Engineering Structures 162 (2018) 226–244

Contents lists available at ScienceDirect

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

Short communication

Damage and collapse mode of existing post tensioned precast concrete T

bridge: The case of Petrulla viaduct
⁎
Laura Anania , Antonio Badalà, Giuseppe D'Agata
Department of Civil and Environmental Engineering, University of Catania, Italy

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

Keywords: This paper describes an investigation into the collapse mode of a single-span, segmental post-tensioned concrete
Post-tensioned concrete beam bridge caused both by the effect of chloride induced corrosion in Prestressed Post Tensioned Bridge and by some
Collapse mechanism initial faults during tendons design. Collapse was determined by the breakage of the tendons and the wires inside
Corrosion the tendons appear completely rusted.
Manufacturing lacks
A numerical investigation was also carried out in order to determine the local mechanisms that led to the
collapse of a segment of the Petrulla Viaduct in the south of Italy (Sicily) after a widespread beam damage was
found. The main damage is the consequence of both an inadequate grout cementitious mortar filling, capable of
introducing chemicals by which corrosion travelled along the tendons, and an inadequate distance among the
ducts employed. The latter produces rotation of the wings around a point in the transversal section.

1. Introduction
Many of the bridges built in Italy are Prestressed or Post Tensioned
(PT) concrete girders where the tendons are the main load-carrying
elements. The advantages of PT bridges compared to bridges con-
structed using conventional reinforcement include greater span length,
structural efficiency, reduced materials, and a more streamlined ap-
pearance. Although prestressed concrete members were generally
manufactured with concrete of relatively higher strength, time has
shown that they are subject to the same adverse effects of reinforcement
corrosion as reinforced concrete members are. Corrosion can take many
forms beyond that which occurs with a great loss of material; for ex-
ample corrosion may occur by the interaction of corrosion and me-
chanical stress to produce a failure by cracking. This type of failure is
known as stress corrosion cracking, often abbreviated to SCC [1,2]. SCC
is an insidious form of corrosion; it produces a marked loss of me-
chanical strength with little metal loss; the damage is not obvious to
casual inspection and the stress corrosion cracks can trigger mechanical
fast fracture and catastrophic failure of components and structures.
Besides, when cracking is clearly a result of hydrogen embrittle-
ment, is called as hydrogen attack. In fact, hydrogen dissolves in all
metals to a moderate extent. It is a very small atom that fits in between
the metal atoms in the crystals of the metal [4]. Consequently it can
diffuse much more rapidly than larger atoms.
Hydrogen tends to be attracted to regions of high triaxial tensile
stress where the metal structure is dilated as the regions ahead of cracks
or notches that are under stress. The dissolved hydrogen then assists in
the fracture of the metal.
Documented cases of prestressed tension members failure as a result
of corrosion make this a most pressing problem. Since prestressed
concrete members rely on the tensile strength of the prestressing steels
to resist loads, loss of even few wires or strands per member [3] could
result catastrophic.
Damage during service, frequently appearing after several years or
decades of exploitation, is usually most consequential. The lacking or
insufficiently alkaline protection already from the beginning or its loss
due to carbonation and/or depassivation after chloride attack are the
major causes of later damage or even of a failure. Responsible for that
are usually the failures caused by shortcomings in planning and/or
execution as well as inaccurate or inefficient structural measures.
Execution faults and construction errors concern e.g. the injection of the
ducts with mortar in case of post tensioned concrete (a mortar-free
section of tendons is exposed to the risk of corrosion, when penetration
of moisture is possible and oxygen can enter the duct space), the con-
crete technology (too small concrete cover and too low concrete
quality, under certain conditions not protect. The prestressing tendons,
the procedure (technology) of the production of the structural elements,
as well as waterproof sealing (not present or damaged) and drainage
(damaged), can strongly affect the durability of PT elements.
The use of poor quality materials and poor construction practices
accelerate and aggravate the problem of corrosion, so that actually, the
annual cost of corrosion was estimated to be very high [5–8].

⁎
Corresponding author.
E-mail addresses: lanania@dica.unict.it (L. Anania), abadala@dica.unict.it (A. Badalà), gdagata@dica.unict.it (G. D'Agata).

