Construction and Building Materials 160 (2018) 598–609

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Study of the loss of bond in reinforced concrete specimens with

accelerated corrosion by means of push-out tests
Beatriz Sanz a,⇑, Jaime Planas a, José M. Sancho b
a
Universidad Politécnica de Madrid, Dep. de Ciencia de Materiales, ETS de Ingenieros de Caminos, Profesor Aranguren 3, 28040 Madrid, Spain
b
Universidad Politécnica de Madrid, Dep. de Estructuras de Edificación, ETS de Arquitectura, Avda. Juan de Herrera 4, 28040 Madrid, Spain

h i g h l i g h t s

Loss of bond due to reinforcement corrosion is studied by means of push-out tests.

The specimens are slices of concrete prisms reinforced with a smooth steel tube.

The prisms are uniformly corroded in accelerated tests up to several corrosion depths.

Adhesion and friction are the main resisting mechanisms during push-out of the tube.

Corrosion depth affects residual stress and a dilatant behavior is detected.

a r t i c l e i n f o a b s t r a c t

Article history: In this work, push-out tests are presented to study loss of bond caused by corrosion of steel in reinforced
Received 21 April 2017 concrete. The specimens are concrete prisms reinforced with a smooth steel tube, thus, interlocking of bar
Received in revised form 2 August 2017 ribs is avoided, and adhesion and friction are analyzed. The prisms were uniformly corroded up to various
Accepted 17 November 2017
corrosion depths. During push-out, widening of the main crack developed during accelerated corrosion
was measured. For the corrosion levels applied in this work, the corrosion level influences the residual
stress at the end of the test, and the stress and crack widening are related, showing a dilatant behavior.
Keywords:
Ó 2017 Elsevier Ltd. All rights reserved.
Reinforced concrete
Accelerated corrosion
Bond
Cracking
Fracture mechanics

1. Introduction mens with deformed bars at four stages of corrosion, by means of

Bond between steel and concrete is crucial for an adequate
behavior of reinforced concrete structures. However, when corro-
sion of the steel occurs, an oxide layer develops that modifies the
properties of the steel-concrete interface and diminishes the net
cross-sectional area of reinforcement, thus, reducing the ultimate
strength of affected concrete members and the service life of the
structure [1]. Moreover, cracking and spalling of the concrete cover
can occur, due to the volumetric expansion of the oxide with
respect to the sound steel [2,3], resulting in loss of confinement
and, thus, loss of bond.
From previous studies found in the literature, essential aspects
of the influence of corrosion on bond behavior have been disclosed.
For example, Al-Sulaimani et al. [1] studied the behavior of speci-
⇑ Corresponding author.
E-mail addresses: beatriz.sanz@upm.es (B. Sanz), jaime.planas@upm.es
(J. Planas), jose.sancho@upm.es (J.M. Sancho).
concentric pull-out tests and bending tests, and found that for
specimens with a low degree of corrosion, bond strength (the max-
imum pure shear stress transmitted between the steel and the con-
crete) increases, due to the generation of a firmly adhered rust
layer, which increases the roughness of the surface; however, for
high corrosion levels, bond strength decreases, due to development
of a heavy oxide layer with a lubricating effect, degradation of bar
ribs, which diminishes interlocking and bearing of bar deforma-
tions against concrete, and formation of longitudinal cracks, which
diminishes confinement. Further work developed by Almusallam
et al. [4] showed results in agreement with those in [1], and
assessed that cracks developed in corrosion tests and rib profile
degradation affect bond strength. They also explained that bond
strength increases in pre-cracked specimens may be due to an
increase in the reactionary confinement of the bar in the concrete,
as the corrosion product develops an expansive mechanical pres-
sure on the surrounding concrete, and reported that hoop tensile

