Soil Dynamics and Earthquake Engineering: Mariano Angelo Zanini, Klajdi Toska, Flora Faleschini, Carlo Pellegrino

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering


journal homepage: http://www.elsevier.com/locate/soildyn

Seismic reliability of reinforced concrete bridges subject to environmental


deterioration and strengthened with FRCM composites
Mariano Angelo Zanini a, *, Klajdi Toska a, Flora Faleschini a, b, Carlo Pellegrino a
a
Department of Civil, Environmental and Architectural Engineering, University of Padova, via Francesco Marzolo 9, Padova, Italy
b
Department of Industrial Engineering, University of Padova, via Gradenigo 6a, Padova, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: Ensuring adequate seismic reliability levels for existing infrastructure components is a key issue for owners in
Bridge earthquake-prone countries. In particular, bridges may suffer relevant damage in case of a seismic event, and this
Deterioration issue can be magnified due to deterioration phenomena induced by environmental agents, like CO2 emissions or
FRCM
chlorides. Among the available techniques, seismic retrofitting can be pursued through the use of composite
Seismic retrofit
Seismic reliability
materials. For a proper bridge management policy, infrastructure owners need to know at which time instant
scheduling restoration interventions on bridges, and such issues have to be fixed with a reliability-based metric.
The present study numerically investigates the effectiveness of the use of fabric reinforced cementitious matrix
(FRCM) systems in the seismic retrofitting of an existing multi-span simply supported (MSSS) reinforced concrete
(RC) bridge subject to aging. Fragility curves are first derived on the basis of refined non-linear time-history
analyses performed on the “as built” configuration. Fragilities are further computed for different combinations of
deterioration scenarios and seismic retrofitting schemes with increasing number of FRCM layers. Time-variant
seismic reliability profiles are thus assessed, and reliability gains achievable at different time instants with the
implementation of the FRCM retrofitting scenarios are quantified in order to provide useful information for the
infrastructure owner decision-making.

1. Introduction replacement, which is largely less sustainable from both environmental


and financial point of views. Different seismic strengthening techniques
Multi-Span Simply Supported (MSSS) Reinforced Concrete (RC) can be adopted for increasing strength, ductility and stiffness of existing
bridges constitute a relevant portion of roadway infrastructure in Italy RC bridges. Among others, from the ‘90s, the use of composite materials
and other European countries. Such infrastructure components are has become a convenient solution for RC elements.
subject to aging and environmental deterioration mainly due to high Several researchers experimentally investigated the efficacy of fiber
concentrations of CO2 emissions and chlorides action. In addition, the reinforced polymer (FRP) confinement provided by externally bonded
increase of traffic volumes, and the evolution of building codes pre­ FRP sheets onto concrete surface in the seismic behavior of RC elements,
scriptions have highlighted a growing number of them to be no more observing significant strength and ductility gains in the retrofitted
able to ensure safety margins required by current codes and guidelines. specimens with respect to the reference ones [3–9]. Ye et al. [10] tested
Many of these structures are also deficient in terms of lateral load car­ eight square columns with the aim to investigate the ductility
rying capacity, since they were designed - especially in medium seis­ enhancement provided by CFRP (carbon FRP). Experimental results
micity areas – without seismic resistance criteria and with insufficient allowed to calibrate a confinement factor for CFRP and an equivalent
shear capacity due to lack of transversal reinforcement in vertical ele­ transversal reinforcement index to be used for design purposes. Colomb
ments. The combination of all these factors may result in sudden failures et al. [11] investigated the effectiveness of CFRP shear reinforcements
in case of earthquake occurrence [1,2], and for such reasons a growing with various fully or partially wrapped solutions in specimens lacking
interest on the implementation of seismic retrofitting actions is sufficient transverse steel reinforcement, showing that ductility of the
emerging among infrastructure owners. Such intervention represents a fully wrapped columns increased. This result was attributed to plastic
cost-effective alternative with respect to bridge demolition and hinges shift in the foundation, due to the considerable over-strength of

* Corresponding author.
E-mail address: marianoangelo.zanini@dicea.unipd.it (M.A. Zanini).

https://doi.org/10.1016/j.soildyn.2020.106224
Received 17 December 2019; Received in revised form 31 March 2020; Accepted 7 May 2020
Available online 18 June 2020
0267-7261/© 2020 Elsevier Ltd. All rights reserved.
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

the wrapped zone, while any buckling of compressed rebars was avoi­ the configurations with taller piers and higher skew angles.
ded. The authors suggested also to not oversize the FRP reinforcement However, in the last decade, the use of fabric reinforced cementitious
since it may imply column over-strength and node failures. Ozcan et al. matrix (FRCM) systems is become an effective alternative solution to the
[12,13] studied the influence of using CFRP anchor dowels, resulting in use of classic FRPs. FRCM technique allows to overcome FRP drawbacks
an improvement of the seismic behavior of rectangular RC columns mainly associated with the use of epoxy resins, i.e. high costs, poor
wrapped with CFRPs. They proposed also a drift-based design equation performance in case of high temperatures, inability to apply it on wet
for sizing FRP wraps on shear deficient rectangular columns. Promis surfaces, incompatibility with some types of substrate materials,
et al. [14] and Promis and Ferrier [15] developed a novel performance through the replacement of organic (epoxy resins) with inorganic
index based on structural damage, aimed at evaluating the efficiency of (mortar) matrix [29]. The inorganic matrix is incombustible and capable
external FRP reinforcement as a function of the axial loading, shape of to maintain the system mechanical properties even in high tempera­
the strengthened RC column, presence of FRP bars and main mechanical tures. In addition, unlike FRP systems, FRCM ones do not emit toxic
characteristics of the FRP wraps. Rodrigues et al. [16] performed an substances when exposed to high temperatures and are not affected by
experimental campaign on nine rectangular RC columns strengthened UV light [30,31]. When retrofitting with FRCM techniques, eventual
with CFRP and steel jackets. The aim of the research was to develop a damage that may occur in the retrofitted element clearly manifests itself
procedure for calibrating the cyclic behavior in case of biaxial loading, in the mortar jacket, making inspection, in case of a seismic event, quick
considering diagonal and diamond horizontal displacement paths. and easy. On the other hand, inspection of elements retrofitted with FRP
Ouyang et al. [17] investigated the use of basalt FRPs (BFRPs) as jackets is rather difficult as the internal damage is not always noticeable
alternative to classic CFRPs for seismic strengthening of square RC onto the FRP surface [29].
columns. This choice was linked to basalt low environmental impact and Limited studies can be found in literature on the seismic response of
price, and high mechanical characteristics, which make it a RC elements retrofitted with FRCMs: Bournas et al. [32] tested three
cost-effective solution and a promising alternative to conventional full-scale deficient RC columns strengthened with both CFRP and FRCM
CFRPs. Wang et al. [18] analyzed the role of the direction of the hori­ wraps in the plastic hinge regions and subject to cyclic loading, showing
zontal load on the seismic response of ten rectangular RC columns similar results between the specimens retrofitted with the two different
strengthened with CFRPs. Results highlighted how shear strength in solutions. In Bournas et al. [33], the authors experimentally demon­
non-principal directions of both bare and retrofitted specimens can be strated a better performance in preventing local rebar buckling in
predicted starting from the predicted strength in the principal di­ specimens confined with FRCMs than CFRP ones, mainly attributed to
rections, combining them with the ellipse equation. the capacity of FRCM to resist local stresses. Similar results were illus­
Other studies were focused on the experimental assessment of the trated by Ref. [34,35] on columns subject to cyclic loading and
performance of seismic damaged columns repaired with FRP wraps strengthened in flexure with near surface mounted (NSM) bars coupled
[19–21]. He et al. [22] tested three-half scale severely damaged square with FRCM wraps for the increase of confinement.
RC bridge columns by artificially fracturing and buckling steel rebars. The above context highlights a substantial lack of studies dealing
They investigated the effectiveness of the CFRP strengthening solution with the numerical assessment of the effectiveness in the use of FRCM
coupled with a specific system designed to anchor longitudinal CFRP systems for the seismic retrofitting of existing RC bridges and no refer­
strips to the column footing. Lavorato and Nuti [23] tested a set of bridge ence can be found dealing with the quantification of the gain in terms of
piers in 1:6 scale with pseudo-dynamic tests in order to reproduce po­ seismic reliability with respect to the as-built configuration for common
tential seismic damage and subsequently repaired the specimens with bridge schemes can be found. In order to fill this gap, the present study
CFRPs. describes the results of an extensive numerical analysis carried out to
Instead, few numerical studies were conducted on the seismic investigate the effectiveness of seismic retrofitting with FRCM systems
response of columns or entire bridges retrofitted with FRPs, in order to for an Italian existing MSSS RC-bridge. The paper details the workflow
quantify, in probabilistic terms, the beneficial effect of such retrofitting and models that can be generally used to derive time-variant seismic
intervention. Billah and Alam [24] numerically investigated the per­ reliability profiles over the entire service-life for a RC bridge. Such
formance of a three-column bridge bent retrofitted with different tech­ profiles represent a key information for infrastructure owners for
niques, namely CFRP wraps, steel jackets, concrete jackets and ECC rationally design and schedule the FRCM strengthening scheme within
(Engineered Cementitious Composite) jacketing, via the execution of the bridge maintenance planning. The work is organized as it follows:
incremental dynamic analysis (IDA) approach and the subsequent Section 2 briefly recaps main methods and models to be used for the
computation of fragility curves. The best performances were displayed assessment of seismic reliability of structural systems, as well as dete­
for the CFRP and ECC solutions. Fakharifar et al. [25] analyzed the rioration models for aging simulation and analytical models of FRCM
effectiveness of three jacketing solutions on the seismic response of a RC passive confinement action. Section 3 presents the analyzed bridge case-
multi-span continuous bridge, initially damaged by the occurrence of a study, whereas Section 4 shows the results of the numerical analysis,
mainshock, and then repaired and attacked again by an aftershock namely fragility curves derived for different aging scenarios both for the
event. IDA approach was used obtaining similar results among the tested as-built configuration and those retrofitted with varying FRCM layouts.
solutions. Parghi and Alam [26] performed a fractional factorial analysis Finally, Section 5 illustrates the time-variant seismic reliability profiles
aimed at investigating the impact of several parameters like material and discusses the results of the following cost-benefit analysis, whereas
properties, longitudinal and transversal reinforcement ratios, external Section 6 lists main concluding remarks of this study.
confinement, axial load, shear span-depth ratio on lateral load capacity,
ductility and failure mode of a single column RC bridge pier. Nonlinear 2. Methods & models
pushover analyses were carried out on piers modelled with fiber-based
approach. More recently, the same authors [27] carried out IDA and The present Section details main methods and models herein adopted
derived fragility curves considering 20 ground motion records, showing to quantify seismic reliability of structural systems, accounting also for
the impact of the abovementioned parameters on the resulting fragility aging and FRCM seismic retrofitting.
estimates. Mahboubi and Shiravand [28] numerically analyzed different
continuous-span RC bridge configurations with increasing skew angles 2.1. Seismic reliability analysis
up to 30� and retrofitted with CFRP, via the execution of non-linear
time-history analyses. The aim of the research was to quantify seismic In the context of the Performance-Based Earthquake Engineering
damage by using displacement-, energy-, and stiffness-based indices, (PBEE) framework [36], the occurrence of main earthquakes at the
showing how the effect of CFRP confinement significantly increased for construction site is assumed to be a Homogenous Poisson Process (HPP).