https://doi.org/10.1016/j.engstruct.2018.02.039
Received 22 November 2016; Received in revised form 4 January 2018; Accepted 12 February 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
L. Anania et al.
So, a brittle collapse of the bridge deck may be caused by a number
of design deficiencies (individually or combined), namely: (i) in-
adequate duct venting, (ii) incomplete duct filling, (iii) strands pressing
against the duct interior surface, (iv) strand congestion, (v) subsidence,
or (vi) poor consistency with segregation [13,14].
In Italy, research on steel corrosion [9] is more focused on re-
inforced concrete (RC) structures than on Post tensioned concrete
structures. PT structures adequately designed and constructed have
been generally considered highly durable because the prestressing
tendons would be protected from corrosion by the duct filling. But,
some deterioration problems due to corrosion have been discovered in
some existing bridges, raising serious concerns about the long-term
durability [10] of PT bridges. In this kind of concrete structures, the
high stress level in the tendons strongly modifies the steel corrosion
process [11,12].
Macrocell corrosion with a local anode and a large cathode fre-
quently occurs in chloride induced corrosion of rebars in concrete and is
responsible for very high local corrosion attacks and reduction in cross-
section. In PT beam with inadequate grout and protection of the ten-
donds, the process is equal to the mild reinforcement corrosion: in fact,
the mild reinforcing in concrete is normally protected from corrosion by
the passive film formed at the steel/concrete interface inside the alka-
line cementitious matrix. But, if the passivity is compromised, than
active corrosion may occur at an unacceptable rate. For instance, this
passivation can be eliminated either by a decrease in the pH value
(pH < 9) due to carbonation, or by the presence of chloride salts,
which initiate an expansive corrosion of the reinforcing steel and
eventually damage the surrounding concrete.
The importance of chloride ion in the corrosion of steel in concrete
has been developed into the concept of chloride threshold level.
Chloride threshold level is defined as the chloride content at steel depth
which is necessary to sustain local passive film breakdown and to in-
itiate corrosion process [15–17]. So, chloride threshold level can re-
present a key element in predicting the life service of structures exposed
to environmental conditions [18].
Concrete structures such as bridges, buildings and other reinforced
concrete structures might suffer severe damages due to macrocell cor-
rosion [19–21].
2. The case studied: Petrulla bridge collapse
The case studied is a forensic analysis for the collapse of the Petrulla
viaduct. The viaduct is constituted by n. 13 simple supported deck with
span variable from 35 to 50 m long. It is located in south of Italy, in in a
city, Licata, of Sicily island at both ‘Salso river’ and sea proximity. The
name of the river derived by the Italian word ‘salsedine’ that means
saltiness. The environmental condition can be considered as aggressive.
The viaduct was built in the early ‘80s. The original as-built project
were approved in 1985. It was designed according to the Italian Code in
force at the time of the executive project (D.M. 1980) [28]. The viaduct
was mainly devoted to vehicular traffic concerning to the transport of
raw materials for agriculture as fertilizers and similar. In fact, it is in
countryside, far away from urban center. The Fig. 1 shows the global
aerial view of the bridge and its location to the river proximity.
So this viaduct is constituted by multiple bridge span. Each single
span is made up of four longitudinal post tensioned double tee beams.
The superstructure of the viaduct is supported by ten piers. The sche-
matization of the typical deck span is represented in Fig. 2. Concrete
slab completes the bridge deck.
In August 2014, a span of the viaduct suddenly collapsed. The col-
lapse mode consisted in forming a plastic hinge in the mid-span of a
bridge beam as shown in Figs. 3 and 4.
3. On site inspection of Petrulla viaduct
The inspection, after collapsing, revealed fractured strands and
227
Engineering Structures 162 (2018) 226–244
failure of the prestressing tendons as a consequence of their dete-
rioration through corrosion (Figs. 5 and 6). In the mid span of the
collapsed span, the complete dissolution of some strands in tendons
numbered as No. (number) 4 and 5 can be observed. But the collapse
was due also to some design lacks in the PT beam (i.e.: no gap presence
among the ducts and inappropriate grout for prestressed concrete ten-
dons). As reported in Fig. 5 the tendon numbered as 4 and 5 and ori-
ginally composed by 12 strands, now present only 7 strands. The
breakage of the tendons produced the expulsion of the ducts anchorages
at the head beam due to the release of elastic energy (Figs. 7 and 8).
4. Analysis of Petrulla viaduct executive project
According to the original project, the single longitudinal beam was
pre-stressed by means of five tendons realized by twelve strands of
0.6 in. each. So according to Standard Production Italian Specifications
Law Act 1086-1971 and Following Revisions for the 7 wire strands, the
strand nominal diameter and the nominal area will be 15.20 mm and
140 mm2 respectively. Then, each tendon exhibits a cross section (or
equivalent diameter) of 1680 mm2. The tendons layout is derived from
the original executive project and represented in Fig. 9 in which the
position of each tendon (in figure called ‘cavo’ according to the Italian
language) and the corresponding anchor-head regions are represented.
It’s worth to note that for the “cavo 4” and “cavo 5” (called tendon No.4
and tendon No. 5 in the following) the anchor-head region is on the
extrados of the deck beam. Fig. 10, shows the strand disposition in the
mid span. The pretension of the tendons was applied in progressive
steps i.e. No. 1–3, tendons were fully tensioned at the beginning, before
the bridge erection, while No. 5–6 tendons were tensioned up to 35% of
the specified tensile strength after the transversal beams construction.
The other 65% of the specified pretension was applied after the casting
and hardening of the deck slab.
During the construction, tendons no. 4 and no. 5 of Fig. 10 were
tensioned at the end, to achieve 100% of the specified pre-stress to be
applied in the tendons. To this aim some slots at the extrados of the
deck slab had to be filled by grout after tensioning (Fig. 11). Thus, a
volume of potential accumulation of water for saturation of the con-
crete, is created and this allows the entrance of aggressive ions.
This is partly due to the shrinkage during the curing of the concrete
used to fill the holes, which results in the formation of cold joints at the
interface between the old and new concrete. This is necessary to com-
plete the existing slab, from which bleed water infiltrates and perco-
lates, accumulating in the tank arranged in the prefabricated beam in
the cap, up to the saturation of the mortar injection of the cables.
From the original project, in the cross section at the mid span of the
PT beam, the tendons, placed in a duct of 70 mm in diameter, converge
into a narrow space (Fig. 10), which does not meet the minimum gap
required by the international codes (ACI 318 – R 18.16, Grout for
bonding prestressed tendons) [24]. Therefore, a right constipation of
the concrete during the cast is not ensured. This is unable to develop
and disseminate the tangential stresses due to the adherence [22,23].
According to Eurocode 2 [25], in fact, for ducts of 70 mm, the required
gap among the tendons must not be less than the duct diameter. In the
case studied, no gap was found among the tendons and this caused the
formation of voids as shown in Figs. 12 and 13. As will be explained and
shown in detail in the section devoted to the FEM analysis, these gaps
caused the implosion of the cross section of the deck, with inward ro-
tation of the lower flanges, resulting in expulsion for tensile of the mild
reinforcement placed along the section web. A uniform rust distribution
can be noted along both the tendons and the duct.
The internal tendons located inside the structural concrete section,
are housed in corrugated metal ducts, and bonded to the structural
concrete by means of cementitious grout. In the past, there have been
no problems due to using grout, capable of filling the interstitial spaces
between tendons and ducts for the bonding of prestressing tendons,
with portland cement as the cementing material capable of preventing
L. Anania et al.
Salso River
Mediterranean sea
Fig. 1. Global aerial view of the bridge.
any introduction of chemicals listed as harmful to tendons.
Nowadays, International Standards list the basic requirements for
the grout for prestressed concrete structures [26,27]. In fact, while the
first line of corrosion protection of grouted PT tendon relies on ade-
quate sealing of ducts from external sources of corrodants (i.e., water,
air, Cl–, and carbon dioxide), the second line of corrosion protection
relies on direct contact of strands with a cementitious grout for which a
high pH (> 13) is maintained. In situ inspection, on the collapsed span
shows the presence of an inadequate grout, consisting of a clear mortar
(Fig. 15) and not of portland cement (dark grey color) as required by
international standards. As we can observe in the collapsed bridge beam
span, the mortar was completely crumbled or absent in many points
(Fig. 14), because it lacked consistency. This would have been im-
possible for a cement mortar of suitable resistance (not less than
30 Mpa). This kind of material favored the collection of chloride and
moisture inside the tendons, especially in the presence of air due to the
voids as well as to the aggressive environment. Corrosion induced a
variation of the cross-sectionional of the individual tendons in PT
Fig. 2. Typical deck span of the bridge (geometrical characteristics are in cm).
228
Engineering Structures 162 (2018) 226–244
Collapsed Span
Petrulla Viaduct
bridge, which induced variations in tensile forces among the tendon
and neutral axis depth with a conseguent failure of one or more strands
and a re-distribution of the applied load.
An important warning sign of the imminent collapse could have
been noticed by a severe and characteristic crack pattern, present in all
the other adjacent bridge spans [26]. Figs. 16 and 17 show the cracking
induced by the implosion of the section, premonitory of collapse, due to
the combined effect of both the corrosion and internal voids because of
no gap among the ducts.
5. Hardened concrete characterization
The main physical-mechanical characteristics of the materials em-
ployed were investigated by means of an accurate campaign of tests.
The object of this investigation was both the collapsed span (No. 10) as
well as the nearby span identified in the project with the number 9. A
set of concrete cores for the estimation of the actual compressive
strength have been tested. To this aim, some concrete cores with
L. Anania et al. Engineering Structures 162 (2018) 226–244

Fig. 3. Collapse mechanism of the bridge.

Fig. 6. Tendons in the bridge after collapse.

Once the cores were drilled, a solution of phenolphthalein was ap-

plied to a fresh fractured core surface in order to measure the carbo-
nation depth. Phenolphthalein is a white or pale yellow crystalline
material. For use as an indicator it is dissolved in a suitable solvent such
as isopropyl alcohol (isopropanol) in a 1% solution. Where the solution
remains colorless, the pH of the concrete is below 8.6, suggesting car-
bonation. A fully-carbonated paste has a pH of about 8.4. Otherwise, if
the indicator turns purple, the pH is above 8.6, and the sample is car-
bonation free. Table 3 reports the carbonation depth measured on the
drilled cores as shown in Fig. 22. The drilled cores were then cut and
prepared for mechanical test as shown in Fig. 23.

