.0
FPS 5.3 5000 f3 0.03 0.075 0.0055 5.0
FPS 5.4 f4 0.04 0.100 0.0055 5.0
FPS 5.5 f5 0.05 0.125 0.0055 5.0

In order to ensure a great variety of possible seismic scenarios, a large number of un-scaled
records were downloaded from the Pacific Earthquake Engineering Research Center (PEER)
database and then subdivided into groups taking into account their “pulse-level” and fundamental
period. In this regard four ranges of the predominant period Tsv were established covering the whole
range of periods of the prototypes previously defined:
(1) Tsv ≤ 2.0 sec;
(2) 2.0 <Tsv ≤ 3.0 sec;
(3) 3.0 <Tsv ≤ 4.0 sec;
(4) Tsv> 4.0 sec.
Moreover, basing on the values of PIk and on the comparison with the classification proposed by
Baker (2007) for “pulse” and “no pulse” earthquakes, three ranges were defined to distinguish
between earthquakes corresponding to a different content of a dominant strong velocity pulse:
(1) no-pulse PIk< 0.40;
(2) weakly pulse 0.40 ≤ PIk ≤ 0.70;
(3) pulse PIk> 0.70.
Twenty-four records were selected and subdivided in twelve homogeneous groups according
to the ranges previously defined (Table 2). Among records selection criteria, also a limit for the
lowest usable frequency l.u.f <0.2 Hz was established. This limit corresponds to the minimum
natural frequency of all considered FPS isolators.
6
 Table 2 - Characteristic parameters of the selected records
Tsv range Tsv PIk l.u.f.
Type PEER file name Event
(sec) (sec) (-) (Hz)
RSN3860_CHICHI.05_CHY008N Chi-Chi, 1999 0.37 0.33 0.075
no pulse
RSN3858_CHICHI.05_CHY004N Chi-Chi, 1999 0.34 0.38 0.075
RSN496_NAHANNI_S2330 Nahanni, 1985 0.52 0.56 0.125
Tsv< 2 sec weakly pulse
RSN3846_CHICHI.03_CHY008W Chi-Chi, 1999 1.52 0.55 0.063
RSN451_MORGAN_CYC285 Morgan Hill 1984 0.83 0.86 0.125
pulse
RSN766_LOMAP_G02090 Loma Prieta, 1989 1.47 0.84 0.075
RSN2938_CHICHI.05_CHY016N Chi-Chi, 1999 2.34 0.29 0.075
no pulse
RSN3844_CHICHI.03_CHY004N Chi-Chi, 1999 2.69 0.34 0.038
RSN827_CAPEMEND_FOR000 Cape Mendocino, 1992 2.56 0.51 0.070
2 <Tsv< 3 sec weakly pulse
RSN3844_CHICHI.03_CHY004W Chi-Chi, 1999 2.90 0.59 0.038
RSN292_ITALY_A-STU270 Irpinia, 1980 2.82 0.82 0.125
pulse
RSN171_IMPVALL.H_H-EMO270 Imperial Valley, 1979 2.94 0.85 0.100
RSN2102_DENALI_NOAA-90 Alaska, 2002 3.43 0.24 0.026
no pulse
RSN297_ITALY_B-BIS270. Irpinia Eq, 1980 3.83 0.39 0.163
RSN827_CAPEMEND_FOR090 Cape Mendocino, 1992 3.08 0.46 0.070
3 <Tsv< 4 sec weakly pulse
RSN2695_CHICHI.04_CHY016W Chi-Chi, 1999 3.82 0.48 0.050
RSN181_IMPVALL.H_H-E06230 Imperial Valley, 1979 3.40 0.89 0.063
pulse
RSN182_IMPVALL.H_H-E07230 Imperial Valley, 1979 3.27 0.85 0.075
RSN3851_CHICHI.04_CHY004W Chi-Chi, 1999 5.07 0.36 0.100
no pulse
RSN2100_DENALI_K205-90 Alaska, 2002 5.05 0.35 0.017
RSN2115_DENALI_PS11-66 Alaska, 2002 5.76 0.47 0.130
Tsv> 4 sec weakly pulse
RSN1170_KOCAELI_MCD090 Kocaeli, 1999 5.88 0.59 0.075
RSN1148_KOCAELI_ARE090 Kocaeli, 1999 5.31 0.70 0.088
pulse
RSN179_IMPVALL.H_H-E04230 Imperial Valley, 1979 4.08 0.76 0.063