2
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Under this hypothesis, and not considering damage accumulation in and β stands for the standard deviation of the demand conditioned on im
structures, the process of events causing the structural failure is also and can be estimated from the regression of the seismic demands as:
represented by an HPP, whose unique parameter, the failure rate λf , can sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
be used for computing the failure probability in any time interval. Pn
ða þ b⋅lnðimi ÞÞ�2
i¼i ½lnðedpi Þ
Particularly, it depends on the hazard curve λim , representing the seis­ β¼ (6)
n 2
micity on a specific site, and on fragility curve P½fjim�, being the proba­
bilistic structural behavior of a specific structure. The failure rate λf is Once estimated the failure rate λf , based on the HPP assumption it is
thus computed by applying the Total Probability Theorem: possible to quantify the probability of failure for a given time window of
Z interest t due to earthquake occurrences as follows:
λf ¼ P½f jim�⋅jdλim j (1)
PE;f ¼ 1 e λf ⋅T
(7)
im

and further derive the seismic reliability index βE;t in accordance to


where jdλim j is the derived from the seismic hazard curve, and P½fjim�
reliability analysis theory with the following transformation:
characterizes the seismic fragility of a given structural system. Current

state-of-the-art approaches for the computation of λim are based on the βE;t ¼ Φ 1 PE;f (8)
so-called Probabilistic Seismic Hazard Analysis (PSHA) [37,38], which
associates to each im value, the corresponding annual rate of events
exceeding it at the site where the structure is located. This calculation 2.2. Deterioration modelling
involves the three main components that contribute to define the seis­
micity of a specific area, namely the earthquake occurrence model, the Most of the bridges in Italy, but also across Europe, have already
seismogenic model and the ground motion attenuation law. From λim , overcome or are near to exceed their service life. In many cases, these
jdλim j is easily obtained as: bridges are located in highly aggressive environments, as in coastal re­
dλim gions or in cold mountainous areas, where airborne chlorides or de-icing
jdλim j ¼ dðimÞ (2) salts are prone to attack the structures. Together, seismic hazard with
dðimÞ
aggressive environment make the overall risk even higher. Among the
representing the mean number of earthquakes per year producing a deterioration processes of RC structures, it is worth noting that
shaking of exactly im; the minus sign is required by the negative slope of chlorides-induced corrosion is the most important one, both because of
the hazard curve. λim is commonly computed via the PSHA integral as: the severity of its effects and the large amount of structures hit by this
Z mmax;i Z rmax;i phenomenon [43,44].
X
nSZ
A lot of effort has been done in the last years by the research com­
λim ¼ νmmin;i P½IM > imjm; r�fMi ðmÞfRi ðrÞ dm dr (3)
i¼1 mmin;i rmin;i munity well describe deterioration phenomena in RC structures, leading
to models as Life-365, CHLODIF, and Duracrete ones [45–47]. The last
where νmmin;i is the rate of occurrence of earthquakes greater than a one, developed within the European project entitled ‘‘Probabilistic
suitable minimum magnitude mmin;i of the ith seismogenic zone (SZ), Performance-Based Durability Design of Concrete Structures’’, gives a
fMi ðmÞ is the magnitude distribution for the ith SZ and fRi ðrÞ is the dis­ full probabilistic method for the calculation of structural service life. The
tribution of the source ith-to-site distance. Finally, P½IM > imjm; r� rep­ fib Bulletin 34 [48] is majorly based on the DuraCrete and DARTS model,
resents the probability to exceed im, for a given combination of which is a revised version of DuraCrete [47,49,50].
magnitude m and distance r, whose distribution parameters are usually Concerning chlorides-induced corrosion, the triggering of this pro­
provided by a suitable assumed Ground Motion Prediction Equation cess is generally modelled with a physical process using Fick’s 2nd law
(GMPE). of diffusion [51], which becomes for the 1D case:
As regards the fragility curve P½fjim�, it represents the probability to
reach and exceed a given damage state level (failure), conditioned on a δCðx; tÞ δ2 Cðx; tÞ
¼ Dc (9)
specific ground motion intensity measure IM ¼ im. The calibration of δt δx2
fragilities is commonly based on results obtained from the execution of a where Cðx; tÞ is the chloride ion concentration at a distance x from the
set of non-linear time-history analyses (NLTHAs). Several theoretical concrete surface after t years and D is the chloride diffusion coefficient.
models and procedures have been proposed in literature for the cali­ The solution to Eq. (9), assuming a constant concentration of chlorides
bration of P½fjim� parameters; among all the Incremental Dynamic near the surface and diffusion coefficient, can be written as [48]:
Analysis [39], the Cloud-Analysis [40], the Multi-Stripes Analysis [41] � � �
and the Incremental Modal Pushover Analysis [42]. The Cloud Analysis x
Cðx; tÞ ¼ Cs 1 erf pffiffiffiffiffiffiffi (10)
approach has proven to be less onerous in computational terms given 2 Dc t
that the resulting fragility curve takes origin from a sample of n ground
motions intensities ½im1 ; im2 ; :::; imn � and the corresponding sample of where Cs is the chloride content at the surface of the concrete and erf is
structural responses ½edp1 ; edp2 ;:::; edpn �. Compared to the IDA method, the Gaussian error function. Then, two successive phases can be easily
the Cloud Analysis approach is less time consuming, since the investi­ distinguished, namely corrosion initiation and corrosion propagation
gated structure is subject to a limited set of n unscaled ground motions phases. The initiation phase is the time interval needed for the loss of the
records. In this case, the fragility function assumes the following form: protective passive film around the reinforcement rebars. In this case, the
initiation phase is considered complete when chlorides concentration at
P½f jim� ¼ P½EDP > edpjim� ¼ 1 P½EDP � edpjim� ¼ 1 the reinforcement surface reaches a critical value Ccrit . A lot of research
� �
lnðedpÞ lnðedpÞ has been done on the determination of Ccrit and different threshold
Φ (4) values have been proposed [52]. In the present work, 0.11% concrete
β
weight has been assumed. The corrosion initiation time can be estimated
In Eq. (4), edp is the median threshold value of the assumed struc­ as:
tural limit state, and edp represents the median estimate of the demand � � �� 2
that can be computed with a ln-linear regression model, as: T¼
x2
erf 1
Cs Ccrit
(11)
4Dc Cs
lnðedpÞ ¼ a þ b⋅lnðimÞ (5)
Once the initiation phase is over, propagation starts. In this phase, it