5.1. Compressive strength of the concrete

The characteristic strength fck of drilled core samples, having a

Fig. 4. Brigde beam: mid span fracture.
Fig. 5. Fractured strands on Tendons No. 4 and 5 conditions after collapse.
100 mm nominal diameter were drilled and extracted (Fig. 18) directly
both from the collapsed span (No. 9) and the nearby span (No. 10) as
highlighted in Fig. 19 and according to the plan described below in
Table 1 (Figs. 20 and 21) (see Table 2).
nominal diameter equal to the height between 100 and 150 mm, can be
assumed as equivalent to the cubic strength Rck related to a specimen of
150 mm side. This, in accordance with the specification of the section
C.11.2.6 of the Italian Guidelines (Instruction for the application of
D.M. 14 January 2008 Italian Code for the Constructions), regarding
the evaluation of the mechanical characteristics of hardened concrete
by means of destructive tests.
Besides, according to the specification of UNI EN 12504-1:2002,
UNI EN 12504-2:2001, UNI EN 12504-3:2005, UNI EN 12504-4:2005
[26], some main factors must also be considered when relating the core
strength to the Actual strength such as: direction of drilling, shape,
capping, effect of drilling, curing of the core, moisture etc.
In this case studied, the cores extracted from the collapsed span
show a quite high cubic compressive strength; with a minimum value
equal to 57.9 MPa and a maximum one equal to 82.0 MPa. The cubic
estimated average strength value, of the concrete employed in the
collapsed span, is equal to 69.78 MPa even if the correction regardless
of the damage is neglected.
Tables 4 and 5 report the laboratory values obtained directly by the
tests, where Mv is the Volumic Mass, Dmax is the maximum diameter of
the aggregate; W is the sample weight; D is the core diameter; H is the
core height; A is the cross section area, C is the load applied; Rc is the
Compressive strength measured.
According to the mandatory rules code in force at the time of the
executive project (D.M. 2008-section 11.2.6 [28]), the average value of
the compressive strength of a core Rcm, determined in accordance with
EN 12504-1, is typically lower than the equivalent strength of a stan-
dard cube or cylinder specimen matured under laboratory conditions.
The mandatory rules, in fact, fix a value not less than 85% of the

229
L. Anania et al. Engineering Structures 162 (2018) 226–244

Fig. 7 and 8 . (7) Expulsion of the ducts head anchorages. (8) Global view of the expulsion at the head of the girder beam.

Fig. 9. Prescribed pattern of the tendons along the PT beam in the original project (cavo = Tendons).

Cold joints

Fig. 11. Detail of the head anchorage holes for the tendons numbered as 4 and 5.

Fig. 10. Mid span cross section of the PT beam derived from the original project
(cavo = Tendon).

230
L. Anania et al.
Fig. 12. No gap among the tendons.
Fig. 13. Congestion of strands in the mid span of the bridge beams.
Fig. 14. View of the tendons in collapsed span.
average value defined in the design phase Rcm,design. In the case
studied the compressive strength Rck was equal to 40 MPa.
According to 2008 Italian Specifications an average compressive
cubic strength of the core (Rcm) defined by the following formula (1) is
allowed:
R cm = 0.85 × R cm,design = 42.16 MPa
where Rcm,design can be evaluated from the formula (2):
Engineering Structures 162 (2018) 226–244
Fig. 15. View of the non-Portland cement grout for the bonding of tendons and advanced
corrosion of both duct and strands.
Fig. 16. Cracking in the other bridge spans.
Fig. 17. Detailed view of vertical mild reinforcement expelled by the rotation of the lower
flange.
R cm,design = R ck + 8/0.83 = 40 + 9.6 = 49.6 MPa (2)
In which the 0.83 coefficient takes into account the correlation
between cubic and cylindrical strength.
In the specific case of the collapsed span an average cube strength
(1) equal to 69.78 was found. The actual strength Rcm, span
n.10 = 69.78 MPa is greater than 85% of the average cubic compressive
strength of the design.
231
L. Anania et al. Engineering Structures 162 (2018) 226–244

Fig. 18. Sample extraction of the collapsed span (a) slab, (b) I beam.

1
2
3 4
11
5 6 7 8 9 10

Fig. 19. Bridge span investigated.

Table 1
Localization of in site drilled core for the compressive strength estimation-collapsed span.
Collapsed span 10
Sample for compressive strength assessment
n.3 concrete drills (D = h=100 mm) on n. 3 beams;
n.3 concrete drills (D = h=100 mm) on the slab;
Samples for Elastic Modulus determination (UNI 6556)
n.3 concrete drills (D = 60; h = 150 mm) on n.3 beam;
n.3 drills (D = 60; h = 150 mm) on the slab
The concrete used was of excellent quality, and would correspond to
a class of resistance far superior to that provided for the project (Rck
40 MPa), equivalent to approximately Rck 72 MPa. A low dispersion in
the results is visible in the sample of the collapsed span. This is prob-
ably due to a not continuous control of the concrete batching at the
time of the construction.
Similar considerations can be made also for spans that have not yet
collapsed, where resistance values not lower than 48.9 MPa were ob-
tained, with an average value of 68.75 MPa corresponding to theore-
tical cubic design strength of 71.0 MPa.
C1-C2-C3
S1-S2-S3 5.2. Young modulus assessment
E1-E2-E3 The determination of the E secant modulus of elasticity, is carried
Es1-Es2-Es3 out in accordance with UNI 6556/76 [29], by the measurement of the
deflection increments corresponding to the application of loading and
unloading cycles (at least three) up to a maximum test strength equal to
about one third of the compressive strength of the concrete. To de-
termine secant elastic modulus we starts from a base stress σ0, equal to
1/10 of the maximum stress of σn test. The experimental data is sum-
marized in Tables 6 and 7, where abbreviation D, H and A have the
same meaning given above and σi are the intermediate stress values
recorded during the tests.

Fig. 20. Localization of the sample collapsed span (No.10).

232
L. Anania et al. Engineering Structures 162 (2018) 226–244

Fig. 21. Localization of the sample on span No. 9.

Table 2
Localization of in site drilled core for the compressive strength estimation- collapsed span.

Span no 9:

Sample for compressive strength assessment

n.3 concrete drills (D = h=100 mm) on no 3 beams; C4-C5-C6
n.3 concrete drills (D = h=100 mm) on the slab S1-S2-S3

Samples for Elastic Modulus determination (UNI 6556)

No. 3 concrete drills (D = 60; h = 150 mm) on the slab E4-E5-E6
n.3 drills (D = 60; h = 150 mm) on the slab Es4-Es5-Es6

Table 3
Carbonation depth on drilled cores.