https://doi.org/10.1016/j.conbuildmat.2017.11.093
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609
stresses widen the corrosion cracks. An increase in bond strength
for a given small amount of corrosion and a decrease for severe
corrosion were also reported for bars corroded under natural con-
ditions before being cast in concrete by Maslehuddin et al. [5], for
specimens corroded by immersion in a solution of Ca(OH)2 by Fu
and Chung [6], and for specimens corroded under a controlled con-
stant current by Coccia et al. [7]. Other authors have investigated
influence of stirrups for specimens reinforced with deformed bars
and with smooth bars, as Fang et al. [8], and the influence of cor-
roded stirrups, as Hanjari et al. [9] and Juarez et al. [10], finding
that loss of bond due to corrosion is smaller for specimens with
stirrups, since they are the main source of confinement after crack-
ing of the cover, or have focussed their attention on the influence of
the corrosion method, as Choi et al. [11]. As a summary, three main
resisting mechanisms are identified that contribute to bond
strength: adherence, which is the chemical bond strength between
the steel and the concrete and is limited by the weakest of the
strength of the steel-oxide-concrete layers, friction, which depends
on the surface characteristics and on the confinement, and geomet-
rical interlock, due to steel ribs. However, differences are found in
the values reported in those works for the critical corrosion level,
maximum bond stress and slip of the bars, which may be explained
by the differences in the conditions of the tests and in the geome-
try of the specimens, which also determine the resisting mecha-
nism contributing to bond strength. For example, most of those
works utilized pull-out tests based on the ASTM-C234 standard
[12], either directly [1,5–7], or with modifications [8,9], but other
types of test were used as well trying to reproduce more closely
the actual state of stress in the structure, such as bending tests
[1,10] and cantilever bond tests [4]. As a result, parameters
obtained from a given experiment may not be adequate to repro-
duce the behavior in other situations in which the dominating
mechanisms of bond potentially differ from those in the
experiment.
Another important aspect in the tests is the condition during
accelerated corrosion. As pointed out by Prieto et al. [13], current
densities much higher than those observed in natural conditions
may yield unrealistic values of bond strength, since density of cur-
rent affect the type of oxide generated, as indicated in reference
studies of corrosion-induced cracking [14]. In this regard, in [15]
limitations of accelerated corrosion tests in reproducing the actual
behavior of the reinforcement were pointed out and it was recom-
mended that such tests should be avoided, except for fundamental
research on the corrosion process, as in the case of the current
paper. Besides, other variables such as the humidity and the tem-
perature also influence the corrosion process. In summary, corro-
sion introduces additional uncertainties in the determination of
bond-slip relationship, as explained in [16]. Therefore, it is essen-
tial to perform tests with repeatable conditions.
To decouple the various resisting mechanisms and to study the
effect of corrosion on them, Cairns et al. [17] designed a test to con-
trol the amount of confinement, using specimens with lateral steel
plates, to which controlled normal and shear forces were applied,
and calculated the friction coefficient between the steel and the
concrete, finding that the prime factor for changes in bond due
to corrosion may be the confinement pressure of the bar, rather
than changes in friction characteristics. As another example,
Ouglova et al. [18] designed specimens to avoid lateral confine-
ment, with a polished bar as reinforcement and with a window
to measure the displacement of the bar by digital image correla-
tion, detecting interface opening normal to the interface for high
corrosion levels. However, further work is required to characterize
individually the effect of corrosion on the mechanisms contribut-
ing to bond, which motivated this work.
In the current paper an experimental study is offered with the
main purpose of providing fundamental research on loss of bond
599
between the steel and the concrete due to reinforcement corrosion,
focusing on adhesion and friction. In particular, bond behavior is
studied by means of push-out tests of smooth reinforcement, using
the experimental device developed in [19]. The specimens are con-
crete slices cut from prisms reinforced with a calibrated steel tube,
which were corroded under a constant current in such a manner
that uniform corrosion was expected to occur along the tube
length, as explained in [20,21]. Since in the cut slices the steel tube
does not protrude from the concrete, push-out instead of pull-out
was selected [19], which has the advantage of allowing using a
simple loading device, as explained later in the paper. Moreover,
since the reinforcement in the tests is a smooth tube, instead of a
bar, interlocking between bar ribs and concrete is avoided, and
adhesion and friction can be studied. During the tests, the applied
load, the displacement of the tube and the additional opening of
the main crack which was developed through the cover during
accelerated corrosion were recorded. This allowed analyzing the
maximum shear stress, residual stress and loss of confinement of
the specimens. Results are presented for specimens corroded up
to three different corrosion depths, as well as for virgin specimens.
It should be noted that the reinforcement of this work differs
from that of concrete structures; thus, different values of the shear
stress may be obtained. The geometry of this reinforcement,
though, was essential in this study to disclose essential aspects
of the bond behavior of the steel-oxide-concrete system. Further-
more, the presented method could be applied to study the behavior
of specimens for other test conditions.
This paper is organized as explained next: following to this
introduction, Section 2 presents the main aspects of push-out tests
and the design of the experimental devices, Section 3 describes the
procedure of preparation of the slices for bond tests, including the
technique used for accelerated corrosion of the specimens, Sec-
tion 4 discusses the results, and Section 5 summarizes the conclu-
sions of this work. Finally, details on the processing of results are
given in Appendix A.
2. Basis of push-out test
2.1. Context
Push-out tests presented in this work have been carried out
within the framework of a general study of cracking of concrete
due to reinforcement corrosion [19–21]. In that study, concrete
prisms reinforced with a calibrated steel tube, as those sketched
in Fig. 1(a) and (b), were artificially corroded under a constant cur-
rent, using the experimental setup presented in [21], with such
conditions that uniform corrosion was expected to occur along
the length of the tube, as further explained in Section 3. After the
tests, each specimen was cut into five slices, as sketched in Fig. 1
(c); the two external slices were discarded, and the central slice,
labeled as 3 in the figure, was subjected to the push-out test. The
two remaining slices were impregnated with fluorescent resin, fol-
lowing the method presented in [20], to study the crack pattern of
the specimens, which is out of the scope of this paper. During
accelerated corrosion, a main crack developed through the cover,
together with several thin secondary cracks, as sketched in Fig. 1
(d), which was assessed from the inspection of the impregnated
slices under ultraviolet light. In parallel to the experiments,
numerical simulations of the accelerated corrosion tests were con-
ducted, using a model that reproduces the cohesive fracture of con-
crete by means of elements with an embedded adaptable crack
[22] and the mechanical behavior of oxide by means of expansive
joint elements [20]. The comparison of the numerical and experi-
mental curves of crack width, variation of inner diameter and vari-
ation of inner volume of the tube permitted disclosing the best
600 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609
90
14.5
(a)
90
50 50
(b)
Fig. 1. Sketch of the specimens before accelerated corrosion and dimensions in mm (a) and (b), cutting of the specimens into slices after accelerated corrosion (c), and cross-
sectional area of the slice used in push-out tests (d). Normally, the central slice—labeled as 3 in (c)—was used in push-out tests, while the slices labeled with 1 and 5 were
discarded, and the slices labeled with 2 and 4 were used in other tests out of the scope of this paper.
values for the oxide parameters developed under those test condi-
tions [21].
2.2. Geometry of the specimens
Fig. 1(a) and (b) show the dimensions of the prisms from which
the slices used in push-out tests were obtained. The prisms had a
cross-section of 90  100 mm2 and a height of 90 mm. They were
reinforced with a seamless precision carbon steel tube compliant
with the EN 10305-1 standard [23], 1.0 mm thick and 20 mm in
diameter. The tube had its axis lying in one of the planes of sym-
metry of the prism, and a cover of 20 mm, to enforce a dominant
mode of failure during accelerated corrosion, consisting of a single
main crack through the cover [20,21].
After accelerated corrosion, each prism was cut into slices, as
explained in Section 2.1 and sketched in Fig. 1(c). Normally the
central slice of each prism was used in push-out tests. It had an
average thickness of 20 mm, and a cross-section as that sketched
in Fig. 1(d), with a main crack at the cover and several thin sec-
ondary cracks surrounding the specimen, and the dimensions indi-
cated in Fig. 1(b).
2.3. Design of push-out test
Since the specimens in the bond tests are concrete slices and
anchoring of the tubes is not possible, a device was specially
designed and built to push the tube out of the concrete, as sketched
3
5
1mm thick
steel tube
(c)
20
20
A B
main crack thinner cracks
(d)
in Fig. 2. This configuration has the advantage of allowing using a
simple loading device, which minimizes influence of anchoring
on the results. Besides, since the tube is smooth, interlocking
between bar ribs and the concrete is avoided during the tests, thus,
bond between the steel and the concrete can be determined with-
out influence of the geometry of the reinforcement, focusing on
adhesion and friction. It should be noted, though, that other con-
current mechanisms, such as interlocking and stirrup confinement,
should be considered when studying the response of concrete
structures.
During the test, the displacement of the tube relative to the con-
crete was measured using a linear variable differential transducer,
which will be referred in the text as the lvdt. Besides, the widening
or additional opening of the main crack developed during acceler-
ated corrosion through the cover was measured using a longitudi-
lvdt
Fig. 2. Sketch of push-out tests.
 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609 601

