See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/274833230

Influence of corrosion on the bond strength of steel rebars in concrete

Article  in  Materials and Structures · December 2014

DOI: 10.1617/s11527-014-0518-x

CITATIONS READS

50 635

3 authors:

Simona Coccia Stefania Imperatore

University of Rome Tor Vergata Università degli studi Niccolò Cusano
36 PUBLICATIONS   201 CITATIONS    63 PUBLICATIONS   282 CITATIONS   

SEE PROFILE SEE PROFILE

Zila Rinaldi
University of Rome Tor Vergata
83 PUBLICATIONS   847 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Seismic behaviour of corroded structures View project

Glass FRP reinforcement in tunnel applications View project

All content following this page was uploaded by Stefania Imperatore on 17 March 2016.

The user has requested enhancement of the downloaded file.

Influence of corrosion on the bond strength
of steel rebars in concrete

Simona Coccia, Stefania Imperatore &

Zila Rinaldi

Materials and Structures

ISSN 1359-5997

Mater Struct
DOI 10.1617/s11527-014-0518-x

1 23
Your article is protected by copyright and all
rights are held exclusively by RILEM. This e-
offprint is for personal use only and shall not
be self-archived in electronic repositories. If
you wish to self-archive your article, please
use the accepted manuscript version for
posting on your own website. You may
further deposit the accepted manuscript
version in any repository, provided it is only
made publicly available 12 months after
official publication or later and provided
acknowledgement is given to the original
source of publication and a link is inserted
to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at link.springer.com”.

1 23
 Author's personal copy
Materials and Structures
DOI 10.1617/s11527-014-0518-x

ORIGINAL ARTICLE

Influence of corrosion on the bond strength of steel rebars

in concrete
Simona Coccia • Stefania Imperatore •

Zila Rinaldi

Received: 24 June 2013 / Accepted: 19 December 2014

Ó RILEM 2014

Abstract The problem of the durability of reinforced 1 Introduction

concrete structures is arisen, dramatically, in the last
decades, showing that one of the most dangerous Corrosion phenomena and related effects, such as size
degradation phenomena is connected to the rebars reduction in both rebars and strands, bond decay at
corrosion. When active corrosion grows the oxides steel–concrete interface, and cracking in the surround-
occupy a volume of about 2–4 times, with respect to the ing concrete, can be considered one of the main cause
sound bar, causing a loss of bond, cracking and cover of degradation in reinforced concrete structures. The
spalling. In this paper the corrosion effects at the steel– relevance of this topic, with reference also to its
concrete interface are analysed. At this aim, pull-out economical implications, is witnessed by the many
tests are carried out, with different corrosion levels, for theoretical and experimental research projects docu-
the definition of the bond-slip constitutive relation- mented in the literature [1–7].
ships of the damaged rebars. Furthermore, an analyt- The corrosion phenomenon affects not only the
ical model is developed for evaluating the internal geometrical and mechanical properties of the steel
pressure due to corrosion products, for concrete rebars [8–11], but also the bond characteristics at the
specimens without stirrups, in order to provide a steel—concrete interface and a correct evaluation of
procedure for the definition of the maximum bond as a the behaviour of reinforced concrete structures cannot
function of the corrosion percentage. Finally the neglect this aspect, necessary to ensure the composite
analytical results are validated through a comparison interaction of the two materials. For very low stress,
with the experimental outcomes. bond strength is assured from chemical adhesion
between steel and concrete. Once slip occurs, bond is
Keywords Corrosion  Bond behavior  mainly given by mechanical interlock between ribs
Experimental tests  Analytical model and concrete. The oxides formation causes a volume
increase that generates radial pressure around the bars,
and can create cracking due to tensile stresses in the
surrounding concrete. Furthermore a deterioration of
S. Coccia  Z. Rinaldi (&)
Department of Civil Engineering, University or Rome the rebars surface, with ribs degradation, occurs that
‘‘Tor Vergata’’, via del Politecnico 1, 00133 Rome, Italy can generate the reduction of the interlocking forces
e-mail: rinaldi@ing.uniroma2.it between the deformed bars and the concrete. As a
matter of fact, for very low values of corrosion, the
S. Imperatore
University ‘‘Niccolò Cusano’’, Via Don Carlo Gnocchi, increase of the surface roughness and the oxides
00166 Rome, Italy presence around the bars can improve the bond
 Author's personal copy
Materials and Structures