Sample Sample H Carbonation Sample Sample H Carbonation depth

[mm] depth [mm] [mm] [mm]
Fig. 23. Samples ready for test.
A B A B

C1 320 30 30 S1 360 0 25 apparent that the tensile strength of concrete is almost equal to 1/16 of
C2 180 25 25 S2 330 7 10
the compression one. Figs. 24 and 25 show, respectively, a specimen in
C3 200 30 25 S3 370 8 30
C4 160 38 – S4 360 6 –
testing machine and a typical broken sample after the test.
C5 180 35 – S5 330 6 –
C6 240 35 35 S6 370 6 –
6. Chloride content in the grout in the ducts

A B
Fig. 22. Example of Phenolphthalein on the drilled core.
The value of the modulus of elasticity becomes useful in the cal-
culation process.
5.3. Tensile strength estimation on drilled core
The determination of the value of the tensile strength of the con-
crete drilled cores were carried out according to the UNI EN 12390-6
[30] Tensile splitting strength of test specimens. Tables 8 and 9 summar-
ized the data obtained for the collapsed span and No. 9, respectively
where F is the load applied and fct is the tensile strength measured. It is
The grout inside the ducts might also act as a secondary protection
to the PT tendon. To this aim, a necessary condition for strand pro-
tection is that Cl is maintained below a critical concentration–, either as
a background grout contaminant or from an external source.
The European Standards UNI EN 206 at sect. 5.2.6 list a range be-
tween 0.1 and 0.2 wt (wt) percent cement as the acid soluble Cl− limit
for PT grout or prestressed concrete. American Standard [31,32] lists
this upper limit as 0.08 wt% cement. This threshold was also found in
the literature [33,34]. In sufficient concentration, Cl– facilitates corro-
sion of PT strand by transitioning steel from a passive state, for which
corrosion rate is negligible, to an active one where corrosion rate may
be unacceptably high and reducing grout electrical resistivity. The aim
of this section is to highlight possible deficiencies of the grout employed
during the bridge construction. From the standpoint of strand corro-
sion, a worst case scenario arises in situations where the grout is sub-
jected to repetitive wetting (presumably from periodic water infiltra-
tion from a source external to the duct) and drying. In the case studied,
during the rains periodic water infiltrations occur from the cold joint at
the anchor-head region of tendons No. 4 and 5. Asphalt drainage waters
transport hydrocarbon deposits and various chemicals which percolates
along the tendons.
According to UNI 9944 [35] and UNI 196-2 [36] the sulfates and
chlorides content can be assessed. All samples were dried in an oven at
105° ± 5°, and then cooled and treated with an acid solution. In the
specific case the sample showed a concentration of chlorides of 0.24%
weight cement, higher than both the European and American standard
limits. Samples also show a concentration of sulfates equal to 1.77
lower that UNI EN 197-1:2006 threshold.

233
L. Anania et al. Engineering Structures 162 (2018) 226–244

Table 4
Ultimate strength for drilled concrete cores- collapsed span.

Sample Physical features Geometrical features Mechanical features

Mv [kg/m3] Dmax [mm] W [g] D [mm] H [mm] A [mm2] C [N] Rc [N/mm2]

C1 2245 24 1460 94.7 92.3 7044 424,200 60.2

C2 2315 22 1536 94.7 94.2 7044 448,800 63.7
C3 2263 26 1498 94.7 94.0 7044 408,000 57.9
S1 2321 22 1537 94.7 94.0 7044 549,000 77.9
S2 2384 22 1579 94.7 94.0 7044 542,400 77.0
S3 2346 24 765 74.3 74.3 4336 355,700 82.0

7. Mild reinforcement and strands mechanical characterization
Samples of both mild reinforcement bars and tendons were taken
from the viaduct in order to carry out a series of tests aimed to the
assessment of the tensile strength, the nominal diameter as well as for
chemical investigations.
7.1. Tensile strength determination
For the mild reinforcement a yielding strength fy higher than
600 MPa was measured. This value is in good accordance with modern
steel reinforcement belonging to B450C class according to Italian
Codes.
Three samples of the prestressing steel 1000 mm long, were taken
from three parts of the collapsed span, to measure the nominal diameter
(Aeq), ultimate tensile strength (fpt), 0.2% proof strength, (fp(0.2), 1%
proof strength (fp(1)), the modulus of Elasticity (E) as well as elongation
(Agt). The latter mechanical characteristics, were assessed by an ex-
tensometer gauge 600 mm long. The data obtained are summarized in
Table 10 where the symbol Tn=1,2,3 indicates to the sample tested.
Sample T3 was taken from most damaged strand of the broken tendons
at the mid span of the beam. The least damaged part of the broken
strand was selected to this aim, samples T1 and T2 were taken from the
head of the girder collapsed beam (Fig. 26).
A drastic cross section area reduction is evident in the T3 sample as
shown in Fig. 27. Moving away from the damaged zone there was a
reduction up to 26.42% of the original area of the some strands found in
situ and of the cross section area (equivalent diameter) of the tendon.
7.2. SEM analysis on the strands
In the specific case the SEM analysis carried out on the extracted
samples was finalized to demonstrate the presence of sodium chloride
as consequence of a misuse, at the construction time, of not appropriate
water in the grout mixture. The SEM analysis, in fact, aims to perform a
chemical analysis of the sample and it is performed by means of a
scanning electron microscope. A beam of electrons accelerated by an
electric field, and focused by an electromagnetic lens system, strikes the
sample and explores a specific surface by means of a scanning device.
These signals include secondary electrons and backscattered electrons;
the former are commonly most valuable for showing morphology and
topography on samples, while the latter are most valuable for illus-
trating contrasts in composition in multiphase samples. X-ray genera-
tion is produced by inelastic collisions of the incident electrons with
electrons in discrete orbital (shells) of atoms in the sample.
Characteristic X-rays are produced for each element in a mineral that is
“excited” by the electron beam.
The report of data obtained for the sample extracted from the col-
lapsed span, is in Figs. 28 and 29. The presence of steel oxide confirms
that an active corrosion occurred on the tendons; this was accelerated
by the complete dissolution of the ducts, so many strands get particu-
larly severe conditions due to the air-water interface caused by the
voids among the tendons. Therefore, poor intergranular or transgra-
nular cracks are visible in the microstructure of the fractured surface,
such as that stress corrosion cracking phenomena can be considered as
not responsible of the collapse of the bridge [37]. In fact, Stress Cor-
rosion Cracking (SCC) determines a minimal overall loss of the mate-
rial. On the contrary, in the case study, a great loss of the overall ma-
terial was found.
8. FEM Analysis
The study and the comprehension of the collapse mechanism of the
span, as well as the mechanisms of spread damage due to the implosion
of the PT beam, was carried out by means of a non-linear FEM analysis,
using the SAP 2000 software [38].
The numerical investigation was carried out by referring to two
main models where many different cases were considered for each
model.
The mechanical properties implemented in the numerical model
were derived directly from the tests carried out on the concrete cores,
the mild reinforcement and the strands.
The first model considers the whole 3-dimensional structure of the
PT beam. Firstly, in order to validate the original project results, the
bridge was firstly analyzed by accounting for the design values of the
employed materials, then by considering the average material strength
under the effect of the corrosion as obtained from laboratory tests.
The second model studied only the transversal section of the studied
beam was developed with the average material strength and accounting
for the corrosion effect and the mistakes noted after the collapse

Table 5
Ultimate strength for drilled concrete cores-span No. 9.