nal extensometer, which will be referred as the crack width
extensometer.
A more detailed sketch of the test device is shown in Fig. 3. To
design the parts, some of the principles of compression tests found
in the ASTM C39 standard [24] were taken into account. A steel
cylinder (c) pushed the steel tube and had a guide to keep it cen-
tered. An auxiliary part (a) joined the device to a load cell through
a threaded joint. A ball joint (b) was placed between the pusher
and the auxiliary part to ensure a good contact of the pusher with
the steel tube, in case the sides of the slice were not strictly paral-
lel. To accommodate the ball, a conical seating was machined at
the bottom of Part (a) and at the top of Part (c). Part (e) supported
the slice contacting the concrete over a ring, and was centered with
the steel tube; a central drill allowed accommodating the lvdt,
which was held by a brace (f) and fixed by a screw; a circumferen-
tial groove accommodated the tube when pushed-out and acted as
a mechanical stop to protect the transducer from an excessive dis-
placement. The set rested over Part (g), which had a lateral groove
to access the transducer. Complementarily, a plate (d) was placed
below the slice as a security element in case the slice was broken.
All the parts were fabricated in steel for tempering. More details in
the experimental devices can be found in [19].
Fig. 4 shows a general view of the tests (a) and a detail of the
crack width extensometer (b). It should be noted that the lvdt
was fixed to the test device using a brace and a screw, and the
crack width extensometer was clamped to the slice using elastic
bands.
It should be noticed that the dimensions of the ring in contact
with the concrete were calculated so that the zone in compression
was the minimum without introducing damage into the concrete,
and the inner edge of the ring was as close as possible to the
steel-concrete interface; similarly, the dimensions of the pusher
were calculated so that the loaded cross-sectional area of steel
was the largest compatible with the tolerances of the tube; hence
the importance of using precision steel tubes as reinforcement.
Another observation is that the lvdt measured the displacement
between Points C and D in Fig. 3, which were located as near as
possible to the sections of steel and concrete being loaded (Lines
A and B), thus elastic deformation of the devices can be neglected.
Finally it should be noted that in this test both the reinforce-
ment and the concrete are compressed, in contrast with typical
pull-out tests. According to [16] the stress state is important for
the serviceability conditions, e.g. for evaluating crack widths; how-
ever, the tests in this work are intended to investigate the adher-
ence behavior of the steel-concrete interface and to obtain
information that could be applied to models that take into account
the actual geometry of real structures; thus, ideal conditions are
sought for a proper interpretation of the experimental results. It
should also be noted that the embedment length in this work is
of the order of the outer diameter of the reinforcement, which is

Fig. 3. Sketch of the devices designed for the push-out test: parts joining the load machine (a) and (g), ball joint (b), parts loading over the steel (c) and the concrete (e) and
auxiliary elements for security (d) and to hold the lvdt transducer (f). The displacement of the tube was measured between the points C and D, which were as near as possible
to the sections of steel and concrete being loaded (A and B).
602 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609
Fig. 4. General view of a push-out test (a) and detail of the extensometer to measure the additional opening of the main crack (b).
much smaller than that in the works mentioned in [16], and uni-
form stress is assumed along the reinforcement, as commented
on next section.
2.4. Test procedure and analysis of results
During the test, the applied load, the displacement of the tube
and the additional opening of the main crack were continuously
recorded. The test was run under displacement control, using the
lvdt output as the feed-back signal.
The stress s at the steel-concrete interface was computed from
the records of load and displacement, assuming uniform stress
along the length of the tube, and taking into account the actual sur-
face of contact between the steel and the concrete, as:
P during casting and curing, while keeping a length of 80 mm cen-
s¼
pDðL  dÞ
where P is the load, D is the outer diameter of the steel tube, L is the
average length of the tube, and d is the displacement of the tube,
which was measured by the lvdt. See Appendix A for details in
the processing of results.
From complementary simulations, which are out of the scope of
this paper, it was assessed that stress was uniform at the steel-
concrete interface from the peak to the end of the test. The simu-
lations were carried out within the finite element framework COFE
(Continuum Oriented Finite Element), assuming elastic behavior for
the steel and for the concrete, and a cohesive behavior for the
interface.
3. Experimental work
3.1. Materials and fabrication of specimens
Ten reinforced concrete prisms as those described in Section 2.2
were used in this work, cast from a single batch.
A porous concrete was fabricated in laboratory conditions using
Portland Cement I 52.5 R, defined in the European Standard EN 197-
1 [25], with a water-to-cement ratio of 0.5. Siliceous aggregate and
sand were used with a maximum size of 8 mm, and calcium chlo-
ride was added to the mixture to produce depassivation of the
steel. Table 1 displays the mix proportions by weight of cement,
which were as those in [20,21], except for the amount of superplas-
ticizer and for not incorporating set-retarding admixture.
Prior to casting, the tubes of reinforced specimens were cleaned
following the method described in [26]; they were immersed and
ultrasonically agitated during 30 s in an inhibited acid solution,
then rinsed with demineralized water, neutralized by immersion
in a solution of calcium bicarbonate, rinsed again with demineral-
ized water, and dried by immersion in propanol. With that treat-
ment, the initial adherence conditions between the steel and the
concrete can be assumed to be the same in all the specimens,
avoiding influence of the degree of cleanliness. The ends of the
tubes were coated with enamel to prevent them from corrosion
ð1Þ
tered at the tube free to corrode.
Complementary specimens were fabricated from the same
batch to characterize the fracture behavior of the concrete at the
age of 28 days. Particularly, the following specimens were cast:
six standard cylinders 150 mm in diameter and 300 mm in height,
and four standard beams with a cross-section of 100  100 mm2
and 500 mm in length. From those, three cylinders were used in
standard compression tests to measure the compressive strength
f c , following the ASTM C39 standard [24], three cylinders were
used in diagonal compression tests to determine the tensile
strength of concrete f t , following the ASTM C496 standard [27]
and additional recommendations proposed by Rocco in [28], and
the beams were notched and used in stable three-point bending
tests, to compute a bilinear approximation of the softening curve,
following the method proposed in the final report to RILEM TC
187-SOC by Planas et al. [29]. Table 2 displays the fracture param-
eters obtained.
All the specimens were cast in steel molds. In the case of the
reinforced concrete prisms, polyethylene parts with the appropri-
ate holes were used to hold the steel tube in place. The specimens
were kept in the molds for 24 h in laboratory conditions, then
taken out of the molds and stored in a bath of lime saturated water
in a temperature-controlled chamber at 20 °C until testing time.
 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609 603