3
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

is important to determine the velocity of the corrosion process in order 2.3. FRCM modelling
to estimate the steel cross-section loss in time. Usually, chlorides-
induced corrosion is characterized by pitting corrosion, which is local­ There has been an intensive research aimed at defining a suitable
ized and generally faster than carbonation-induced corrosion, which is prediction model for assessing the response of concrete members
instead generally considered as homogenous on reinforcement surface. confined by composite systems, starting from Spoelstra and Monti [59]
The reinforcement diameter reduction can be estimated as: and Lam and Teng [60] ones for FRP-confined concrete. Particularly, the
former is based on the assumption that, given a certain lateral strain, the
DðtÞ ¼ D0 rcorr ðt Ti Þ (12)
relationship between the incremental axial stress and strain is the same
where D0 is the initial diameter of the steel bar, t is the age of the bridge, as that of a concrete element confined by a constant pressure. Such
Ti is the corrosion initiation time and rcorr is the corrosion rate. The pressure can be evaluated starting from the knowledge of the lateral
reduced section area AðtÞ can then be computed as: strain offered by the composite hoop. In practical terms, both the axial
strength fcc and the axial strain εcc are typically expressed as a function of
8 π the confining pressure exerted by the jacketing system, through some
> nD2o for t � Ti
>
>
>
>
4 coefficients that should be calibrated by case:
>
< 2π D0 � �m
AðtÞ ¼ nDðtÞ for Ti < t � Ti þ (13) fcc fl;e
>
> 4 rcorr ¼ 1 þ k1 (17)
>
> fco fco
>
>
: D0
0 for t � Ti þ � �n
rcorr fl;e
εcc ¼ εc0 þ k2 (18)
Even though the corrosion rate is a time-dependent parameter, it is fco
generally assumed to be constant during the service life of a structure. In Eqs. 17 and 18, the ratio fcc/fc0 defines the axial strength gain, with
Vu and Stewart [53] developed a time-dependent corrosion rate model the axial strength being fcc and fc0 is the unconfined concrete compres­
based on some limited marine exposure data reported by Ref. [54]. The sive strength, εcc is the axial strain associated to fcc, fl,e is the effective
rate of corrosion can be estimated based on the corrosion current in­ lateral confining pressure, calculated as in Eq. (19), and lastly k1, k2, m, n
tensity icorr;0 [μA/cm2] at the beginning of the propagation phase, which are the empirical constants.
is estimated as:
ρf ⋅Ef ⋅ke ⋅εf
� � 1:64 fl;e ¼ (19)
37:8 1 wc 2
icorr;0 ¼ (14) Eq. (19) requires the knowledge of fiber elastic modulus Ef, the axial
x
rigidity of the composite ρfEf, ke is the effectiveness coefficient, generally
where wc is the water to cement ratio of the concrete mix, and x is the provided by single authors, and accounting for the variation of the
concrete cover depth. The time-dependent corrosion intensity is then confinement pressure in square and rectangular cross-section specimens
given by the following equation: respect to circular ones, the design tensile strain in the jacket εf and the
confinement reinforcement ratio ρf, computed as:
icorr ðtÞ ¼ 0:85icorr;0 ðt Ti Þ 0:29
(15)
4nf tf
ρf ¼ (20)
where t is the age of the structure and Ti the initiation time. Given icorr ðtÞ D
[μA/cm2], the intensity corrosion rate rcorr [mm/year] is simply ob­
tained by a conversion factor of 0.232 [53]. Since the time-dependent where nf is the number of layers of FRP, tf is the thickness of each layer
corrosion rate model is based on marine exposure data, in the present and D is the diameter of the specimen in case of circular section column,
work a relatively high constant corrosion rate rcorr of 0.254 is assumed as or the diagonal for rectangular/square sections. Similar models have
by Refs. [55]. The corrosion process affects also the strength of rein­ been provided for FRCM-confined specimens, which are based on the
forcement steel [56]. The residual capacity of corroded bars can be previous work carried out to assess the effectiveness of FRP jacketing
estimated by: systems, as thoroughly reviewed by Fossetti el at [61]. Particularly, the
� � recent CNR-DT-215 Guidelines [62] provides the following confinement
0:5ðA0 AðtÞÞ model for RC sections:
fy ðtÞ ¼ 1 fy;0 (16)
A0 � �2=3
fcc fl;e
¼ 1 þ 2:6 (21)
where A0 is the initial area, AðtÞ is the residual area at the time t and fy;0 fco fco
is the initial yielding strength of the bars.
Realistic corrosion modelling is a complicated task, especially in RC where the effective lateral confining pressure fl,e is calculated according
structures. It strongly depends on local phenomena as well as on con­ to Eq. (21), being the effectiveness coefficient ke equal to 1 for specimens
struction characteristics that can promote the development of deterio­ with circular section continuously wrapped, and computed as follows
ration. Corrosion of reinforcing bars may lead also to concrete spalling for a generic rectangular section:
and loss of bond between concrete and reinforcement. Concrete cracking b’2 þ h’2
and spalling can reduce the corrosion initiation time or increase the kH ¼ 1 (22)
3⋅b⋅h
corrosion velocity. On the other hand, loss of bond strength, especially
for elements with insufficient confinement reinforcement [57], may where b’ and h’ are respectively the effective width and height of the
significantly affect structure performance. However, given the lack of specimen section, estimated as follows:
robust models able to capture this and other secondary effects, and in
order to avoid huge computational efforts linked to the implementation b’ ¼ b 2rc (23)
of stochastic approaches aimed to simulate random fields in order to
capture uncertainties in spatial distribution of deterioration phenomena h’ ¼ h 2rc (24)
[58], it was decided to not include them in the following computations,
with rc standing for the corner radius, that has to be at least equal to 20
take advantage of using a uniform corrosion model.
mm, as prescribed by Ref. [56] in order to avoid stress concentrations at
the section edges able to induce a premature failure of the reinforcement

4
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 1. Bridge case study and FRCM seismic retrofitting protocol considered.

system. The computation of the effective lateral confining pressure fl,e As regards the definition of the cyclic behavior of FRCM confined
requires also the definition of the design tensile strain in the jacket εf, concrete elements under cyclic compressive loading, few studies can be
calculated as follows: found in literature. Colajanni et al. [64] studied the behavior of
� � medium-size concrete columns wrapped with Carbon FRCM under
ε
εf ¼ min kmat ⋅ ηa ⋅ fu ; 0:004 (25) monotonic and cyclic compressive loads. The authors came to the
γm
conclusion that the stress-strain curve of FRCM confined concrete col­
umns under monotonic loading is always the same as the envelope curve
where εfu is the ultimate FRCM strain, γm is the material partial safety
of the cyclic stress-strain curve in the pre-peak branch while in the
factor equal to 1.5, ηa characterizes the environmental exposure class
post-peak range when softening behavior is exhibited, the curve for the
ranging between 0.7 for aggressive environmental conditions to 0.9 for
cyclic load is slightly lower than the monotonic one. The strength
indoor spaces and depending on fibers type, and kmat is a dimensionless
degradation for small strains is negligible. As strain increases, strength
coefficient accounting main material properties of the inorganic matrix:
degradation tends to increase for both specimens with medium and high
� �3=2
4⋅nf ⋅tmat ⋅fc;mat lateral confinement pressure. Failure of FRCM confined columns, for
kmat ¼ 0:217⋅ �1 (26) both monotonic and cyclic loads, is less abrupt compared with that of
D⋅fcd
FRP confined columns due to the slowly progressing fracture of indi­
In Eq. (26), nf is the number of layers of fiber reinforcement, tmat and vidual fiber bundles which is in accordance with earlier studies. Taking
fc,mat are the inorganic matrix thickness and characteristic compressive into account the above considerations from previous research and
strength, D is the diameter of the circular section, or alternatively the considering that relevant progress is yet to be made in this topic, the
diagonal dimension for square/rectangular sections, and fcd is the cy­ robust and well-acknowledged Mander model [65] for the concrete
lindrical design compressive strength of the unconfined concrete material was adopted in this paper to model the cyclic behavior of FRCM
specimen. confined concrete. The backbone monotonic stress-strain curve for the
The ultimate axial compressive strain of confined concrete εccu is model was thus computed in accordance with CNR-DT-215 and
instead computed following the ACI549.4R-13 Guideline for design and ACI549.4R-13 Guidelines [62,63].
construction of externally bonded FRCM systems [63] as:

f
� �0:45 �
ε 3. Case study application
εccu ¼ εc 1:5 þ 12Kb l;e’ fe’ (27)
fc εc
In this section, a detailed description of the bridge case study as well
where εc is the compressive strain level in the concrete, ε’c is the as the methodology adopted for the probabilistic characterization of its
seismic behavior via the construction of a set of fragility curves, is
compressive strain of unconfined concrete corresponding to fc’ , with an
provided. The adopted structure is a multi-span simply supported bridge
upper bound limitation to 0.01 prescribed in order to prevent excessive
built in the municipality of Tolmezzo, northeastern Italy. The bridge was
cracking and the resulting loss of concrete integrity, and Kb is an effi­
designed according to the old Italian building code [66], considering an
ciency factor, equal to 1 for circular specimens, and computed for
equivalent static horizontal force of about 10% of its gravitational load.
rectangular sections as follows:
The bridge is composed by two spans of 24 m length each, for a total
� �2
Ae h length of 48 m. In each span, six prestressed reinforced concrete beams
Kb ¼ (28)
Ac b with I-shaped cross section of 1.2 m height support the concrete deck
(12 m wide and 20 cm thick). A pair of neoprene bearings of 3-cm
with thickness, with a squared size of 30-cm side, interposed between the
�� � �� � deck and the pier system sustains each beam. The decks are supported at
b
h ðh 2rc Þ2 þ h
b ðb 2rc Þ2 the mid length of the bridge by a reinforced concrete frame pier with two
Ae 1 ρg square columns of 1 m side each and height of 5.5 m, whereas the
(29)
3Ag
Ac
¼
1 ρg foundational system consists in nine piles for each column. Each RC
column of the frame pier is reinforced with 36 longitudinal steel rebars
where Ag and Ac are respectively the gross and the net cross-sectional of 20 mm diameter uniformly spaced, and with 4-legs stirrups of 10 mm
area of the specimen, Ae is the effectively confined concrete, and ρg diameter, 250 mm spacing and a concrete cover equal to 40 mm. Fig. 1
stands for the ratio of the area of longitudinal steel reinforcement (As) to shows main geometrical features of the case study. As regards materials,
the gross cross-sectional area Ag of the specimen. concrete has a characteristic compressive cubic strength of fck ¼ 25 MPa
(C25/30), with an elastic modulus equal to Ec ¼ 30472 MPa, while a

5
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 2. Adopted bridge FE model strategy.

