Received: 2 February 2021 Revised: 18 May 2021 Accepted: 18 June 2021

DOI: 10.1002/suco.202100065

TECHNICAL PAPER

An experimental study on seismic performance levels

of highly corroded reinforced concrete columns

Alper Celik | Hakan Yalciner | Atila Kumbasaroglu _

| Ahmet Ihsan Turan

Department of Civil Engineering,

Erzincan Binali Yıldırım University,
Abstract
Erzincan, Turkey In the literature, only one empirical model is available as a nondestructive
method for the prediction of seismic performance levels of corroded reinforced
Correspondence
Hakan Yalciner, Department of Civil concrete (RC) columns as a function of the initial corrosion crack width at
Engineering, Erzincan Binali Yıldırım lower corrosion levels. Because of the ruptured transverse reinforcement bars
University, Erzincan, Turkey.
at higher corrosion levels, the structural behavior may turn brittle in terms of
Email: hakan.yalciner@emu.edu.tr
shear failure. Therefore, in this study, higher corrosion levels for a different
Funding information concrete strength level from that empirical model were studied. To do this,
Scientific Research Commission of
Erzincan Binali Yıldırım University,
four RC columns were subjected to accelerated corrosion, and the widths of
Grant/Award Number: BAP-FBA- initial corrosion cracks were measured. The corroded RC columns were then
2020-659 tested under combined constant axial load and cyclic lateral displacement
excursions. After the cyclic loading test, the actual corrosion levels at each
reinforcement bar were obtained by extracting the reinforcement bars from
the concrete. Test results showed that the prediction of seismic performance
levels of corroded RC columns based on initial corrosion crack widths were
limited owing to the nonlinear increase in the crack width with the increase in
the corrosion levels. New empirical models were developed to predict the
remaining energy capacities and seismic performance levels of the corroded
RC columns.

KEYWORDS
corrosion, crack width, cyclic loading, reinforced concrete column, reinforcement bar,
seismic performance level

1 | INTRODUCTION developed to predict the mechanical properties of cor-

roded bars,1,2 time to cracking of concrete,3–5 bond–slip
The degradation of the mechanical properties of both relationships,6-8 probabilistic service life prediction of
steel bars and concrete comprises several corrosion corroded RC members,9–11 effects of corrosion on the
effects that reduce the seismic performance levels of the structural capacity of concrete members by considering
corroded reinforced concrete (RC) members. Therefore, either uniform or pitting corrosion,12–15 and corrosion
analytical and empirical models have been widely repair techniques.16,17
However, in situ data collections by considering all
the effects of corrosion are complex. How to predict and
Discussion on this paper must be submitted within two months of the
print publication. The discussion will then be published in print, along
reflect the effects of the degree of reinforcement corro-
with the authors’ closure, if any, approximately nine months after the sion in corroded reinforced structures have been a subject
print publication. of discussion for several decades. The removal of the total

Structural Concrete. 2021;1–19. wileyonlinelibrary.com/journal/suco © 2021 International Federation for Structural Concrete. 1
2 CELIK ET AL.

or partial concrete cover depth of an RC column as well absence load), a number of researchers have attempted to
as the detection of the amount of corrosion levels and focus on the prediction of the structural capacity of cor-
type (i.e., uniform or pitting), which may show differ- roded RC beams as a function of the corrosion crack wid-
ences along the length of a reinforcement, is not easy for ths.27 Moreover, Khan et al.25 notably stated that the
existing RC buildings. Consequently, the prediction of crack width of longitudinal corrosion cracks is indepen-
the degradation of mechanical properties of concrete and dent of the steel cross-section loss in the stirrups. Con-
reinforcement bars, bond–slip relationships, and then tradicting results for the initial corrosion crack widths of
structural modeling of such effects may overburden the RC beams still exist without considering other test
structural engineers. The experimental study conducted parameters, such as the effects of steel ratios and distance
by Chung et al.18 on corroded RC beams and the theoreti- between the steel bars on crack widths (not considering
cal study for corroded RC columns by Zhang et al.19 also crack widths on the compression zone, and thus the
highlighted that for accurate structural assessment and reduction in the ultimate strain in the concrete). How-
exact modeling of corroded RC members, the changes in ever, only one empirical model exists for the prediction
the development length of the reinforcement bars must of the structural capacity of corroded RC columns as a
be predicted. While the theoretical studies to predict the function of the measured initial corrosion crack width.
development length of the corroded RC columns are Yalciner et al.28 corroded 25 RC columns at three con-
new19 and because no experimental models are available crete strength levels (9, 27, and 37 MPa) and tested
yet, another major effect of corrosion, which is the degra- them under cyclic loading at two axial load ratios (0.20
dation in the plastic hinge length, remains a topic of and 0.40). In that study, all the RC columns were bro-
study for the accurate modeling of corroded RC mem- ken, and the longitudinal and transverse reinforcement
bers. Owing to the lack of modeling of the plastic hinge bars were extracted after the cyclic lateral displacement
length of corroded RC columns, studies must still use the tests to obtain the actual corrosion levels, at which the
plastic hinge length models of uncorroded RC columns initial corrosion crack widths were measured after the
for corroded RC columns.20,21 Therefore, the problem of accelerated corrosion method. Yalciner et al.28 found a
the practical structural assessment of RC concrete struc- strong relationship between the energy absorption
tures due to the complex effects of corrosion of reinforce- capacities and recorded initial corrosion crack widths
ment bars has received researchers’ attention. for the prediction of seismic performance levels of cor-
Visual inspection can be counted as the most practical roded RC columns. In contrast to the study by Yalciner
technique for the structural assessment of corroded RC et al.,28 in this study, a different concrete strength level
members. As the reinforcement bars tend to corrode, cor- was used, and the reinforcement bars were corroded to
rosion products fill the spaces in the concrete. As indi- higher corrosion levels to investigate the effect of buck-
cated by Bažant,22 the volume of rust is generally two to ling of reinforcement bars on the structural capacity
four times larger than the volume of the original rein- and ensure the validity of the model proposed by
forcement bar. Restricted space in concrete due to corro- Yalciner et al.28
sion products and volumetric expansion with internal
radial pressure on the concrete surface cause cracking
and spalling of the concrete cover depth. In addition, 2 | RESEARCH SIGNIFICANCE
Yalciner et al.6 showed that the effects of the volumetric
expansion by corrosion products differ at low and high The latest empirically developed model by Yalciner
strength levels of the concrete and must be considered et al.,28 which is included as a single model in the current
for the proposed models. Because of the complex behav- literature, is limited to the average corrosion levels rang-
ior of corrosion and lack of modeling of corrosion effects, ing between 0% and 12.36% at stirrups. In that study,
the first sign of corrosion in situ, which is cracking of owing to the lower corrosion levels than those considered
concrete, has attracted the attention of researchers to in the present study, no rupturing of stirrups was
develop simple models for the prediction of the structural observed. Furthermore, the buckling of the reinforce-
capacity of RC members based on measured initial corro- ment bars, which is another effect of corrosion, did not
sion crack widths. Therefore, several experimental stud- occur because of lower corrosion levels during combined
ies on corroded RC beams have been conducted to cyclic lateral displacement excursions. The different con-
predict the crack width of corroded RC beams as a func- crete strengths and corrosion levels used in the present
tion of the corrosion levels or vice versa.23–25 Although study provide further insights and validation of the model
Malumbela et al.26 highlighted a critical review for the by Yalciner et al.28 through an experiment considering
loading systems used for corroded RC beams in terms of the mentioned contradicting results on initial corrosion
the crack width (i.e., corroded RC beams under service or crack widths.
CELIK ET AL. 3

