Engineering Structures 216 (2020) 110397

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Corrosion effects on the seismic response of existing rc frames designed T

according to different building codes
⁎
Luisa Bertoa, Silvia Caprilib, Anna Saettaa, Walter Salvatoreb, Diego Talledoa,
a
DCP – University IUAV of Venice, Dorsoduro 2206, Venice 30123, Italy
b
Department of Civil and Industrial Engineering, Largo L. Lazzarino 1, 56122 Pisa, Italy

A R T I C LE I N FO A B S T R A C T

Keywords: Durability problems significantly affect the structural performance of reinforced concrete (RC) buildings: when
Seismic assessment exposed to aggressive environmental conditions, both steel reinforcing bars and concrete undergo relevant al-
Corrosion terations of their mechanical performance (strength and deformation) leading to a general decrease of the
RC frames bearing capacity of structural elements and, more in general, of the whole construction. Material deterioration
Material degradation
can be responsible for the shift of the collapse mode from ductile/flexural mechanisms to brittle/shear ones in
Collapse mechanism
relation to the entity of corrosion attack, the materials used (both concrete strength class and reinforcement
grade) and the structural details assumed according to the design. As well known, the Italian building stock is
made up of structures realized, for the majority, without following specific seismic design features and to resist –
mainly – gravitational vertical load combinations, until the introduction of the actual code after the dramatic
earthquake event of L’Aquila 2008. For this reason, the deep knowledge of the structural performance of RC
buildings realized neglecting (or simplifying) seismic action in presence of corrosion attack becomes a topic of
relevant importance to determine possible solutions and improvement at section, element and whole building
level. In the present paper, the effects of corrosion on two case study buildings designed following actual and
past Italian codes for constructions (i.e. with and without respecting the capacity design approach) are then
presented. The structural assessment in aggressive environmental conditions is performed adopting different
entity and distribution scenarios of corrosion, taking into consideration the decrease of strength and deformation
capacity of constitutive materials and the resulting loss of confinement affecting RC sections, allowing to de-
termine the attainment of the ductile and brittle mechanisms and the modification of the structural performance.

1. Introduction predict corrosion consequences also in case of new buildings realized

following seismic detailing rules and using pre-defined material prop-
Durability problems due to aggressive environmental conditions erties, protecting them against the unexpected failures caused by cor-
represent one of the most relevant issues affecting the structural per- rosion deterioration.
formance of Reinforced Concrete (RC) buildings. Corrosion leads to the It is well known that corrosion products (i.e. rust products) lead to
alteration of the mechanical properties of materials (both concrete and the increase of the rebars’ volume with resulting cracking and spalling
steel rebars), with the associated decrease of the bearing capacity of of the concrete cover and reduction of the rebar’s cross section. Several
sections, elements and whole structure. Even if several strategies exist formulations exist in the current scientific literature to estimate the
to face up the problem by reducing corrosion effects on materials (e.g. deterioration of concrete due to reinforcement corrosion, e.g. [4–10].
by introducing new grades less affected by durability degradation Concerning reinforcements, beside the well-known cross section
[1,2]), studies are still on going to evaluate the consequences of ag- reduction, the decrease of the deformation capacity of rebars due to
gressive environmental conditions on RC buildings. corrosion is evident. Experimental results highlighted residual values of
Corrosion usually affects structures realized in detrimental con- the elongation to maximum load ( Agt ) rather equal to the 50% of the
struction sites (e.g. coastal areas, industrial sites, etc.) and character- sound performance [11–15], a lower but anyway relevant decrease of
ized by poor-quality materials [1,3], that is the case, for example, of old ultimate elongation and a negligible deterioration of strength, both at
existing RC constructions. By the side, technicians shall be able to yielding (R e ) and ultimate (Rm) level [16,17]. In some cases, ductility

⁎
Corresponding author.
E-mail address: [email protected] (D. Talledo).

https://doi.org/10.1016/j.engstruct.2020.110397
Received 14 July 2019; Received in revised form 16 December 2019; Accepted 17 February 2020
0141-0296/ © 2020 Elsevier Ltd. All rights reserved.
L. Berto, et al. Engineering Structures 216 (2020) 110397

levels achieved after corrosion exposure are even below the threshold both concrete and steel together with reduced time effort [11,12].
fixed for reinforcements in new constructions by Eurocode 8 [18] in Different procedures can be able to reproduce different environmental
terms of deformation capacity ( Agt ) and hardening ratio (Rm / R e ), evi- scenarios (e.g. localized/pitting or uniform corrosion) and the best
dencing the need to go deeper inside the problem [19]. The decrease of technique needs to be determined as function of the research aims. For
ultimate elongation of stirrups acts on RC sections’ confinement: to- example, recently, Meda et al. [25] performed experimental cyclic tests
gether with the progressive reduction of stirrups’ diameter, the dete- on corroded RC columns reproducing corrosion through impressed
rioration of the deformation capacity due to corrosion leads to the loss current density procedures, neglecting stirrups’ degradation oppor-
of confinement and to the following decrease of strength and ductility tunely protecting them. Results highlighted a 30% decrease of the ul-
properties at section and element level. timate force of corroded elements and a reduction of global ductility up
Besides, aggressive environmental conditions affect bond proper- to 50% for a corrosion level of about 20% in terms of mass loss even
ties; the modification of the commonly adopted bond-slip relationship neglecting the loss of confinement due to stirrups’ deterioration. The
is usually neglected for lower entity of corrosion, due to the apparent importance to consider the effects of corrosion on strength and mainly
increase of the cross section of the rebar generating, together with on the displacement capacity of RC columns has been evidenced in
friction, a fictitious increase of bond strength. High levels of corrosion several recent experimental tests, also with reference to pitting corro-
need to account for the significant deterioration of bond strength (e.g. sion and well confined specimens designed to behave in a ductile
[20,21]) and the modification of the constitutive bond stress-slip law manner (e.g. [31–33]).
(e.g. [22–26]), under both monotonic and cyclic conditions. In general, what can be stated is that the deep knowledge of ag-
Basing on the above-mentioned corrosion effects, the decrease of the gressive environmental conditions on buildings’ performance is funda-
structural seismic performance of RC structures are often related to the mental to guarantee adequate safety levels and to foresee specific
shift of the collapse modality. Modern RC buildings are designed to protection/prevention precautions at material, section and element le-
develop a global ductile failure mechanism with plastic rotations ex- vels.
pected in beams, preventing shear collapse by adopting the capacity In the present paper the influence of the aggressive environmental
design procedure and respecting specific requirements for executive conditions on the seismic performance of RC structures have been
details and material characteristics [18]. The effects of corrosion on analyzed. Aspects related to the alteration of the whole structural
steel and concrete (e.g. reduction of ductility, confinement, strength, ductility and to the modification of the modality of failure have been
etc.) can impair structural response with decrease of strength and more highlighted and related to the entity of corrosion attack and to its lo-
importantly with reduction of structural displacement capacity, often cation in the building.
leading to the change of collapse modality and to unexpected failures. To achieve these objectives, different case study buildings, re-
This issue was highlighted in some numerical studies aimed to capture presentative of the typical constructions realized before and after the
the effect of corrosion on the seismic performance of the entire building introduction of mandatory rules for buildings in seismic regions (i.e.
[e.g. [14,27]]. Such problem can be even higher in presence of con- neglecting or respecting the modern principles of the capacity design
structions where no specific attention was devoted to these aspects: this approach and detailing rules) have been designed and subjected to
is the case, for example, of existing RC buildings characterized by the nonlinear pushover analyses able to simulate their structural response
strong beam/weak column hierarchy typical of design performed only under increasing horizontal actions.
towards vertical loads, frequently exposed to brittle soft-story collapses. The influence of corrosion attack has been introduced in the ana-
Stating what herein presented, there is a clear need to know the lyses through the reproduction of their effects on both steel reinforcing
structural response evolution of buildings in presence of different de- bars (i.e. reduction of cross section, decrease of ductility – based on
gradation scenarios combined with the occurrence of a generic seismic experimental tests), concrete (i.e. deterioration of mechanical perfor-
event, also in a probabilistic framework [e.g. [28–30]]. mance accounted for through a coupled mechanical - environmental
To analyze the effects of corrosion on RC structures, both experi- damage model) and loss of confinement. Different entity of corrosion
mental and numerical studies are needed. From the material point of attack and different distribution of corroded elements (corrosion sce-
view, accelerated corrosion procedures represent a valid tool to com- narios) have been considered to represent real conditions of RC build-
bine a meaningful estimation of residual mechanical performance of ings.