FeB44k steel type has been assumed for both longitudinal and trans­ following steps:
versal reinforcing bars, with characteristic tensile strength of fyk ¼ 430
MPa and an elastic modulus equal to Es ¼ 210000 MPa. The bridge was a) concrete surface damping, to homogeneously hydrate the concrete
tested with a set of bi-directional time-history ground shakings in its “as support;
built” configuration (i.e. at the age of construction) and considering b) application of a first layer of mortar onto the surface, with an average
several combinations of deterioration and seismic retrofitting scenarios. thickness of 3 mm;
As regards the seismic retrofitting protocol, integral wrapping of RC c) placement of the first layer of fiber onto the mortar surface, pushed
columns with carbon FRCM was considered as suitable solution to to better adhere with the matrix;
enhance bridge seismic reliability, taking advantage of the additional d) application of a new layer of mortar onto the carbon fiber.
confinement effect given by the intervention, as also suggested in other
practical applications (e.g. Ref. [67]). Fig. 1 schematically represents the Steps c) and d) have to be repeated n-1 times in order to apply n
considered seismic retrofitting protocol, showing also the strengthening layers of FRCM. The retrofitting protocol can be integrated in the Step a)
procedure to be used, as detailed in recent literature research [68–71]. with other work phases aiming to remove deteriorated portions of
In detail, the application procedure can briefly described with the concrete cover, and cleaning and application of protective coatings to

Table 1
Main modelling parameters.
Element type Direction Relevant stiffness/force/displacement values

Neoprene pad ke [kN/mm] kp [kN/mm] Fy [kN]


Longitudinal 4.5 0 183.3
Transversal 4.5 0 183.3
Gap k1 [kN/mm] k2 [kN/mm] Δ1 [mm] Δ2 [mm]
Longitudinal 1102 374.7 25 27.5
Soil k1p [kN/mm] k2p [kN/mm] k3p [kN/mm] Δ1p [mm] Δ2p [mm] Δ3p [mm]
Longitudinal 21.6 10.8 3.2 8 28 80
Abutment piles ke [kN/mm] kp [kN/mm] Δ1 [mm] Δ2 [mm]
Longitudinal 20.4 3.7 7.6 25.4
Transversal 20.4 3.7 7.6 25.4
Foundations ke [kN/mm] ke [kN/mm] kr [kN*mm/rad]
Longitudinal 56 Transversal 56 Rotational 609000000

6
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Table 2
Selected 2-D earthquake records.
Record name Record ID Date Mw PGAx (cm/s2) PGAy (cm/s2)

1.L’Aquila Mainshock IT0792 April 06, 2009 6.3 644 535


2.L’Aquila Mainshock IT0789 April 06, 2009 6.3 395 434
3.South Iceland (aftershock) 006334 June 21, 2000 6.4 411 373
4.L’Aquila Mainshock IT0790 April 06, 2009 6.3 437 479
5.Northern Italy IT0049 June 17, 1976 4.5 71 82
6.Friuli IT0077 September 11, 1976 5.8 229 127
7.Southern Italy IT0231 January 16, 1981 5.2 107 67
8.Umbria-Marche 3rd shock IT0491 October 14, 1997 5.6 43 37
9.Garfagnana IT0157 June 07, 1980 4.6 35 59
10.App. Lucano IT0607 September 09, 1998 5.6 40 33
11. South Iceland (aftershock) 006349 June 21, 2000 6.4 805 729
12. Firuzabad 007162 May 19, 1995 5.2 1000 1010
13. Gazli 000074 May 17, 1976 6.7 707 335
14. Erzincan 000535 March 13, 1992 6.6 478 403
15. South Iceland 006277 June 17, 2000 6.5 278 509
16. Racha (aftershock) 000501 May 03, 1991 5.6 494 348
17. Pyrgos 000558 March 26, 1993 5.4 165 432
18. Kalamata (aftershock) 000419 September 15, 1986 4.9 328 150
19. NE of Banja Luka 005651 August 13, 1981 5.7 355 274
20. Ionian 006131 April 24, 1988 4.8 86 273
21. Kozani (aftershock) 006093 May 19, 1995 5.2 185 262
22. Faial 007329 July 09, 1998 6.1 362 397
23. Oelfus 005030 November 13, 1998 5.1 144 135
24. Mt. Hengill Area 005149 August 24, 1997 4.9 170 100
25. Mouzakaiika 000566 June 13, 1993 5.3 44 145
26. Mouzakaiika 000566 June 13, 1993 5.3 168 174
27. Kremidia (aftershock) 002025 October 25, 1984 5.0 152 187
28. Friuli (aftershock) 000707 September 11, 1976 5.3 73 78
29. Izmit (aftershock) 006440 November 07, 1999 4.9 310 346
30. Strait of Gibraltar 000878 January 04, 1994 4.9 24 60

break off further development of corrosion on the internal model was implemented in Midas Civil software [72] to properly char­
reinforcement. acterize main structural features needed for the following structural
A preliminary modal analysis was developed to derive main dynamic seismic reliability analysis. As regards RC pier columns, distributed
features of the bridge. Main vibration modes have periods equal to plasticity fiber-section elements were used, whereas elastic elements
0.689 s in longitudinal direction (participating mass equal to 98.15%), were adopted for modelling both the pier cap beam and the decks,
0.644 s in transversal direction (participating mass equal to 95.35%) and considering a unique element with equivalent elastic properties. Sig­
0.533 s with torsional direction around the vertical axis (participating nificant efforts were spent in detailing FE model in proximity to the joint
mass equal to 81.25%). between two spans, at the abutments as well as the foundational system.
A set of rigid links was implemented in order to connect the equivalent
3.1. Fragility analysis deck beam to each of the six bearing zones, both at the abutment and on
the pier cap. Fig. 2 shows a sketch where more detailed information is
Seismic fragility analysis was conducted performing a set of non- visible on the three main modelling details.
linear time history analyses (NLTHAs) on a 3-D finite element (FE) The abutment consisted in a set of series/parallels bi- and tri-linear
model of the analyzed two-span simply supported bridge. The 3-D FE links able to capture the resulting non-linear behavior due to the

Fig. 3. Response spectra of the selected seismic records.

7
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 4. Cloud analysis regression (left) and resulting fragility curves (right) for both ductile and shear failure mechanisms – “as built” configuration.

combined abutment-soil interaction, abutment piles contribution, as modelling detail regards the columns foundational system, modelled as
well as the impact between deck and abutment (i.e. gap) and the pres­ a set of two linear springs for capturing both longitudinal and trans­
ence of neoprene pads. In particular, in the longitudinal direction, pas­ versal stiffness contributions, and coupled with a rotational spring.
sive soil effect (i.e. soil that is pressurized by the abutment) was Lastly, the third modelling detail was related to the deck-to-pier node
modelled with a tri-linear behavior, while the active stiffness was not connection, where two neoprene pads are present (one supporting left-
accounted in order to represent the separation between soil and abut­ side span beam, one the right-side span beam), coupled in series with
ment: this component was neglected in transversal direction, reasonably a gap element able to capture potential pounding phenomena between
assuming that the abutment piles provided all the stiffness. Potential the two adjacent spans. Table 1 lists in detail all the assumed modelling
pounding between the abutment and deck was accounted with the gap parameters.
bilinear link. As regards the neoprene pads, they were modelled as As regards RC columns, fiber sections were discretized into steel and
bilinear links, considering two node link elements in order to combine concrete fibers (confined and cover concrete) and modelled using the
both longitudinal and transverse behavior of the bearing. The second Mander concrete model [65], while the Menegotto-Pinto steel model