3 | MATERIALS AND METHODS follows: yield strength of 500 MPa, rupture strength of
590 MPa, and yielding and rupture strains of 0.0022 and
3.1 | Experimental program 0.1156, respectively. The calculated elastic modulus of the
reinforcement bars was 2
105 MPa. Ready-mixed con-
In the current study, the specimens were designated as crete was used for all RC columns, where the concrete was
RCA,x%,y%, where A indicates the number of specimens, poured at the same time and had the same concrete qual-
and x and y represent the average corrosion levels of the ity. The average mechanical properties of the cylindrical
longitudinal bars and stirrups, respectively, (for example, concrete specimens were obtained on the day of the cyclic
RC1,0%,0%). loading tests. The determined average cylindrical compres-
sive strength of the concrete (fc) was 15 MPa, and the
corresponding split tensile strength was 1.5 MPa. Concrete
3.2 | Sectional and material properties of a low-strength level was selected because of the follow-
ing two reasons. First, considering that the strength levels
The sectional properties of RC columns were constructed of concrete decrease because of time-dependent effects of
as described by Yalciner et al.28 and summarized here. As corrosion in the age of a structure29 where the corrosion
shown in Figure 1, all the RC columns had a square cross- and seismic risks are higher for older RC buildings. Sec-
section of 300
300 mm2 with a 2600-mm length. Eight ond, because the volumetric expansion due to the corro-
16-mm deformed longitudinal bars were used for testing sion products and the consequent cracking of concrete are
the RC columns. The concrete cover depths used for the based on the tensile strength of concrete, an experimental
RC columns were the same and equal to 35 mm. Tensile study on the low-strength level of concrete is needed to
and compression tests were conducted to obtain the validate the previous experimental studies.
mechanical properties of the reinforcement bars and con-
crete. The average mechanical properties of the deformed
16-mm-diameter reinforcement bars were recorded as 4 | AC CELERATED C OR ROS ION
METHOD

Before the installation of the longitudinal bars and stirrups

and application of the accelerated corrosion method, all
reinforcement bars were cleaned by brushing. Next, a bal-
ance between two points of load cells was used to record
the initial mass of each reinforcement bar. As shown in
Figure 2, after recording the initial mass of the reinforce-
ment bars, 4-mm-diameter stainless copper wires sur-
rounding the diameter of the steel bars were tied to each
longitudinal bar and stirrup prior to pouring the concrete.
During the application of copper wires, they were
uncoated only at the contact area of the reinforcement
bars to prevent the loss of applied current. In addition, as
shown in Figure 3, the concrete corrosion pool was iso-
lated with a plastic membrane to prevent any loss in the
applied current.
In Figure 3, an adjustable direct current power supply
of 60 V with a range of 0–10 A was used for the aging
process of the RC columns. The applied artificial corro-
sion current was 1.5 A during the accelerated corrosion
process. The positive terminal (i.e., anode) for the aging
process included the extended copper wires from the
reinforcement bars. In addition, 3-mm stainless copper
plates were placed parallel to the RC columns in the cor-
rosion pool, which acted as the negative terminal (cath-
F I G U R E 1 Section properties of the RC columns. Bs, back ode). One of the objectives of the present study was to
side; Fs, front side; Ls, left side; RC, reinforced concrete; Rs, corrode the reinforcement bars to levels higher than the
right side corrosion levels studied by Yalciner et al.28 To achieve
4 CELIK ET AL.

F I G U R E 2 Copper wires used

for positive terminal

products. In the present study, the influence of pitting

corrosion and spatial variability was not considered when
the corrosion rates were used for artificial corrosion. It
should be noted that alternative methods for accelerating
the corrosion of steel in concrete have been reported in
the literature, where their effects lead to different corro-
sion distribution on the surface of reinforcement bars.
While the whole surface of reinforcement bars is cor-
roded using the impressed current technique, localized
corrosion highly occurred because of a natural corrosive
environment. Although the impressed current technique
of corrosion acceleration is an effective and quick method
for accelerating chloride-induced corrosion, it may not
fully capture the naturally induced corrosion, particularly
FIGURE 3 Accelerated corrosion setup in term of time-dependent models. Obtained actual corro-
sion levels in the present study with a short period might
be counted as higher than naturally corroded RC mem-
higher amounts of corrosion levels, digital voltmeters bers found in the available literature (e.g., Zhou et al.).30
were used for each corroded RC column to record the The obtained test results in this paper do not consider the
passing current at 1-min intervals. Theoretical mass influence of pitting corrosion and spatial variability.
losses were monitored according to Faraday's law. Note However, it was also believed that for in-service struc-
that Faraday's law was used only to monitor and estimate tures, it is not easy to obtain the exact distribution of
the time of the applied current. The actual corrosion localized corrosion to model them along the length of the
levels were obtained by breaking the concrete and reinforcement bars. Therefore, further studies are needed
extracting all the longitudinal and transverse reinforce- to correlate the corrosion crack widths and corrosion
ment bars following the cyclic loading tests and levels under a natural corrosive environment for different
reweighing each reinforcement bar. Equation (1) was axial load ratios acting on the concrete with different ten-
used to determine the percentage of corrosion levels: sile strengths for the concrete.

mi mf
CL ¼ x 100, ð1Þ
mi 5 | MEASUREMENT OF INITIAL
COR ROS ION CRACK WIDTHS
where mi is the original mass of the reinforcement bars
prior to corrosion, and mf is the final mass of the rein- The initial corrosion crack widths at the reinforcement
forcement bars following the removal of the corrosion bars, numbered as 1 and 7 (see Figure 1), were measured
CELIK ET AL. 5