Sample Physical features Geometrical features Mechanical features

3 2
Mv [kg/m ] Dmax [mm] W [g] D [mm] H [mm] A [mm ] C [N] Rc [N/mm2]

C4 2337 24 1536 94.0 94.7 6940 394,900 56.9

C5 2293 22 1507 94.0 94.7 6940 451,800 65.1
C6 2249 26 1516 95.2 94.7 7118 348,000 48.9
S4 2469 22 1595 93.2 94.7 6822 510,000 74.8
S5 2385 22 1601 95.0 94.7 7088 613,100 86.5
S6 2352 24 1579 95.0 94.7 7088 569,500 80.3

234
L. Anania et al. Engineering Structures 162 (2018) 226–244

Table 6
Elastic modulus for drilled concrete cores-collapsed span.

Sample Geometrical features σ1 σ2 σ3 σ4 E

D [mm] H [mm] A [mm2] 6.6 [MPa] 11.1 [MPa] 15.7 [MPa] 20.2 [MPa] Modulus [MPa]

E1 74.3 185.1 4335.8 30,528 27,567 26,761 26,260 27,779

E2 74.3 185.1 4335.8 24,969 24,451 25,040 25,337 24,949
E3 74.3 185.1 4335.8 21,596 22,682 22,811 23,030 22,530
Es1 74.3 185.1 4335.8 48,215 45,939 44,261 42,571 45,246
Es2 74.3 185.1 4335.8 32,972 30,743 29,592 28,865 30,543
Es3 74.3 185.1 4335.8 35,108 31,698 30,887 30,210 31,976

8.1. The 3-D bridge model Table 8

Tensile Strength for drilled concrete cores-collapsed span.
A 3-D reference model was calibrated according to the loads and the
Sample Geometrical features Mechanical features
results derived from the as-built calculations.
Loads applied are those listed in Section 3 of the Italian Standard D [mm] H [mm] A [mm ] 2
F [N] fct [MPa]
D.M. 1980 [30]. According to the listed combination of loads. This re-
B1 102.0 94.7 9659 75,000 4.9
ference model was calibrated with the as-built calculations.
B2 102.2 94.7 9678 69,300 4.6
A series of numerical models, in which the mechanical character- B3 102.6 94.7 9716 66,100 4.3
istics of the materials employed (concrete, steel and strands) were as-
sumed to be equal to the as-built project ones, aimed to validate the
calculation carried out by the designer. The data obtained proved that Table 9
at the time calculations the strengths in play in the tendons were the- Tensile Strength for drilled concrete cores-span No. 9.
oretically correct both at the initial stage and at the service stage. There
Sample Geometrical features Mechanical features
was, however, an incompatibility between the provisions in theory and
what was actually feasible and executed. For example, the structural D [mm] H [mm] A [mm2] F [N] fct [MPa]
designer valued precisely, the stress in concrete, with a surprising ac-
B4 94.7 94.0 8902 64,000 4.6
cordance to the three-dimensional model, and each type of immediate
B5 94.7 94.1 8911 58,000 4.1
and time dependent losses and the relaxation in the steel. Mistakenly, B6 94.7 95.1 9006 59,000 4.2
the designer did not consider that the inserting of five 5 ducts of 70 mm
wide in the position planned in the executive project could not comply
with the limit distances among the ducts as defined in the code.
In all 3D models both the longitudinal and transversal beams, which
define the bridge structure, were modelled by using “shell layered” fi-
nite elements. These permit us to take into account also the presence of
the steel reinforcement in the concrete. The tendons were modelled by
Frame finite elements connected to the shell elements by links simu-
lating the grout interface region. During each stage of non linear nu-
merical analysis the stiffness of these links were changed such as to
simulate the degradation of the mechanical characteristics of the grout
employed and consequently the different adherence conditions at the
interface with the tendons.
A global view of the model is reported in Fig. 30. Fig. 31 shows the
shape of the modelled tendons of the studied bridge beam.

8.2. Model for materials

The mechanical behavior of the concrete, was modelled by Mander’s

unconfined constitutive law material in which characteristic values
Fig. 24. Tensile test set up.
were derived from direct tests on drilled core samples. The non linearity
of the constitutive model was used even accounting for the creep, fluage
and relaxation. The Elastic modulus adopted for the beams was

Table 7
Elastic modulus for drilled concrete cores- span No. 9.

m Geometrical features σ1 σ2 σ3 E

2
D [mm] H [mm] A [mm ] 6.6 [MPa] 11.1 [MPa] 15.7 [MPa] Modulus [MPa]

E4 74.3 185.1 4335.8 28,733 27,391 26,893 27,672

E5 74.3 185.1 4335.8 25,643 26,830 25,695 26,056
E6 74.3 185.1 4335.8 23,992 22,378 22,800 23,037
Es4 74.3 185.1 4335.8 34,111 34,688 34,035 34,278
Es5 74.3 185.1 4335.8 35,029 37,089 35,704 35,941
Es6 74.3 185.1 4335.8 34,877 32,723 33,688 33,763

235
L. Anania et al. Engineering Structures 162 (2018) 226–244

Fig. 25. Sample after test.

Table 10
Parameter of the strands in situ.

Sample Deff Aeff [mm2] fp(0.2) fp(1) fpt [MPa] Agt [%] E [MPa]
[mm] [MPa] [MPa]

T1 13.39 140.78 1786 1728 1836 1.8 193,171

T2 13.41 141.29 1743 1708 1859 4.8 197,032
T3 12.87 130.03 993 1046 1074 0.6 183,197

Fig. 28. Strand T4 -D position – Collapsed span-Lowest tendon.

E = 25.000 MPa the compressive strength was equal to Rcm, span n.

10 = 69.78 MPa and tensile strength to fct = 4.6 MPa.
Fig. 26. Tendons samples.
Mild reinforcement constitutive law adopted, was SAP A615G60
material, adapted by the user by which the averaged values derived
from the tests. A yielding strength of 430 MPa was adopted.
Further, the stress strain law used for the strands and tendons was
SAP A416GR270 material, switched by the user with the ultimate
tensile strength (fpt), 0.2% proof strength (fp(0.2), 1% proof strength
(fp(1)), the modulus of Elasticity (E) obtained from the laboratory tests
carried out on the samples.
The grout interface was simulated by the links of which stiffness was
assigned accounting for the mechanical properties of a calcareous
mortar employed in the as- built bridge. Corrosion resistance of the
prestressing steel, in fact, depends upon the grout material [39]. The
poor quality of the mortar was proved byits brittle behavior. Similar
mortars typically of Eastern Sicily were object of a wider study by
Anania & Co. It can be assumed that for a calcareous mortar an initial
compressive strength not higher than 0.5 MPa [40,41] to be reduced by
35% for environmental moisture conditions [42]. Thus, a compressive
strength between 0.25–0.30 Mpa seems to be acceptable at the end of
the numerical analysis. Other parameters can be assumed as follow:
Fig. 27. Strand details. ultimate strain ɛu = 3.5%0 (as concrete), secant modulus of elasticity
E = 75–85 MPa, shear modulus G = 65 MPa. The correlated shear

236
L. Anania et al. Engineering Structures 162 (2018) 226–244

Fig. 29. Strand T5 – Collapsed span-Tendon in C Position.