8
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

where μds;i is the relative DS threshold and ki is a factor that considers the
admissible plastic displacement in each DS, with ki ¼ 0, 0.3, 0.6, 0.8 and
1.
The yielding and ultimate displacement capacity values (μy ; ​ μu Þ
were been derived performing a set of pushover analyses at 5� horizontal
inclination steps, in order to characterize the displacement thresholds as
a function of the horizontal inclination of the ground shaking. It should
be noted that the ultimate displacement capacity values were derived
from the capacity curves at a drop of 15% of the peak value in the
pushover curve. As regards the definition of the damage states, the
abovementioned four DSs were herein considered, as commonly done in
scientific literature (e.g. Ref. [76,77]), and deriving their respective
time-variant DS thresholds in accordance with [78].
60 NLTHAs were performed applying the selected 30 ground motions
Fig. 5. System fragility curve - “as built” configuration. in X-Y and Y-X directions respectively to account for the intrinsic un­
certainty on the directionality of the ground shaking, thus deriving for
[73] was adopted for both longitudinal bars and transversal stirrups. As each analysis a maximum horizontal displacement resultant. The DS
regards potential reinforcement buckling phenomena that can affect displacement threshold, for each deteriorated and retrofitted scenario, is
corroded RC elements subject to cyclic loads, even if some refined cyclic not constant but an angle-variant value, so each threshold was derived
models have been recently proposed in literature [44,74], it was herein from the pushover capacity curve carried out in same angle in which the
assumed to not consider such issue in order avoid a too onerous model maximum displacement resultant was recorded. Based on such data, a
detailing and at the same time allow a comparability of results between kinematic displacement index was considered as relevant engineering
the investigated deterioration and retrofitting scenarios, given that no demand parameter and computed for each NLTHA as the ratio between
literature evidence is still currently available on the effectiveness of maximum displacement μ recorded in each analysis and the displace­
FRCM confinement in avoiding bar bucking phenomena. ment threshold defined for each damage state μDsi . Hence fragility curves
A set of 30 natural seismic records was collected from the European were derived via the use of the Cloud Analysis method [40], with
Strong Motion Database [75] selected in order to be consistent with reference to the abovementioned damage states, based on the regression
disaggregation results obtained from PSHA, and further used in the models calibrated on the basis of the cloud of points ln(im) – ln (μ=μDsi )
execution of the NLTHAs. Table 2 lists main information of each record reported in a bi-logarithmic plane. In addition, for shear failure mech­
providing also magnitude, X- and Y- horizontal peak ground accelera­ anism a single fragility curve, which coincides with the collapse damage
tion (PGA) values. Fig. 3 shows response spectra of the horizontal state, was computed. In this case, the ratio of maximum shear at the base
components of the selected seismic records, with the mean response of the column VSd and the shear strength of the element VRd was
spectrum compared with the uniform hazard spectrum of the site of considered. Fig. 4 shows respectively results of Cloud Analysis regres­
interest, highlighting a good spectral matching. The vertical component sion data and the resulting set of four fragility curves for the ductile
of the ground motions was not considered in the present case study. failure and the one related to the shear one, with the resulting mean μ
and standard deviation σ values.
3.1.1. Damage state considerations Lastly, considering as relevant condition for a structural assessment
NLTHAs were subsequently run extracting the maximum 2-D hori­ the Ultimate Limit State (ULS), Complete DS and shear fragility curves
zontal displacement of the control point, which was set at the top of the considered to take into account both ductile and shear failure mecha­
pier columns. To compute fragility curves, damage state thresholds have nisms, were condensed in a unique ULS fragility curve via system reli­
to be defined. Generally, in bridge fragility analysis the so-called kine­ ability theory. Fig. 5 shows the resulting system fragility curve.
matic ductility index μ, as defined by Choi [76], is used. The kinematic
ductility index μ is defined as the ratio between the ultimate horizontal 4. Results
displacement at the top of the pier Δu , and the horizontal displacement
of the same point corresponding to the steel yielding at the pier-base In this section, results obtained from 16 different deterioration and
cross-section Δy . According to Choi [77], reference kinematic ductility FRCM retrofitting scenarios listed in Table 3, are illustrated.
index threshold values are μdsi ¼ 1, 2, 4, 7, respectively for Damage State
(DS) i ¼ 1, 2, 3 and 4, namely Slight, Moderate, Extensive and Complete
damage. While kinematic ductility index well describes damage states in 4.1. Deterioration scenarios
standard cases, it lacks in getting the time-variant capacity for deterio­
rated cases and capacity gains when retrofitting scenarios are run. As a Based on bridge location, a highly corrosive condition due to de-icing
matter of fact, yielding displacement, the unique parameter that defines salts was considered as discussed in Section 2.2 and used for estimating
kinematic ductility index, is little affected when retrofitting the element its effects on reinforcement strength and cross-section losses. In partic­
by FRCM confinement, while the ultimate displacement capacity is ular, given the different concrete cover values of the stirrups and lon­
conversely highly enhanced. Cheng et al. [78] studied seismic fragilities gitudinal bars, the corrosion initiation time was estimated in 13 and 28
for RC bridge columns affected by corrosion considering a time-variant years, respectively. Accordingly, four deterioration scenarios were
capacity index. The authors suggested that the ultimate top displace­ considered, i.e. 20-, 30-, 40- and 50-years, in addition to the above­
ment can be assumed as the sum of the yielding displacement and the mentioned “as-built” configuration, corresponding to the non-
succeeding plastic displacement, as: deteriorated structure (i.e. t ¼ 0 years). NLTHAs were rerun consid­
ering the same 60 ground shaking time histories, damage states and
μu ¼ μy þ μp (30) failure mechanisms (i.e. ductile and fragile) and monitoring the same
Five different displacement and drift thresholds were then defined to engineering demand parameters described in Section 3.1.1. Fig. 6 shows
describe five different damage limit states as: the yielding and ultimate displacement capacity radial diagram derived
performing a set of pushover analyses at 5� horizontal inclination steps
μDs;i ¼ μy þ ki μp (31) for each deterioration scenario, and further used for the computation of
the time-variant fragilities.

9
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Table 3
Deterioration and retrofitting scenarios.

10
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 6. Time-variant radial yielding (a) and ultimate capacity (b) displacement values [in cm] – deterioration scenarios.

Fig. 7 shows the fragility curves obtained for the ductile mechanism.
It can be seen that the structure becomes more fragile while aging. For
the first 30 years, the differences in the fragility are mainly due to the
loss of the confining reinforcement which affects concrete strength and
ultimate strain values. In the 40- and 50-years scenarios, longitudinal
bars are also deteriorated while the stirrups contribution to the concrete
confinement becomes almost negligible, leading to increasing fragilities.
The Slight DS for the first three scenarios remains almost the same, since
it is determined more from the yielding strength of the longitudinal bars
than from concrete strength reduction.
Fig. 8 portraits the time-variant fragilities obtained for the shear
failure mechanism, showing resulting curves that get lower as the bridge
ages and the deterioration goes on. Even though this result may seem
contradictory, it is mainly due to the fact that the analyzed bridge lacks
all the code detailing and limitations that recent building codes provide.
This means that the shear resistance of the pier elements is guaranteed
by the concrete element without considering shear reinforcement. Since
the effects of the deterioration mechanism considered in this paper Fig. 8. Time-variant fragilities for the deteriorated bridge – shear fail­
ure mechanism.

Fig. 7. Time-variant fragilities for the deteriorated bridge – ductile failure mechanism.

11
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

4.2. Retrofitting scenarios – fragilities

The effectiveness of four retrofitting scenarios, confining the pier


elements using 2, 4, 6 and 8 FRCM layers and applied at different time
instant was investigated in the present study. A carbon fabric FRCM
solution was considered, characterized by a nominal thickness of 0.047
mm, tensile strength of 4700 MPa and elastic modulus of 240 GPa. The
average thickness of the jacket in case of 2-layers solution was about of
8 mm, reaching an overall thickness around 30 mm for the 8-layers
strengthening protocol. The overlapping length of the fiber was set
equal to 30 cm, whereas the corner radius was assumed equal to 50 mm
in order to avoid stress concentrations at the section edges. Four
different retrofitting scenarios were characterized by increasing fiber
reinforcement ratios respectively equal to 0.027%, 0.053%, 0.08% and
0.106%. Such retrofitting strategy was selected also because it is
intended to provide additional confinement in order to balance that
Fig. 9. Time-variant system fragilities for the deteriorated bridge. originally provided by steel stirrups and lost due to corrosion phenom­
ena: deterioration scenarios have in fact highlighted how stirrups are
mainly regards reinforcement bars (the effects on concrete itself are early subject to corrosion if compared to longitudinal reinforcement due
mostly negligible until concrete cover is not spalled), the shear resis­ to their lower concrete cover, showing at the end of the considered
tance of the piers remains the same during its service life. On the other service life remaining cross-sections almost negligible, whereas for the
hand, the moment yielding point decreases in time. This means that also longitudinal bars values almost equal to 75% of the initial area.
the maximum shear acting in the piers decreases in time, hence the ratio The contribution provided by the retrofitting system was assessed by
VSd=V , considered in the fragility curves, decreases in time. the scientific models currently embodied in the Italian CNR-DT-215
Rd
Hence, based on the previous results, Fig. 9 reports the time-variant guidelines [62]. Fig. 10 presents the time-variant radial diagrams of
ULS system fragilities derived according to system reliability theory for the ultimate displacement capacity computed for each retrofitting sce­
the considered deterioration scenarios. nario respectively at 20, 30, 40 and 50 years, whereas Figs. 11–14 show

Fig. 10. Time-variant radial ultimate displacement capacity [in cm] – retrofit scenarios at 20 (a), 30 (b), 40 (c) and 50 years (d).