prior to the loading tests and following the completed the present study, both the vertical and horizontal jacks
accelerated corrosion process. To do this, a crack- comprised hinge systems; thus, the nominal load effect
detection microscope was used and the initial corrosion and eccentricity were prevented. The lateral displacement
crack widths at 200-mm intervals along the length of the was measured using a linear variable differential trans-
L1 and L7 reinforcement bars were measured. In this pro- former (LVDT#1), which was placed to the left end of the
cess, the corrosion products that filled the crack widths beam at the level of the applied horizontal load. Figure 5
were removed using an air compressor. shows the loading protocol used in the current study in
accordance with FEMA 461.31 Figure 5 shows that three
complete cycles using a triangular waveform (i.e., the
6 | L O A D I N G SE T U P loading and unloading) were applied for each roof drift
ratio.
Figure 4 shows the schematic of the experimental test
setup. To prevent the horizontal and vertical displace-
ments of the column foundation, two 4-mm steel plates 7 | RESULTS A ND DISCUSSION
were used to mount the perimeter of the foundation. The
foundation surrounded with steel plates was then 7.1 | Initial corrosion crack widths and
anchored to the strong floor by using six 16-mm-diameter actual corrosion levels
steel profiles.
A constant axial load ratio of 0.40 was applied to each After the completed accelerated corrosion process and
tested RC column. The axial load that was placed on the before the cyclic loading tests, the initial corrosion crack
RC beams was applied with two hydraulic jacks, each widths of reinforcement bars 1 and 7 (see Figure 1) were
with a 1000 kN capacity. For the present study, reduced measured, as shown in Figure 6. Although same amount
axial loads at the level of plastic deformations during the of current was applied at the same RC column, more cor-
cyclic loadings were not re-increased to that of applied rosion levels occurred at the reinforcement bars of L1
axial load level. The lateral load was applied with a and L7 compared with L8. Two main reasons can be
600 kN hydraulic jack that was fixed to the reaction wall. counted for this situation. The first one was the corrosive
The applied lateral and axial loads were monitored using behavior resulting from the corrosion acceleration using
a load cell and two pressure transducers, respectively. In an impressed direct current technique may be different
from that resulting from the corrosion acceleration under
an artificial climate environment. The second reason was
because of the positions of RC columns in the corrosion
pool. The copper plates surrounding the RC columns
were closer to those reinforcement bars of L1 and L7,
thus the path of applied current caused to have more visi-
ble initial corrosion crack widths at L1 and L7.
To obtain the corrosion levels related to the initial
corrosion crack widths, all the RC columns had to be bro-
ken to extract the longitudinal bars and stirrups after
cyclic lateral displacement tests. Figure 7 shows the
breaking of the concrete and extracted reinforcement
bars from the RC columns.
Then, as shown in Figure 8, once the reinforcement
bars were fully extracted, mechanical cleaning
(i.e., brushing the steel bars) was applied to each rein-
forcement bar to remove the corrosion. All longitudinal
reinforcement bars and stirrups were then reweighted to
determine the actual corrosion levels using Equation (1).
The obtained distribution of corrosion levels of the longi-
F I G U R E 4 Loading test setup: 1: Computer, 2: Data loggers, 3: tudinal bars and stirrups at each RC column are shown
Uninterruptible power supply, 4: RC column, 5: LVDT#1 for lateral in Figures 9 and 10, respectively. The gravimetric test
displacement, 6: Axial load, 7: Lateral hydraulic jack (600 kN), 8: results are summarized in Table 1.
Vertical hydraulic jack (1000 kN). LVDT, linear variable differential It should be noted that the given corrosion levels in
transformer; RC, reinforced concrete Table 1 are not the corrosion levels corresponding to the
6 CELIK ET AL.

crack widths, as shown in Figure 6. To determine the cor- 11 parts, where the initial corrosion crack widths were
rosion levels caused by the related crack widths in recorded after the accelerated corrosion process.
Figure 6, the longitudinal bars of L1 and L7 were cut into Figure 11 shows a selected cut piece of reinforcement
bar, and Table 2 summarizes the gravimetric test results
of the cut bars with respect to the corresponding initial
120 6

100
corrosion crack widths.
80 4 The constructed initial corrosion crack widths in
60 Figure 6 and the corresponding actual corrosion levels
Displacement (mm)

40 2
are depicted in Figure 12. The experimental test results

Drift raito (%)

20

0 0
by Yalciner et al.28 and the maximum crack widths
-20 obtained by Liu et al.32 are also included in Figure 12.
-40 -2 An empirical model was developed by Yalciner
-60
et al.28 to predict the reduction in the cross-sectional area
-80 -4

-100 (ΔAs) of the longitudinal bars based on the measured ini-

-120 -6 tial corrosion crack widths (W cr Þ, using Equation (2):
0 10 20 30 40 50 60 70 80 90 100

Number of cycle
W cr
ΔAs ¼ ð2Þ
FIGURE 5 Loading protocol 0:046
W cr þ 0:058

F I G U R E 6 Crack patterns B:
beam, F: foundation
CELIK ET AL. 7

F I G U R E 8 Obtaining actual corrosion levels: (a) mechanical

cleaning and (b) weighing of the reinforcement bars
F I G U R E 7 Process for obtaining actual corrosion levels:
(a) breaking of RC columns; (b) extracting the reinforcement bars.
RC, reinforced concrete
corrosion levels of the longitudinal bar of RC5-L7 (see
Table 2) at a distance of 1000 and 1200 mm were 100%,
the measured crack widths were not higher than those of
Similarly, Equation (2) was used in the present study, the other measured crack widths at lower corrosion
and Figure 13 depicts the ratios of the predicted experi- levels. Therefore, the model proposed by Yalciner et al.28
mental values. is limited to the corrosion levels ranging from 0.0% to
As shown in Figure 13, although the compressive 20% used for nondestructive methods. However, as
strength of the current study was within the limits of the shown in Figure 13, there is a fitting region to improve
model proposed by Yalciner et al.,28 the developed model the proposed Equation (2) for higher corrosion levels.
underestimated the current experimentally measured Even if the fitting region in Figure 13 is used to improve
reduction in the cross-sectional area. This was because of the previous model, the main problem is to determine
the limited crack widths of the concrete after the corro- the corrosion level at which the corrosion rate continues
sion propagation process. As shown in Figure 12, consid- to increase at a constant crack width. Although an
ering the present study and the data obtained from increased amount of corrosion level may not be predicted
Yalciner et al.28 and Liu et al.,32 once the concrete was for the constant crack width, the considered corrosion
cracked and reached a certain value (i.e., depending on levels up to 20% can aid in rapid decision-making of the
the diameter of the steel bars), the crack width of the structural performance of corroded buildings as a nonde-
concrete remained constant while the corrosion level structive method.
(i.e., the reduction in cross-sectional area) continued to
increase. This can be explained by the diffusion of the
corrosion products from the concrete during the corro- 7.2 | Hysteresis loops
sion propagation process. The reduced radial stress on
the concrete due to the diffusion of corrosion products The hysteretic loops of the tested RC columns are
caused the measured crack widths to spread, as shown in depicted in Figure 14 and the test results are summa-
Figure 12. As shown in Figures 11 and 12, although the rized in Table 3. The corrosion levels at the two
8 CELIK ET AL.

60 F I G U R E 9 Corrosion
distribution at the longitudinal bars

50

40
Corrosion level (%)

30

20

10

0
L1 L2 L3 L4 L5 L6 L7 L8 L1 L2 L3 L4 L5 L6 L7 L8 L1 L2 L3 L4 L5 L6 L7 L8 L1 L2 L3 L4 L5 L6 L7 L8
RC2 RC3 RC4 RC5
Longitudinal bars

90 F I G U R E 1 0 Corrosion
distribution at the stirrups
80

70

60
Corrosion level (%)

50

40

30

20

10

0
S7
S9
S1
S3
S5
S7
S9

S1
S3
S5

S1
S3
S5
S7
S9

S1
S3
S5
S7
S9
S11
S13
S15
S17

S11
S13
S15
S17

S11
S13
S15
S17

S11
S13
S15
S17

RC2 RC3 RC4 RC5

Stirrups

TABLE 1 Gravimetric test results

Initial mass (g) Actual mass loss (g) Actual corrosion level (%) ΔAs (mm2)

Specimen S L S L CLS CLL ΔAsS ΔAsL ΣΔAsL

RC1 – – 0.0 0.0 0.0 0.0 0.0 0.0 0.0
RC2 8029.0 32487.5 4991.5 27480.5 37.83 15.41 18.98 30.98 247.86
RC3 8080.0 32375.5 4538.0 25836.0 43.84 20.20 21.94 40.68 325.45
RC4 8186.5 32394.0 2909.0 23527.0 64.47 27.37 32.40 55.10 440.86
RC5 8007.0 32586.5 2978.5 23350.5 62.80 28.34 31.54 56.76 454.12