Table 1 completeness of the text, while the details can be found in

Mix proportions of concrete by weight of cement.
Water Cement Sand Aggregate CaCl2 Superplasticizer
0:5 1 2:85 2:05 0:03 0.0055
Table 2
Mechanical parameters of the fabricated concrete, where f c is the compressive
strength, f t the tensile strength, E the elastic modulus, GF the fracture energy, w1 the
horizontal intercept of the initial branch of the softening curve with the abscissas
axis, and rk the stress of the kink point of the bilinear curve.
f c (MPa) f t (MPa) E (GPa) GF (N/m) w1 (lm) rk (MPa)
48.2 ± 2.8 3.33 ± 0.04 34.2 ± 1.4 110 ± 19 20.6 0.521
3.2. Accelerated corrosion of reinforced concrete prisms
After curing for at least three months, the specimens were arti-
ficially corroded using the impressed current technique [2,30]. The
specimens were corroded under a constant current of 400 lA/cm2,
up to a given corrosion depth, calculated according to Faraday’s
law, assuming 100% of effectiveness of the current. In particular,
three series of tests were carried out with three specimens being
corroded in parallel, up to final corrosion depths of 38, 25 and
13 lm. Table 3 displays the specimens and their nominal corrosion
depth. For future convenience, other results that will be explained
later are included in the table.
It should be noted that the applied density of current is much
higher than the values observed in real structures [14]. Thus, other
oxides may develop different from those formed under natural
conditions. In order to reproduce the behavior of real structures,
in the literature it is generally accepted to limit the current density
to 200 lA/cm2, as reported in [16]. Accordingly, many works are
found in which the applied current was below that limit
[9,13,17], although we would like to mention other reference
works in which the current density was much higher than that,
as for Example 500 lA/cm2 in [18], 2 mA/cm2 in [1] and even more
than 10 mA/cm2, as reported for 12 mm diameter bars with 102
mm embedment length corroded under a constant current of 0.4
A in [4]. This paper, though, does not intend to provide values for
the actual structures, but to provide fundamental research on bond
behavior of the steel-oxide-concrete system. In the present exper-
iments, the corrosion rate was selected to minimize the testing
time until a well-defined pattern of cracks was obtained, which
was deemed essential to study loss of confinement during push-
out tests. Such rate falls within the range for which Faraday’s law
has been shown to hold for steel corrosion in chlorinated concrete
[30,31].
The experimental set-up for accelerated corrosion was as that
presented in [21]; only its main aspects are explained next for
[19,21]. During the test, the specimen was submerged in a water
bath containing calcium chloride, resting over one of the sides per-
pendicular to the axis of the steel tube, for the sake of repeatability
of the conditions in all the tests. To provide electrical insulation of
the end of the tube that would be submerged, a PVC cap was
bonded surrounding it, and an epoxy layer was applied to the faces
of concrete perpendicular to the tube, to enforce plane field of cur-
rent during the test. With those precautions, uniform corrosion
was expected to occur along the length of the tube. A main crack
developed through the cover, and several secondary cracks sur-
rounding the reinforcement, as reported in [20] for tests with sim-
ilar specimens and conditions, and sketched in Fig. 1(d). For
specimens with smaller corrosion, fewer secondary cracks devel-
oped, and the width of the main crack was smaller.
The width of the main crack at the middle section of the spec-
imen, which was so-called main CMOD, was continuously recorded
using an adapted extensometer; particularly, the crack opening
was measured between the points labeled as A and B in Fig. 1(d),
at 5 mm from the concrete surface, for a safe anchor of the exten-
someter. Other measurements were carried out regarding the inner
dimensions of the tube, which are out of the scope of this paper
and for which details can be found in [21,19]. Finally, the electrical
potential, the circulating current and the temperature were
recorded using standard instruments. From the records of electri-
cal current, the corrosion depth at each time was computed, apply-
ing Faraday’s law.
The loss of weight of the steel after corrosion was measured for
the tubes of the slices used in bond tests, after being totally
pushed-out. For the measurement, the bases of the tubes were
sandpapered in a polishing machine to ensure perpendicularity
with the axis of the tube and the oxide was removed following
the procedure described in the ASTM-G01 standard [32]. Fig. 5
shows pictures of the corroded tubes for each final corrosion depth
in order to illustrate the corrosion morphology. Fig. 5(a), (b) and (c)
show the tubes before oxide removal and after being totally
pushed out of the concrete. It should be noted that part of the oxide
was adhered to the concrete surface. Fig. 5(d), (e) and (f) corre-
spond to the same tubes after oxide removal. It can be observed
that the corrosion attack affected the entire surface of the steel
tubes.
Table 3 displays, for each specimen, the absolute loss of weight
and the relative loss referred to the initial weight of the tube, the
experimental corrosion depth, the relative loss of diameter, and
the crack width at the end of accelerated corrosion tests. In the
case of specimen P10, which was not subjected to accelerated cor-
rosion, the average of the measurements of its three slices is dis-
played. For comparative purposes, the relative loss of cross-
sectional area corresponding to a solid bar with the same outer
diameter as the tube is included in the table. It would represent

Table 3
Data of accelerated corrosion tests: specimen, theoretical corrosion depth x, experimental loss of weight Dw ~ g , relative loss of weight Dw~ g =w0 , experimental corrosion depth xg
determined by gravimetry, relative loss of diameter DD=D0 , relative loss of cross-sectional area DA=A00 calculated for a fictitious section of a solid bar with the same outer diameter
as the tube, and main CMOD at the end of the test.

Specimen x (lm) ~ g (g/mm)

Dw ~ g =w0 (%)
Dw xg (lm) DD=D0 (%) DA=A00 (%) main CMOD (mm)