All the records were scaled to the same value of the peak ground acceleration PGA=0.3g
selected to be higher than the maximum critical acceleration acr of the selected FPS isolators.
The nonlinear dynamic analyses were conducted using the software OpenSees v.2.4.5
[McKenna et al., 2000]. Assuming a rigid body behaviour of the superstructure, the friction
pendulum isolators is modelled as an elastic-plastic SDOF system with an initial stiffness k1, a
secondary stiffness k2, and the yield strength Fy (see Fig. 2). In this regard, the friction pendulum
(FP) element available in the code is able to reproduce both the dependence of the restoring
stiffness k2 and the friction coefficient μ respectively on the instant vertical load N acting on the
device and the velocity v at the sliding surface according to Eq. 5. Moreover, the dynamic response
of the system is calculated integrating well-acknowledged equations of motion [Nagarajaiah et al.,
1991].

Figure 2 - Initial and post-yielding stiffness model for numerical analysis

The yielding displacement dy, beyond which the actual sliding of the isolator starts, and the
corresponding force Fy are given by:
Q
dy = (11)
k1 − k 2

7
 F y = k1 ⋅ d y (12)

In order to minimize the elastic displacements, the initial stiffness was set to k1 =100k2.

5 RESULTS

For each considered FPS prototype, twenty-four nonlinear analysis were performed by
applying all the seismic input reported in Table 2. The displacements time histories were calculated
in order to estimate the maximum dmax and residual dres displacements at the end of the earthquake.
Fig. 3 shows for the isolator FPS 3.3 (Req=3000mm, μLV=0.03, μHV =0.075, α = 0.0055s/mm)
the differences between the displacement time histories and the force-displacement loops relevant
to ground motions with different values of the kinetic Pulse Index. Specifically the responses to a
“no-pulse” (RSN2102_DENALI_NOAA-90), a “weakly-pulse” (RSN827_CAPEMEND_FOR090),
and a “pulse” (RSN181_IMPVALL.H_H-E06230) earthquake are reported respectively in Fig. 3. In
the case on the right the presence of a pulse in the displacements time history is quite evident.
Moreover, it can be seen that oscillatory quakes ("no-pulse") are characterized by a sensibly higher
number of hysteretic cycles with respect to weakly and “pulse-like” motions.

Figure 3 - Displacement time-histories (up) and hysteretic loops (down) of prototype FPS 3.3
(Req=3000mm, μLV =0.03, μHV =0.075, α = 0.0055s/mm) subjected to “no-pulse” (left), “weakly-
pulse” (middle), and “pulse” (right) quakes

Fig. 4 (left) shows the variation of the maximum displacement dmax with the ratio Tratio between the
predominant period of the quake Tsv and the period of the device Tis. The Tratio was aimed to identify near-
resonant conditions ideally corresponding to unit value of this parameter. This is quite evident in the graph
where higher values of dmax occur for values of Tratio close to 1. The same near-resonant effect can also be
observed for residual displacements dres at the end of the seismic shaking (Fig. 4, right).

8
 Figure 4 - Near-resonance effect on the maximum (left) and residual (right) displacement

The sensitivity of the maximum displacements dmax to the frictional behaviour at the sliding interface
is shown in the Fig. 5 (left): as expected, lower maximum displacements were computed for high-friction
(f5) devices with respect to the ones obtained for low-friction (f1) isolators.
Moreover, Fig. 5 (right) shows the dependence of the restoring capability, here referred as the ratio dres/ dmax,
on the frictional properties of the device: it can be noted that high-friction devices (f5) exhibit a lower re-
centring capability with respect to low-friction (f1) isolators.

Figure 5 - Comparison between the maximum displacements (left) and the restoring capability
(right) of low-friction (f1) and high-friction (f5) FPS isolators

In order to evaluate the possible sensitivity of both the maximum dmax and the restoring capability
dres/ dmax to “pulses-like” seismic excitations, Fig. 6 shows these parameters as a function of the previously
defined pulse index PIk (Eq. (9)). A clear trend, proving the dependence of the examined properties on the
“pulse-level” of the seismic event, was not found.