12
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 11. 20-years old bridge: fragilities for different FRCM retrofitting scenarios – ductile failure mechanism.

the resulting fragility curves for the ductile failure mechanism. assumed interventions would give a major contribution with respect to
It is clearly shown that the number of layers used in the reinforce­ the concrete-only shear resistance, just in the case of retrofitting with #6
ment highly affects the fragility of the structure: in particular, the #8 and #8 FRCM layers, for the same reasons described in x4.1 when
layer scenario is by far the best performing case, especially for extensive looking at deterioration scenarios outcomes.
and complete DSs. It can be observed also how significant improvements Also in these cases, ULS system fragilities have been derived ac­
can be achieved independently by the time at which the intervention is cording to system reliability theory and shown in Fig. 16 for the
made. considered FRCM retrofitting scenarios implemented at different time
As regards shear failure mechanism fragilities, Fig. 15 shows how the instants.

Fig. 12. 30-years old bridge: fragilities for different FRCM retrofitting scenarios – ductile failure mechanism.

13
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 13. 40-years old bridge: fragilities for different FRCM retrofitting scenarios – ductile failure mechanism.

5. Discussion less differences than when comparing varying intervention schemes


carried out at the same time instant.
While it appears obvious the effect of the number of confining layers Only the #8 layers retrofitting scenario displays a different trend:
of the columns to reduce the fragilities, the same thing cannot be there, the intervention at 40 years results almost as fragile as the one at
immediately stated when looking at the influence of the age at which the 20–30 years of service life. Such result can be explained because at 40
retrofitting is done. Figs. 17–20 compare fragility curves related to the years the cross-section reduction of the longitudinal is still not so rele­
ductile failure mechanisms for the same retrofitting scenario carried out vant, with few effects on the yielding moment of the pier. However, the
at different time instants. It can be noted how the fragility of late ret­ section benefits by greater ductility (Fig. 10) and therefore it presents
rofitting scenarios are higher than early ones, even though they display greater ultimate displacement capacity. Hence, this particular case

Fig. 14. 50-years old bridge: fragilities for different FRCM retrofitting scenarios – ductile failure mechanism.

14
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 15. Fragilities for different FRCM retrofitting scenarios – shear failure mechanism at 20 (a), 30 (b), 40 (b) and 50 years (d).

seems showing in such a way a sort of balancing between strength loss Fig. 22b portraits the time-variant seismic reliability profile and
and ductility gain. gains linked to the investigated retrofitting scenarios with reference to
Analogous considerations can be done looking at the shear failure the shear failure mode. Conversely to the ductile failure mechanism, it is
mechanism (Fig. 21): it can be noticed that the highest fragilities, almost worth nothing in this case how such profile shows an overall increasing
for every retrofitting scenario, are those related to the retrofitting sce­ trend over time, with an almost constant profile up to 30 years, followed
narios carried out at the age of 30 years. by a more pronounced increasing linear trend up to 50 years. Such
This outcome can be justified by the fact that at this time instant, the seemingly unexpected development is a consequence of the corrosion
analyzed aging bridge has still almost all its longitudinal bars but the initiation in longitudinal bars implying a progressive reduction of the
confinement effect of the stirrups is very reduced. The yielding moment pier bending moment resistance that, as previously stated, in turn results
of the pier at 30 years, and therefore the maximum shear acting on the in a reduction of its maximum allowable shear action. As regards the
pier, is definitely greater than the yielding moment at 40 or 50 years. On effectiveness of the retrofitting protocols with respect to shear failure
the other hand, shear resistance, calculated as the contribution of steel mechanism, both #2 and #4 layers retrofitting scenarios seem to have a
reinforcement and fiber reinforcement, results in almost the same values negligible influence on seismic reliability index estimated after the
at 30–50 years, since the contribution of steel reinforcement is almost upgrading. On the other hand, both #6 and #8 layer retrofitting pro­
null. This makes the shear fragility, for retrofitting scenarios at the age of tocol exhibit an initial decreasing trend up to 30 years, mainly attrib­
30, the highest among than the other considered cases. utable to the loss of transversal steel reinforcement, followed by an
opposite increasing slope with almost proportional gains up to the end of
service life reasonably attributable to the concurrent corrosion initiation
5.1. Time-variant seismic reliability profiles and propagation along the longitudinal rebars.
Based on these failure mechanism outcomes, Fig. 22c illustrates the
Finally, a seismic reliability analysis was carried out in order to es­ time-variant seismic reliability profile and gains linked to the investi­
timate the gains in seismic reliability achievable when implementing at gated retrofitting scenarios with reference to system reliability. The
different time instants one of the proposed FRCM retrofitting scenarios. reliability profile for the “as built” configuration is characterized as
The seismic hazard curve for the site of interest was derived from the expected by lower seismic reliability indexes, with a value of 2.92 for the
MPS04 hazard maps [79]. One-year seismic reliability indexes βE;1 were configuration not affected by deterioration and a monotonically
computed as shown in Section 2.1 for the “as-built” configuration as well decreasing profile down to 2.83 at the end of bridge service life. Reli­
as for each of the investigated deterioration and FRCM retrofitting sce­ ability gains associated to each retrofitting protocol are more hetero­
narios. Fig. 22 shows the resulting seismic reliability profile during the geneous, as result of the combination of both ductile and shear failure
service life of the bridge and reliability gains achievable at different time mechanisms.
instants with the implementation of each FRCM retrofitting scenario. Finally, a simplified cost-benefit analysis was carried out in order to
As regards the ductile failure mechanism (Fig. 22a), the time-variant derive some graphical outputs able to support decision making of
seismic reliability profile is characterized by a decreasing trend in time, infrastructure owners dealing with seismic retrofitting optimal sched­
and with gains almost proportional to the number of layers considered in uling for the specific case-study herein analyzed. Both replacement and
the #2, #4 and #6 layers retrofitting scenarios. On the other hand, the retrofitting costs were analytically estimated considering regional unit
implementation of #8 layer retrofitting protocol results in seismic reli­ price lists. A unit replacement cost of about 10 300 €/m2 was assumed,
ability indexes almost constant with values around 3.35 until 40 years, thus resulting for the 576 m2 deck in a total bridge replacement cost of
whereas a significant reduction of its effectiveness is observed if 7480 800 €. Retrofitting costs were computed for each of the investigated
implemented at 50 years.

15
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 16. System fragilities for different FRCM retrofitting scenarios at 20, 30, 40 and 50 years.

Fig. 17. #2 FRCM layers retrofitting scenario: time-variant fragilities – ductile failure mechanism.

protocols, with a relative incidence on the total replacement cost (i.e. mechanisms as well as to system reliability (c). Results show ΔβE;1
retrofit/replacement cost ratio) of 2.48%, 4.68%, 6.92% and 9.21% for almost always positive, showing higher improvements in terms of
the #2, #4, #6 and #8 layers retrofitting scenarios, respectively. As structural safety for the retrofitting scenarios characterized by the
regards the effectiveness of each retrofitting protocol, the benefit was highest fiber reinforcement ratios and their earlier implementation.
estimated in terms of percentage variation of the seismic reliability As regards ductile failure mechanism (Fig. 23a), the distinction be­
index with respect to the “as built” scenario, ΔβE;1 . Fig. 23 portraits tween the considered time instants is more pronounced for the #2, #4
dimensionless cost-benefit analysis outcomes for each investigated and #6 layers protocols, whereas similar benefits are achieved for the
configuration, with reference to both ductile (a) and shear (b) failure #8 layers solution up to the 40 years of service life. Results show also

16
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 18. #4 FRCM layers retrofitting scenario: time-variant fragilities – ductile failure mechanism.

Fig. 19. #6 FRCM layers retrofitting scenario: time-variant fragilities – ductile failure mechanism.

how the #2, layers protocol is not sufficient to equalize “as built” seismic #6 and #8 layers solutions.
reliability if implemented at 40 years. Fig. 23b highlights an opposite Lastly, Fig. 23c portraits dimensionless cost-benefit analysis results
trend for shear failure mechanism, where the effectiveness of the ret­ with reference to system reliability, to be considered as reference case in
rofitting protocols is more pronounced as the bridge ages, with a the context of a decision-making process. Results show how a slightly
maximum value of ΔβE;1 around 15% for the #8 layers solution at 50 less proportional linear relationship exist between and cost and benefits,
years. Results show also how for this failure mode, the effectiveness of that tends to be more parabolic-like as the bridge ages. A maximum
the #2 and #4 layers protocols is negligible, whereas a more signifi­ benefit ΔβE;1 of 10% can in fact achieved with an investment charac­
cantly marked improvement in safety margins can be achieved with the terized by a retrofit/replacement cost ratio of 9.21% (i.e. #8 layers

17
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 20. #8 FRCM layers retrofitting scenario: time-variant fragilities – ductile failure mechanism.

Fig. 21. #2, #4, #6, #8 FRCM layers retrofitting scenarios: time-variant fragilities – shear failure mechanism.

protocol) if it is implemented at 20 years, whereas decreasing benefits layers at 50 year” option.


can be obtained if such intervention is postponed over time with a
minimum benefit of about 7.5% for the 50 years scheduling. In addition, 6. Conclusions
the use of dimensionless cost-benefit analysis results allow to make a
comparison of the effectiveness of various retrofitting strategies imple­ The present work analyzed the effectiveness of pier confinement
mented at different time instants: for sake of example, the solution “#2 with FRCM as seismic retrofitting strategy for an existing MSSS RC
layers at 20 years” has almost the same effectiveness of the one “#4 bridge, with the aim to quantify benefits achievable from the imple­
layers at 50 year”, if the latter is anticipated at 20 years (i.e. “#4 layers mentation of a set of retrofitting schemes at different time instants.
at 50 year”) results in slightly lower benefits than those linked to a “#6 Several deterioration scenarios were investigated in order to simulate

18
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

Fig. 22. Seismic reliability profile during the service life of the bridge and gains related to the implementation of different FRCM retrofitting scenarios: ductile failure
mechanism (a), shear failure mechanism (b) and system reliability (c).