Fig. 30. Global view of the collapsed bridge model.

237
L. Anania et al.
Fig. 31. Tendons modelling by frame element.
stiffness of the links with a step of 2 m along the tendons can be eval-
uated from literature relationship kGA/s, where k is a constant equal to
5/6, and A is the area of the grout inside the duct, defined by the dif-
ference between the duct area and the tendon area. A final adherence
value equal to 166 kN/mm can be assumed by referring to a stiffness of
the link inputted with a one meter step.
8.3. Load applied
sect. 3.1 of the Bridge Design Specifications code in force at the time
of the executive project, requires to consider all the following load
paths:
Permanent loads are defined as loads and forces that are either
constant or varying over a long time interval upon completion of con-
struction. They include dead load of structural components and non-
structural attachments (g1), dead load of wearing surfaces and utilities
(g2), downdrag forces (ɛ1), force effects due to creep force effects
caused by shrinkage (ɛ2).
Transient loads are defined as loads and forces that are varying over
a short time interval. A transient load is any load that will not remain
on the bridge indefinitely.
This includes vehicular live loads (q1) and their secondary effects
including dynamic load allowance (q2), braking force (q3), centrifugal
force (q4). Additionally, there are pedestrian live loads, force effects due
to uniform temperature, and temperature gradient, force effects due to
settlement (SE), wind loads on structure and live loads (q5), earthquake
loads (q6), friction forces (q7), collision forces (q8).
Dead and live weight are essentially vertical loads, whereas forces
from nature may be either vertical or horizontal.
The Vehicular live loads to be considered are those indicated below
as q1A,B,C,D,E,F:
q1a and q1b are design lane loads to be applied centrally on a single
lane. According to sect. 3.4.1. of Italian code DM 1980, it must be
assess on a lane 3.50 mt large, as:
q1A = 28.9 + 52/L kN/m when L < 40 m;
in the case studied the span length was L = 38 m, so
q1A = 40.26 kN/m. Now, considering a lane 3.50 m large, the final
value is, 11.96 kN/m2;
q1B = 22.3 – L/500 when 15 < L < 400 m;
in the case studied the span length was L = 38 m, so
q1B = 21.54 kN/m. Now, considering a lane 3.50 m large, the final
value is 6.03 kN/m2.
The q1C is a three-axle vehicle, 550 kN, q1D is a three-axle vehicle,
310 kN, q1E is a two-axle vehicle, 10 kN to be applied on 0.7 m2 and
q1F = 4 kN/m2 is live pedestrian load considered as uniform load
accounting for the presence of large crowds, parades, and regular
use of the bridge by pedestrians.
Pretension in the tendons was applied in the case studied, according
to the original project, by considering the loss of stresses after 30 years
from the bridge construction time.
238
Engineering Structures 162 (2018) 226–244
The value applied in the SAP model was equal to −5.75E−3 m/m
in terms of strain load, corresponding to a stress of 1150 MPa.
The Limit States of sect. 3.13 of the D.M. 2 Agosto 1980 are require
to combine the individual loads with specific load factors to achieve
design objectives.
8.4. Modelling of the collapsed PT bridge element
The study of such a complex model allowed us to better understand
the real behavior that leads to the bridge span collapse as the con-
sequence of the corrosion of the strands and as the resulting redis-
tribution of the stresses inside the beam. Under these conditions, in fact,
the allowable tensile strength of the concrete material employed is
overcome, resulting in a brittle fracture mechanism where the strands
are definitively broken due to the very low amount of mild steel re-
inforcement. In other terms the beam realized with its faults doesn’t
show any post peak inelastic resource. Collapse mechanism occurs
mainly due to load transferring from oxidized strands to the concrete,
than for the overcoming of the tensile strength in the concrete. Once the
crack was formed the load was transferred to the reduced cross section
of the strands which suddenly collapsed.
The model run with a Non linear Staged construction analysis; it en-
ables the definition of a sequence of the design construction stages in
which structural systems and load patterns are added or removed, and
time-dependent behaviors are evaluated, including creep, shrinkage,
aging (change in elastic modulus with age), and tendon relaxation.
Material and geometric nonlinearity may be applied to staged con-
struction.
Further, staged construction is considered as part of a sequence of
nonlinear static analysis load cases. For linear load cases, the structural
stiffness at a given construction stage may serve as the basis for ana-
lysis.
For instance, initial stages considered were: (i) a first stage where
the beams were added, (ii) a second stage where pretension was as-
signed at the inner tendons, (iii) a third stage where the bridge slab was
added, (iv) a fourth stage where the transverse beams were added, (v) a
fifth stage where the loads were assigned to the tendons No. 4 and 5 (vi)
a sixth stage where live loads were applied on the slab. According to the
determination of Italian standard of 1980, a live distributed load of
14.48 kN/m2 was applied on the first lane while a live distributed load
of 73,256 kN/m2 was applied in the second lane as represented in
Figs. 32 and 33. Those live static load amounts include also the per-
centage of the vehicular dynamic load allowance, as specified by sect.
3.1 of the D.M. 2 August 1980 in order to capture the “bouncing” effect
and the resonant excitations due to moving trucks. Besides the uni-
formly distributed load of 2.6 kN/m2 was also considered.
Then the other stages must be considered: (i) a seventh stage where
creep is considered for fifty years, (ii) 8th stage where both the nominal
diameters of the tendons were reduced to 90% of the original one to
simulate the corrosion of the strands and the links stiffness were
changing to simulate the grout mortar degradation.
The cross section change was applied to the portion of the strands
L. Anania et al.
Fig. 32. Load case stage 3–live load applied on single lane of the bridge deck.
indicated in Fig. 35 where corrosion is concentrated as the evidence of
the in situ inspection.
After this stage, if the numerical analysis converges, a new stage is
considered by reducing gradually, the cross section of the tendons from
90% up to 39%. Sap code automatically calculate their relaxation. At
each stage the analysis continues from the state at end of previous non
linear stage as schematized in Fig. 34.
More than fifty models were studied by varying the number of the
corroded tendons (only the No. 1–2–3- or all the tendons). Fig. 35 in-
dicated the portion of tendons where the corrosion was changed during
the numerical analysis. The percentage of the cross section area in re-
spect to the non-damaged one, and the stiffness of the links due to the
deterioration of the mortar grout were modified with a consequent
reduction of the adherence between strands and concrete.
No convergence is found when the cross section of the tendon be-
comes equal to 39% of the original one, never under only dead load.
The reduction of 61%, equal to 660 mm2 corresponding to an equiva-
lent tendon diameter of Φ 29 mm and an estimated adherence value of
16.5 ton/mm per meter, agrees with the tendon residual diameter di-
rectly evident during on site inspection. So, it can be said that it
Fig. 33. Live load applied on single lane of the bridge deck.
239
Engineering Structures 162 (2018) 226–244
represents the closest mechanism to the real collapse mode
In the case cited above, the longitudinal stress contour reported in
Fig. 36 is obtained at a step load very close to the collapse. The cyan
zone shows where the tensile stresses overcome the characteristic ten-
sile strength of the concrete employed. In Fig. 37 a longitudinal shear
gradient is visible with a change of the sign of the shear strength. That
occurs in the same zone where a great damage was observed in the real
structure.
8.5. Local damage model
The model used for the non linear FEM analysis (Fig. 38) represents
half of the lower portion of the cross section of the PT concrete beam,
symmetrical with respect to the vertical axis. That is, a portion of a
thickness of 20 cm correspondent to the step of the vertical mild re-
inforcement in the span, provided in the as-built project drawings and
detected in situ. The model was implemented by Sap2000 [38] Code
using Shell Layered elements for the section and frame element for the
mild reinforcement. They are connected by Links gap reacting only in
compression. In order to highlight the post-expulsion of the vertical
L. Anania et al.
Fig. 34. Example of load case where the tendons cross sections was changed.
Fig. 35. Portion of the tendons under corrosion.
bars as well as the complex phenomena of damage observed in situ, the
concrete cover was not considered because it was carbonated, fragile
and very thin.
The constraints (in green) restore the symmetry conditions. In the
local model, loads have been applied in order to simulate the real
conditions of stresses, induced in the analyzed beam part, both by the
reactive load of the post tensioned tendons (pushing from the bottom
upwards) and by the shear stresses due to shear and torsion. The former
is equivalent to a nodal vertical load of 5.8 kN each. The latter equal to
0.3 MPa were applied as surface loads acting in the face of the cross
section. They were assessed by the difference of the maximum shear
pattern, highlighted in Fig. 36 with the cyan1 color, between section
20 cm distant. The horizontal shear stresses were neglected.
The constitutive laws of the materials employed in the local model
are exactly the same as the global model of the bridge.
Fig. 39 gives back principal tensile stresses in the concrete. A
maximum value of 4.916 Mpa, higher than the experimental average
1
For interpretation of color in Fig. 36, the reader is referred to the web version of this
article.
240
Engineering Structures 162 (2018) 226–244
LICATA
value for the concrete reported in Table 8, was reached.
So cracks occured along the bridge beam as documented by Figs. 40
and 41.
The cracking pattern of Figs. 40 and 41 is also proved by the de-
formed scheme of Fig. 42, where the expulsion of the stirrup bars is
evident in yellow color. A detailed view of the expulsed stirrups along
the bridge span is represented in Fig. 42.
The numerical investigation confirms the nature of the severe da-
mage pattern found in all the spans of the Petrulla viaduct. Moreover,
the gradual corrosion of both ducts and strands (Figs. 43 and 44), to-
gether with the deterioration of the inadequate mortar grouted in the
duct, produced the following effects: (i) an increase of the voids for the
strands corrosion and reduction of the tendon cross section up to 39%
of the as built bridge tendons; (ii) a reduction in the upward thrust
provided by the tendons, due to the great loss of the pre-stress force
associated with re-distribution of tension; (iii) increasing of the shear
stress in the beam section, due to the same stress re-distribution and to
the decreasing of the vertical load component offered by pre-stressed
tendons.
L. Anania et al. Engineering Structures 162 (2018) 226–244