Fig. 1. Methodology adopted in the present paper.

2
L. Berto, et al. Engineering Structures 216 (2020) 110397

Fig. 2. Schematic drawings of the two designed buildings (a) Plan view; (b) Vertical section.

The methodology followed in the present work is briefly summar- structural elements (e.g. span length of beams, height and number of
ized in Fig. 1. columns), functional destination (commercial buildings) and are lo-
cated in L’Aquila (Italy). Fig. 2a and Fig. 2b depict respectively the plan
type and the vertical sections of the buildings.
2. Design of RC frames with different strategies Two different standard codes were used for the design, i.e. the
current Italian technical standard for constructions (D.M.17/01/2018,
2.1. General characteristics of the buildings [34] – in the following NTC2018) and the standard used till 2009, that
was the date when D.M.14/01/2008 (the first version of NTC2018)
Two case study Moment Resisting Frame buildings were designed became valid, i.e. D.M.16/01/1996 [35] (in the following D.M 1996).
and analyzed for the assessment of the detrimental effects of corrosion In the following, the two case study buildings will be named Building A
on their global structural behavior. They are 5-story buildings with total (designed with NTC2018) and Building B (designed with D.M.1996).
height equal to 19 m, owing the same architectural disposition of

3
L. Berto, et al. Engineering Structures 216 (2020) 110397

Concrete class C30/37 was assumed. Steel reinforcing ribbed bars 0,80

Sa(T) [g]
grade FeB44k and B450C were respectively used in the design of DL

Building B and Building A. It is worth noting that, according to pro- 0,70 LS - q=1,0
duction, the two steel grades own the same characteristic yielding LS - q=5,85
strength, equal to about 450 MPa: this means that the material, except 0,60
for the tag adopted, is the same for the two case studies.
Gravitational loads were defined in relation to the typology of 0,50
structural and not structural elements (slabs, roof, internal and external
infills, equipment, etc.): story slabs like Predalle H24-i50, resulting in a 0,40
vertical load equal to 3.35 kN/m2 (G1), were considered. Additional
vertical loads were assumed equal to 2.75 kN/m2 (G2). 0,30
For live loads Eurocode 0 [36] was followed with specific reference
to national annexes, in relation to commercial destination of the 0,20
building, i.e. 5.0 kN/m2 (Q1); for roof slab the live load assumed was
equal to 0.50 kN/m2 (not practicable roofs). Wind and snow actions 0,10
were evaluated and included in the design according to the funda-
mental combination provided by the Standard Codes. 0,00
0,0 1,0 2,0 3,0 4,0
T [s]
2.2. Design philosophy
Fig. 3. Response spectra (LS and DL limit states) used for the design of building
A.
The design of Building A and Building B differs for the approach
pursued with reference to the behavior under seismic actions. Building A
was designed following the capacity design approach insight of a global 2.2.2. Building B
ductile collapse mechanism that is achieved through the scheme ‘weak According to the old Italian seismic building code D.M.1996 and
beam-strong column’ and the adoption of very strict details’ requirements being the structure characterized by plan and height regularity, the
concerning stirrups’ spacing, reinforcements’ amounts, etc. Building B, structural design was performed considering, separately, bi-dimen-
on the other hand, was designed without seismic detailing. Besides, it sional plane frames representatives of the two main directions of the
shall be also noted that the definition of seismic action and the rules of building. A linear static analysis, one for each main direction, was
load combination for the design of the buildings followed the pre- executed. The resulting horizontal seismic action (i.e. base shear) was
scriptions imposed by the corresponding design regulations, i.e. defined as Fh = C·R·I ·W , being C the seismic intensity coefficient –
NTC2018 and Eurocode 8 [18] for Building A and D.M.1996 for Building defined basing on the seismicity level of the construction site, R the
B as better clarified in the following. response coefficient relative to the considered direction, I the seismic
protection coefficient and W the total weight. For the considered
building site (characterized by a high seismicity level S equal to 9.0
2.2.1. Building A
according to that past national classification), the above-mentioned
The building was designed for High Ductility Class (HDC); the re-
coefficients were, respectively, equal to C = 0.07, I = 1.00 and
sponse spectrum was based on the effective reference life of the struc-
R = 1.00. This action was combined in the seismic load combination
ture (VR), defined as function of the nominal life (VN) and of the use
defined in D.M.1996 where an amplification factor of 1.5 for seismic
coefficient (CU). The nominal life VN was assumed equal to 50 years, CU
action is considered together with the amplified value of vertical loads
was unitary, and the resulting reference life VR was 50 years.
adopted at Ultimate Limit State. No specific mandatory detailing rules
Life Safety limit state (LS), concerning strength, and Damage
were prescribed for seismic areas; even if some additional indications
Limitation limit state (DL), concerning stiffness and displacements,
were provided, it was unusual for designers their application.
were considered. The return periods of seismic action (TR) were equal
Therefore, in the present case study such details were not encountered,
to 75 and 475 years respectively for DL and LS. Soil category B was
to represent the ‘typical’ building designed according to D.M. 1996.
assumed. Since Building A was designed following the capacity design
It is worth noting that, considering the differences between the
approach, the elastic spectral acceleration can be reduced by behavior
standard codes for the design of the two buildings in terms of seismic
factor accounting for ductile resources. In the present case, the max-
action and design load combination, the main direction of the floor
imum q-factor allowed by NTC2018 (and Eurocode 8) for RC MRF
system was slightly varied, as shown in Fig. 4, rotating several story
structures equal to 5.85 was adopted, Fig. 3.
slabs in order to partially unload beams from gravitational loads. In
According to what foreseen by actual standards, a three-dimen-
such a way it was possible to realize elements that better satisfy the
sional model was used for the design; a linear dynamic analysis with
strength requirements, minimizing – where possible – their size and
reduced response spectrum was performed to achieve design actions on
adopting more realistic sections and corresponding reinforcements.
elements. The model was realized with SAP2000® software, using frame
In the Section 4, the results of the numerical analyses are referred to
elements for both beams and columns; the contribution of story slabs
the frame (in the XZ plane) marked by green color in Fig. 4, which is
was considered through the adoption of rigid diaphragm. Vertical loads
characterized by the same vertical loads acting on the beams for the two
were directly applied on story slabs as uniformly distributed actions and
Buildings. Fig. 5 and Table 1, together with Figs. 6, 7 and 8, summarize
torsional moments were introduced for the effects of accidental ec-
the typical sections adopted for structural elements and their corre-
centricity, equal to the 5% of the maximum dimension of the building
sponding longitudinal and transversal reinforcements in the case of
in the perpendicular direction respect to the seismic action according to
Building A and Building B. Several differences can be appreciated: in
NTC2018 and Eurocode 8.
general, D.M.1996 design approach leaded to larger beams with higher
The cracking phenomena of concrete at LS limit state were included
amount of reinforcements and, on the contrary, to smaller columns.
through the reduction of the stiffness of primary structural elements.
This is the consequence of the above described different design method
The stiffness was reduced in relation to the axial force and to the type of
for seismic action and load combination.
the elements themselves (e.g. [37–39]): the 50% of the uncracked
stiffness was adopted for beams, while for columns, characterized by a
significant axial load, the reduction amounted to 30%.

4
L. Berto, et al. Engineering Structures 216 (2020) 110397

Fig. 4. Floor type of the two designed buildings.

3. Modelling of corrosion effects reinforcing steel occurs, i.e. when the chlorides’ concentration at the
levels of rebars attains a critical threshold or when the carbonation
The effects of corrosion at material level (both steel rebars and depth reaches the reinforcing bars. The time for this event to occur,
concrete) can be introduced in the numerical model as function of the named initiation time (tin ), mainly depends on the concrete cover (i.e. its
depth of corrosion Px (t ) , defined as the loss of the rebar radius. Px (t ) can property and thickness) and on the diffusion process of concrete. Both
be evaluated as the integral over the propagation time of the corrosion empirical and mathematical methods, formulated in physical - chemical
rate Vcorr . The propagation time starts when the depassivation of terms, with different levels of complexity are available in literature for

Fig. 5. Distribution of beams’ sections in the typical XZ frame.