behaviour. When the phenomenon increases, the ribs corrosion level increases, both for confined and
height reduces up to their complete disappearance, and unconfined specimens. The authors highlighted the
the oxides, laid up at the concrete-steel interface, degradation of strength was less definite in confined
favour the slips, and can cause the formation of radial elements, whose collapse was due to the bar slippage.
cracks, with loss of bond. In this condition the failure On the contrary the unconfined specimens suffered a
mechanism is strictly connected with the confinement splitting failure. Finally the authors provided some
level exerted on the steel rebars. In case of low relationships between the maximum bond stress and
confinement a sudden splitting failure can occur, while the compressive strength of the concrete and the
on the contrary, if a suitable confinement is present, corrosion level. A wide experimental program was
the radial cracks cannot propagate up to the cover, and carried out by Fang et al. [20], with pull-out tests, on
the pull-out mechanism is more probable. both smooth and deformed /20 bars, artificially
Studies conducted by Auyeung [12] assessed that corroded with a current variable between 0 and 2 A
loss of bond is very critical; for a 2 % of diameter loss, (i.e. current density in the range 0–637 mA/cm2). The
he found an experimental bond reduction of about authors found an increase of bond strength for plane
80 %. The problem of the evaluation of the corrosion bars and corrosion levels up to about 5 % in mass loss,
effects on the bond behaviour has been analysed by while a decrease of strength was measured in
different authors both with experimental [13–21] and deformed bars, for corrosion levels of about 5 and
theoretical studies [22–24]. These last ones are based 9 %. The experimental outcomes were also compared
on models, that starting from the evaluation of the with numerical results.
stress state caused by the oxides at different corrosion Anyway the results available in literature, are
level, lead to the estimate of the bond strength. characterised by a wide dispersion, as highlighted in
Numerical FEM models are also available in order to Fig. 1, where the maximum bond strength, experi-
catch the main parameters affecting the phenomenon mentally found by some of the cited authors, is plotted
[25–29]. versus the corrosion level, expressed in terms of
More in detail, experimental tests were carried out percentage of mass loss. This result is due to the
by Al-Sulaimani et al. [13] in order to evaluate the different test set-up, different kind of specimens and to
bond properties at different corrosion levels, for the different current densities applied to accelerate the
different cover—steel diameter ratio. Both pull-out corrosion [31]. Indeed, high densities accelerate the
test and beam tests were performed. The current process, making faster the test time, but obstruct the
density was set equal to 2 mA/cm2. The obtained diffusion of the corrosion products in the concrete;
results highlighted an increase of the bond stress for according to [32] a suggested value of the current
corrosion levels up to about 1 % in mass loss; for density should be of about 0.05 mA/cm2.
higher corrosion levels a sharp reduction of bond In the present paper the influence of the corrosion
strength was measured (of about 50 % for mass loss of on the bond strength is analysed through experimental
5–6 %). Similar results were found by Almusallam and analytical models. Pull-out tests are carried out,
et al. [17] who studied, through the cantilever bond with different corrosion levels, in order to evaluate the
test, the bond properties for different corrosion levels. complete bond-slip relationship, and to acquire all the
The current intensity for the accelerated corrosion information necessary for the calibration of an
process on [12 rebars was assumed equal to a 0.4 A analytical procedure. The proposed model extends
(i.e. current density equal to about 350 mA/cm2). An the studies by Tepfer [33] and Den Uijl and Bigaj [34],
increase of bond (maximum value about 17 %) was developed for sound bars, to the case of corroded
measured for corrosion levels up to 5 % in mass loss rebars, by introducing the effect of the internal
(i.e. up to the specimen cracking). A sharp decrease of pressure due to both the pull-out action and the
strength was observed with bond reduction of about corrosion products, in the definition of the bond
80 % for mass loss of about 10 %. Results of pull-out strength. The model is based on the analysis of a thick-
tests are further reported in Lee et al. [30]. Current walled cylinder and this approach has been already
density of about 752 mA/cm2 was adopted for accel- followed in previous papers. In particular Bazant [35],
erating the corrosion process. Once again the results Bhargava et al. [24], Liu and Weyers [36] and more
shows the reduction of the bond strength when the recently Balafas and Burgoyne [37] and Lu et al. [38]
 Author's personal copy
Materials and Structures

Al Sulaimani et al., 1990 (D10)

26 Al Sulaimani et al., 1990 (D14)
24 Al Sulaimani et al., 1990 (D20)
Auyeung et al., 2000
22
Lee et al., 2002 (W/C=45%)
20 Lee et al., 2002 (W/C=55%)
18 Lee et al., 2002 (W/C=65%)
Bond strength [MPa]

Fang et al., 2004 (D1)

16
Fang et al., 2004 (S1)
14 Horrigmoe et al., 2007 (i=0.05mA/cm 2)
12
Horrigmoe et al., 2007 (i=0.25mA/cm 2)
Horrigmoe et al., 2007 (i=0.10mA/cm 2)
10

0
0 1 2 3 4 5 6 7 8 9 10 11 12
Corrosion level [%]

Fig. 1 Results obtain from different authors

devoted their studies to the prediction of the time to

cracking of the concrete cover, due to the corrosion
products. In Wang and Liu [22, 23], the thick-walled F
cylinder was adopted for the development of a bond
strength model, accounting for the corrosion products.
100

The presented model, differently from the other

cited ones, accounts for the contemporaneous pre-
sence of the corrosion and of the pull-out actions, as
specified in the following.
150

70

2 Experimental investigation

An experimental program was carried out in the

40

Laboratory of Structures and Materials of the ‘‘Tor

Vergata’’ University, in order to evaluate the influence
of the corrosion on bond, with reference to unconfined 150
specimens. The average bond behavior was studied
12

through pull-out tests. Twelve specimens, composed

Ø

by cubes (Fig. 2, side equal to 150 mm) reinforced

with a [12 mm deformed steel rebar (B450C,
fy = 507.5 N/mm2), with a bonded length of 70 mm
were tested. The unbonded region was obtained by
using a PVC pipe. The mix design used for concrete is l/2 l/2
shown in Table 1. The mean compressive strength
measured on 6 cubes was equal to 29 MPa. Fig. 2 Specimen geometry
 Author's personal copy
Materials and Structures

Table 1 Concrete mix Mix design (kg/m3) Slump (mm)

design
Cement Aggregates (mm) Water Plasticizer (cl)
Fine (0–4) Medium (4–8) Coarse (8–12)