Figure 6 - Maximum displacements and restoring capability as a function of the pulse index PIk:
comparison between low-friction (f1) and high-friction (f5) isolators

9
Recent studies [Medeot et al., 2004, Katsaras et al., 2008], later endorsed in the European Standards
EN 15129 [CEN, 2009], assuming the requirement (dres/dmax)<0.5 for a good re-centring of base-isolated
structures, suggested the following re-centring criterion:

d max
≥ 0.5 (13)
d rm

Obtained numerical results confirmed the reliability of this criterion as withnessed by Fig. 7. Indeed, all the
points on the right of the vertical line (dmax/drm)=0.5 fullfill the re-centring requirement (dres/dmax)<0.5.

Figure 7 - Re-centring capability dres/ dmax as a function of the ratio dmax/ drm

6 A CASE STUDY

In this section the results of a shake table investigation are reported to demonstrate the
consistency of the proposed numerical formulation with the experimental behaviour of isolation
systems with FPS.
In this regard, several unidirectional shake table tests were carried out at the laboratory of the
Department of Structural Engineering of the University “Federico II” of Naples, Italy [Quaglini, et
al., 2014].
The test mock up consisted of a one-storey steel structure at a third-length scale: a vertical
frame with dimensions 2.5x2.5 m in plant and 2.9 m height was fixed to a rectangular 2.5x2.5 m
base frame (Fig. 8).
The whole structure was supported by four seismic isolation bearings; moreover a reinforced
concrete slab (dimensions 2.1x2.65x0.250 m) was placed on the roof of the structure and a rigid
reaction mass consisting of 40 concrete blocks (150x235x305 mm) was placed at the level of the
base frame. The total mass of the isolated structure was 8200 kg. Taking advantage of the
symmetry of the prototype structure, each isolator supported a quarter of the total weight.
Three isolation layout, each one comprised of four identical FPS bearings, were assessed in
the tests. Three different models of FPS bearings were accounted for (Table 3).
Low Friction (LF) and Medium Friction (MF) specimens consisted of double FPS bearings,
each one including two pairs of curved surfaces with identical radius R1=R2=770 mm and a slider
10
with height h=55 mm, corresponding to an equivalent radius Req=R1+R2–h=1475 mm. The two
models differed for the material of the sliding pads: lubricated virgin PTFE resin was used for the
Low Friction specimens, and a PTFE – bronze compound lubricated with silicon grease was used
for the Medium Friction specimens.
The High Friction (HF) model consisted of a single FPS, with radius of curvature of the
primary curved surfaces R1=1270 mm and radius of the secondary surfaces R2=300 mm; the height
of the slider was h=70 mm, corresponding to an equivalent radius Req=1500 mm. The sliding pad
was made of the same PTFE – bronze compound used for the MF model, but not lubricated.

Figure 8 - Test mock up

The steel frame was modelled in OpenSees using 3D beam elements: fixed restraints were
supposed for the connection between the nodes and a rigid diaphragm with lumped mass was
introduced to simulate the concrete slab of the floor and the base horizontal frame. Moreover, four
friction pendulum (FP) elements were introduced at the base of each column simulating the
behaviour of the tested isolator prototypes. In this regard, the characteristic parameters of the sliding
materials used in the tests are reported in Table 3 and were assessed by means of a custom biaxial
machine at the Materials Testing Laboratory of Politecnico di Milano [Quaglini et al., 2012].

Table 3 -Tested FPS prototypes and their design parameters

Req k2 Tis bearing pressure µ ST µ LV µ HV α
FPS model
(m) (kN/m) (sec) p (MPa) (-) (-) (-) (s/mm)
LF 1.485 13.54 2.44 7.10 0.020 0.008 0.039 0.003
MF 1.485 13.54 2.44 7.10 0.065 0.027 0.200 0.002
HF 1.500 13.41 2.46 3.98 0.082 0.032 0.265 0.002

The selected ground motions used as input for the shake table tests are listed in Table 4:
every ground motion record was compressed in time by a factor of √3 to satisfy the similitude
requirements and scaled to the design peak ground acceleration (PGA) of 0.259 g.