5
L. Berto, et al. Engineering Structures 216 (2020) 110397

Table 1
Comparison of sections’ dimensions [cm] and reinforcements for columns.
Building A (NTC2018) Building B (DM1996)

Columns Sect. Stirrups Long. reinf. Columns Sect. Stirrups Long. reinf.

Floor I 40 × 80 ϕ10/5 12ϕ16 + 4ϕ18 Floor I 40 × 65 ϕ10/20 16ϕ20

Floor II 40 × 70 ϕ10/10 10ϕ16 + 4ϕ18 Floor II 40 × 65 ϕ10/15 16ϕ20
Floor III 40 × 60 ϕ10/10 8ϕ16 + 4ϕ18 Floor III 40 × 50 ϕ10/15 14ϕ20
Floor IV 40 × 60 ϕ10/10 8ϕ16 + 4ϕ18 Floor IV 40 × 50 ϕ10/20 14ϕ20
Floor V 40 × 60 ϕ10/10 8ϕ16 + 4ϕ18 Floor V 40 × 50 ϕ10/25 14ϕ20

evaluating tin (e.g. [5,40,41]). Once rebars are depassivated, steel cor- properties of reinforcing steels, joining an efficient estimation of cor-
rosion may start and proceed with a corrosion rate (Vcorr ) depending on rosion degradation to relatively low required time.
several factors, such as the availability of oxygen and the electrical Table 2 shows the monotonic tensile experimental tests’ results for
resistance of concrete, temperature and relative humidity, etc. Gen- rebars in sound conditions. Results are presented in terms of strength
erally, Vcorr varies with time due to the evolution of corrosion process as and deformation; yielding (R e ) and tensile strength (Rm ) values refer to
well as to the climatic conditions; in many cases, Vcorr can be assumed the rebar’s cross section before corrosion, whose diameter (ϕreal ) has
constant and evaluated through the Faraday’s law starting from a value been assessed through weight and length measurements on reference
of current density icorr [μA/cm2], that can be achieved as function of the samples. Deformations are presented in terms of Agt (elongation corre-
exposure class or predicted using empirical models when direct mea- sponding to maximum load) and A5 (ultimate deformation), evaluated
sures from inspection are not available. Predictive models for the onset through manual measurements according to EN15630-1:2010 [44];
of corrosion tin and the corrosion rate Vcorr (or the current density icorr ) EN6892-1:2009 [45]).
are provided in the current scientific literature [42,43]. Table 3 shows the results of monotonic tensile tests on corroded
Basing on Faraday’s law, a simplified relationship to estimate the rebars, whose corrosion entity was evaluated in terms of percentage
depth of corrosion on reinforcements is the following, where all the Mass Loss (ML). Residual values of the mechanical performance are
terms have been before explained: presented, being evaluated according to:

Px = 0.0116icorr (t − tin ) (1) PIcor

PIres (%) = ·100
PIsound (2)
where PIres is the residual value of the generic Performance Indicator (PI)
3.1. Steel deterioration representing a specific mechanical performance parameter (i.e. de-
formation corresponding to maximum load or ultimate elongation),
Corrosion effects on steel rebars can be simply translated in the assessed as the ratio between the experimental value of the PI coming
reduction of the cross section and in the decrease of the deformation from tensile test on corroded specimen and the average value achieved
capacity, being the other mechanical properties (such as strength) not from sound condition (respectively Agt,sound and A5,sound).
significantly affected by aggressive environmental conditions. In general, conditions achieved through salt spray chamber proce-
To model in a reliable way corrosion influence on rebars, the results dure are comparable to uniform carbonation phenomena; in such case,
of experimental tests in sound (Table 2) and corroded (Table 3) con- the progressive loss of rebar diameter after the initiation of corrosive
ditions performed by Caprili et al. [11,16,17] were used. Experimental process can be related to the corrosion depth Px (t ) by simple geometric
data were provided by different experimental tests’ campaign per- relationship:
formed within the framework of a European research project funded by
the Research Fund for Coal and Steel called ‘Rusteel’[16], positively ∅res = ∅sound ·(1 − 2Px (t )) (3)
concluded in 2013. Monotonic and cyclic tests were performed on un- where ∅res is the residual value of bar diameter and ∅sound the initial
corroded/reference and corroded reinforcement of different steel one. In case of localized or non-uniform corrosion attack, a pitting
grades, selecting the most common around the European market. factor may be introduced in Eq. (3) instead of the factor 2, ([46,47]).
Tensile tests on two sets of B450C ϕ 16 mm rebars (namely, in the fol- Concerning deformation capacity, basing on the results of the ex-
lowing, ‘A’ and ‘B’), the most common reinforcement diameter for or- perimental tests presented in Table 3, the following relationships for the
dinary constructions, were executed. Corrosion was reproduced evaluation of the residual values ( Agt , res and A5, res ) were proposed:
through accelerated tests in salt spray chamber, a good tool to evaluate
the effects of aggressive environmental conditions on the mechanical Agt , res = (−0.83ΔA + 0.46) Agt , sound (4)

Fig. 6. Beams’ sections for Building A.

6
L. Berto, et al. Engineering Structures 216 (2020) 110397

Fig. 7. Beams’ sections for Building B.

A5, res = (−2.29ΔA + 0.93) A5, sound (5) Table 2

Results of tensile tests on uncorroded reinforcing bars (sound condition).
being ΔA the reduction of cross section area expressed by
ΔA = 1 − Ares / Asound . Specimen ϕreal [mm] Re [MPa] Rm [MPa] Rm/Re [-] Agt [%] A5 [%]
These relationships were obtained by means of linear regression of
Sound A-3 16.2 517.8 615.4 1.19 13.8 25.4
the experimental results of Table 3, as shown in Fig. 9; it is worth noting Sound A-4 16.2 507.4 610.4 1.20 15.0 25.1
that these predictive equations can be reliably used only for the in- Sound A-5 16.2 516.3 613.7 1.19 12.0 –
vestigated range of corrosion level (i.e. ΔA within the range of about Sound B-1 15.9 489.2 601.6 1.23 14.8 25.4
5–15%), losing their meaning for negligible values of area reduction, as Sound B-2 15.8 487.9 591.4 1.21 14.4 24.5
Sound B-3 15.9 483.6 593.7 1.23 14.9 28.4
represented by dashed lines and grey intervals in Fig. 9. Sound B-4 15.9 484.1 600.3 1.24 18.5 28.3
Sound B-5 15.9 483.8 586.9 1.21 15.8 26.8
3.2. Concrete deterioration Sound – average values 496.3 601.7 1.21 14.9 26.3

The volume expansion due to the formation of the oxides in the

corrosion process may cause the cracking of the concrete surrounding (or pseudo-time step in case of static analysis) as:
the bars, up to the complete spalling of the concrete cover, with a r + (t ) = max( max σ¯ + (s ); f 0+ )
consequent reduction of strength and stiffness of the concrete material. s ∈ [0, t ] (7)
These aspects were included in a coupled mechanical-environ- r − (t ) = min( min σ¯ − (s ); f 0− )
mental damage model, developed by some of the Authors, and recently s ∈ [0, t ] (8)
enriched and reformulated to consider the concrete degradation due to
These two threshold variables represent the maximum value
different environmental attacks (e.g. aggressive species like sulphate
reached respectively by the positive and negative effective stress during
and calcium chloride, freezing-and-thawing cycles, or due to cracking
loading history, once the initial threshold is reached. The initial values
caused by corrosion phenomena). The model was developed within the
of the threshold variables are, respectively, f 0+ for tension, which can be
framework of strain-based formulations and adopts the following con-
assumed as the tensile strength ft of concrete ( f 0+ = ft ) and f 0−for
stitutive law: compression, which is set as a fraction of compressive strength fc , being
+
σ = (1 − denv −
)(1 − d+) σ¯ + + (1 − denv )(1 − d−) σ¯ − (6) generally f 0− = −(0.4 ÷ 0.65) fc .
+ −
The evolution of damage variables was computed according to the
d+
where , and d− denv , are respectively the two mechanical and the
denv following equations, where the one relative to the mechanical damage
two environmental damage variables which affect the positive and variable d− was re-formulated to allow including the effect of confine-
negative parts of the effective stress tensor σ̄ + and σ̄ −. The detailed ment according to Mander’s model [52]:
description of the model and some examples of validation are provided
in [7,48–50]. f 0+ ⎡ r+ ⎤
d+ = 1 − exp ⎢A+ ⎜⎛1 − + ⎟⎞ ⎥
In the present work, the model was implemented in OpenSees [51] r+ f0 ⎠
⎣ ⎝ ⎦ (9)
as uniaxial material in order to be adopted in non-linear beam-column
elements with fiber section approach. The elastic domain was re-
⎧ f 0− r−
presented by two variables, named r + and r −, accounting for the ⎪1 − (1 − A− ) − A− exp ⎡B − ⎛1 −
⎜
⎞ ⎤ ifε ≥ εcu
⎟

d− r− ⎢ f 0−
memory of the material (increasing of elastic threshold due to accu-
=
⎨ ⎣ ⎝ ⎠⎥⎦
⎪ 1 ifε < εcu (10)
mulation of damage), defined in the 1D formulation, for each time step t ⎩

Fig. 8. Columns’ sections for Building A and Building B.