350 1,100 520 130 245 180 102

In order to avoid cracking due to shrinkage, the length of 70 mm, the pressure of the oxides causes the
specimens were kept moist by wet towels covered with cracking of the concrete (Fig. 4). Nevertheless it is
layers of cellophane, then cured in water for 28 days. worth mentioning that not always a visible crack
The accelerated corrosion was impressed by means
of an anodic current of about 100 lA/cm2 for each
specimen (Fig. 3). The time necessary for obtaining
the required corrosion level was evaluated through a
variation of the Faraday’s law, in agreement with
Eq. 1, by calibrating the coefficient k with preliminary
tests:
k  mloss  nspecimen  CFaraday
time½sec ¼ ; ð1Þ
current½A  Mspecimen
where Mspecimen is the molar mass of the reinforcing
bar (equal to 55.8 mol), nspecimen is its valence equal to
2, CFaraday is the Faraday constant (equal to 96,485 C/
mol) and k is the constant accounting for the
possibility that the corrosive process doesn’t start
immediately (i.e. equal to 1 for bare bars, and higher
than 1 for the embedded specimens). Preliminary
calibration tests shown that, for the analysed speci-
mens in which the corrosion process was delayed due
to the concrete presence around the rebars, the value of
k was about 1.5 [11].
Different corrosion levels were considered
(expressed in terms of weight reduction), in agreement
with the ones analysed in Al-Sulaimani et al. [13], in
order to catch both the uncracked and cracked stages.
In the analysed specimens, characterized by a bonded

Fig. 4 Crack patterns a during corrosion process; b before the

Fig. 3 Accelerated corrosion test setup bond test; c during the bond test; d at the collapse
 Author's personal copy
Materials and Structures

(a) wire displacement transducer (Fig. 5b). Two LVDTs

(Fig. 5c), connected at the upper end of the bar
(immediately below the jaws) allow an additional
measure of the displacement, which contains, besides
LVDTs a rate of slip, the elastic deformation of the reinforcing
rebar. The load is measured by a 100 kN cell.
For each corrosion level the obtained results are
summarized in Table 2, and plotted in terms of load-
wire slip curves and bond stress-slip curves in Fig. 6.
In the analysed specimens the cracking of the
concrete cover appears for corrosion levels of about
1 % in terms of mass loss. Lower corrosion levels lead
(b) (c)
to an increase of the bond strength (see Fig. 6). In this
case the collapse mechanism is characterized by a
pull-out failure.
For higher corrosion values, a sudden deterioration
of the bond characteristics occurs and the failure mode
change by pull-out to splitting.
It can be noted, from Fig. 6, that the peak value of
the bond stress, in the corroded curves, moves toward
lower slip values, and the initial modulus (bond stress/
slip) increases in the corroded specimens. The values
Fig. 5 Test set up a and measurement instruments, b wire of the peak bond stress related to corroded specimens,
transducer for the measure of the slip, c LVDT for the measure
of the elastic deformation of the reinforcing rebar
normalized with respect to the sound one, are plotted
versus the corrosion level (expressed as a percentage
appears, even if the rebar is subjected to significant of mass loss) in Fig. 7. In order to clarify the
corrosion, due to the properties (mainly porosity) of experimental pattern a non linear regression is
the concrete matrix. performed and superimposed to the experimental
After the corrosion process, the specimens are results. It can be noted the increase of bond stress up
subjected to pull-out tests in control displacement. The to 60 % for a mass loss of about 0.5 % and the
test setup is shown in Fig. 5 together with the measure following sharp reduction. For corrosion levels of
instrumentation. The bar slip is measured through a about 1.5 % the bond stress is reduced of about 40 %.

Table 2 Results by Specimen Measured mass loss (%) Loadmax (kN) Slippeak (mm) smax (MPa)
average bond tests
M1 0.00 29.58 2.15 11.21
M2 0.10 37.25 1.77 14.13
M3 0.46 46.21 1.42 17.56
M4 0.80 43.98 1.76 16.74
M5 0.88 40.48 1.19 15.42
M6 0.94 30.81 1.31 11.74
M7 1.00 29.23 1.78 11.14
M8 1.24 24.92 1.67 9.51
M9 1.34 22.73 2.03 8.68
M10 1.52 20.98 1.36 8.02
M11 2.80 12.40 1.05 4.70
M12 3.20 18.04 1.27 6.95
M13 3.92 13.36 0.64 5.17
 Author's personal copy
Materials and Structures

Fig. 6 Experimental Load-

slip and bond—slip curves

1.80 cracking
1. evaluation of the pressure (pcorr) due to corrosion
1.60
1.40
post -cracking only, by extending the models [33, 34];
1.20 2. evaluation of the radial displacement due to
1.00 corrosion (Urs,corr), as a function of the corrosion
τc/τs

0.80 penetration (X), according to the model proposed

0.60 by Wang and Liu [22, 23] (Sects. 3.1, 3.2, 3.3);
0.40
3. evaluation of the radial displacement due to pull-
0.20
0.00
out force (Urs,nc) (Sect. 3.4);
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 4. evaluation of the combination of the corrosion and
Mcorr [%] pull-out actions in the definition of the total radial
displacement and total pressure (Sect. 3.4);
Fig. 7 Experimental normalised peak bond stress versus
corrosion level (mass loss) 5. definition of the bond strength as a function of the
corrosion level (Sect. 3.4).
Finally the obtained experimental results,
The Young’s modulus of the rust products is
expressed in terms of non-dimensional bond strength
considered equal to the reinforced steel one [24],
versus mass loss, are superimposed to other ones
because of the uncertainties on this mechanical
available in literature, in Fig. 8.
property ([14, 15, 39, 40]).
The innovation of the proposed analytical model
lays in points 4 and 5, in which the contemporaneous
3 Analytical model
presence of the corrosion, accounted for in [22, 23]
and of the pull-out actions [33, 34] is considered.
The proposed analytical model is based on the studies
developed by Tepfer [33] and by Den Uijl and Bigaj
3.1 Uncracked stage
[34] for the bond modeling of sound rebars, and it is
extended to the case of corroded reinforcement. As a
In the uncracked stage the corrosion pressure (pcorr) and
matter of fact, in the response of a thick-walled
the related radial displacement (Urs,corr) (Fig. 9) can be
cylinder proposed in [33, 34], an internal pressure, due
evaluated according to Timoshenko [41], in the hypoth-
to the corrosion action, is introduced and evaluated,
esis of elastic behaviour of the cylinder and neglecting
with reference to the uncracked, partly-cracked and
the Poisson effects, as reported in Eqs. 2 and 3,
entirely-cracked stages.
 2 
In particular the analytical procedure, described in c1  rs2
detail in Sects. 3.1, 3.2, 3.3 and 3.4, follows the pcorr ¼ Ec ers 2 ; ð2Þ
c1 þ rs2
subsequent steps:
 Author's personal copy
Materials and Structures