Table 4 - Ground motions used in the test program and relevant characteristics
PGA Scale Factor Tsv PI
Earthquake name
(m/s2) (-) (sec) (-)
Bingol 2.92 0.87 0.40 0.65
Friuli 3.50 0.73 0.52 0.77
Campano-Lucano 1.78 1.43 1.58 0.70

11
 For reasons of synthesis, only the results related to the Campano Lucano shake table test are here
presented; indeed, having the same the Tratio closest to 1, it produced the largest isolator displacements. In
this regard, Figures 9-11 show the comparison between recorded experimental FPS displacements and
the ones calculated by the software.

Figure 9 - Comparison between experimental (EXP) and numerical (FEM) displacements at

base (FPS displacements) - LF isolator, Campano Lucano earthquake

Figure 10 - Comparison between experimental (EXP) and numerical (FEM) displacements at

base (FPS displacements) - MF isolator, Campano Lucano earthquake

12
 Figure 11 - Comparison between experimental (EXP) and numerical (FEM) displacements at
base (FPS displacements) - MF isolator, Campano Lucano earthquake

A good agreement between numerical and experimental displacement time-histories during the strong-
motion stage, in particular for the peak displacement dmax (which is the most important design information),
was reached for all FPS isolators.
There is still a gap between numerical and experimental values of residual displacements dres,
especially for MF and HF devices. Among possible causes of this discrepancy there are: (1) excessive
simplification in the superstructure modelling (mechanisms of structural damping not considered); (2)
influence of static friction (μST), not implemented in the FE model, on the response of the FPS isolator
during the free oscillations at the end of seismic excitation (coda stage).

7 CONCLUSIONS

In the present paper some preliminary results of a parametric study on the re-centring
capability of FPS isolators are presented. The sensitivity of the maximum and the residual
displacements induced by a seismic ground motion to the characteristic parameters of the device
(restoring stiffness and frictional damping) and to the features of the ground motion (predominant
period and “pulse-level”) was investigated.
Results show that:
• the maximum displacement dmax decreases at the increase of the frictional damping at high velocities
(represented by µ HV);
• the re-centring capability (referred as dres/ dmax) decreases with the increasing of the frictional damping;
• both maximum dmax and residual dres displacements significantly increase when the predominant period of
the quake Tsv approaches the effective period of the device;
• there is not a clear trend proving the dependence of both maximum dmax and residual dres displacements on
the “pulse-level” of the seismic event.

13
References

Anderson, J.G. [2004] “Quantitative measure of the goodness-of-fit of synthetic seismogram”,