Abbreviations: CLL (%), average actual corrosion levels at longitudinal bars; CLS (%): average actual corrosion levels at stirrups; L, total mass of eight
longitudinal bars; RC, reinforced concrete; S, total mass of 18 stirrups; ΔAsS, average loss in cross-sectional area of a stirrup in a column; ΔAsL, average loss in
cross-sectional area of longitudinal bars in a column; ΣΔAsL, cumulative loss in cross-sectional area of longitudinal bars in a column.
CELIK ET AL. 9

resulted in the decrease of the structural performance of

the corroded RC columns. Because the initial masses of
the stirrups were less than those of the longitudinal bars at
the same corrosion levels, the rupturing of the stirrups
was observed at all corroded RC columns. The rupturing
of the stirrups occurred along the length of the plastic
hinges of the corroded RC columns. Because of the
spalling of the concrete, rupturing of the stirrups, and con-
sequent reduction in the confinement effects due to the
transverse reinforcement bars, the longitudinal reinforce-
ment bars of the corroded RC columns were observed to
buckle. Moreover, as shown in Figure 15, as the corrosion
levels increased, the buckling length of the longitudinal
FIGURE 11 Corroded and cut bars bars also increased owing to the reduction in the diameter
of the longitudinal reinforcement bars and the absence of
transverse reinforcement bars under the same applied
loading directions of the RC columns are also shown axial load ratio. In addition, corrosion also caused the deg-
in Figure 14. The experimentally obtained cyclic loops radation of the plastic deformation, thus highly affecting
were idealized for the required comparisons. For this, the structural performance of the corroded RC column. As
the yield displacement (Δy) and corresponding load shown in Figure 15, the plastic hinge length of the cor-
(Fy) of all the RC columns were calculated according roded RC columns increased with the increase in the
to the procedure described by Sezen and Moehle33 The amount of corrosion levels compared with uncorroded RC
ultimate load (Fu) was obtained when the given set of column. In other words, corroded RC columns prema-
three cycles dropped to 85% of the maximum turely dissipated their energy capacities at lower drift
lateral load. ratios and at increasing lengths of the plastic hinges. In
According to the obtained hysteresis test results, as Figure 15, the average plastic hinge length of uncorroded
shown in Figure 14, the number of cycles decreased RC column was equal to 14.2 cm and well-agreed
as the corrosion levels increased. Consequently, the according to the defined plastic hinge length by Park and
decrease in the number of cycles resulted in the decrease Paulay34 (i.e., 15 cm). Four damage sides (i.e., red lines in
in the total energy absorption capacities of the corroded Figure 15) of each RC column were measured, and mea-
RC columns. In addition, considering the unilateral sured average lengths of stubs (i.e., blue lines in Figure 15)
(i.e., the test results discussed only for a positive or nega- were subtracted from that of damage side lengths to obtain
tive loading direction) and bilateral (i.e., the structural the average plastic hinge lengths. The determined average
capacity considering both loading directions) failure plastic hinge lengths of corroded RC columns RC2, RC3,
criteria, the maximum yield, and ultimate strengths of RC4, and RC5 were 43.7, 50.2, 20.3, and 43.9 cm, respec-
the corroded RC columns were decreased as the corro- tively. The measured average stub length of uncorroded
sion levels were increased. RC column was equal to zero and 10.4, 11.2, 2.5, and
Among the corroded RC columns, the energy absorp- 8.5 cm for the corroded RC columns from RC2 to RC5.
tion capacity of RC3,20.20%,43.80% was greater than that of
other corroded RC columns with an increased number
of loading steps. This was because of the premature foot- 7.3 | Seismic performance levels of
ing failure at RC3,20.20%,43.80%, which decreased the applied corroded RC columns
axial load during the experimental test. Therefore,
RC3,20.20%,43.80% exhibited a more ductile behavior than the The seismic performance levels of each tested RC col-
other corroded RC columns. One of the main degradations umn were evaluated according to the energy-based
of the structural capacity of corroded RC columns was due method. The main reason for considering the energy-
to the spalling of concrete under the influence of lateral based comparison was the misleading of other methods
loading. Figure 15 shows the crack patterns of the RC col- (i.e., the displacement ductility) for corroded RC col-
umns after the cyclic loading test. Based on the lateral load umns, which was explained in detail by Li et al.35 and
effect, the initial corrosion cracks further expanded and Yalciner and Kumbasaroglu36 As shown in Table 3, the
caused spalling of the concrete cover depth. Degradation yield strength and corresponding displacements of the
in the mechanical properties of both the steel bars and corroded RC columns were reduced with the increase
concrete with a reduction in the bond strength capacity in the corrosion levels. Because the reinforcement bars
 10

TABLE 2 Gravimetric test results of the cut bars

RC2 RC3 RC4 RC5

L1 L7 L1 L7 L1 L7 L1 L7

Length CL Wcr CL Wcr CL Wcr CL Wcr CL Wcr CL Wcr CL Wcr CL Wcr

75 18.65 0.67 26.65 0.58 16.43 0.41 41.32 0.53 18.65 0.50 45.77 0.47 55.55 0.2 59.55 0.24
150 25.32 0.66 33.32 0.81 23.54 0.65 30.65 0.67 33.77 0.29 59.99 0.80 42.66 0.22 53.77 0.25
200 33.61 0.53 36.29 0.9 15.50 0.57 36.96 0.50 29.59 0.47 43.67 0.63 28.92 0.77 63.79 0.45
400 25.80 0.38 34.51 0.75 16.75 0.80 54.27 0.43 22.61 0.28 25.96 0.14 16.25 0.32 44.72 0.27
600 32.83 0.70 21.78 0.37 44.72 0.85 47.24 0.40 12.23 0.42 18.26 0.28 23.95 0.41 33.17 0.26
800 21.61 0.60 34.51 0.26 35.51 0.91 32.16 0.55 67.34 0.32 30.49 0.8 46.23 0.44 61.14 0.25
1000 17.25 0.56 34.51 0.12 64.15 0.97 45.23 0.40 65.33 0.89 51.26 0.54 93.30 0.90 100.0 0.25
1200 21.44 0.30 30.99 0.66 77.22 0.55 18.76 0.53 36.35 0.54 41.54 0.77 40.37 0.31 100.0 0.36
1400 31.49 0.26 28.98 0.98 57.45 0.80 23.45 0.75 45.06 0.19 46.23 0.45 24.46 0.81 94.64 0.31
1600 13.74 0.41 29.65 0.44 43.89 0.70 52.09 0.55 44.72 0.19 52.09 0.94 27.97 0.75 51.09 0.75
Average 24.17 0.51 31.12 0.59 39.52 0.72 38.21 0.53 37.56 0.41 41.53 0.58 39.96 0.51 66.19 0.34

Abbreviations: CL (%), corrosion level at cut bars; RC, reinforced concrete; Wcr (mm), initial corrosion crack width.
CELIK ET AL.
CELIK ET AL. 11

1.4

1.2

1
Crack width (mm)

0.8

0.6

0.4

0.2

0
0 10 20 30 40 50 60 70 80 90 100
Corrosion level (%)