Al Sulaimani et al., 1990 (D10)

1.8 Al Sulaimani et al., 1990 (D14)
Al Sulaimani et al., 1990 (D20)
1.6 Auyeung et al., 2000
Lee et al., 2002 (W/C=45%)
1.4 Lee et al., 2002 (W/C=55%)
Lee et al., 2002 (W/C=65%)
1.2 Fang et al., 2004 (D1)
Fang et al., 2004 (S1)
1 Horrigmoe et al., 2007 (i=0.05mA/cm2 )
Horrigmoe et al., 2007 (i=0.25mA/cm2)
0.8 Horrigmoe et al., 2007 (i=0.10mA/cm2)
This paper
0.6

0.4

0.2

0
0 1 2 3 4 5
Corrosion level [%]

Fig. 8 Experimental results: non-dimensional bond—corrosion level: comparison with literature (Tc corroded bond, Ts sound,
uncorroded bond)

c1

Urs Urs VNP

rs rs
VUrs
χ
pcorr

Fig. 9 Uncracked thick-walled cylinder

being rs and c1 the internal and the external radius of related to the radial displacement and to the diameter
the cylinder, respectively, ers the hoop strain in the loss, due to the corrosion, must be completely
concrete at the internal radius (Fig. 9), and Ec the occupied by the rust (Fig. 9), and then:
concrete Young’s modulus. Voss ¼ VUrs;corr þ VNP ; ð5Þ
The hoop strain can be express as a function of the
radial displacement Urs,corr as follow: where the volume of oxides (Voss), which forms as a
  result of a depth of corrosion attack v, is:
2p rs þ Urs;corr  2prs
ers ¼ ð3Þ
2prs Voss ¼ a1 pð2rs v  v2 Þ; ð6Þ
and then: being a1 the volumetric ratio between oxide and iron.
Values of a1 suggested by Bhargava et al. [24] are
Urs;corr ¼ rs ers : ð4Þ
reported in Table 3.
It is assumed that the radial elastic deformation of The volume corresponding to the radial displace-
both rust and steel can be neglected with respect to the ment Urs,corr is:
concrete one. For this reason, due to the different 2
densities of the rust and the steel, the sum of the spaces VUrs;corr ¼ pð2rs Urs;corr þ Urs;corr Þ: ð7Þ
 Author's personal copy
Materials and Structures

Table 3 Volumetric ratio between the oxide and iron [24]

FeO Fe3O4 Fe2O3 Fe(OH)2 Fe(OH)3 Fe(OH)33H2O

a1 (Volumetric ratio) 1.80 2.00 2.20 3.75 4.2 6.4

Fig. 10 Hoop stresses pattern in thick-walled cylinder: cracking onset

The volume of the oxides that exert no pressure 3.2 Partly-cracked stage
(VNP) corresponding, in this phase, to the volume of
the corroded part of the steel bar is: In this stage, the cylinder can be subdivided in two parts,
delimited by the critical radius rcr (Fig. 11a). The first
VNP ¼ pð2rs v  v2 Þ: ð8Þ one, close to the bar (Fig. 11c), is cracked, while the
Therefore, the compatibility condition provides the other (uncracked) has an elastic behavior (Fig. 11b).
relationship between the corrosion attack and the The total corrosion pressure is equal to:
radial displacement: pcorr ¼ pEL NL
corr þ pcorr ; ð12Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
Urs;corr þ 2rs Urs;corr where pEL NL
corr and pcorr are the radial pressures at the
v ¼ rs  rs2  : ð9Þ radius rs and they represent the contributions of the
a1  1
uncracked and cracked cylinders to the corrosion
The uncracked stage ends when the hoop elastic pressure, respectively.
stress in the radius r = rs reaches the concrete tensile If the cracked portion is isolated, both the external
strength (Fig. 10). In this elastic stage, the hoop stress pressure due to the uncracked concrete pi and the internal
at the radius rs is equal to: corrosion pressure act on it (Fig. 11c). The pressure pi is
  evaluated on the elastic cylinder with thickness equal to
r2 c2 c1-rs, (Fig. 11b), in which the hoop internal stress is
rt;rs ¼ 2 s 2 pcorr 1 þ 21 : ð10Þ
c1  rs rs equal to the peak tensile strength fct (Fig. 10).
The internal pressure can be obtained from Eq. 2:
Corrosion attack corresponding to this first crack-
 2 2

ing is equal to: c1  rcr
pi ¼ fct 2 2
; ð13Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi! c1 þ rcr
e2 þ 2ecr
vcrack ¼ rs 1  1  cr : ð11Þ and for the equilibrium condition of the cracked
a1  1 cylinder:
 Author's personal copy
Materials and Structures