7
L. Berto, et al. Engineering Structures 216 (2020) 110397

Table 3
Results of tensile tests on corroded specimens and residual values of deformation capacities.
Spec Tcorr ML Re Rm Agt A5 Agt, sound A5, sound Agt, res A5, res
[days] [%] [MPa] [MPa] [%] [%] [%] [%] [%] [%]

A-1 90 14.6% 481.4 599.5 4.3 15.4 13.6 25.3 31.7 61.0
A-2 90 6.1% 484.4 598.0 4.4 15.6 32.4 61.8
A-3 90 8.7% 499.8 610.5 5.1 16.6 37.6 65.7
A-4 90 6.9% 497.4 607.9 5.7 17.8 42.0 70.5
A-5 90 8.3% 480.9 600.0 5.1 14.1 37.6 55.8
A-6 90 17.3% 502.8 613.8 5.5 16.3 40.6 64.5
B-1 90 7.3% 420.9 534.7 6.5 18.9 15.7 26.7 41.3 70.8
B-2 90 6.8% 418.8 534.6 4.6 25.8 29.1 96.6
B-3 90 6.1% 417.6 535.7 5.9 23.4 37.4 87.7
B-4 90 7.5% 419.3 537.2 6.3 17.1 40.0 64.3
B-5 90 7.9% 418.0 516.1 5.1 20.4 32.3 76.5
B-1 45 4.2% 434.3 543.9 8.3 22.4 15.7 26.7 52.8 84.0
B-2 45 4.0% 465.4 557.7 7.2 23.9 45.8 89.6
B-3 45 4.5% 456.3 555.7 7.5 24.4 47.7 91.5
B-4 45 3.6% 476.6 562.0 7.8 25.5 49.6 95.7
B-5 45 3.9% 454.5 555.8 6.9 23.1 43.9 86.8

Fig. 9. Residual values of (a) Agt and (b) A5 in relation to cross section reduction ΔA for steel grade B450C.

being A+ the parameter controlling the softening branch of tensile degraded concrete respectively; k is a coefficient depending on bar
constitutive law, and A− and B − the parameters governing the shape of roughness and diameter (Capè [54] suggested k = 0.1), ε1 is the average
the compressive constitutive law related to the peak stress fc , the peak tensile strain in cracked concrete, bo is the original section width, nbars is
strain εc0 , and the ultimate strain εcu including possibly the effect of the number of compressed bars, and νrs is the ratio of volumetric ex-
confinement. pansion of oxides assumed equal to 2 as to Molina et al.[55].
Finally, it was assumed that damage surface corresponds also to the Similarly, in tension the environmental damage variable can be
plastic surface. Consequently, the development of damage was con- evaluated as:
sidered simultaneous with the accumulation of inelastic strain, whose
evolution can be described by the following relation: + f 't
denv =1−
ft (14)
ε ̇p = βH (d )̇ ε ̇ (11)
where ft and f 't are the mean tensile strength for sound and degraded
being H the Heaviside function and β the plastic accumulation para- concrete respectively, which can be evaluated, in absence of experi-
meter. mental results, as in this case, from the corresponding value of com-
The concrete degradation due to corrosion attack was considered by pressive strength, according to Model Code 2010 [56].
defining two environmental damage variables, one in compression and Finally, concerning the confined concrete stress-strain relationship,
one in tension, depending on corrosion depth Px (t ) evaluating according the confinement effect due to transverse reinforcement is affected both
to Eq. (1). In compression, starting from the formulation proposed by by the mechanical volumetric ratio of stirrups (i.e.ωw = ρw f yw / fc where
Coronelli and Gambarova [53] for the evaluation of residual strength of ρw is the volumetric ratio of stirrups and f yw their yield strength) and by
compressed cracked concrete, the environmental damage variable can their elongation at failure A5 . The stirrups’ corrosion leads to a decrease
be defined as: of the volumetric ratio, due to the cross-section reduction and a de-
f 'c crease of the elongation at failure. Both these aspects were included in
− kε1/ εc o
denv =1− = the concrete confinement model adopted in this work to assess the
fc 1 + kε1/ εc o (12)
confined concrete stress-strain relationship.
being:
3.3. Environmental degradation scenarios
nbars ·[2π·(νrs − 1) Px ]
ε1 =
bo (13)
Several possible environmental degradation scenarios were con-
where fc and f 'c are the mean compressive strength for sound and sidered, different both for corrosion level and for their spatial

8
L. Berto, et al. Engineering Structures 216 (2020) 110397

distribution among the elements of the RC frames. confined concrete, they are evaluated with the mechanical- environ-
The effects of corrosion on the structural behavior of the frame were mental damage model described in Section 3.2, as shown, as an ex-
analyzed assuming three current intensities: icorr = 0.2–0.5–0.8 μA/cm2 ample, in Fig. 10 and Fig. 11 for the columns of first floor of both
and two different exposure periods: t = 50 years (corresponding to the buildings. These degraded laws were applied in the concrete zone
nominal life adopted in the design of the studied buildings) and surrounding the rebars. From Fig. 10 it is possible to observe that un-
t = 100 years. For all the study cases, with reference to the concrete confined concrete is significantly degraded for the three corrosion le-
class, the concrete cover and the environmental condition, the initiation vels and there are no differences between Building A and Building B, as
time tin was assumed equal to 14 years, according to [29]. To account expected being the geometry and reinforcement pattern very similar. It
for the spatial distribution of the corrosion among the elements of the can be seen that the degradation affects both the initial stiffness and the
RC frame, different “distribution scenarios” were analyzed. strength of concrete according to Eqs. (6) and (12).On the other side, as
Each “corrosion scenario” is named with a first label corresponding it can be seen in Fig. 11, for confined concrete the effect of corrosion is
to the spatial distribution scenario and a second label corresponding to more evident in case of Building B for which the columns are char-
the corrosion level. In detail, concerning the spatial distribution of acterized by a lower volumetric ratio of stirrups and consequently by a
corrosion two scenarios were considered: lower level of confinement (lower peak stress fc , and lower peak and
ultimate strain).
– Distribution scenario D1: uniform corrosive attack on the columns of
the ground floor of the buildings. 4. Seismic performance and ageing of RC frames
– Distribution scenario D2: uniform corrosive attack on the perimeter
columns of the buildings. 4.1. Numerical modelling of the structures