Liu et al. (2017) fc= 46 MPa Yalciner et al. (2019) fc= 9 MPa Yalciner et al. (2019) fc= 27 MPa
Yalciner et al. (2019) fc= 37 MPa Present study fc= 15 MPa

FIGURE 12 Distributions of initial corrosion crack widths

1.4

1.2
ΔAs Predicted/ΔAs Experimental

0.8

0.6

0.4

0.2

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Wcr (mm)

FIGURE 13 Prediction of initial corrosion crack widths

were corroded, the premature yielding of the reinforce-

ment bars and damage to the concrete due to the wider
opened crack widths during loading reduced the dis-
placement at the yield strength (Δy Þ of the corroded RC
columns. Furthermore, the calculated earthquake indica-
tors of the corroded RC columns by using such
displacement methods (e.g., displacement ductility
ratio, μD ¼ ΔΔuy ) cause an increase in the ductility ratios
of corroded RC columns. Therefore, Yalciner and
Kumbasaroglu36 highly recommended the use of an
energy-based method (i.e., μE ¼ EEuy ) to assess the seismic
performance levels of corroded RC structures. The cumu-
lative dissipated energy of each tested RC column was
calculated as follows:

E ¼ Σni¼1 E i , ð3Þ

where E is the sum of all hysteretic loops, and Ei is calcu-

lated from the energy dissipated at each positive and neg-
ative cyclic loop. In Equation (3), the ultimate dissipated
energy of the tested RC columns was determined by the FIGURE 14 Hysteresis loops
12 CELIK ET AL.

TABLE 3 Experimental test results

Positive loading direction Negative loading direction Bilateral failure criteria

Specimen Fy (kN) Fu (kN) Δy (m) Δu (m) Fy (kN) Fu (kN) Δy (m) Δu (m) Fy (kN) Fu (kN) E (kN m)
RC1 48.285 52.162 0.024 0.083 36.289 35.91 0.021 0.098 42.287 46.013 40.22
RC2 44.347 47.739 0.016 0.040 32.412 35.13 0.018 0.042 38.379 41.439 12.83
RC3 39.258 45.195 0.009 0.045 37.925 31.442 0.019 0.045 35.350 39.076 14.01
RC4 38.955 40.530 0.015 0.040 39.318 31.382 0.012 0.041 35.168 37.925 9.66
RC5 37.622 40.590 0.010 0.040 36.531 31.261 0.014 0.045 34.441 35.926 11.13

Abbreviations: E, cumulative energy-dissipation capacity at ultimate load; RC, reinforced concrete.

sum of all loading cycles until the lateral load of the RC strength levels of 9 and 27 MPa. As shown in Figure 17,
columns was reduced to 85% of the maximum load. as the concrete strength level of the model by Yalciner
As shown in Figure 16, as the corrosion levels et al.28 was increased by 37 MPa, premature energy dissi-
increased, the ultimate energy-dissipation capacity of pation occurred for the same drift ratios owing to the
the corroded RC columns decreased. The reduction in increased brittleness of the concrete.
the ultimate energy-dissipation capacities of the cor- Two empirical models were proposed by Yalciner
roded RC columns was significant, and it was 68%, et al.28 to predict the percentage of the energy capacity of
65%, 75%, and 72% for RC2,15.40%,37.80%, RC3,20.20%,43.80%, uncorroded RC columns as a function of plastic deforma-
RC4,27.40%,64.50%, and RC5,28.30%,62.80%, respectively, tion. In that study, the developed models were based on
compared to that of RC1,0.00%,0.00%. One of the main dissipated energies determined based on the roof drift
reasons for the reduction in the energy capacities of ratios with an interval of 0.1% and limited to the
corroded RC columns was further widening initial cor- drift ratio of 2%. Therefore, in the current study, the
rosion crack widths. While the crack widths were authors developed the following empirical models to pre-
increased and concrete cover depth was damaged, dict the percentage of energy capacity of uncorroded RC
absorbed strain in reinforcement bars were reduced columns by considering the current and experimental
with the degradation of bond–slip relationships prior data obtained from Yalciner et al.28
to compressive strain was reached to its ultimate
capacity. E uncorroded ð%Þ ¼ 100  11
δ1:5 for 9 MPa ≤ f c ≤ 27 MPa,
ð4Þ

7.3.1 | Developed energy model for E uncorroded ð%Þ ¼ 100  22
δ1:5 for f c ≥ 27 MPa, ð5Þ
uncorroded RC columns

To determine the seismic performance levels of the where Euncorroded is the remained energy capacity of
tested RC columns, as a first step, the cumulative uncorroded RC columns and δ is the drift ratio. The data
energy-dissipation capacity of each RC column was obtained using the proposed model and the experimental
determined for each step of the hysteric loading, up to data obtained from Yalciner et al.28 were inserted into
85% of the maximum load. Then, the percentages of the the newly proposed models. Figure 18 shows the ratio of
remaining energy capacities in the uncorroded and cor- the predicted values to the experimental values. In
roded RC columns with respect to the cumulative dissi- Figure 18, the newly proposed models, as shown in Equa-
pated energy were determined. Figure 17 shows the tions (4) and (5), captured both experimental test results
percentage of the remaining energy capacity of the well, with the lowest and highest ratios of 0.79 and
uncorroded RC column in the present study 1, respectively.
(fc = 15 MPa) and the experimental data obtained from
the study by Yalciner et al.28 at three concrete strength
levels of 9, 27, and 37 MPa. 7.3.2 | Developed energy model for corroded
In Figure 17, the percentage of energy capacity of the RC columns
uncorroded RC column (fc = 15 MPa) obtained using
the proposed method was similar to that of the energy Figure 19 shows the percentage of remaining energy
capacities obtained by Yalciner et al.28 for concrete capacity of the corroded RC columns in the present study
CELIK ET AL. 13

F I G U R E 1 5 Crack patterns and

measured plastic hinge lengths of
the corroded RC columns. RC,
reinforced concrete

and experimental data obtained from the study by concrete with the higher strength of 15 MPa for
Yalciner et al.28 The corrosion levels obtained from the the uncorroded RC column (i.e., RC1,0.00%,0.00%) had a
study by Yalciner et al.28 are also shown in Figure 18 for higher energy capacity than the concrete with the low
comparison. strength of 9 MPa used by Yalciner et al.28 However, as
An important result was found when the higher shown in Figure 19, compared to the concrete with the
strength concrete used in the present study (f c = 15 MPa) lower strength level (i.e., f c = 9 MPa), the higher strength
and the lower strength concrete (f c = 9 MPa) used by level of the concrete (i.e., f c = 15 MPa) with higher corro-
Yalciner et al.28 were compared in terms of corrosion. As sion levels caused the premature exhaustion of the struc-
shown in Figure 17, for up to 2% of the drift ratio, the ture's energy capacities at the same drift ratios. In other
14 CELIK ET AL.