P1 38 0.01998 4.28 40.4 0.405 0.807 0.296

P2 38 0.01991 4.27 40.3 0.403 0.805 0.277
P3 38 0.02006 4.30 40.6 0.406 0.811 0.309
P4 25 0.01412 3.03 28.6 0.286 0.571 0.228
P5 25 0.01458 3.13 29.5 0.295 0.590 0.251
P6 25 0.01628 3.49 33.0 0.330 0.658 0.236
P7 13 0.00919 1.97 18.6 0.186 0.372 0.129
P8 13 0.00962 2.06 19.5 0.195 0.389 0.122
P9 13 0.00901 1.93 18.2 0.182 0.364 0.134
P10 0 0.00050 0.108 1.0 0.0102 0.0204 —
604 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609
Fig. 5. Tubes before oxide removal (a, b, c) and after oxide removal (d, e, f) corresponding to specimens with 38 lm (a, d), 25 lm (b, e) and 13 lm (c, f) of corrosion depth.
the loss of weight of a solid bar with the same corrosion depth as
the tubes. This value should be considered rather than the loss
referred to the initial weight of the tubes when comparing the cor-
rosion level of the specimens with other data in the literature.
It should be noted that in most of the cases the experimental
corrosion depth was greater than the theoretical one, calculated
according to Faraday’s law by assuming uniform corrosion along
the length of the tube, but was within the ranges reported by
Alonso et al. [14]; that phenomenon was explained by an acidifica-
tion of the media, which produces spontaneous corrosion, and by
parts of metal that are not electrolytically dissolved and spall out
from the metal surface. The initial loss of weight during curing
was of the same order as the error in the measurements, and can
be neglected. Therefore, Faraday’s law was demonstrated to be
applicable in these tests.
3.3. Obtention and preparation of slices
After corrosion, each reinforced concrete prism was cut into five
slices, as sketched in Fig. 1(c), using a low-speed radial diamond
saw. As previously explained, the two external slices of each prism,
which were cut with an average thickness of 10 mm, were dis-
carded to avoid border effects. From the remaining slices, which
had an average thickness of 20 mm, normally the central slice
was used in the bond tests presented in this work. As a particular
case, a prism non-subjected to accelerated corrosion (Specimen
P10) was cut into slices, and the three inner ones were used in
bond tests. The slices were labeled with the name of the corre-
sponding prism, followed by a number ranging from 1 to 5, as
sketched in Fig. 1(c), indicating the position from the top face, tak-
ing as a reference the position of the prism during accelerated
corrosion.
To improve the smoothness and parallelism of the surface of the
cuts, which is essential to introduce proper loads in push-out tests,
the surfaces of the slices were ground using a grinding machine.
Next, the surfaces of the slices were sandpapered to eliminate
burrs at the steel-concrete interface that were produced during
cutting and grinding, since those could increment friction during
push-out, leading to incorrect results. The slices were sanded in a
polishing machine, using 80 grit silicone carbide discs and refriger-
ated with plain water, until the steel-concrete interface was
exposed.
Then the slices were saturated until constant weight, following
the ASTM C642 standard [33]. To avoid corrosion of the exposed
surface of steel during saturation, rubber caps were inserted at
the tube ends and a silicone ring was applied around them. The
water baths were kept in a temperature-controlled chamber at
20 °C until testing time.
Finally, prior to the tests, the rubber caps were removed from
the slice and the bases of the tubes were cleaned in case oxide
developed, maintaining the concrete surface constantly humid by
spraying water until the end of the test.
3.4. Particular aspects of push-out tests
In this work, nine push-out tests have been carried out using
the slices of specimens subjected to accelerated corrosion, with
various corrosion depths. Additionally, three tests have been car-
ried out using slices of a prism not subjected to accelerated corro-
sion, to obtain a reference value of the bond properties on non-
corroded slices. Table 4 displays a list of the tests and the charac-
teristics of the slices; for convenience of the text, results of the
tests are also displayed in the same table, as explained in the next
section.
The linear variable differential transducer measuring the dis-
placement of the tube was a transducer SCHLUMBERGER model
AG1.0, prepared by INSTRON to be plugged in INSTRON testing
machines, with a range of 1 mm. The crack width extensometer
 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609 605

Table 4
List of push-out tests and main results: corrosion depth x during accelerated corrosion, initial length of the tubes L, initial stiffness s, load Pu , shear displacement du , shear stress su
and additional crack opening Dwu corresponding to the peak, and shear stress s1000 and additional crack opening Dw1000 corresponding to a shear displacement of 1000 lm.

Specimen x (lm) L (mm) s (MPa/mm) P u (N) du (lm) su (MPa) Dwu (lm) s1000 (MPa) Dw1000 (lm)
P1-3 38 20.54 35.27 3940 46.7 3.06 18.8 0.210 395
P2-3 38 19.17 92.26 4000 37.3 3.33 19.2 0.428 370
P3-3 38 19.56 58.02 3530 46.9 2.88 21.8 0.455 367
P4-3 25 19.28 59.11 3260 42.2 2.69 15.7 0.642 266
P5-3 25 19.46 84.69 3280 37.2 2.69 19.8 0.560 308
P6-3 25 20.01 68.83 3500 39.1 2.79 16.6 0.694 293
P7-3 13 19.14 67.28 3860 40.6 3.21 13.1 1.32 144
P8-3 13 19.80 61.98 3450 48.7 2.78 23.2 1.15 155
P9-3 13 19.63 69.48 3750 43.6 3.04 19.3 1.22 166
P10-2 0 20.69 64.12 4250 50.0 3.28 3.30 0.610 2.25
P10-3 0 20.81 70.34 3890 44.1 2.98 2.97 0.549 1.74
P10-4 0 20.88 79.05 3710 39.1 2.83 2.14 0.609 1.48