Breaking zones.
Tensile cracks due
to the overcoming of
the tensile resistance
of the concrete.

LICATA

Fig. 36. Normal stresses at load step close to the collapse.

Fig. 37. Longitudinal shear stress gradient on the global model of the collapsed bridge.

Fig. 38. Scheme of the portion of the cross sectional studied.

241
L. Anania et al. Engineering Structures 162 (2018) 226–244

Fig. 39. Principal tensile stress-global and detailed view.

Fig. 40. Cracking on the bridge span due to tensile tress and to the expulsion of the stirrups.

Fig. 41. Zoom on the expulsed stirrups.

242
L. Anania et al. Engineering Structures 162 (2018) 226–244

Stirrup

Fig. 42. Deformed scheme of the portion studied.

9. Conclusions

The application of post tensioned concrete technology in civil

structures is relatively recent. Therefore, the existing post-tensioned
concrete members in such structure are still relatively young, and the
corrosion and the concrete deterioration problems associated with this
type of concrete members have became clearly evident only in the early
1980s. However, Post Tensioned (PT) tendons can be susceptible to
corrosion and ultimate failure if physical deficiencies (PDs) or chemical
deficiencies (CDs) are present.
Global behavior and local mechanism which caused the collapse of
the Petrulla viaduct is explained in this paper. Local mechanism was
deducted by means of the study of the mid-span cross section. This
study was carried out by taking into account both the design faults in
the executive project related to no gap presence among the ducts and
the induced corrosion of the tendons.
In fact, the realization of the anchor-head region related to no. 4 and
no. 5 tendons resulted in the preparation of rectangular holes in the
predalles employed for the construction of the deck slab.
These holes are the preferential points of infiltration of water from
the road super-structure that leaks both outside the beam and inside the
Fig. 43. Advanced deterioration of the tendon with a drastic reduction of the cross sect.
interstitial space between duct and strands until reaching, by infiltra-
tion through porous concrete, the other tendons numbered as 1–2–3,