Concerning the entity of corrosive attack, six different levels ob- For the numerical modelling of RC frames of the case study build-
tained combining the three values of icorr with the two different ex- ings OpenSees [51] was adopted, directly implementing the con-
posure period t were analyzed. In the following, the results of three stitutive laws for materials and degradation described in Section 3.
different levels, considered the most representative ones, are described The constitutive law presented in Section 3.2 for concrete was im-
in detail: plemented in its 1-D formulation as a Uniaxial Material; for steel re-
inforcing bars the effects of corrosion (i.e. reduction of cross section and
‐ Corrosion Level CL1: icorr = 0.5 μA/cm2, t = 50 years. ductility) were implemented and the Menegotto and Pinto law [57]
‐ Corrosion Level CL2: icorr = 0.5 μA/cm2, t = 100 years. (named Steel02 in OpenSees) was adopted.
‐ Corrosion Level CL3: icorr = 0.8 μA/cm2, t = 100 years. Beams and columns were modelled as force-based nonlinear beam-
column elements based on the iterative flexibility formulation with the
It is worth noting that the corrosion level CL1 may be considered integration method proposed by [58] (also known in OpenSees as
representative also for the corrosion level icorr = 0.2 μA/cm2 and “Beam With Hinges”). According to this method, the element is re-
t = 100 years, since they produce a very similar depth of corrosion presented with six integration points, two at the element ends re-
Px (t ) . On the other hand, the results of less severe corrosion levels were presented in this work as fiber sections and four elastic sections for the
not reported in the present paper since no structural effects were ob- interior part of the element.
served. To simplify the process of definition of fiber section, a specific
The effects of the different corrosion levels on steel reinforcement software capable to export in “tcl” format the section for OpenSees
for longitudinal bars and stirrups are summarized in Table 4, where Px framework was developed. The software allows to calibrate the damage
represents the depth of corrosion, Ares is the reduced cross section area constitutive law for concrete accounting for confinement due to stirrups
as function of corrosion and εsu (i.e. in terms of Agt ) is the maximum and for the effect of corrosion with time on material and section
deformation capacity evaluated according to Eq. (4) for longitudinal properties.
bars and Eq. (5) for stirrups (i.e. in terms, in this case, of A5 ). As already Concerning plastic hinge length, the formulation proposed by [59]
mentioned, different deformations were considered in the modelling of was adopted being expressed as:
longitudinal and transversal reinforcements: in particular, for long-
Lpl = 0.08LV + 0.022dbl f y (15)
itudinal rebars the deformation at maximum load is the relevant
parameter to describe the ductile flexural behavior, while in case of where LV is the shear span length, dbl is the mean diameter of long-
stirrups the ultimate deformation directly influences the confinement itudinal bars and f y [MPa] is yield strength of rebars. The effect of
action till failure occurs. corrosion can be profitably accounted for in Eq. (15) considering the
In the numerical model the effects of corrosion on steel reinforce- loss of rebar diameter defined in Section 3.1. According to [60] this
ment were implemented considering the reduced cross section area Ares formulation can be suitably used with a detailed confinement model
and reduced value of εsu calibrated from the experimental results of tests like Mander.
described in Section 3.1, without variations in the stress strain law Nonlinear static analyses were performed on the models of sound
which has been assumed elastic-plastic with strain hardening. Con- structures and for different degradation scenarios, using a distribution
cerning the stress-strain relationship for degraded unconfined and load pattern proportional to first modal shape. P-Δ effects were

Table 4
Effects of corrosion on steel reinforcement for the different corrosion level.
Longitudinal bars Stirrups

Px Ares εsu (Agt ) Ares εsu (Agt ) Ares εsu (Agt ) Ares εsu (A5 )
Corrosion Level [mm] [mm2] [%] [mm2] [%] [mm2] [%] [mm2] [%]

Sound – 201 14.90 254 14.90 314.0 14.90 78.5 17.00

CL1 0.2 190.7 6.25 242.8 6.32 301.2 6.38 72.1 12.67
CL2 0.5 176.8 5.39 227.0 5.55 293.5 6.07 63.7 8.48
CL3 0.8 162.9 4.54 211.3 4.79 266.0 4.99 55.5 4.43

9
L. Berto, et al. Engineering Structures 216 (2020) 110397

Fig. 10. Effects of corrosion on unconfined concrete for column of first floor for Buildings A and Building B.

Fig. 11. Effects of corrosion on confined concrete for column of first floor: (a) Building A; (b) Building B.

and columns). According to Eurocode 8-3 [61] the total chord rotation
capacity (considering both the elastic and the inelastic part) at ultimate
limit state (Near Collapse, NC) for members with only vertical shear
reinforcement was evaluated with:

0.225 0.35 ⎛αρ f yw ⎞

1 max(0.01; ω') ⎤ ⎛min ⎛9; L v ⎞ ⎞
⎜ sx f ⎟
θu = 0.016(0.3υ) ⎡ fc 25⎝ c ⎠
γel ⎢
⎣ max(0.01; ω) ⎥ ⎦ ⎝ ⎝ h ⎠⎠
(16)

In which γel is equal to 1.5 or 1.0 respectively for primary and sec-
ondary elements – being secondary elements the ones devoted only to
carry on vertical loads and not seismic action, υ is the compression
stress normalized to fc , ω and ω' the mechanical reinforcement ratios of
the tension and compression longitudinal reinforcement, α the con-
finement effectiveness factor, ρsx the ratio of transverse steel parallel to
the direction of loading, f yw and fc the strength of stirrup and concrete
Fig. 12. Building A – Sound condition. Base shear vs. top displacement curve. respectively. For members without detailing for earthquake resistance,
as for Building B, the value given by (16) must be divided by 1.2. In this
considered in the analyses. formulation the effect of corrosion was introduced properly modifying
ω , ω' and fc values. The assessment with respect to Life Safety (LS) limit
state refers to a chord rotation capacity equal to 3/4θu .
4.2. Assessment of the structural capacity Since the validity of formula (16) has not been proved for corroded
elements, to have a further insight of the structure response of the end
Capacity checks for ductile mechanisms were performed at the sections, especially in case of corrosion, the deformation demand was
element level, evaluating the chord rotation demand and the corre- compared with the limit value of strain assumed for confined concrete
sponding capacity at the ends of each structural element (both beams and steel, respectively equal to εccu and εsu = Agt, suitably reduced as a

10
L. Berto, et al. Engineering Structures 216 (2020) 110397

Table 5
Building A: attainment of the limit conditions in the first elements for columns and beams for different degradation scenarios in terms of base shear Vbase (expressed in
kN) and corresponding top displacement utop (expressed in mm).
Scen. CRACK YIELD εs ≥ εsu εc ≤ εccu θ ≥ θu V ≥ VR

Column Beam Column Beam Column Beam Column Beam Column Beam Column Beam
Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase
(utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop)

Sound 479.59 (18) 276.28 (7) 1320.65 (77) 774.53 (34) – – – 2129.43 (427) 2099 (380) 2008.18 (455) – –
D1.CL1 389.08 (13) 272.85 (7) 1274.21 (73) 778.42 (35) 2028.45 (445) – – 1986.30 (403) 2066 (390) 1986.30 (403) – –
D1.CL2 366.95 (10) 269.39 (7) 1224.32 (69) 776.33 (35) 2017.90 (388) – – 2026.30 (402) 2017.90 (388) 2007.19 (437) – –
D1.CL3 303.92 (8) 266.34 (7) 1182.92 (66) 763.33 (35) 1913.73 (330) – – 1990.83 (429) 1973 (390) 1973.51 (445) – –
D2.CL1 391.71 (13) 275.14 (7) 1298.59 (75) 767.84 (34) 2076.59 (453) – – 2115.64 (429) 2095.65 (389) 2095.65 (389) – –
D2.CL2 349.84 (10) 273.78 (7) 1265.81 (72) 779.92 (35) 2090.38 (401) – – 2002.92 (404) 2081.56 (386) 2086.03 (392) – –
D2.CL3 340.42 (9) 272.62 (7) 1242.84 (70) 760.92 (34) 2042.46 (347) – – 2077.95 (424) 2066.00 (392) 2065.37 (394) – –

Fig. 13. Building A – Sound condition. Frame state corresponding to a top displacement of about: (a) 200 mm; (b) 500 mm (failure).

concrete cracked in shear depending on fc (assumed equal to 0.5 in

these study cases) and θ is the angle of the inclined struts in the web
(assumed equal to 45°, that is cotgθ = 1), Asw is cross-sectional area of
the shear reinforcement and s is the spacing of the stirrups. The effect of
corrosion was included in the reduction of Asw and fc as well.