45 1.8

1.6
40

E (Predicted)/E (experimental)
Eu 1.4

35 1.2

1
30
0.8

0.6
Emax
Energy (kN.m-Joule)

25
0.4

20 0.2

0
15 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Eu Drift ratio (%)
Eu
10 Eu Yalciner et al. (2019), fc= 27 MPa Yalciner et al. (2019), fc= 9 MPa
Eu Present study, fc= 15 MPa Yalciner et al. (2019), fc= 37 MPa
Emax
Emax Emax
5 Emax
Ey F I G U R E 1 8 Validation of the proposed energy models for
Ey
Ey Ey Ey uncorroded RC columns. RC, reinforced concrete
0
RC1 RC2 RC3 RC4 RC5
CLL= 0.00% CLL= 15.41% CLL= 20.20% CLL= 27.37% CLL= 28.34%
CLS= 0.00% CLS= 37.83% CLS= 43.84% CLS= 64.47% CLS= 62.80%

Specimen Equations (6) and (7) captured both experimental test

results well with regression coefficients of 87% and 93%,
F I G U R E 1 6 Energy capacities of the corroded RC columns: Ey
respectively. Note that the proposed models in Equations
represents the cumulative energy capacity at the yield load, Emax
(6) and (7) were based on the actual corrosion levels
represents the cumulative energy capacity at the maximum load,
because of the described phenomena regarding the
and Eu represents the cumulative energy capacity at the ultimate
load. RC, reinforced concrete
nonlinear increase in the initial corrosion crack widths
with an increase in the corrosion levels. Therefore, the
following models are suggested to be used for destructive
100
methods at higher corrosion levels.
Percentage of remain enegy capacity (%)

90

E corroded ð%Þ ¼ 110  26
δ  CLL for 9 MPa ≤ f c ≤ 27 MPa:

80

70

60
ð6Þ
50

40 Ecorroded ð%Þ ¼ 117  43
δ  3
CLL for f c ≥ 27 MPa: ð7Þ

30

20

10
The proposed model in the present study considers
0
both not buckling (i.e., Yalciner et al.28) and buckling
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
of reinforcement bars (i.e., present study) for given
Drift ratio (%)
Present study, fc= 15 MPa, CLL= 0.00% Yalciner et al. (2020), fc= 9 MPa, CLL= 0.00% lower and higher corrosion levels, respectively. As it is
Yalciner et al. (2020), fc= 27 MPa CLL= 0.00% Yalciner et. al (2020), fc= 37 MPa, CLL= 0.00% well known, even the buckling of uncorroded rein-
forcement bars is based on bar diameters, applied axial
F I G U R E 1 7 Percentage of energy capacities for the
load ratios, and particularly confinement effects by
uncorroded RC columns. RC, reinforced concrete
stirrups. Owing to the unknowns and difficulties
involved in corrosion, different studies are also
words, the comparison between low-strength concrete required to exactly define the limit state of corrosion
and that with almost two times greater strength levels to initiate the buckling of reinforcement bars at
(i.e., comparison between 15- and 9-MPa concrete) with different intervals of stirrups, bar diameters, and axial
higher corrosion levels indicated that the effect of corro- load ratios for the same amount of corrosion level. The
sion levels in reinforcement bars is more important than comparison of the current conditions of the stirrups
the concrete strength levels for structural performance. with those used in the previous experimental stud-
The following empirical model was used to predict the ies36,37 showed that at corrosion levels of more than
percentage of the remaining energy capacity of a cor- 15% at the longitudinal bars, the stirrups may rupture
roded RC column as a function of the plastic deformation owing to the lower initial mass. Therefore, corrosion
by considering the current and obtained experimental levels must be higher than 15% at the longitudinal bars,
data from Yalciner et al.28 The newly proposed models in and the reduction in confinement effects by stirrups
CELIK ET AL. 15

100

90 Define Desired Target Drift Ratio (δtarget )

80
Percentage of remain enegy capacity (%)

70

60 Define Corrosion Level (CLL)

50

40

30 Predict Remained Energy for Corroded RC Columns (Ec) by

using Eqs. (6) and (7)
20

10

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 Insert (Ec) Into Eqs. (4) - (5) and Determine New Drift Ratio (δnew)
Drift ratio (%)
Yalciner et al. (2019), A7, fc= 9 MPa, CLL= 2.86% Yalciner et al. (2019), B7, fc= 27 MPa, CLL= 1.79%
Yalciner et al. (2019), A8, fc= 9 MPa, CLL= 4.21% Yalciner et al. (2019), B8, fc= 27 MPa, CLL= 2.45%
Yalciner et al. (2019), A9, fc= 9 MPa, CLL= 4.33% Yalciner et al. (2019), B9, fc= 27 MPa, CLL= 3.9%
Yalciner et al. (2019), A10, fc= 9 MPa, CLL= 4.82% Yalciner et al. (2019), B10, fc= 27 MPa, CLL= 5.91% F I G U R E 2 1 Simplified procedure to predict the new target
Present study RC2, fc= 15 MPa, CLL= 15.4% Yalciner et al. (2019), C7, fc= 37 MPa, CLL= 1.83%
Present study RC3, fc= 15 MPa, CLL= 20.2% Yalciner et al. (2019), C8, fc= 37 MPa, CLL= 3.18% drift ratio of the corroded RC columns. RC, reinforced concrete
Present study RC4, fc= 15 MPa, CLL= 27.4% Yalciner et al. (2019), C9, fc= 37 MPa, CLL= 3.78%
Present study RC5, fc= 15 MPa, CLL= 28.3% Yalciner et al. (2019), C10, fc= 37 MPa, CLL= 6.02%

F I G U R E 1 9 Percentage of energy capacities for the corroded

RC columns. RC, reinforced concrete data (i.e., corrosion levels) found in the literature were
input into Equations (6) and (7) to predict the ultimate
drift ratio. As only the ultimate energy capacities and
1.8
1.7
corresponding drift ratios of previous studies37–42 were
1.6 available, it was assumed that the remaining energy
1.5
1.4 capacity was equal to zero when the drift ratio reached
1.3
85% of the corresponding maximum load. Figure 20
δu (Predicted)/ δu (Experimental)