243
L. Anania et al.
Fig. 44. Absence of the grout inside the duct.
when approaching the mid-span. So, the induced corrosion on the
strands produces a reduction of the tendons cross section and a loss of
the pre-stress in the tendons. Concrete has to face this event with a
redistribution of the stresses in the cross section, but the presence of
initial imperfections in designing, creates voids among the tendons
which permit the rotation of the lower flanges of the cross section. The
tensile strength of concrete is overcome and cracks occur.
The damage phenomena does not occur during the initial and ser-
vice stage of the bridge, thanks to the effect of reactive loads (shear load
due to the prestressing force) offered both by the tendons and by the
adhesion between the latter and the duct. It is subsequently powered by
the corrosion of the tendons joined to the prestress losses, the latter,
also increased by the consequent not negligible overall loss of material
of the strands cross section. It is well known that one of the primary
causes of tendon failure in anchor assemblies is the failure of the grout
protecting the tendons. Voids occurring at the time of construction
discussed above, allow moisture and water to penetrate into the PT duct
and corrode the cables. The presence of an inappropriate grout also
eliminates any protection of the cable. In the specific case the presence
of chlorides in grouts represents a contribution to tendon failure.
Thus, an increase of the shear stress gradients in the cross section
occured, induced by the shear and torsion produced by global external
loads. The same gradients caused the inward rotation of the bottom
flanges with the conseguent crack patterns visible in the other bridge
span. Flexural failure occurs suddenly.
This study was undertaken to help bridge owners to better under-
stand the cracking condition of the bridge during its livespan.
References
[1] Arup H, Parkins RN. Stress corrosion research. NATO; 1979.
[2] Ashby MF, Jones DRH. Introduction to properties and applications, engineering
materials, 2 vols. Pergamon; vol. I, 1980, vol. 2, 1986.
[3] Newman RC. In: Marcus P, Oudar J, editor. Corrosion mechanisms in theory and
practice. Dekker; 1995.
[4] Perrin M, Gaillet L, Tessier C, Idrissi H. Hydrogen embrittlement of prestressing
cables. Corros Sci 2010;52(6):1915–26.
[5] Sreedharan E. India to invest Rs 200,000 crore in Metro rail in 10 years. The
Economic Times; 2009.
[6] Hays FG. Now is the time. World Corrosion Organization; 2011. < http://www.
corrosion.org/ > [last accessed on March 7, 2013].
[7] Lieser MJ, Xu J. Composites and the future of society: preventing a legacy of costly
corrosion with modern materials. Owens Corning; 2010. < http://www.
ocvreinforcements.com/pdf/library/CSB_Corrosion_White_Paper_Sept_10_10_
English.pdf > [last accessed on March 7, 2013].
[8] Castel A, Vidal T, Zhang R, François R, Sirivivatnanon V. Corrosion initiation and
propagation in the service life of reinforced concrete in chloride environment.
Proceedings of the international RILEM workshop on integrated service life mod-
elling of concrete structures. 2007. 239_46.
244
Engineering Structures 162 (2018) 226–244
[9] Nürnberger U. Corrosion induced failure mechanisms of prestressing steel. Mater
Corros 2002;53:591–601.
[10] Whitmore D, Arnesen T, Pailes B. Investigating and resolving bridge grouted PT-
strand corrosion problems. In: Structures congress 2017 – ascelibrary.org; 2017.
[11] Cavell DG, Waldron P. A residual strength model for deteriorating posttensioned
concrete bridges. Comput & Struct 2001;79:361–73.
[12] Vu A, Castel A, François R. Global behaviour prestressed beam after broken pre-
stressing cables by corrosion. 1st international conference on modern design, con-
struction and maintenance of structures. 2007.
[13] Godart B. Status of durability of post-tensioned tendons in France. Durability of
post-tensioned tendons. In: Proceedings of FIB workshop, Ghent, Belgium; 2001.
p. 25.
[14] Fuzier JPh. Development of grout and grouting techniques. Durability of post-ten-
sioned tendons. In: Proceedings of FIB workshop, Ghent, Belgium; 2001. p. 173.
[15] Choi YS, Kim JG, Lee KM. Corrosion behavior of steel bar embedded in fly ash
concrete. Corros Sci 2006;48:1733–45.
[16] Pour-Ghaz M, Burkan IO, Ghods P. The effect of temperature on the corrosion of
steel in concrete. Part 1: simulated polarization resistance tests and model devel-
opment. Corros Sci 2009;51:415–25.
[17] Ann KY, Song HW. Chloride threshold level for corrosion of steel in concrete. Corros
Sci 2007;49:4113–33.
[18] Polder RB, Peelen WHA. Characterisation of chloride transport and reinforcement
corrosion in concrete under cyclic wetting and drying by electrical resistivity. Cem
Concr Compos 2002;24:427–35.
[19] Chaussin R, Chabert A. Strategies for improvement—approach in France. Durability
of post-tensioned tendons. In: Proceedings of FIB workshop, Ghent, Belgium; 2001.
p. 235.
[20] Bertolini L, Carsana M. High pH corrosion of prestressing steel in segregated grout.
In: Modeling of corroding concrete structures, RILEM series 5. Cedex, France; 2011.
p. 147–58.
[21] Lau K, Lasa I, Parades M. Corrosion development of PT tendons with deficient grout:
corrosion failure in ringling causeway bridge. Draft. Gainesville (FL): Florida
Department of Transportation State Materials Office; 2011.
[22] Minh Ha, Mutsuyoshi Hiroshi, Niitani Kyoji. Influence of grouting condition on
crack and load-carrying capacity of post-tensioned concrete beam due to chloride-
induced corrosion. Constr Build Mater 2007;21(2007):1568–75.
[23] Rinaldi Zila, Imperatore Stefania, Valente Claudio. Experimental evaluation of the
flexural behavior of corroded P/C beams. Constr Build Mater 2010;24:2267–78.
[24] ACI Committee 222. Guide to design and construction practices to mitigate corro-
sion of reinforcement in concrete construction. Report no. 222.3R-11. American
Concrete Institute; 2011.
[25] EN 1992-1-1. Eurocode 2: design of concrete structures Part – 1-1: general rules and
rules for buildings.
[26] UNI EN 12504-1 (Prelievo sul calcestruzzo nelle strutture – Carote – Prelievo, esame
e prova di compressione). UNI EN 12390-1 (Prova sul calcestruzzo indurito –
Forma, dimensioni ed altri requisiti per provini e per casseforme).
[27] D.M. 2 Agosto 1980 Gazzetta Ufficiale 10 Novembre 1980, n. 308 Criteri generali e
prescrizioni tecniche per la progettazione, esecuzione e collaudo di ponti stradali.
[28] D.M. 14 gennaio 2008 (G.U. 4 febbraio 2008 n. 29 - S. O. n. 30) Nuove Norme
Tecniche per le Costruzioni.
[29] UNI 6556:1976. Tests of concretes – determination of static modulus of elasticity in
compression; 1993.
[30] UNI EN 12390-1-2 Prova sul calcestruzzo indurito.
[31] American Concrete Institute. Protection of metals in concrete against corrosion. ACI
222R-01. Farmington Hills, MI Farmington Hills, MI; 2010.
[32] European Committee for Standardization. Grout for prestressing tendons—basic
requirements. European Standard EN 447, Brussels, Belgium; 2007.
[33] Alonso C, Andrade C, Castellote M, Castro P. Chloride threshold values to de-
passivate reinforcing bars embedded in a standardized OPC mortar. Cem Concr Res
2007; 30: 1047. [Importante per stabilire le soglie].
[34] Sanches AN, Sagüés AA. Chloride threshold dependence on potential in reinforced
mortar. Paper no. C2012-0001728. Presented at CORROSION/12. Houston (TX):
NACE International; 2012.
[35] UNI 9944:1992 Corrosion and protection of reinforcing steel in concrete – de-
termination of the carbonation depth and of the chlorides penetration profile in
concrete.
[36] UNI EN 196–2:2013 method of testing cement – Part 2: chemical analysis of cement
– Italian Standards.
[37] Haskins Richard, White Barry, Ebeling Robert, Evans James. Relating corroded
seven-strand, posttensioned cable cross-sectional properties to load capacity. 3909
Halls Ferry Rd., Vicksburg (MS 39180, USA): U.S. Army Engineer Research and
Development Center; 2016.
[38] SAP 2000 – computers and engineering – software and consulting.
[39] Karthikeyan, Palaniswamy N. Evaluation of performance of grout materials in
protection of prestressing steel. Int J Electrochem Sci 2008;3(2008):315–24.
[40] Anania L, D’Agata G. Limit analysis of vaulted structures strengthened by an in-
novative technology in applying C-FRP. Constr Build Mater J 2017;145:336–46.
http://dx.doi.org/10.1016/j.conbuildmat.2017.03.212 [ISSN: 0950-0618].
[41] Anania L, Badalà A. The use of carbon fiber cementitious matrix connection for the
strengthening of multiple-leaves masonry. In: Kumar K, editor. Proceeding of
MuRiCo 2017 Bologna 28-June. Reinhardstrasse 18 (CH-8008 Zurich, Switzerland):
Trans Tech Publications Ltd; 2017.