4.3. Structural analyses: Results

Relevant results are presented for both the case studies in terms of
base shear (Vbase) – top displacement (utop) curves; the results obtained
for X direction are reported, referring to the frame marked by green
color in Fig. 4. The achievement of ductile or brittle failures in the first
elements according to Eqs. (16) and (17) are reported in the Vbase – utop
curve (using for chord rotation capacity a yellow circle and for shear
capacity a black cross), indicating the family of elements where the
limit is reached (C for column and B for beam). In addition, other limit
Fig. 14. Comparison of base shear vs. top displacement curve for Building A conditions have been introduced, i.e. the achievement of the limit value
and Building B. of concrete and steel strain (respectively using magenta and cyan dia-
mond), the reaching of cracking and steel yielding limits (respectively
function of corrosion level. using a blue and a green square).
As far as the brittle mechanisms are concerned, shear demand was It is worth noting that the attainment of ductile or brittle failures in
compared with shear capacity for all elements evaluated according to the first element is not assumed corresponding to failure of the whole
Eurocode 8 [18] and Eurocode 2 [62], that for members with vertical structure, due to the ‘conventional’ nature associated with the ultimate
shear reinforcement is the following: limit condition. For this reason, the Vbase – utop curves are not stopped
when the first limit condition is reached but they continue up to a loss
zAsw f yw ⎞ of at least 15% of the peak base shear capacity. In correspondence of
V = min ⎛⎜zbw α c, w υ1 fc cotgθ ; ⎟ cotgθ
s (17) the different steps of the loading history, the state of the frame is re-
⎝ ⎠
presented in order to locate elements where the limit conditions are
where z is the lever arm of internal forces (in this study assumed equal reached, making easier the interpretation of failure mechanisms.
to0.9d ), bw is the width of the section, α c, w is the coefficient taking ac-
count the state of the stress in the compression chord, function of the 4.3.1. Sound condition
normalized value of axial stress, υ1 is a strength reduction factor for The results obtained for Building A in sound conditions are reported

11
L. Berto, et al. Engineering Structures 216 (2020) 110397

Table 6
Building B: attainment of the limit conditions in the first elements for columns and beams for different degradation scenarios in terms of base shear Vbase (expressed in
kN) and corresponding top displacement utop (expressed in mm).
Scen. CRACK YIELD εs ≥ εsu εc ≤ εccu θ ≥ θu V ≥ VR

Column Beam Column Beam Column Beam Column Beam Column Beam Column Beam
Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase Vbase
(utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop) (utop)

Sound 540.69 (20) 282.69 (9) 1894.27 (97) 1597.21 (77) – – 2549.01 (273) – 2498.09 (198) – 1528.54 (73) –
D1.CL1 364.47 (12) 274.75 (9) 1772.87 (91) 1573.32 (78) – – 2318.30 (221) – 2273.77 (161) – 1422.91 (69) –
D1.CL2 324.53 (11) 266.01 (9) 1619.57 (84) 1573.21 (81) – – 2076.55 (178) – 2022.62 (130) – 1266.33 (62) –
D1.CL3 286.59 (10) 286.59 (10) 1479.38 (78) 1552.27 (83) – – 1857.89 (143) – 1757.40 (103) – 1116.69 (55) –
D2.CL1 395.19 (13) 278.00 (9) 1833.14 (94) 1573.44 (77) – – 2443.05 (270) – 2349.84 (166) – 1522.61 (74) –
D2.CL2 334.53 (11) 274.31 (9) 1771.89 (91) 1554.26 (77) – – 2357.24 (257) – 2200.54 (141) – 1504.14 (74) –
D2.CL3 331.36 (11) 271.69 (9) 1713.69 (88) 1540.42 (77) – – 2299.05 (247) – 2040.84 (120) – 1424.53 (70) –

in Fig. 12 in terms of base shear – top displacement curve super-

imposing the points corresponding to the attainment of the limit con-
ditions above described, whose values are summarized in Table 5. As
expected, the structural response is ductile, with a roof drift of about
2.6% and a ductility factor close to 4.5 (evaluated as the ratio between
ultimate and yielding displacement) The mechanism is favorable with
plastic hinge formation first at beams and then on the columns’ base
sections (Fig. 13a), before the reaching of maximum base shear. When
the structure approaches the peak base shear, the strain in the most
compressed confined concrete fiber of some beams reaches ultimate
compressive strain εccu (Fig. 13b). Considering the post-processing
verifications, shear failure of beams and columns are not detected,
while ductile failure (i.e. reaching of chord rotation capacity) is at-
tained in some base columns and some beams. Nevertheless, strains in
both concrete and steel fibers do not reach maximum value in any
column.
Fig. 15. Building B – Sound condition. Base shear vs. top displacement curve. Fig. 14 shows the comparison of the results obtained for Building A

Fig. 16. Building B – Sound condition. Frame state corresponding to a top displacement of about: (a) 75 mm; (b) 275 mm; (c) 335 mm (failure).

12
L. Berto, et al. Engineering Structures 216 (2020) 110397

Fig. 17. Building A – Sound and Corroded conditions. Base shear vs. top dis- Fig. 19. Building B – Sound and Corroded conditions. Base shear vs. top dis-
placement curve for degradation scenarios D1.CL1, D1.CL2 and D1.CL3. placement curve for degradation scenarios D1.CL1, D1.CL2 and D1.CL3.

and Building B in sound conditions. The response of Building B is sig- concrete in one column of ground floor (Fig. 16b). At incipient failure
nificantly less ductile than in the previous case, with a roof drift of (Fig. 16c), almost all the columns have reached their shear capacity
about 1.7% and a ductility factor close to 3 corresponding to a ductility and, at ground floor, the ultimate compressive strain εccu .
difference of around 40%. Building B is, otherwise, stronger in terms of It is moreover evident the tendency to develop a soft story me-
maximum base shear (about +20% compared with Building A), ac- chanism with plastic hinges at both ends of the columns of the ground
cording to the different design philosophy adopted. Building B yields floor.
later than Building A, but the damage evolves faster. Peak displacement This behavior is opposite to the case of Building A, characterized by
(i.e. the displacement corresponding to peak base shear) is almost 40% a higher ductile response with diffused plastic hinges among all the
less than peak displacement of Building A: this means that the secant structural elements and the attainment of the ultimate concrete com-
stiffness at peak point in Building B is twice the value achieved in case of pressive strain εccu only in few beam elements of the lowest level.
building designed following capacity design approach, even if the in-
itial elastic stiffness is lower than that of Building A by about 20% (see
also Tables 5 and 6). Looking at the force displacement curves, it can be 4.3.2. Corroded condition
then concluded that Building B is stronger and stiffer than Building A Fig. 17 shows the base shear – top displacement curves for Building
although less ductile, as expected from the differences in design phi- A subjected to degradation distribution scenario D1 (i.e. uniform cor-
losophy. rosive attack on the columns of the ground floor of the buildings), with
As far as the attainment of the limit conditions is concerned, different values of current intensity and exposure times. It can be ob-
Figs. 15, 16 and Table 6 show that Building B exhibits an undesired served that the strength of the frame is reduced of about 6% in the
failure sequence with the achievement of shear capacity first in the worst case and the initial stiffness of about 4%. The occurrence of limit
columns close to the elastic threshold of the structure with the si- condition for the different degradation scenarios is reported in Table 5.
multaneous development of plastic hinges in only few beam elements at The comparison of the failure mechanisms in sound (Fig. 13) and in
the lower levels, corresponding to a top displacement as low as 75 mm corroded conditions (Fig. 18) does not highlight relevant differences in
and to a roof drift of about 0.4% (Fig. 16a). Increasing the load, the the structural performance except for the achievement of εsu at the base
maximum chord rotation is reached in a column of the ground level for of all the columns at ground floor even for low values of corrosion
200 mm of displacement, evidencing once again the brittle behavior of entity (Fig. 18a). As much as the entity of corrosion increases, the at-
the structure. tainment of εsu occurs for decreasing value of displacement. These re-
In correspondence of a displacement of about 275 mm (roof drift of sults agree with experimental findings which evidence failure of long-
about 1.4%), the ultimate compressive strain εccu is reached in confined itudinal reinforcement bars for exceedance of the steel ultimate
elongation in case of corroded well-confined columns (e.g. [31,33]). It

Fig. 18. Building A – Corroded condition. Frame state at incipient failure for degradation scenario: (a) D1.CL1; (b) D1.CL3.

13
L. Berto, et al. Engineering Structures 216 (2020) 110397

Fig. 20. Building B – Corroded conditions. Frame state at incipient failure in degradation scenario: (a) D1.CL1, (b) D1.CL3.

Fig. 21. Building B – (a) Floor displacement profile (b) inter-story drift distribution for a top displacement equal to 5 mm (before cracking).