Fig. 11 Partly-cracked stage model

2prs pEL
corr ¼ 2prcr pi : ð14Þ
Finally, the corrosion pressure in equilibrium with
pi is:
 2 2

rcr c1  rcr
pEL
corr ¼ f ct : ð15Þ
rs c21 þ rcr
2

The cracked contribution to the corrosion pressure

(pNL
corr) is derived from the radial pressure in a generic
radius r, evaluated as follows:
Z
1 rcr Fig. 12 Entirely-cracked stage model
rr ¼ rt;r ðr Þdr; ð16Þ
r r
where rt,r is the hoop stress at a generic radius r.
The softening branch of the concrete constitutive In the following, only the first branch of the
tensile law is [34]: softening curve (Fig. 10) is used. This limitation may
r w be considered true for the usual concrete cover [34].
¼ ai þ bi i ¼ 1; 2; ð17Þ To correlate the crack opening w to the radius r the
fct w0
following hypotheses are accepted:
where:
– constant total elongation of the generic ring [42];
1f – hoop strain at the radius r equal to the strain related
a1 ¼ 
w to the peak tensile stress in the concrete ecr
(Fig. 11a).
f
a2 ¼  In this case the total elongation is equal to:
1w
Dtot ¼ nwðr Þ þ 2pecr r ¼ 2pecr rcr ; ð18Þ
b1 ¼ 1
where n is the number of cracks, and:
f
b2 ¼ 2pecr rcr  2pecr r
1w wðr Þ ¼ : ð19Þ
n
w, f are the coordinates of the intersection point of the
two softening lines (Fig. 10); w is the crack opening; The contribution of the cracked cylinder to the
w0 is the crack opening at a zero tensile strength corrosion pressure is obtained through integration of
(Fig. 10). the Eq. 16 with r = rs:
 Author's personal copy
Materials and Structures

Z
1 rcr fictitious radius greater than the concrete cover
pNL
corr ¼ rt;r ðr Þdr (Fig. 12). Following the procedure adopted for the
rs "rs
 2  # partially cracked stage, the values of pressure and
pecr rcr rcr
¼ fct a1 rs  1 þb1 1 : radial displacements are obtained for both the soften-
nw0 rs rs
ing branches (i = 1,2):
ð20Þ Z  
1 c1 2pecr rcr
In the same way, through the integration of the hoop pcorr;i ¼ rt;r ðr Þdr ¼ fct ai þ bi
rs rs nw0
strain, the radial displacement is obtained:   "  ##
  2    c1 pecr rs c1 2
2
c1  rcr rcr   1  ai 1 ;
Urs;corr ¼ ecr rcr 1 þ 2 ln rs nw0 rs
c þr 2 rs
  1 cr  ð26Þ
rcr rs
þ b1 ln 1þ   
rs rcr 2prcr
   ð21Þ Urs;corr;i ¼ ecr rcr þ ai ecr þ bi
prcr a1 ecr rcr nw0
þ 2 ln    
nw0 2 rs c1
 2 !#  c1 ln
rs
þ rs  c1
rs rs    
3 þ 4  : pecr c1
rcr rcr ai 2c21 ln þ rs2  c21 :
2nw0 rs
Finally, imposing that the corrosion products ð27Þ
spread into the cracks, the volume VNP is equal to:
In order to have a simpler solution, it is assumed
VNP ¼ pð2rs v  v2 Þ þ nwm lcr ; ð22Þ that all the concrete in the cracked cylinder is entirely
being wm the mean crack opening and lcr the crack in the stage that can be described either by the first or
length. by the second branch of the softening curve [34].
The corrosion attack can be obtained as a function Finally the corrosion attack may be calculated with
of the radius rcr: Eq. 23 where the mean crack opening is equal to:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  
2 w rs þ Urs;corr;i þ wðc1 Þ
Urs;corr þ 2rs Urs;corr þ nwm lcr =p wm ¼
v ¼ rs  rs2  ;  2 
a1  1 2rcr  rs þ Urs;corr;i  c1
¼p ecr ð28Þ
ð23Þ n
where the displacements Urs,corr are evaluated with and the crack length is:
Eq. 21, the crack opening length is equal to:  
  lcr ¼ c1  rs þ Urs;corr;i : ð29Þ
lcr ¼ rcr  rs þ Urs;corr ð24Þ
3.4 Bond model
and the mean crack opening wm is evaluated with Eq. 19:
  When the rebar is subjected to corrosion, and no slip is
w rs þ Urs;corr applied, the corrosion only determines, due to the
wm ¼
2   oxides expansion, radial displacements Urs,corr and
1 2prcr ecr  2p rs þ Urs;corr ecr pressure pcorr, assessable with the procedure described
¼ : ð25Þ
2 n in the previous Sections. When a pull-out force is
The values of pcorr and v are obtained by varying rcr applied to an un-corroded bar, considering the above
from rs to c1. cited cylinder model, a radial pressure pr,nc takes place
on the concrete around the bar, giving rise to a radial
3.3 Entirely-cracked stage displacement Urs,nc. These last variables can be
evaluated according to the classical bond modeling
In this stage the entire cover is cracked, and the by [33, 34], and then by adopting the analytical
corrosion pressure pNL
corr is evaluated by means of a formulations of Sects. 3.1, 3.2 and 3.3, in which pr,nc
 Author's personal copy
Materials and Structures

oxides expansion for a given corrosion level; K is the

secant stiffness of the pressure-radial displacement
curve for the fixed corrosion level, evaluated as:

pcorr
K¼ : ð31Þ
Urs ;corr
Finally, for a generic corrosion penetration v, the
radial stress rrad can be written as:
rrad ¼ pr;nc þ pcorr  Dp ð32Þ
where pr,nc is the radial pressure in absence of
Fig. 13 Radial displacements due to oxides (Urs, corr), and to corrosion; pcorr is the radial pressure due to the
bond action (Urs, nc)
expansive oxides; Dp is the oxides pressure variation
and Urs,nc have to replace the terms pr,corr and Urs,corr, caused by the increase of the radial displacements,
respectively. defined by Equation [30].
When the rebar is subjected to both axial pull-out Once known the radial pressure in the concrete
force and corrosion, the radial displacement due to the cover, the bond stresses are evaluated by means of the
oxides (Urs,corr) adds to the radial displacement due to Mohr–Coulomb criterion:
the bond action (Urs,nc) (Fig. 13). Then, when the
Urs,corr is higher than Urs,nc the oxides can expand in a s ¼ rrad  tan u þ c: ð33Þ
further space defined by DU = (Urs,nc–Urs,corr) and a The decay of the friction coefficient and the
reduction of the corrosion pressure occurs. adhesion strength, due to corrosion, can be given by
The proposed model, then, accounts for the con- linear formulations, whose parameters can be set
temporaneous presence of the corrosion and of the according to Coronelli [26]:
pull-out actions by introducing a correlation between
the corrosion penetration and the total radial displace- l ¼ tan u ¼ 0:37  0:26ðv  vcr Þ ð34Þ
ment due to both the cited phenomena (equations [30]
and [32]). As a result the loss of pressure (Dp) is c ¼ 3:41  21:21ðv  vcr Þ ð35Þ
introduced and defined, in a simple way as:
being the corrosion penetration and cr the cracking
Dp ¼ K  ðUrs;nc  Urs;corr Þ; ð30Þ corrosion penetration related to the cover through-
cracking (, cr in mm and c in MPa).
where Urs,nc is the radial displacement without corro-
sion; Urs,corr is the radial displacement due to the

Fig. 14 Comparison 20
between analytical and
18 Analytical model
numerical results (n = 2,
a1 = 3.75) 16 with cohesion
14
Analytical model with
τ [ MPa]

12 cohesion decay
10

8
6
Analytical model
4
without cohesion
2 Experimental results
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Μ loss [%]
 Author's personal copy
Materials and Structures

Table 4 Comparison between experimental and analytical results

% Mass Experimental Analytical curves (A) jEAj Min
Errors (%): E
loss data (E) (%)
With Cohesion Without With Cohesion Without
cohesion decay cohesion cohesion (%) decay (%) cohesion (%)

0.0 11.2 9.159 9.159 5.408 17.99 17.99 – 17.99

0.10 14.1 14.982 14.895 11.231 6.08 5.47 – 5.47
0.46 17.5 18.43 17.958 14.679 5.20 2.50 – 2.50
0.80 16.7 16.609 15.885 12.858 0.39 4.73 – 0.39
0.88 15.3 15.778 14.978 12.027 2.81 2.41 – 2.41
0.94 11.7 14.114 13.174 10.363 – 12.78 11.28 11.28
1.00 11.1 13.276 12.272 9.525 – 10.74 14.05 10.74
1.24 9.4 11.615 10.491 7.864 – 11.04 16.77 11.04
1.34 8.6 10.776 9.597 7.025 – 11.36 18.48 11.36
1.52 8.0 10.776 9.597 7.025 – 20.65 11.68 11.68
2.80 4.7 9.066 6.96 5.315 – 48.53 13.42 13.42
3.20 6.9 8.997 6.592 5.246 – 3.62 23.30 3.62
3.92 5.1 8.908 5.978 5.156 – 18.02 1.79 1.79
Average 6.49 13.06 13.85 7.98

30 ∅ = 10 mm 30
Analytical model with i = 0.05 mA/cm2
cohesion (n=3; α1 = 3.75) 25
25
Analytical model with
cohesion (n=2; α1 = 3.75)
20 20
τ [MPa]
τ [MPa]

15 15

10
10
Analytical model without
5 Analytical model without
5 cohesion (n=3; α1 = 3.75) cohesion (n=2; α1 = 3.75) Experimental results
Experimental results 0
0 0 1 2 3 4 5 6 7 8
0 1 2 3 4 5 6 7 8 M loss [%]
M loss [%]
Fig. 16 Comparison between analytical and experimental
Fig. 15 Comparison between analytical and experimental results in [18]
results in [13]

(Ec = 30,279 MPa and fct = 2.3 MPa). The constitu-

3.5 Model validation tive relationship in tension is defined by assuming the
values of 0.25 and 0.33 for the coefficients f and w
The proposed analytical model has been validated respectively (detail of Fig. 10). The crack opening w0
firstly through a comparison with the experimental (related to a null tensile strength, see Fig. 10) is
results shown in Sect. 2. assumed equal to 0.25 [43].
The cubic concrete strength in compression is According to the experimental evidence, it is
assumed equal to 29 MPa, according to the mean assumed that two cracks form in the specimen
experimental value (Sect. 2). The Young’s modulus (Fig. 4). Since the tested specimens were fully
(Ec) and the tensile strength (fct) are evaluated immersed in water during the accelerated corrosion
with reference to the Ec2 code formulations process, in the analytical simulation it is supposed that
 Author's personal copy
Materials and Structures

30 Analytical model with expressed as percentage of mass loss, in the hypoth-

cohesion (n=2; α1 = 3.75) i = 0.1 mA/cm2
25 eses of full cohesion, absence of cohesion, and with
Analytical model with cohesion decay given by Eq. 35. The experimental
20 cohesion (n=2; α1 = 4.2)
τ [MPa]

outcomes are also highlighted in the same figure. It can

15 be clearly noted that the experimental dots are very
10 close to the curve with corrosion decay up to mass loss
5
of about 1 %. For higher corrosion levels, the exper-
without cohesion without cohesion
(n=2; α1 = 4.2) (n=2; α1 = 3.75)
imental outcomes lay on the diagram related to the
0
0 1 2 3 4 5 6 7 8 absence of cohesion, showing that no gradual decay of
M loss [%] this property occurs in the analysed case. The error, in
percentage, is reported in Table 4. It can be noted the
Fig. 17 Comparison between analytical and experimental average value is less than 13 % and the minimum one
results in [18]
is about 8 %.
The effectiveness of the proposed analytical model
Fe(OH)3 can form (a1 = 3.75 as reported in Table 3). is further shown through a comparison with other
In Fig. 14 the maximum bond stress obtained with the experimental data available in literature [13, 18]. In
analytical model is plotted versus the corrosion level, particular the results of the analytical simulation of the