was a longitudinal extensometer INSTRON model 2620-602, with a
range of ±2.5 mm, and a gauge-length of 50 mm. The extensome-
ters were calibrated before each test, checking that the maximum
error was less than 0.2% of their range of measurement.
The test procedure was as explained next. Firstly, a preload was
applied to accommodate the parts loading over the concrete and
the steel (Parts (c) and (e) in Fig. 3), and to center the devices in
the steel tube. The preload was between the 5 and 10% of the max-
imum load, which was up to 300 N for the specimens in this work.
Then the test was run under displacement control. The displace-
ment rate at the beginning of the test was 15 lm/min, which
was calculated following recommendations of three-point bending
tests [29], with the criterion that the maximum load was reached
within 3–5 min. The displacement rate was doubled after 300 lm
of displacement of the tube. The test was stopped after 1.85 mm
of displacement of the steel tube. During the test, the concrete sur-
face was maintained constantly humid by spraying water.
Unloading and reloading were carried out in six tests (corre-
sponding to specimens P5-3, P6-3, P8-3, P9-3, P10-3, P10-4), after
reaching the peak, at 200 and 600 lm of displacement of the tube.
The displacement rate during unloading and reloading was 15 lm/
min.
4. Results and discussion
4.1. Analysis of stress versus displacement curves
Fig. 6(a) shows the resulting curves of nominal shear stress ver-
sus displacement of the corroded slices. The curves are plotted up
to a displacement of 1000 lm, although the final displacement in
the tests was nearly 2000 lm, since no significant changes
occurred from that displacement on. For all the specimens, the
curves display an initial steep slope, raise up to a peak, and then
progressively descend with a decreasing slope; the residual
strength is considerable after a shear displacement well over 1
mm. It indicates that after reaching the peak, stress is still trans-
mitted between the steel and the concrete, as reported in previous
works [1,4,7–9].
No significant differences were found in the stress and displace-
ment of the peak regardless of the corrosion depth, which indicates
that no significant changes in the adherence between the steel and
the concrete occurred for the values of corrosion reported in this
work. The maximum shear stress of the specimens with 25 lm of
corrosion depth was slightly smaller than that of the other speci-
mens, but this difference was within the experimental error and,
thus, was not considered to be significative. Noteworthy differ-
ences, though, were found in the evolution of the stress-
displacement curves after the peak for specimens with different
corrosion depth: for specimens with smaller corrosion depth, the
residual stress at the tail of the curve was higher. It could be
explained by differences in the roughness of the contact surface,
but the main reason might be the difference in the confinement
of the specimens, as will be discussed later.
Table 4 displays, for each specimen, the initial stiffness, the val-
ues at the peak of the load, the displacement, the shear stress and
the additional crack opening, and the values corresponding to a
shear displacement of 1000 lm of the shear stress and the addi-
tional crack opening. Table 5 records the corresponding average
and standard deviation values for each corrosion depth. For future
convenience, also the results of specimens not subjected to accel-
erated corrosion are displayed in those tables. It should be noticed
that the maximum stress was reached within 37.24 and 50.02 lm
of displacement of the tube. For those values, the error for calculat-
ing the stress considering the initial surface of steel in contact with
the concrete is less than 0.25%, and could be neglected. However, at
the end of the test, the error would be 10%, calculated for twenty-
millimeter length tubes and for a tube displacement of 2000 lm;
thus, considering the actual contact surface is essential for the
geometry of the specimens in this work.
It should be noted that the values reported in Table 5 differ sen-
sibly from those in the works mentioned in Section 1 [1,4–
11,13,16–18], due to the differences in the test configuration and
the test conditions, mainly the absence of bar ribs, the level of cor-
rosion, and the confinement of the specimens, which strongly
influences bond behavior.
Fig. 6(b) shows the curves of stress versus displacement, for
specimens not subjected to accelerated corrosion. A maximum
stress similar to that of corroded specimens was obtained, which
indicates that adherence between the steel and concrete was not
modified for the levels of corrosion covered in this work. Notwith-
standing, for non-corroded specimens, a sudden drop in stress
occurred after the peak with a marked instability, as indicated in
the curves by dotted lines, followed by a steady descent. The stress
after the drop was smaller than that of corroded specimens, which
could be due to an increase in roughness of corroded specimens, as
pointed out in [1]. Accordingly, the stress at the end of the tests of
non-corroded specimens was smaller than that of specimens with
13 lm of corrosion depth; however, it was greater than that of
specimens with 38 and 25 lm of corrosion depth, manifesting that
other phenomena, mainly loss of confinement, prevailed over vari-
ation of surface roughness at the end of the tests for those speci-
mens, as also pointed out in [17]. It should be noted that the
maximum stress obtained for non-corroded specimens differs from
the values found in the literature; for example, it was smaller than
that reported in [1], but greater than that in [18], for reasons sim-
ilar to those explained for corroded specimens.
To obtain complementary information of the displacement of
the tube, unloading and reloading were carried out in two
606 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609

Fig. 6. Curves of shear stress versus shear displacement for specimens with 38, 25 and 13 lm of corrosion depth (a) and for specimens not subjected to accelerated corrosion
(b).

Table 5
Mean and standard deviation of the results of push-out tests for each corrosion depth x, of the initial stiffness s, the load Pu , the shear displacement du , the shear stress su and the
additional crack opening Dwu corresponding to the peak, and the shear stress s1000 and additional crack opening Dw1000 corresponding to a shear displacement of 1000 lm.

x (lm) s (MPa/mm) P u (N) du (lm) su (MPa) Dwu (lm) s1000 (MPa) Dw1000 (lm)
38 61.9 ± 29 3820 ± 250 43.6 ± 5.5 3.09 ± 0.23 19.9 ± 1.6 0.364 ± 0.13 377 ± 15
25 70.9 ± 13 3350 ± 130 39.5 ± 2.5 2.72 ± 0.056 17.4 ± 2.2 0.632 ± 0.068 289 ± 21
13 66.2 ± 3.9 3680 ± 210 44.3 ± 4.1 3.01 ± 0.22 18.5 ± 5.1 1.23 ± 0.085 155 ± 11
0 71.2 ± 7.5 3950 ± 280 44.4 ± 5.4 3.03 ± 0.23 2.80 ± 0.60 0.589 ± 0.035 1.82 ± 0.39

Fig. 7. Zoom of the unloading and reloading curves of specimens P6-3 (a) and P9-3 (b).
 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609 607

specimens with 25 lm of corrosion depth (P5-3 and P6-3), in two
specimens with 13 lm of corrosion depth (P8-3 and P9-3) and in
two slices of a prism non subjected to accelerated corrosion
(P10-3 and P10-4), as seen in Fig. 6(a) and (b). Fig. 7 presents
zooms of the unloading and reloading branches for specimens
P6-3 and P9-3. Regardless of the corrosion depth, the unloading
curves were almost elastic, and permanent displacement of the
tube was observed. The reloading curves were elastic up to approx-
imately 75% of the stress before the unloading, but then smoothly
curved, until recovering the monotonic curve after a relatively
large displacement of the order of 25 to 100 lm. These curves indi-
cate that after reaching the maximum stress, displacement of the
tube is permanent, therefore debonding has occurred, and, thus,
stress is mainly due to friction between the steel and the concrete.
4.2. Analysis of the crack opening during the tests
During push-out tests, it was observed that the main crack
which developed at the concrete cover during accelerated corro-
sion widened, as illustrated in Fig. 8 for one of the slices. Such an
additional opening was quantified by the measurement of the
crack width extensometer.
Fig. 9(a) displays the curves of additional crack opening versus
displacement of the tube for corroded specimens. It is observed
that the width of the main crack increased monotonically during
the test, with a slow growth before the displacement correspond-
ing to the maximum stress, and then a significant increase,
followed by a continuous decrease in the slope of the curves. It
should be noted the similarity with the results reported for direct

Fig. 8. Detail of one of the slices with 38 lm of corrosion depth, after sandpapering and before push-out (a), and after push-out (b).