Fig. 23. Dimensionless cost-benefit analysis results for different FRCM retrofitting scenarios: ductile failure mechanism (a), shear failure mechanism (b) and system
reliability (c).

the decay of seismic safety margins due to the development of deterio­ estimate the effectiveness of each retrofitting solution when imple­
ration phenomena inducing corrosion on steel stirrups and longitudinal mented at a certain time instant and thus support the decision-making
rebars. For each deterioration scenario, four retrofitting schemes con­ process of infrastructure owners, who have to schedule over time
sisting in a growing number of FRCM layers were taken into account, seismic retrofitting interventions and therefore need to know how each
quantifying their associated improvement in terms of seismic reliability. solution can be effective in upgrading structural safety margins against
Analyses were carried out accounting for bidirectional effects of ground seismic actions.
shaking via the characterization of time-variant radial ultimate The present work has therefore the merit to illustrate the general
displacement capacity diagrams able to account differences in non- approach that has to be pursued in order to derive time-variant reli­
linear behavior of the investigated structure when subjected to ground ability profiles and dimensionless cost-benefit charts, and for such
shaking with various angles of incidence. Two failure modes (i.e. ductile reason it has to be intended to be an exhaustive reference study. On the
and fragile ones) were accounted in the development of time-variant other hand, it has to be underlined how results are strictly linked to the
seismic fragilities and coupled with system reliability theory. In partic­ modelling assumptions, deterioration and retrofitting models, time in­
ular, it was illustrated how shear failure modes can significantly impact stants and retrofitting protocols investigated, as well as to the specific
on seismic reliability estimates, thus reducing safety margins and better case-study for what concerns to the structural type, geometry, materials
capturing a more realistic structural behavior. The time-variant seismic and seismicity. Future research will be oriented at filling these gaps,
reliability profile was first derived showing an overall worsening of the particularly with the adoption of more refined deterioration models
seismic reliability index due to the development of deterioration phe­ aimed at capturing bucking phenomena and rebars slippage, as well as
nomena, and gains achievable via the implementation of the considered on issues related to the interaction between corroded existing steel
retrofitting protocols at different time instants were estimated. Lastly, a reinforcement and newly applied external fiber reinforcement compos­
cost-benefit analysis was carried out in order to derive dimensionless ites. Lastly, a wider range of seismic retrofitting interventions, envi­
cost-benefit diagrams highlighting how they can be used in order to ronmental scenarios, structural types and sites with different seismic

19
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

hazard levels will be investigated. [25] Fakharifar M, Chen G, Sneed L, Dalvand A. Seismic performance of post-mainshock
FRP/steel repaired RC bridge columns subjected to aftershocks. Composites Part B
2015;72:183–98.
CRediT authorship contribution statement [26] Parghi A, Shahria Alam M. Seismic behavior of deficient reinforced concrete bridge
piers confined with FRP – a fractional factorial analysis. Eng Struct 2016;126:
Mariano Angelo Zanini: Conceptualization, Methodology, Formal 531–46.
[27] Parghi A, Shahria Alam M. Seismic collapse assessment of non-seismically designed
analysis, Validation. Klajdi Toska: Methodology, Software. Flora circular RC bridge piers retrofitted with FRP composites. Compos Struct 2017;160:
Faleschini: Conceptualization, Validation, Formal analysis. Carlo Pel­ 910–6.
legrino: Supervision. [28] Mahboubi S, Shiravand MR. Failure assessment of skew RC bridges with FRP piers
based on damage indices. Eng Fail Anal 2019;99:153–68.
[29] Triantafillou TC, Papanicolaou CG. Shear strengthening of reinforced concrete
Appendix A. Supplementary data members with textile reinforced mortar (TRM) jackets. Mater Struct 2006;39(1):
93–103.
[30] Bisby LA, Stratford TJ, Smith J, Halpin S. FRP versus Fibre Reinforced
Supplementary data to this article can be found online at https://doi. Cementitious Mortar Systems at Elevated Temperature. Structural Faults þ Repair
org/10.1016/j.soildyn.2020.106224. 2010. 201013pp. Edinburgh, Scotland, 15th-17th June.
[31] Bisby LA, Roy RC, Ward M, Stratford TJ. Fibre reinforced cementitious matrix
systems for fire-safe flexural strengthening of concrete: pilot testing at ambient
References temperature. In: 4th Int. Conf. Advanced Composites in Construction, Edinburgh,
1st-3rd September; 2010. p. 449–60.
[1] Han Q, Du X, Liu J, Li Z, Li L, Zhao J. Seismic damage of highway bridges during [32] Bournas DA, Lontou PV, Papanicolaou CG, Triantafillou TC. Textile-reinforced
the 2008 Wenchuan earthquake. Earthq Eng Eng Vib 2009;8(2):263–73. mortar versus fibre-reinforced polymer confinement in reinforced concrete
[2] Ab�e M, Shimamura M. Performance of railway bridges during the 2011 Tohoku columns. ACI Struct J 2007;104(6):740.
earthquake. J Perform Constr Facil ASCE 2014;28(1):13–23. [33] Bournas DA, Triantafillou TC, Zygouris K, Stavropoulos F. Textile-reinforced
[3] Camata G, Spacone E, Zarnic R. Experimental and nonlinear finite element studies mortar versus FRP jacketing in seismic retrofitting of RC columns with continuous
of RC beams strengthened with FRP plates. Composites Part B 2007;38(2):277–88. or lap-spliced deformed bars. J Compos Construct 2009;13(5):360–71.
[4] Al-Sunna R, Pilakoutas K, Hajirasouliha I, Guadagnini M. Deflection behaviour of [34] Bournas DA, Triantafillou TC. Flexural strengthening of RC columns with NSM FRP
FRP reinforced concrete beams and slabs: an experimental investigation. or stainless steel. ACI Struct J 2009;106(4):495–505.
Composites Part B 2012;43:2125–34. [35] Bournas DA, Triantafillou TC. Biaxial bending of reinforced concrete columns
[5] Ozbakkaloglu T, Lim JC. Axial compressive behavior of FRP-confined concrete: strengthened with externally applied reinforcement in combination with
experimental test database and a new design-oriented model. Composites Part B confinement. ACI Struct J 2013;110(2):193–204.
2013;55:607–34. [36] Cornell CA, Krawinkler H. Progress and challenges in seismic performance
[6] Acciai A, D’Ambrisi A, De Stefano M, Feo L, Focacci F, Nudo R. Experimental assessment. PEER Cent News 2000;3(2):1–3.
response of FRP reinforced members without transverse reinforcement: failure [37] Cornell C. Engineering seismic risk analysis. Bull Seismol Soc Am 1968;58(5):
modes and design issues. Composites Part B 2016;89:397–407. 1583–606.
[7] De Diego A, Arteaga A, Fernandez J. Strengthening of square columns with [38] McGuire RK. Probabilistic seismic hazard analysis and design earthquakes: closing
composite materials. Investigation on the FRP jacket ultimate strain. Composites the loop. Bull Seismol Soc Am 1995;85(5):1275-1284.
Part B 2016;162:454–60. [39] Vamvatsikos D, Cornell CA. Applied incremental dynamic analysis. Earthq Spectra
[8] Del Zoppo M, Di Ludovico M, Balsamo A, Prota A, Manfredi G. FRP for seismic 2004;20(2):523–53.
strengthening of shear controlled RC columns: experience from earthquakes and [40] Jalayer F, Cornell CA. Direct probabilistic seismic analysis: implementing non-
experimental analysis. Composites Part B 2017;129:47–57. linear dynamic assessments. Stanford University, 2003; 2003.
[9] Chen W, Pham TM, Sichembe H, Chen L, Hao H. Experimental study of flexural [41] Baker JW. Efficient analytical fragility function fitting using dynamic structural
behavior of RC beams strengthened by longitudinal and U-shaped basalt FRP sheet. analysis. Earthq Spectra 2015;31(1):579–99.
Composites Part B 2018;134:114–26. [42] Bergami AV, Forte A, Lavorato D, Nuti C. Proposal of a incremental modal
[10] Ye LP, Zhang K, Zhao SH, Feng P. Experimental study on seismic strengthening of pushover analysis (IMPA). Earthqu Struct 2017;13:539–49.
RC columns with wrapped CFRP sheets. Construct Build Mater 2003;17(6–7): [43] Nuti C, Briseghella B, Chen A, Lavorato D, Iori T, Vanzi I. Relevant outcomes from
499–506. the history of Polcevera Viaduct in Genova, from design to nowadays failure.
[11] Colomb F, Tobbi H, Ferrier E, Hamelin P. Seismic retrofit of reinforced concrete J Civil Struct Health Monitor 2020;10:87–107.
short columns by CFRP materials. Compos Struct 2008;82(4):475–87. [44] Lavorato D, Fiorentino G, Pelle A, Rasulo A, Bergami AV, Briseghella B, Nuti C.
[12] Ozcan O, Binic B, Ozcebe G. Improving seismic performance of deficient reinforced A corrosion model for the interpretation of cyclic behavior of reinforced concrete
concrete columns using carbon fiber-reinforced polymer. Eng Struct 2008;30(6): sections. Struct Concr 2019;2020:1–15.
1632–46. [45] Life 365. Service life prediction model for reinforced concrete exposed to chlorides.
[13] Ozcan O, Binici B, Ozcebe G. Seismic strengthening of rectangular reinforced 2001. Version 1.1.
concrete columns using fiber reinforced polymers. Eng Struct 2010;32:964–73. [46] Oslakovic IS, Bjegovic D, Mikulic D, Krstic V. Development of service life model
[14] Promis G, Ferrier E, Hamelin P. Effect of external FRP retrofitting on reinforced CHLODIF þþ. Computation Modeling of Concrete Structures. CRC Press; 2014.
concrete short columns for seismic strengthening. Compos Struct 2009;88(3): [47] DuraCrete. Probabilistic performance based durability design of concrete
367–79. structures: statistical quantification of the variables in the limit state functions. Rep
[15] Promis G, Ferrier E. Performance indices to assess the efficiency of external FRP No.: Bioprocess Eng 2000;vol. 95-1347:62–3.
retrofitting of reinforced concrete short columns for seismic strengthening. [48] Fib Bulletin N� 34. Model Code for Service Life Design, ISBN 978-2-88394-074-1.
Construct Build Mater 2012;26:32–40. p. 116.
[16] Rodrigues H, Arede A, Furtado A, Rocha P. Seismic behavior of strengthened RC [49] DARTS. Durable and Reliable Tunnel Structures: Deterioration Modelling,
columns under biaxial loading: an experimental characterization. Construct Build European Commission, Growths 2000. 2004. Contract G1RD-CT-2000-00467,
Mater 2015;95:393–405. Project GrD1-25633.
[17] Ouyang LJ, Gao WY, Zhen B, Lu ZD. Seismic retrofit of square reinforced concrete [50] DARTS. Durable and Reliable Tunnel Structures: Data, European Commission,
columns using basalt and carbon fiber-reinforced polymer sheets: a comparative Growths 2000, Contract G1RD-CT-2000-00467. 2004. Project GrD1-25633.
study. Compos Struct 2017;162:294–307. [51] Collepardi M, Marcialis A, Turriziani R. Penetration of chloride ions into cement
[18] Wang D, Wang Z, Yu T, Li H. Seismic performance of CFRP retrofitted large-scale pastesand concretes. J Am Cer Soc 1972;55:534–5.
rectangular RC columns under lateral loading in different directions. Compos [52] Pacheco J, Polder RB. Critical chloride concentrations in reinforced concrete
Struct 2018;192:475–88. specimens with ordinary Portland and blast furnace slag cement. Heron 2016;61
[19] Lehman DE, Gookin SJ, Nacamuli AM, Moehle JP. Repair of earthquake-damaged (2):99–119.
bridge columns. ACI Struct J 2001;98(2):233–42. [53] Vu KAT, Stewart MG. Structural reliability of concrete bridges including improved
[20] Cheng CT, Yang JC, Yeh YK, Chen SE. Seismic performance of repaired hollow chloride-induced corrosion models. Struct Saf 2000;22:313–33.
bridge piers. Construct Build Mater 2003;17(5):339–51. [54] Liu T, Weyers RW. Modeling the dynamic corrosion process in chloride
[21] Saiidi M, Cheng Z. Effectivenss of composites in earthquake damaged repair of RC contaminated concrete structures. Cement Concr Res 1998;28:365–79.
flared columns. J Compos Constr ASCE 2004;8(4):306–14. [55] Enright MP, Frangopol DM. Probabilistic analysis of resistance degradation of
[22] He R, Grelle S, Sneed LH, Belarbi A. Rapid repair of a severely damaged RC column reinforced concrete bridge beams under corrosion. Eng Struct 1998;20:960–71.
having fractured bars using externally bonded CFRP. Compos Struct 2013;101: [56] Du YG, Clark LA, Chan AHC. Residual capacity of corroded reinforcing bars. Mag
225–42. Concr Res 2005;57:135–47.
[23] Lavorato D, Nuti C. Pseudo-dynamic tests on reinforced concrete bridges repaired [57] Fang C, Lundgren K, Chen L, Zhu C. Corrosion influence on bond in reinforced
and retrofitted after seismic damage. Eng Struct 2015;94:96–112. concrete. Cement Concr Res 2004;34:2159–67.
[24] Muntasir Billah AHM, Shahria Alam M. Seismic performance evaluation of multi- [58] Most T, Bucher C. Stochastic simulation of cracking in concrete using
column bridge bents retrofitted with different alternatives using incremental multiparameter random fields. Int J Reliab Saf 2006;1(1/2):168–87.
dynamic analysis. Eng Struct 2014;62–63:105–17. [59] Spoelstra MR, Monti G. FRP-confined concrete model. J Compos Construct 1999;3
(3):143–50.