1.2
1.1
1
depicts the ratios of the predicted values to experimental
0.9
0.8
values by using the present model and the model devel-
0.7 oped by Goksu et al.41
0.6
0.5 Figure 20 shows that the main difference between the
0.4
0.3 model by Goksu et al.41 and the proposed models is that
0.2
0.1 the model by Goksu et al.41 predicts only the ultimate dis-
0
0 5 10 15 20 25 30 35 40 45 50 55 60
placement for a given corrosion level. In contrast, the
Corrosion level (%) present proposed models predict any drift ratio for a
Data form Wang et al. (2003), n= 0.25 inserted into present model Data form Wang et al. (2003), n= 0.34 inserted into present model
given corrosion level based on the considered energy
Data form Wang et al. (2003), n= 0.50 inserted into present model Data form Ma et al. (2012), n= 0.15 inserted into present model
Data form Ma et al. (2012), n= 0.25 inserted into present model Data form Ma et al. (2012), n= 0.40 inserted into present model capacity. As shown in Figure 20, although the model by
Goksu et al.41 was based on ultimate displacement and
Data form Ma et al. (2012), n= 0.75 inserted into present model Data form Ma et al. (2012), n= 0.90 inserted into present model
Data form Meda et al. (2014), n= 0.22 inserted into present model Data form Goksu et al. (2016), n= 0.18 inserted into present model
Data form Yang et al. (2016), n= 0.18 inserted into present model
Data form Ma et al. (2012), n= 0.15 inserted into Goksu et al. (2016)
Data form Zheng et al. (2020), n=0.30 inserted into present model
Data form Ma et al. (2012), n= 0.25 inserted into Goksu et al. (2016)
obtained corrosion levels were within the limits of corro-
Data form Ma et al. (2012), n= 0.40 inserted into Goksu et al. (2016) Data form Ma et al. (2012), n= 0.75 inserted into Goksu et al. (2016) sion levels in other studies, the developed model by Goksu
Data form Ma et al. (2012), n= 0.90 inserted into Goksu et al. (2016) Data form Yang et al. (2016), n= 0.18 inserted into Goksu et al. (2016)
Data form current study, n= 0.40 inserted into Goksu et al. (2016) et al.41 either underestimated or overestimated the previ-
ous and current experimental test results. This is owing to
FIGURE 20 Validation of the proposed model the single axial load ratio (i.e., n = 0.18) for the tested RC
columns by Goksu et al.41 As shown in Figure 20, a single
axial load ratio was used by Goksu et al.41 to accurately
and buckling behavior are needed to be considered for predict the previous experimental studies if the axial load
structural modeling of corroded RC columns. ratios were less than 0.25. However, as the axial load ratios
Several models were proposed to predict the ultimate increased, the performance of their model was reduced.
drift ratios of corroded RC columns as a function of the Figure 20 shows that the ratio of the predicted to experi-
corrosion levels and independent of energy absorption mental values using the approach of Goksu et al.41 varied
capacities. To validate the newly proposed empirical from 0.6 to 1.7. Moreover, when the empirical model by
models (i.e., Equations (6) and (7)), experimental test Goksu et al.41 was revisited, it was found that the ratios of
16 CELIK ET AL.

the predicted values to their corresponding experimental ratio based on Equation (4) is obtained as 2.51%, which
values were low. Although the cyclic loading programs by is the same as that of the collapse performance level
Goksu et al.41 and Yang et al.37 used axial load ratios, sec- based on Vision 2000.43 When the predicted results
tional properties of concrete, reinforcement layouts, and from the two examples are compared with experimen-
concrete strength levels were different than those used in tal test results, the ratio of the predicted value to exper-
the present study, which considered four concrete strength imental data was obtained as 0.93 and 0.94 for the first
levels in the present model and provided superior predic- and second examples, respectively. Note that for the
tions from previous experimental test data. The lowest cal- considered 16-mm-diameter reinforcement bar, the
culated ratio was obtained for the experimental test results model proposed by Yalciner et al.28 predicts a corrosion
by Yang et al.37 and was equal to 0.60 at a corrosion level level of 15% for the measured 0.75-mm corrosion crack
of 16%. This lower prediction of the data could be because width by using the nondestructive method within the
they used a single cycle at each drift ratio, and thus defined limit states of 20% of the corrosion level.
obtained more dissipation energy capacities for corroded
RC columns. When Ma et al.38 increased the number of
cycles to three, the calculated ratios of the predicted values 8 | CONCLUSION
to experimental values ranged between 0.70 and 1 for dif-
ferent applied axial load ratios. The comparison of the pre- The effect of corrosion on the structural behavior of RC
vious model by Goksu et al.41 to predict the ultimate columns was tested to predict the remaining energy-
displacement of corroded RC columns showed the test dissipation capacities and seismic performance levels of
results demonstrated that the applied axial load ratios and RC columns as a function of corrosion levels. The follow-
loading programs affect the performance of that model. ing conclusions can be drawn based on the experimental
The new seismic performance levels of the corroded test results.
RC columns can be determined according to the flow-
chart in Figure 21, which can be explained as follows. • Considering the bilateral failure criteria, the yield and
First, the target drift ratio based on the seismic perfor- ultimate load carrying capacities and ductility ratios of
mance level of the RC column and the corrosion level the corroded RC columns were significantly reduced as
must be defined. Then, the remaining energy capacities the corrosion levels increased.
of the corroded RC columns (Ecorroded) can be predicted • Among the earthquake indicators of structural
according to Equations (6) and (7). These predicted behavior, the present study agreed well with the
energy capacities of the corroded RC columns can facili- results obtained by Li et al.35 and Yalciner and
tate in predicting the new seismic performance levels of Kumbasaroglu,36 in which the energy-based
the corroded RC columns based on the drift ratio ðδnew Þ methods provided superior results to define the deg-
by substituting the predicted Ecorroded into Equations (4) radation of corroded RC columns. There were strong
and (5). relationships between the drift ratios and energy-
The newly developed models in the current study dissipation capacities of corroded RC columns that
were validated through the following example and were signed the energy-based methods for corroded
were compared with the experimental test results. The RC members rather than displacement methods par-
target seismic performance was assumed to be opera- ticularly to determine the ductility ratios of corroded
tional, in which the associated roof drift ratio is less RC members.
than 0.5% according to the defined performance levels • Once the crack width of the concrete reaches its ulti-
by Vision 2000.43 If the average corrosion level in a cor- mate value, the corrosion level increases for the
roded RC column is assumed to be 15% with the con- remaining crack width of the concrete during the
crete strength level of 15 MPa, the remaining propagation period of corrosion. Therefore, it is chal-
percentage of energy capacity of a corroded RC based lenging to predict the exact corrosion levels of rein-
on Equation (6) is determined as 82%. When the forcement bars as a function of the measured initial
predicted remaining energy capacity of the corroded corrosion crack widths. Although crack width
RC column (Ecorroded) is inserted into Equation (4) by remains constant and corrosion levels increase,
working backwards, the new drift ratio is obtained as given distribution of initial corrosion crack widths
1.38%. Thus, the new seismic performance level of the obtained from different studies indicated that
corroded RC column yielded as life safety performance approximately predicted corrosion levels based on
level based on Vision 2000.43 If the same example is measured crack widths provide an idea to determine
considered for the target roof drift ratio of less than the corrosion levels for RC members in term of prac-
1.5% (i.e., life safety), the newly predicted target drift tical engineering.
CELIK ET AL. 17