Fig. 22. Building B – (a) Floor displacement profile (b) inter-story drift distribution for a top displacement equal to 143 mm (attainment of εccu for D1.CL3).

is worth noting that the first attainment of εsu somewhere in the ranges from 23% to 53%.
structure does not directly correspond to the attainment of ultimate In contrast to the case of Building A, Building B does not show at-
displacement of the whole structure due to distribution of load to less tainment of εsu . This is because (due to lower confinement effect) the
stressed longitudinal bars. failure of base section of columns is anticipated by crushing of confined
The results obtained considering the degradation distribution sce- concrete (as evidenced by reaching of ultimate compressive strain εccu )
nario D2 are presented only in Table 5. for both sound and corroded frames for all the corrosion levels (Fig. 20).
In the case of Building B, on the other hand, the effects of corrosion The failure mechanism does not show any relevant difference ranging
are more relevant, as shown in Fig. 19 for the degradation distribution from scenario D1.CL1 to D1.CL3, although failure occurs for progres-
scenario D1, where the strength and ductility drastically drop by in- sively lower top displacement values.
creasing the corrosion entity. The reduction of maximum base shear The effects of this degradation distribution scenario (i.e. corrosive
varies from 9% to 27% moving from lowest to highest entity (i.e. from attack on the columns of the ground floor) can be evidenced in Figs. 21
D1.CL1 to D1.CL3), the corresponding reduction of initial elastic stiff- and 22, which depict the profiles of floor displacement and the related
ness varies from 3% to 9%, and the reduction of ultimate displacement inter-story drift distribution along the building height for two

14
L. Berto, et al. Engineering Structures 216 (2020) 110397

Fig. 23. Building A – (a) Floor displacement profile (b) inter-story drift distribution for a top displacement equal to 5 mm (before cracking).

Fig. 24. Building A – (a) Floor displacement profile (b) inter-story drift distribution for a top displacement equal to 330 mm (attainment of εsu for D1.CL3).

in Figs. 23 and 24, and the effects of the corrosive attack are less evi-
dent, as expected by the design philosophy adopted, characterized by
the capacity design principles with strong columns and weak beams.
As concerns the degradation distribution scenario D2, the structural
behavior is less significantly affected. Nevertheless, in contrast to the
case of Building A, the reduction of both peak base shear and ultimate
displacement is not negligible with values ranging from 4% to 10% in
terms of peak base shear and from 3% to 2% in terms of ultimate dis-
placement (Fig. 25).

5. Conclusions

In the present work the effects of corrosion on the structural per-

formance of RC buildings were investigated; to be representative of the
actual building stock, two case study buildings typical of structures
Fig. 25. Building B – Sound and Corroded conditions. Base shear vs. top dis- designed before and after the adoption of seismic codes based on the
placement curve for degradation scenarios D2.CL1, D2.CL2 and D2.CL3. capacity design philosophy were considered, evidencing differences
concerning size of structural elements, structural details and collapse
significant values of top displacement (before cracking and at the modality.
reaching of ultimate compressive strain εccu , respectively). The results Corrosion effects were represented in terms of decrease of the me-
shown in Fig. 21b reveal that, in the initial elastic phase, the increase of chanical performance of both steel and concrete, adopting a coupled
the corrosion level up to D1.CL3 leads to an increase of the inter-story mechanical – environmental damage model for concrete as well as ca-
drift at ground floor up to 18% even if the elastic stiffness of the librating the reduction of cross section and ductility of reinforcing steel
building decreases only by 9%. As it can be seen in Fig. 22b, at the basing on the results of experimental tests performed in past research
reaching of ultimate compressive strain εccu , the inter-story drift at work. Besides, the modification of confinement due to the loss of ulti-
ground floor increases up to 70%. This result evidences that in the case mate deformation capacity of rebars was accounted for. The most re-
of Building B the corrosion on the ground floor columns leads to a highly levant simplification assumed in the work consisted in the assumption
non–uniform inter-story drift distribution along the building height and of perfect bond condition, sufficiently reliable for low/moderate cor-
a clear formation of a soft story. On the other hand, in case of Building rosion entity (such as in the considered cases). Different degradation
A, the inter-story drift are more uniformly distributed, as it can be seen scenarios, corresponding to different intensities of the corrosion attack
(CL1, CL2, CL3) and to different spatial distribution affecting columns