Fig. 9. Curves of additional crack opening versus displacement (a) and curves of stress versus additional crack opening (b) for specimens with 38, 25 and 13 lm of corrosion
depth.
608 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609
shear tests of dense sand (see [34,35] as reference works), in which
the volume of the sample monotonically increases with the shear
displacement, a phenomenon known as positive dilatancy, and the
maximum strain rate occurs approximately for the displacement
corresponding to the maximum shear stress. Clear influence of
the corrosion level on the additional crack opening was observed,
with higher opening for specimens with greater corrosion depth.
See Table 5 for the additional crack opening corresponding to a
shear displacement of 1000 lm.
For non-corroded specimens, an extensometer was placed at
the same location as in corroded specimens, although they were
not cracked. For clarity of the graphics, the curves of additional
crack opening of these specimens are not drawn in Fig. 9(a). An
opening less than 4 lm was recorded, with a maximum corre-
sponding to the peak of stress, then it decreased 1 lm and
remained constant until end of the test, which indicates that no
cracks developed during push-out.
Along the unloading reloading loops described in Section 3.4,
the variation of crack opening was less than 3 lm, which indicates
that the additional opening of the main crack is permanent.
To study whether there is a correlation between the residual
stress and the additional opening of the main crack during push-
out, the curves of stress versus crack opening of corroded speci-
mens were drawn, as shown in Fig. 9(b). The curves overlapped,
regardless of the corrosion depth of the specimens, assessing that
the stress and the additional crack opening are related. It should
be noted that the scatter in those curves is smaller than that of
the stress-displacement curves of specimens with a given corro-
sion depth (Fig. 6a).
The foregoing results indicate that confinement is essential
after adherence between steel and concrete is broken, in agree-
ment with the results in [17], and explain why residual stress dur-
ing push-out is smaller for specimens with greater corrosion depth
and wider initial cracks; moreover, crack widening during the tests
was also larger for those specimens, which further reduces
confinement.
5. Conclusions
From the foregoing analysis of results of push-out tests carried
out in slices of concrete prisms reinforced with a smooth steel tube
and corroded under a constant current, the following conclusions
have been disclosed:
P (N)
Pu
experimental curve
AB: linear fit
P1 + 0.15Pu B
P1 A A
C δ1
δ (μm)
(a)
Fig. A.10. Example of an experimental curve of load P versus displacement of the tube d obtained in push-out tests (a) and curve after being processed (b).

The maximum shear stress in the tests is not affected by the
corrosion depth for the corrosion levels applied in this work,
which indicates that adherence was similar in all the
specimens.

The residual stress at the end of the tests is influenced by the
corrosion depth. The residual stress of specimens with small
corrosion depth was greater than that of non-corroded speci-
mens, due to an increase in the surface roughness. However,
as corrosion depth increased, the residual stress decreased,
due to loss of confinement of the specimens.

A continuous widening of the main crack developed through the
cover during accelerated corrosion was observed, which resem-
bles the positive dilatant behavior of dense sand. Moreover, the
additional crack opening was found to be related with shear
stress, manifesting that confinement of the specimens is essen-
tial after adherence between the steel and the concrete is lost,
and determines the stress transmitted due to friction. Besides,
crack widening was greater for specimens with higher corrosion
level, which may explain why the residual stress of those spec-
imens was smaller.
Funding
This work was supported by the Secretaría de Estado de Investi-
gación, Desarrollo e Innovación of the Spanish Ministerio de Econ-
omía y Competitividad under the projects BIA2010-18864 and
BIA2014-54916-R.
Acknowledgements
The authors gratefully acknowledge Oficemen, Grupo Cementos
Portland Valderrivas for donating the cement used to fabricate the
specimens of this work.
Appendix A. Treatment of results
After the tests, the results of load, displacement of the tube and
additional opening of the main crack were processed to treat the
records corresponding to the preload, taking as example some of
the steps followed in the procedure of analysis results of three-
point bending tests described in [29,36], and the stress at the
steel-concrete interface was computed, as explained next.
Fig. A.10 shows an example of a load-displacement curve (P vs
P (N)
Pu
experimental curve
preload replaced by AC
C
δ (μm)
(b)
 B. Sanz et al. / Construction and Building Materials 160 (2018) 598–609
d), as obtained from the experiments, and the curve after being
processed. A routine was programmed in C++ language, to auto-
matically process the results, which read a file containing the basic
information about the test and the geometry of the specimen, read
the records of load, displacement of the tube and additional crack
opening, and carried out the next calculations:

First of all, it computed the maximum load Pu and the load P 1
corresponding to the end of the preload d1 , which was indicated
by the user (Point A in Fig. A.10(a)).

Next, it replaced the part corresponding to the preload by a lin-
ear branch as follows: it computed a linear fit of the load-
displacement curve within the range ðP1 ; P1 þ 0:15P u Þ (Points
A and B in Fig. A.10(a)) and replaced the records with P < P 1
by the linear fit (the points below A were replaced by the seg-
ment CA, as seen in Fig. A.10(b)). Then the curve was moved
to the origin (Point C was moved to 0) and finally the linear
approximation was removed from the record (i.e., Point C was
deleted, so the curves began at point A).

The same operations were carried out for the curve of load ver-
sus additional crack opening.