Fig. 18 Analytical (a) 20

maximum bond strength cohesion decay, n=3 α1 = 3.75
versus mass loss; a influence 18
of number of cracks (n);
b influence of type of oxides 16 cohesion decay, n=2
(a1) 14
τ [ MPa]

12

10

8
without cohesion, n=3
6
4

2
without cohesion, n=2
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Μ loss [%]

(b) 20
cohesion decay, α1 = 3.75 n= 2
18

16
14 cohesion decay, α1 = 4.2
τ [ MPa]

12

10

6
4 without cohesion, α1 = 3.75
2 without cohesion, α1 = 4.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Μ loss [%]
 Author's personal copy
Materials and Structures

tests reported in [13] are shown in Fig. 15. It can be The presented model appears effective in catching
clearly noted how the model catches the experimental the phenomenon, as the experimental results are very
pattern, since the experimental dots are well disposed close to the theoretical curves. In particular, the
in the range defined by the analytical curves with full differences between the analytical and experimental
cohesion and without it. In Figs. 16 and 17 the results outcomes, here described and developed by the
pointed out in [18], related to two different current authors, is, in average, less than 13 %. The compar-
densities (0.05 and 0.1 mA/cm2) and higher corrosion ison with other data, available in literature, confirms
levels ranging between 3 and 7 % of mass loss, are this result.
reported and compared with the outcomes of the
analytical model. Due to the uncertainties of the type Acknowledgments The authors are grateful to Serena
Checchi who performs analytical evaluations during her
of oxides, two typical values of a1 are considered (3.75
graduation thesis.
and 4.2). It can be seen that experimental data are
between the curves with and without cohesion
(Fig. 16), or very close to the curve without cohesion
(Fig. 17). References
Finally, the influence of the cracks number (n) and
of the type of formed oxides (a1) on the bond-corrosion 1. Rodriguez J, Ortega L, Casal J, Diez JM (1996) Assessing
relationship is highlighted in Fig. 18. In particular the structural conditions of concrete structures with corroded
reinforcement. In: International conference of concrete in
cases related to the n = 2 (continuous lines) and n = 3 the service of mankind, Dundee, UK, June
(dotted lines) are drawn in Fig. 18a, for a1 = 3.75, 2. Cairns J (1998) Assessment of effects of reinforcement
whilst the influence of the a1 parameter (a1 = 3.75, corrosion on residual strength of deteriorating concrete
continuous lines and 4.2, dotted lines), for n = 2 is structures. In: Proceedings of the first international confer-
ence on behaviour of damaged structures, Rio de Janeiro,
shown in Fig. 18b. It can be noted how both the Niteroi, May
variables can influence the softening branch of the law. 3. Castel A, Francois R, Arligue G (2000) Mechanical
behaviour of corroded reinforced concrete beams—part 1:
experimental study of corroded beams. Mater Struct
33:539–544
4 Conclusions 4. Coronelli D, Gambarova PG (2001) Structural response of
corroded R/C elements: assessment by numerical analysis.
The influence of the corrosion on the bond properties In: Proceedings of the 9th international conference on
of r.c. elements is analysed in the paper, both with structural faults and repair (SFR2001), London, July
5. Coronelli D, Gambarova PG (2004) Structural assessment
experimental and analytical models. of corroded reinforced concrete beams: modeling guide-
On the basis of the experimental evidences, lines. ASCE J Struct Eng 130(8):1214–1224
obtained through pull-out tests on corroded specimens 6. Rinaldi Z, Valente C, Pardi L (2008) A simplified meth-
without stirrups, the following remarks can be drawn: odology for the evaluation of the residual life of corroded
elements. Struct Infrastruct Eng 4(2):139–152
– low corrosion percentage (lower than 0.5–0.6 % in 7. Rinaldi Z, Imperatore S, Valente C (2010) Experimental
mass loss) leads to an increase of the bond strength evaluation of the flexural behavior of corroded P/C beams.
Constr Build Mater 24(11):2267–2278
up to about 50–60 % with respect to the sound 8. Almusallam AA (2001) Effect of degree of corrosion on the
bars; properties of reinforcing steel bars. Constr Build Mater
– higher corrosion entities cause a sharp bond 15(8):361–368
reduction; as a matter of fact for mass loss of 9. Cairns J, Plizzari GA, Du Y, Law DW, Franzoni C (2005)
Mechanical properties of corrosion-damaged reinforce-
about 1.5 % the maximum bond stress is reduced ment. ACI Mater J 102(4):256–264
of about 40 %; 10. Apostolopoulos CA, Papadopoulos MP, Pantelakis SG
(2006) Tensile behavior of corroded reinforcing steel bars
The proposed analytical procedure, based on the BSt 500 s. Constr Build Mater 20:782–789
analysis of a thick-walled cylinder, allows the corre- 11. Imperatore S, Rinaldi Z (2008) Mechanical behaviour of
lation between the corrosion penetration and the corroded rebars and influence on the structural response of
R/C elements. Proceedings of the 2nd International con-
related radial displacement, and then the relationship ference on concrete repair, rehabilitation and retrofitting,
between the bond stress and the corrosion level, Cape Town, South Africa, 24–26 November, CRC Press,
accounting for both the corrosion and pull-out actions. Balkema. ISBN 978-0-415-46850-3
 Author's personal copy
Materials and Structures