20
M.A. Zanini et al. Soil Dynamics and Earthquake Engineering 136 (2020) 106224

[60] Lam L, Teng JG. Design-oriented stress–strain model for FRP-confined concrete. [70] Faleschini F, Gonzalez-Libreros J, Zanini MA, Hofer L, Sneed L, Pellegrino C.
Construct Build Mater 2003;17(6–7):471–89. Repair of severely-damaged RC exterior beam-columns joints with FRP and FRCM
[61] Fossetti M, Alotta G, Basone F, Macaluso G. Simplified analytical models for composites. Compos Struct 2019;207:352–63.
compressed concrete columns confined by FRP and FRCM system. Mater Struct [71] Faleschini F, Zanini MA, Hofer L, Toska K, De Domenico D, Pellegrino C.
2017;50(6):240. Confinement of reinforced concrete columns with glass fiber reinforced
[62] CNR-DT 215/2018. Guidelines for the design, execution and monitoring of static cementitious matrix jackets. Eng Struct 2020;218:110847.
retrofitting interventions with fabric reinforced cementitious matrix systems. 2019. [72] Midas Civil. Integrated solution system for bridge and Civil engineering. https:
p. 85 [Rome, Italy (in Italian)]. //en.midasuser.com/; 2019.
[63] American Concrete Institute. ACI 549.4R-13: Guide to design and construction of [73] Menegotto M, Pinto PE. Method of analysis for cyclically loaded reinforced
externally bonded fabric-reinforced cementitious matrix (FRCM) systems for repair concrete plane frames including changes in geometry and non-elastic behavior of
and strengthening concrete and masonry structures. American Concrete Institute, elements under combined normal force and bending. Proc IABSE Symp Resist
38800 Country Club Drive Farmington Hills; 2013. p. 74. MI 48331. Ultimate Deformability Struct Acted by Well Defned Repeated Loads, Int Assoc
[64] Colajanni P, Fossetti M, Macaluso G. Effects of confinement level, cross-section Bridge Struct Eng, Lisbon, Portugal 1973;13:15–22.
shape and corner radius on the cyclic behavior of CFRCM confined concrete [74] Kashani MM, Crewe AJ, Alexander NA. Nonlinear cyclic response of corrosion-
columns. Construct Build Mater 2014;55:379–89. damaged reinforcing bars with the effect of buckling. Construct Build Mater 2013;
[65] Mander JB, Priestley MJN, Park R. Theoretical stress–strain model for confined 41:388–400.
concrete. ASCE J Struct Eng 1988;114(8):1804–26. [75] Luzi L, Puglia R, Russo E, ORFEUS WG5. Engineering Strong Motion Database,
[66] Ministerial Decree 09/01/1996. Technical standards for design, execution and version 1.0. Istituto Nazionale di Geofisica e Vulcanologia, Observatories &
testing of ordinary and prestressed reinforced concrete structures, and steel Research Facilities for European Seismology. 2016. https://doi.org/10.13127/
structures. 1996 [in Italian)]. ESM.
[67] Buckle IG, Friedland IM, Mander JB, Martin GR, Nutt R, Power M. Seismic [76] Choi E. Seismic analysis and retrofit of mid-America bridges. Ph.D. thesis. Atlanta:
retrofitting manual for highway structures: Part 1-Bridges. Technical Report Georiga Institute of Technology; 2002.
MCEER-06-SP10; 2006. p. 659. [77] Choi E. Seismic fragility of typical bridges in moderate seismic zones. Eng Struct
[68] Gonzalez-Libreros J, Zanini MA, Faleschini F, Pellegrino C. Confinement of low- 2003;26(2):187–99.
strength concrete with fiber reinforced cementitious matrix (FRCM) composites. [78] Cheng H, Li HN, Yang YB, Wang DS. Seismic fragility analysis of deteriorating RC
Compos B Eng 2019;177:107407. bridge columns with time-variant capacity index. Bull Earthq Eng 2019;17(7):
[69] Faleschini F, Zanini MA, Hofer L, Pellegrino C. Experimental behavior of reinforced 4247–67.
concrete columns confined with carbon-FRCM composites. Construct Build Mater [79] Meletti C, Montaldo V. Seismic hazard estimates for different 50-years exceedance
2020;243:118296. probabilities: peak ground acceleration values. DPC-INGV Project; 2007. S1-D2 (in
Italian, last access 30/07/2019).

21

You might also like