15
L. Berto, et al. Engineering Structures 216 (2020) 110397

(D1, D2) were analyzed. properties of corroded steel rebars. Constr Build Mater 2017;148:219–30.
Nonlinear static analyses were performed on bidimensional models [14] Inci P, Goksu C, Ilki A, Kumbasar N. Effects of reinforcement corrosion on the
performance of RC frame buildings subjected to seismic actions. J Perform Constr
representing the main frames of the case study buildings. Structural Facil 2013;27(6):683–96.
analyses performed in sound condition highlighted, as expected, the [15] Finozzi I, Saetta A, Budelmann H. Structural response of reinforcing bars affected by
different behavior of the two considered buildings: in case of adoption pitting corrosion: experimental evaluation. Constr. Build. Mater. 2018;192:478–88.
[16] Salvatore W, Caprili S, Braconi A, Finetto M, Bianco L, Ascanio C, Moersch J,
of capacity design approach, a global ductile collapse mechanism was Apostolopoulos CA, Ferreira Pimenta G. Effects of corrosion on low-cycle fatigue
achieved. On the contrary, the old code design approach leaded to a (seismic) behaviour of high-strength steel reinforcing bars (RUSTEEL) – RFSR-CT-
building with higher strength and stiffness but lower ductility, high- 2009-00023 - Technical Steel Research Series EUR 26687, European Commission -
Directorate General for Research, Bruxelles; 2014.
lighting the development of brittle shear mechanism at failure. [17] Caprili S, Moersch J, Salvatore W. Mechanical Performance vs. Corrosion Damage
The analyses in corroded conditions evidenced the detrimental ef- Indicators for corroded steel reinforcing bars. Adv Mater Sci Eng 2015. Article ID
fects of corrosion in the case of Building B (i.e. in buildings not designed 739625.
[18] UNI EN 1998-1:2005, Eurocode 8 - Design of structures for earthquake resistance-
following the ‘modern’ approach foreseen by current codes), with
Part 1: General rules, seismic actions and rules for buildings; 2005.
progressively higher decrease of the base shear capacity (up to about [19] Braconi A, Braga F, Caprili S, Gigliotti R, Salvatore W. Seismic demand on steel
30%) with the increase of corrosion entity. Buildings designed ac- reinforcing bars in reinforced concrete frame structures. B Earthq Eng
cording to old codes are not able to achieve the ultimate deformation of 2014;12:2633–64.
[20] Almusallam AA, Al-Gahtani AS, Aziz AR, Rasheeduzzafart. Effect of reinforcement,
steel reinforcing bars since – due to lower confinement – the main corrosion on bond strength. Constr Build Mater 1996;10(2):123–9.
mechanism is related to the achievement of the (lower) concrete ulti- [21] Prieto M, Tanner P, Andrade C. Bond response in structural concrete with corroded
mate strain in compression. This highlights the problem of existing RC steel bars. experimental results. In: Andrade C, Mancini G (Eds) Modelling of
Corroding Concrete Structures. RILEM Bookseries, vol 5. Springer, Dordrecht.
constructions, constituting most of building stock, for which the cor- [22] Berto L, Simioni P, Saetta A. Numerical modelling of bond behaviour in RC struc-
rosion attack increases the already significant seismic vulnerability. tures affected by reinforcement corrosion. Eng Struct 2008;3(5):1375–85.
On the other hand, buildings designed according to current stan- [23] Coronelli D. Corrosion cracking and bond strength modeling for corroded bars in
reinforced concrete. Aci Struct. J. 2002;99(3):267–76.
dards for constructions in seismic regions (i.e. respecting the principles [24] Lundgren K. Effect of corrosion on the bond between steel and concrete: an over-
of capacity design and detailing rules) are less affected by the corrosive view. Mag Concr Res 2007;59(6):447–61.
attack considered in this work, as it could be expected by the design [25] Meda A, Mostosi S, Rinaldi Z, Riva P. Experimental evaluation of the corrosion
influence on the cyclic behaviour of RC columns. Eng Struct 2014;76:112–23.
philosophy characterized by the capacity design principles with strong [26] Blomfors M, Zandi K, Lundgren K, Coronelli D. Engineering bond model for cor-
columns and weak beams. In this case, there is a slight decrease of the roded reinforcement. Eng Struct 2018;156:394–410.
base shear capacity and no substantial modifications in the resulting [27] Berto L, Vitaliani R, Saetta A, Simioni P. Seismic assessment of existing RC struc-
tures affected by degradation phenomena. Struct Saf 2008;31:284–97.
ductile failure mechanism, even if corrosion degradation leads to the
[28] Choe DE, Gardoni P, Rosowskya D, Haukaas T. Probabilistic capacity models and
achievement of the ultimate deformation of rebars at the base of the seismic fragility estimates for RC columns subject to corrosion. Reliab Eng Syst Safe
columns even for low values of corrosion entity. 2008;93:383–93.
It is worth noting that further research is necessary in order to [29] Berto L, Saetta A, Simioni P. Structural risk assessment of corroding rc structures
under seismic excitation. Constr Build Mater 2012;30:803–13.
generalize the outcomes of this study including also the effects of pit- [30] Lim S, Akiyama M, Frangopol DM. Assessment of the structural performance of
ting corrosion and of different and more detailed corrosion scenarios corrosion-affected RC members based on experimental study and probabilistic
(e.g. number and localization of pits, distribution of corrosion in the modeling. Eng Struct 2016;127:189–205.
[31] Goksu C, Ilki A. Seismic behavior of reinforced concrete columns with corroded
transverse confinement reinforcement). deformed reinforcing bars. ACI Struct J 2016;113(5):1053–64.
[32] Xia J, Jin WL, Li LY. Performance of corroded reinforced concrete columns under
Declaration of Competing Interest the action of eccentric loads. J. Mater. Civ. Eng. 2016;28(1). 04015087 1–16.
[33] Rajput AS, Sharmab UK. Corroded reinforced concrete columns under simulated
seismic loading. Eng Struct 2018;171:453–63.
The authors declare that they have no known competing financial [34] D.M.17/01/2018 (NTC2018): Norme Tecniche per le Costruzioni, G.U. SO n.42 20/
2/2018 [in Italian].
interests or personal relationships that could have appeared to influ- [35] D.M.16/01/1996: Norme Tecniche per le Costruzioni in zone sismiche G.U. n.29 5/
ence the work reported in this paper. 02/1996 [in Italian].
[36] UNI EN 1990-1:2006 UNI EN 1990:2006, Eurocode – Basis of structural design;
2006.
References
[37] NZS 1170.5:2004, Structural Design Actions, Part 5, Earthquake actions, New
Zealand, Standards New Zealand; 2004.
[1] Caprili S, Salvatore W, Valentini R, Ascanio C, Luvarà G. A new generation of high- [38] NZS 3101:2006, Parts 1 & 2: Concrete Structures Standard, Standards New Zealand;
ductile dual-phase steel reinforcing bars. Constr Build Mater 2018;179:66–79. 2006.
[2] Caprili S, Salvatore W, Valentini R, Ascanio C, Luvarà G. Dual-phase steel reinfor- [39] Petrini L, Pinho R, Calvi GM Criteri di Progettazione Antisismica degli Edifici, IUSS
cing bars in uncorroded and corroded conditions”. Constr Build Mater Press, Pavia, Italy; 2004 [in Italian].
2019;218:162–75. [40] Pedeferri P, Bertolini L. La durabilità del calcestruzzo armato, McGraw-Hill; 2000.
[3] Di Carlo F, Meda A, Rinaldi Z. Numerical evaluation of the corrosion influence on [41] Andrade C, Prieto M, Tanner P, Tavares F, D'Andrea R. Testing and modelling
the cyclic behaviour of RC columns. Eng Struct 2017;153:264–78. chloride penetration into concrete. Constr Build Mater 2013;39:9–18.
[4] Glasser FP, Marchand J, Samson E. Durability of concrete – degradation phenomena [42] Duracrete. The European Union-Brite EuRam III –Modelling Degradation; 1998.
involving detrimental chemical reactions. Cem Concr Res 2008;38(2):226–46. [43] FibBulletin 59 (2011) Condition control and assessment of reinforced concrete
[5] Saetta AV, Schrefler BA, Vitaliani RV. The carbonation of concrete and the me- structures exposed to corrosive environments (carbonation/chlorides).
chanism of moisture, heat and carbon dioxide flow through porous materials. Cem [44] EN ISO 15630-1:2010 Steel for the reinforcement and pre-stressing of concrete –
Con Res 1993;23(4):761–72. Test methods - Part 1: Reinforcing bars, wire rod and wire, 2010.
[6] Saetta A, Vitaliani R. Experimental investigation and numerical modeling of car- [45] EN ISO 6892-1:2009 Metallic materials - Tensile testing - Part 1: Method of test at
bonation process in reinforced concrete structures - Part I theoretical formulation. room temperature.
Cem Concr Res 2004;34(4):571–9. [46] Rodriguez J., Ortega L.M., Casal J., Diez J.M. (1996) Assessing structural conditions
[7] Saetta A. Deterioration of reinforced concrete structures due to chemical-physical of concrete structures with corroded reinforcement. In: Dhir RK, Jones MR, editors.
phenomena: model-based simulation. J Mater Civil Eng 2005;17(3):313–9. Concrete repair, rehabilitation and protection. E&FN Spon, 65–78.
[8] Cabrera JG. Deterioration of concrete due to reinforcement steel corrosion. Cem [47] Finozzi IBN, Berto L, Saetta A. Structural response of corroded RC beams: a com-
Concr Compos 1996;18(1):47–59. prehensive damage approach. Comput Concr 2015;15(3):411–36.
[9] Ožbolt J, OršanícF, BalabanícG, Kušter M. Modeling damage in concrete caused by [48] Berto L, Saetta A, Scotta R, Talledo D. A coupled damage model for RC structures:
corrosion of reinforcement: coupled 3D FE model Int J Fract; 2012. proposal for a frost deterioration model and enhancement of mixed tension domain.
[10] Jamali A, Angst U, Adey B, Elsener B. Modeling of corrosion-induced concrete cover Constr Build Mater 2014;65:310–20.
cracking: a critical analysis. Constr Build Mater 2013;42:225–37. [49] Berto L, Saetta A, Talledo D. Constitutive model of concrete damaged by free-
[11] Caprili S, Salvatore W. Cyclic behaviour of uncorroded and corroded steel re- ze–thaw action for evaluation of structural performance of RC elements. Constr
inforcing bars. Constr Build Mater 2015;76:168–86. Build Mater 2015;98:559–69.
[12] Zhang W, Song X, Gu X, Li S. Tensile and fatigue behavior of corroded rebars. [50] Berto L., Budelmann H., Finozzi I.B.N., Saetta A., Talledo D.A, (2015). Coupled
Constr Build Mater 2012;34:409–17. damage model for RC Elements Assessment Under Environmental Degradation, ACI
[13] Imperatore S, Rinaldi Z, Drago C. Degradation relationships for the mechanical SP-305 Durability and Suistanability of Concrete Structures, SP-305–5.

16
L. Berto, et al. Engineering Structures 216 (2020) 110397

[51] McKenna F, Scott MH, Fenves GL. Nonlinear finite element analysis software ar- plane frames includ-ing changes in geometry and non-elastic behavior of elements
chitecture using object composition. J Comp Civil Eng 2010;24(1):95–107. under combined normal force and bending. Proc., IABSE SympLibson, Portugal
[52] Mander J, Priestley M, Park R. Theoretical Stress-Strain model for confined con- 1973;13:15–22.
crete. J Struct Eng 1988;114(8):1804–26. [58] Scott MH, Fenves GL. Plastic hinge integration methods for force-based beam-
[53] Coronelli D, Gambarova P. Structural assessment of corroded reinforced concrete column elements. J Struct Eng 2006;132:244–52.
beams: modelling guidelines. J Struct Eng 2004;130(8):1214–24. [59] Paulay T, Priestley MJN. Seismic Design of Reinforced Concrete and Masonry
[54] Capè M. Residual service-life assessment of existing R/C structures. MS Thesis, Buildings, Wiley; 1992.
Chalmers University of Technology, Goteborg Sweden and Milan University of [60] Fardis MN. Seismic Design, Assessment and Retrofitting of Concrete Buildings-
Technology, Italy; 1999. based on EN-Eurocode 8, Springer; 2009.
[55] Molina FJ, Andrade C, Alonso C. Cover cracking as a function of bar corrosion: Part [61] EN 1998-3:2005, Eurocode 8—Design of structures for earthquake resistance—Part
II – Numerical model. Mater Struct 1993;26(9):532–48. 3: Assessment and retrofitting of buildings 2005.
[56] Model Code 2010 - Final draft, Vol 1. (2012) ISBN 978-2-88394-105-2. [62] EN1992-1-1:2005 Eurocode 2: Design of concrete structure - Part 1-1: General rules
[57] Menegotto M, Pinto PE. Method of anaysis for cyclically loaded reinforced concrete and rules for buildings.

17