Finally, the stress s at the steel-concrete interface was com-
puted according to Eq. (1).
References
[1] G.J. Al-Sulaimani, M. Kaleemullah, I.A. Basunbul, Rasheeduzzafar, Influence of
corrosion and cracking on bond behavior and strength of reinforced concrete
members, ACI Struct. J. 87 (1990) 220–231.
[2] C. Andrade, C. Alonso, F.J. Molina, Cover cracking as a function of bar corrosion:
part i-experimental test, Mater. Struct. 26 (1993) 453–464.
[3] K. Tuutti, Corrosion of Steel in Concrete, CBI Forskning 82:4, Swedish Cement
and Concrete Research Institute, Stockholm, 1982.
[4] A.A. Almusallam, A.S. Al-Gahtani, A.R. Aziz, Rasheeduzzafar, Effect of
reinforcement corrosion on bond strength, Constr. Build. Mater. 10 (1996)
123–129.
[5] M. Maslehuddin, I.A. Allam, G.J. Al-Sulaimani, A. Al-Mana, S.N. Abduljauwad,
Effect of rusting of reinforcing steel on its mechanical properties and bond
with concrete, ACI Mater. J. 87 (1990) 496–502.
[6] X. Fu, D.D.L. Chung, Effect of corrosion on the bond between concrete and steel
rebar, Cem. Concr. Res. 27 (1997) 1811–1815.
[7] S. Coccia, S. Imperatore, Z. Rinaldi, Influence of corrosion on the bond strength
of steel rebars in concrete, Mater. Struct. 49 (2016) 537–551.
[8] C. Fang, K. Lundgren, L. Chen, C. Zhu, Corrosion influence on bond in reinforced
concrete, Cem. Concr. Res. 34 (2004) 2159–2167.
[9] K.Z. Hanjari, D. Coronelli, K. Lundgren, Severely corroded reinforced concrete
with cover cracking: part 2. Anchorage capacity, in: C. Andrade, G. Mancini
(Eds.), Modelling of Corroding Concrete Structures: Proceedings of the Joint
fib-RILEM Workshop held in Madrid, Spain, 22–23 November 2010, vol. 5,
Springer, Netherlands, Dordrecht, 2011, pp. 207–217, https://doi.org/10.1007/
978-94-007-0677-4_14.
[10] C.A. Juarez, B. Guevara, G. Fajardo, P. Castro-Borges, Ultimate and nominal
shear strength in reinforced concrete beams deteriorated by corrosion, Eng.
Struct. 33 (2011) 3189–3196.
[11] Y.S. Choi, S.-T. Yi, M.Y. Kim, W.Y. Jung, E.I. Yang, Effect of corrosion method of
the reinforcing bar on bond characteristics in reinforced concrete specimens,
Constr. Build. Mater. 54 (2014) 180–189.
[12] ASTM C234, Standard Test Method for Comparing Concretes on the Basis of the
Bond Developed with Reinforcing Steel, ASTM International, West
Conshohocken, PA, 2000.
609
[13] M. Prieto, P. Tanner, C. Andrade, Bond response in structural concrete with
corroded steel bars. Experimental results, in: C. Andrade, G. Mancini (Eds.),
Modelling of Corroding Concrete Structures: Proceedings of the Joint fib-
RILEM Workshop held in Madrid, Spain, 22–23 November 2010, Springer,
Netherlands, Dordrecht, 2011, pp. 231–241, https://doi.org/10.1007/978-94-
007-0677-4_16.
[14] C. Alonso, C. Andrade, J. Rodríguez, J.M. Díez, Factors controlling cracking of
concrete affected by reinforcement corrosion, Mater. Struct. 31 (1998) 435–
441.
[15] A. Poursaee, C.M. Hansson, Potential pitfalls in assessing chloride-induced
corrosion of steel in concrete, Cem. Concr. Res. 39 (2009) 391–400.
[16] G. Mancini, F. Tondolo, Effect of bond degradation due to corrosion – a
literature survey, Struct. Concr. 15 (2014) 408–418.
[17] J. Cairns, Y. Du, D. Law, Influence of corrosion on the friction characteristics of
the steel/concrete interface, Constr. Build. Mater. 21 (2007) 190–197.
[18] A. Ouglova, Y. Berthaud, F. Foct, M. François, F. Ragueneau, I. Petre-Lazar, The
influence of corrosion on bond properties between concrete and reinforcement
in concrete structures, Mater. Struct. 41 (2008) 969–980.
[19] B. Sanz, Experimental and numerical study of cracking of concrete due to
reinforcement corrosion (Ph.D thesis), E.T.S.I. Caminos, Canales y Puertos (U.P.
M.), 2014, URL: http://oa.upm.es/30568/.
[20] B. Sanz, J. Planas, J.M. Sancho, An experimental and numerical study of the
pattern of cracking of concrete due to steel reinforcement corrosion, Eng. Fract.
Mech. 114 (2013) 26–41.
[21] B. Sanz, J. Planas, J.M. Sancho, A closer look to the mechanical behavior of the
oxide layer in concrete reinforcement corrosion, Int. J. Solids Struct. 62 (2015)
256–268.
[22] J.M. Sancho, J. Planas, D.A. Cendón, E. Reyes, J.C. Gálvez, An embedded crack
model for finite element analysis of concrete fracture, Eng. Fract. Mech. 74
(2007) 75–86.
[23] EN-10305-1, Steel tubes for precision applications. Technical delivery
conditions – part 1: seamless cold drawn tubes, European Committee for
Standardization, 2010.
[24] ASTM C39, Standard Test Method for Compressive Strength of Cylindrical
Concrete Specimens, ASTM International, West Conshohocken, PA, 1986,
https://doi.org/10.1520/C0039_C0039M.
[25] EN-197-1, Cement – part 1: composition, specifications and conformity
criteria for common cements, European Committee for Standardization, 2000.
[26] I. Kayafas, Technical note: corrosion product removal from steel fracture
surfaces for metallographic examination, Corrosion, NACE 36 (1980) 443–445.
[27] ASTM C496, Standard Test Method for Splitting Tensile Strength of Cylindrical
Concrete Specimens, ASTM International, West Conshohocken, PA, 1990,
https://doi.org/10.1520/C0496_C0496M-11.
[28] C. Rocco, Influencia del tamaño y mecanismos de rotura del ensayo de
compresión diametral (Ph.D thesis), E.T.S.I. Caminos, Canales y Puertos (U.P.
M.), 1996.
[29] J. Planas, G.V. Guinea, J.C. Gálvez, B. Sanz, A.M. Fathy, Indirect tests for stress-
crack opening curve, in: J. Planas (Ed.), Report 39: Experimental Determination
of the Stress-Crack Opening Curve for Concrete in Tension – Final report of
RILEM Technical Committee TC 187-SOC, RILEM Publications SARL, 2007, pp.
13–29.
[30] T.A. El Maaddawy, K.A. Soudki, Effectiveness of impressed current technique to
simulate corrosion of steel reinforcement in concrete, J. Mater. Civil Eng. 15
(2003) 41–47.
[31] S. Caré, A. Raharinaivo, Influence of impressed current on the initiation of
damage in reinforced mortar due to corrosion of embedded steel, Cem. Concr.
Res. 37 (2007) 1598–1612.
[32] ASTM G1, Standard Practice for Preparing, Cleaning and Evaluating Corrosion
Test Specimens, ASTM International, West Conshohocken, PA, 2003,
https://doi.org/10.1520/G0001-03R11.
[33] ASTM C642, Standard Test Method for Density, Absorption and Voids in
Hardened Concrete, ASTM International, West Conshohocken, PA, 1990,
https://doi.org/10.1520/C0642-97.
[34] D.W. Taylor, Fundamentals of Soil Mechanics, John Wiley & Sons, New York,
1948.
[35] L. Barden, H. Ismail, P. Tong, Plane strain deformation of granular material at
low and high pressures, Géotechnique 19 (1969) 441–452.
[36] A.M. Fathy, B. Sanz, J.M. Sancho, J. Planas, Determination of the bilinear stress-
crack opening curve for normal- and high-strength concrete, Fatigue Fract.
Eng. Mater. Struct. 31 (2008) 539–548.