12. Auyeung Y (2001) Bond properties of corroded reinforce- 28. Kallias AN, Imran Rafiq M (2010) Finite element investi-
ment with and without confinement, PhD Thesis, New gation of the structural response of corroded RC beams. Eng
Brunswick Rutgers, The State University of New Jersey Struct 32(9):2984–2994
13. Al-Sulaimani GJ, Kaleemullah M, Basunbul IA, Rash- 29. Giordano L, Mancini G, Tondolo F (2010) Numerical
eeduzzafar (1990). Influence of corrosion and cracking on interpretation of bond between steel and concrete in pre-
bond behavior and strength of reinforced concrete members. sence of corrosion and cyclic action. Key Eng Mater
ACI Struct J, 87(2):220–231 417–418:349–352
14. Andrade C, Alonso C, Molina FJ (1993) Cover cracking as a 30. Lee HS, Noguchi T, Tomosawa F (2002) Evaluation of the
function of bar corrosion: part I-experimental test. Mater bond properties between concrete and reinforcement as a
Struct 26(8):453–464 function of the degree of reinforcement corrosion. Cem
15. Rodriguez J, Ortega L, Garcia A (1994). Corrosion of Concr Res 32(8):1313–1318
Reinforcing Bars and Service Life of R/C Structures: Cor- 31. Clark LA, Saifullah M (1994) Effect of corrosion rate on the
rosion and Bond Deterioration, Concrete across Borders, bond strength of corroded reinforcement. In: Swamy RN
Proceedings, Odense, Denmark, V. II, pp. 315-326 (ed) Corrosion and corrosion protection of steel in concrete.
16. Cabrera JG (1996) Deterioration of concrete due to rein- Sheffield Academic Press, Sheffield, pp 591–602
forcement steel corrosion. Cem Concr Compos 18(1):47–59 32. Coronelli D (1997) 1997. Bond of corroded bars in confined
17. Almusallam AA, Al-Gahtani AS, Aziz AR, Rasheeduzzafar concrete: test results and mechanical modelling, Studies and
(1996). Effect of reinforcement corrosion on bond strength. Researches 18:137–211
Constr Build Mater 10(2):123–129 33. Tepfers R (1979) Cracking of concrete cover along
18. Horrigmoe G, Sæther I, Antonsen R, Arntsen B (2007) anchored deformed reinforcing bars. Mag Concr Res
Sustainable bridges. Laboratory investigations of steel bar 31(106):3–12
corrosion in concrete Background document SB3.10. Sixth 34. Den Uijl JA, Bigaj AJ (1996) A bond model for ribbed bars
Framework Programme based on concrete confinement. HERON 41(3):206–226
19. Auyeung Y, Balaguru P, Chung L (2000) Bond behavior of 35. Bažant ZP (1979) Physical model for steel corrosion in sea
corroded reinforcement bars. ACI Mater J 97(2):214–220 concrete structures applications. ASCE J Struct Div
20. Fang C, Lundgren K, Chen L, Zhu C (2004) Corrosion 105(ST6):1155
influence on bond in reinforced concrete. Cem Concr Res 36. Liu Y, Weyers RE (1998) Modeling the time-to-corrosion
34:2159–2167 cracking in chloride contaminated reinforced concrete
21. Prieto M, Tanner P, Andrade C (2011) Bond response in structures. ACI Mater J 95(6):675–680
structural concrete with corroded steel bars Experimental 37. Balafas I, Burgoyne CJ (2010) Modeling the structural
Results In: C Andrade and G Mancini (eds) Modelling of effects of rust in concrete cover. ASCE J Eng Mech
corroding concrete structures RILEM Bookseries. Madrid, 137(3):175–185
Spain, 5:231–241 38. Lu C, Jin W, Liu R (2011) Reinforcement corrosion-
22. Wang X, Liu X (2004) Modeling bond strength of corroded induced cover cracking and its time prediction for rein-
reinforcement without stirrups. Cem Concr Res 34:1331–1339 forced concrete structures. Corros Sci 53:1337–1347
23. Wang X, Liu X (2006) Bond strength modeling for corroded 39. Lundgren K (2005) Bond between ribbed bars and concrete.
reinforcements. Constr Build Mater 20:177–186 Part 2: the effect of corrosion. Mag Concr Res 57(7):383–396
24. Bhargava K, Ghosh AK, Mori Y, Ramanujam S (2007) 40. Ouglova A, Berthaud Y, Francois M, Foct F (2006)
Corrosion-induced bond strength degradation in reinforced Mechanical properties of an iron oxide formed by corrosion
concrete—analytical and empirical models. Nucl Eng Des in reinforced concrete structures. Corros Sci 48(12):3988–
237:1140–1157 4000
25. Molina F, Alonso C, Andrade C (1993) Cover cracking as a 41. Timoshenko SP, Goodier JN (1970) Theory of elasticity.
function of bar corrosion: part 2—numerical model. RILEM McGraw Hill Book Co., New York
Mater Struct 26:532–548 42. Van der Veen C (1990) Theoretical and experimental
26. Coronelli D (2002) Corrosion cracking and bond strength determination of the crack width in reinforced concrete at
modeling for corroded bars in reinforced concrete. ACI very low temperatures. Heron 35(2):111
Struct J 99(3):267–276 43. Bažant ZP (2002) Concrete fracture models: testing and
27. Berto L, Simioni P, Saetta A (2008) Numerical modelling of practice. Eng Fract Mech 69(2):165–205
bond behaviour in RC structures affected by reinforcement
corrosion. Eng Struct 30(5):1375–1385

View publication stats