Proceedings of the 13th World Conference on Earthquake Engineering, Vancouver, Canada, 1-6
August, Paper N° 243
AASHTO [2010] “Guide Specification for Seismic Isolation Design”, Third Edition, American
Association of State Highways and Transportation Officials, Washington D.C, USA
Baker, J.W. [2007] “Quantitative Classification of Near-Fault Ground Motions Using Wavelet
Analysis”, Bulletin of the Seismological Society of America, Vol. 97, No. 5, pp. 1486–1501,
October 2007, DOI: 10.1785/0120060255
Bolt, B.A. [1969] “Duration of strong motion”, Proceedings of the 4th World Conference on
Earthquake Engineering, Santiago, Chile, pp. 1304–1315
Calvi, G.M., Pietra, D., Moratti, M. [2010] “Criteri per la progettazione di dispositivi di isolamento
a pendolo scorrevole”, Progettazione Sismica, n.3/2010, IUSS Press
Cardone, D. [2012] “Re-centring capability of flag-shaped seismic isolation systems”, Bulletin of
Earthquake Engineering, 10:1267–1284, DOI 10.1007/s10518-012-9343-1
Cardone, D., Gesualdi, G., Brancato, P. [2015] “Restoring capability of friction pendulum seismic
isolation systems”, Bulletin of Earthquake Engineering, DOI 10.1007/s10518-014-9719-5
Constantinou, M.C., Tsopelas, P., Kasalanati, A., and Wolff, E.D. [1999] “Property modification
factors for seismic isolation bearings”, Report MCEER-99-0012, Multidisciplinary Center for
Earthquake Engineering Research, State University of New York
Dicleli, M., and Buddaram, S. [2005] “Effect of isolator and ground motion characteristics on the
performance of seismic-isolated bridges”, Earthquake Engineering & Structural Dynamics, Wiley,
DOI: 10.1002, 35:233–250
CEN 15129 [2009] “Antiseismic Devices”, Comité Européen de Normalisation, Brussels
CEN 1998 [2005] “Design of structures for earthquake resistance”, Part 2: Bridges, Comité
Européen de Normalisation, Brussels
Hall, J.F., Heaton, T.H., Halling, M.W., Wald, D.J. [1995] “Near-source ground motion and its
effects on flexible buildings”, Earthquake Spectra, 11:569–605
Heaton, T.H., Hall, J.F., Wald, D.J., Halling, M.W. [1995] “Response of high-rise and base-isolated
buildings to a hypothetical MW7.0 blind Thrust earthquake”, Science, 267:206–11
Jangid, R.S. [2005] “Optimum friction pendulum system for near-fault motions”, Engineering
Structures, 01/2005; 27(3):349-359
Jangid, R.S. [2007] “Optimum lead–rubber isolation bearings for near-fault motions”, Engineering
Structures, 01/2007; 29(10):2503-2513
Katsaras, C.P., Panagiotakos, T.B., Kolias, B. [2008] “Restoring capability of bilinear hysteretic
seismic isolation Systems”, Earthquake Engineering & Structural Dynamics, Wiley, DOI: 10.1002,
37:557–575
Medeot, R. [2004] “Re-centering capability evaluation of seismic isolation systems based on energy
concepts”, Proceedings of the 13th World Conference on Earthquake Engineering, Vancouver,
Canada, Paper N° 3106
Medeot, R. [2013] “Re-centring Capability of Seismic Isolation Systems: a Controversial Matter”,
Proceedings of XV conference ANIDIS – L’ingegneria sismica in Italia, Padova, Italy, 30th June –
4th July

14
McKenna, F., Fenves, G.L., Scott, M.H., and Jeremic, B. [2000] “Open System for Earthquake
Engineering Simulation (OpenSees)”, Pacific Earthquake Engineering Research Center (PEER),
University of California, Berkeley, CA
Nagarajaiah, S., Reinhorn, A.M., Constantinou, M.C. [1991] “Nonlinear dynamic analysis of 3D
base-isolated structures”, Journal of Structural Engineering, ASCE, 117(7), 2035–2054
PEER database, http://peer.berkeley.edu/peer_ground_motion_database (last consulted on 4 April
2014)
Quaglini, V., Dubini, P., Poggi, C. [2012] “Experimental assessment of sliding materials for
seismic isolation systems”, Bulletin of Earthquake Engineering, Springer-Verlag, Berlin Heidelberg
(D), Vol. 10: 717-740
Quaglini, V., Gandelli, E., Dubini, P., Vazzana, G., Farina, G. [2014] “Re-centring capability of
friction pendulum system: experimental investigation”, Proceedings of the 2nd ECEES – European
Conference on Earthquake Engineering and Seismology, Istanbul 24-28th August
Shahi, S.K., and Baker, J.W. [2011] “Regression Models for Predicting the Probability of Near-
Fault Earthquake Ground Motion Pulses, and their Period", Proceedings of the 11th International
Conference on Applications of Statistics and Probability in Civil Engineering, August 1-4, 2011,
ETH Zurich, Switzerland
Trifunac, M.D., and Brady, A.G. [1975] “A study of duration of strong earthquake ground motion”,
Bulletin of the Seismological Society of America, vol. 65 (3), pp. 581-626
Zayas, V.A., and Low, S.S., and Mahin, S.A. [1987] “The FPS earthquake protection system”,
Earthquake Engineering Research Center, Report No. UCB/EERC-87/01, Berkeley, California
Zayas, V.A., and Low, S.S., and Mahin, S.A. (1990) “A simple pendulum technique for achieving
seismic isolation”, Earthquake Spectra, Earthquake Spectra, Vol. 6, No. 2, pp. 317-333. DOI:
10.1193/1.1585573

15