• Comprehensive experimental studies are required to structures. Corros Sci. 2011;53:1337–47. https://doi.org/10.
accurately predict the corrosion levels for the mea- 1016/j.corsci.2010.12.026
sured initial corrosion crack widths. For doing this, dif- 6. Yalciner H, Eren O, Sensoy S. An experimental study on the
bond strength between reinforcement bars and concrete as a
ferent axial load ratios, steel ratios, cover-to-bar-
function of concrete cover, strength and corrosion level. Cem
diameters, and distance between bar diameters are Concr Res. 2012;42(5):643–55. https://doi.org/10.1016/j.
needed to be considered in such test programs. Further cemconres.2012.01.003
developed crack width models then provide inputs on 7. Jiang C, Wu YF, Dai MJ. Degradation of steel-to-concrete bond
the proposed model in the present study for predicting due to corrosion. Construct Build Mater. 2018;158:1073–80.
seismic performance levels of corroded RC columns. https://doi.org/10.1016/j.conbuildmat.2017.09.142
• The degradation in the plastic hinge length of the RC 8. Zhou H, Lu J, Xv X, Dong B, Xing F. Effects of stirrup corro-
sion on bond–slip performance of reinforcing steel in concrete:
columns due to corrosion must be considered in future
an experimental study. Construct Build Mater. 2015;93:257–66.
studies of corroded RC columns in terms of modeling
https://doi.org/10.1016/j.conbuildmat.2015.05.122
the plastic hinge properties due to corrosion and 9. Choe DE, Gardoni P, Rosowsky D, Haukaas T. Probabilistic
strengthening, where strengthening is mainly applied capacity models and seismic fragility estimates for RC col-
at the theoretically determined plastic hinge length of umns subject to corrosion. Reliab Eng Syst Saf. 2008;93:
RC columns. 383–93.
• The proposed models may provide guidance for rapid 10. Choe DE, Gardoni P, Rosowsky D, Haukaas T. Seismic fragility
decision-making for corroded RC columns and may estimates for reinforced concrete bridges subject to corrosion.
Struct Saf. 2009;31:275–83. https://doi.org/10.1016/j.strusafe.
enhance the seismic vulnerability assessment in
2008.10.001
earthquake-prone regions. 11. Xu JG, Wu G, Feng DC, Cotsovos DM, Lu Y. Seismic fragility
analysis of shear-critical concrete columns considering corro-
A C K N O WL E D G M E N T sion induced deterioration effects. Soil Dyn Earthquake Eng.
This study was funded by the Scientific Research Com- 2020;134:106165. https://doi.org/10.1016/j.soildyn.2020.106165
mission of Erzincan Binali Yıldırım University under 12. Stewart MG, Al-Harthy A. Pitting corrosion and structural reli-
grant number BAP-FBA-2020-659. ability of corroding RC structures: experimental data and prob-
abilistic analysis. Reliab Eng Syst Saf. 2008;93:373–82.
13. Rodriguez J, Ortega LM, Casal J. Load carrying capacity of concrete
DATA AVAILABILITY STATEMENT
structures with corroded reinforcement. Construct Build Mater.
Data available on request from the authors. 1997;11:239–48. https://doi.org/10.1016/S0950-0618(97)00043-3
14. Alghamdi SA, Ahmad S. Service life prediction of RC structures
ORCID based on correlation between electrochemical and gravimetric
Alper Celik https://orcid.org/0000-0003-3816-680X reinforcement corrosion rates. Cem Concr Compos. 2014;47:
64–8.
Hakan Yalciner https://orcid.org/0000-0002-7289-3384
15. Zhou H, Xu Y, Peng Y, Liang X, Li D, Xing F. Partially cor-
Atila Kumbasaroglu https://orcid.org/0000-0002-6338- roded reinforced concrete piers under axial compression and
4553 cyclic loading: an experimental study. Eng Struct. 2020;203:
_
Ahmet Ihsan Turan https://orcid.org/0000-0003-4865- 109880. https://doi.org/10.1016/j.engstruct.2019.109880
6490 16. Meda A, Mostosi S, Rinaldi Z, Riva P. Corroded RC columns
repair and strengthening with high performance fiber
R EF E RE N C E S reinforced concrete jacket. Mater Struct. 2016;49:1967–78.
17. Chen Y, Yu J, Leung CK. Use of high strength strain-hardening
1. Imperatore S, Rinaldi Z, Drago C. Degradation relationships
cementitious composites for flexural repair of concrete struc-
for the mechanical properties of corroded steel rebars. Con-
tures with significant steel corrosion. Construct Build Mater.
struct Build Mater. 2017;148:219–30. https://doi.org/10.1016/j.
conbuildmat.2017.04.209 2018;167:325–37. https://doi.org/10.1016/j.conbuildmat.2018.
2. Cairns J, Plizzari GA, Du Y, Law DW, Franzoni C. Mechanical 02.009
properties of corrosion-damaged reinforcement. ACI Mater J. 18. Chung L, Cho SH, Kim JHJ, Yi ST. Correction factor suggestion
2005;102:256. for ACI development length provisions based on flexural testing of
3. El Maaddawy T, Soudki K, Topper T. Long-term performance RC slabs with various levels of corroded reinforcing bars. Eng
of corrosion-damaged reinforced concrete beams. ACI Struct J. Struct. 2004;26:1013–26. https://doi.org/10.1016/j.engstruct.2004.
2005;102:649. 01.008
4. Liu, Y. (1996). Modeling the time-to corrosion cracking of the 19. Zhang Y, Bicici E, Sezen H, Zheng S. Reinforcement slip model
cover concrete in chloride contaminated reinforced concrete considering corrosion effects. Construct Build Mater. 2020;235:
structures (Doctoral dissertation, Virginia Tech). https:// 117348. https://doi.org/10.1016/j.conbuildmat.2019.117348
vtechworks.lib.vt.edu/handle/10919/30541 20. Yalciner H, Sensoy S, Eren O. Seismic performance assessment
5. Lu C, Jin W, Liu R. Reinforcement corrosion-induced cover of a corroded 50-year-old reinforced concrete building. J Struct
cracking and its time prediction for reinforced concrete Eng. 2015;141(12):05015001.
18 CELIK ET AL.

21. Berto L, Vitaliani R, Saetta A, Simioni P. Seismic assessment of concrete columns with different maximum amounts of corro-
existing RC structures affected by degradation phenomena. sion of rebar. Construct Build Mater. 2016;121:319–27. https://
Struct Saf. 2009;31:284–97. https://doi.org/10.1016/j.strusafe. doi.org/10.1016/j.conbuildmat.2016.06.002
2008.09.006 38. Ma Y, Che Y, Gong J. Behavior of corrosion damaged circular
22. Bažant ZP. Physical model for steel corrosion in concrete sea reinforced concrete columns under cyclic loading. Construct
structures – theory. J Struct Div. 1979;105(ST6):1137–53. Build Mater. 2012;29:548–56. https://doi.org/10.1016/j.
23. Vidal T, Castel A, François R. Analyzing crack width to predict conbuildmat.2011.11.002
corrosion in reinforced concrete. Cem Concr Res. 2004;34:165– 39. Zheng H, Zheng S, Zhang Y, Cai Y, Ming M, Zhou J. Experi-
74. https://doi.org/10.1016/S0008-8846(03)00246-1 mental investigation on seismic behaviours of reinforced con-
24. Zhang R, Castel A, François R. Concrete cover cracking with crete columns under simulated acid rain environment. Adv
reinforcement corrosion of RC beam during chloride-induced Civil Eng. 2020;2020:15. https://doi.org/10.1155/2020/3826062
corrosion process. Cem Concr Res. 2010;40:415–25. https://doi. 40. Wang X. Research of seismic performance and hysteretic mode
org/10.1016/j.cemconres.2009.09.026 of corroded reinforced concrete members. J Xi'an Univ Archit
25. Khan I, François R, Castel A. Prediction of reinforcement cor- Technol. 2003;33:15–21.
rosion using corrosion induced cracks width in corroded 41. Goksu C, Ilki A. Seismic behavior of reinforced concrete col-
reinforced concrete beams. Cem Concr Res. 2014;56:84–96. umns with corroded deformed reinforcing bars. ACI Struct J.
https://doi.org/10.1016/j.cemconres.2013.11.006 2016;113. https://www.concrete.org/publications/international
26. Malumbela G, Moyo P, Alexander M. Behaviour of RC beams concreteabstractsportal/m/details/id/51689030.
corroded under sustained service loads. Construct Build Mater. 42. Meda A, Mostosi S, Rinaldi Z, Riva P. Experimental evaluation
2009;23:3346–51. https://doi.org/10.1016/j.conbuildmat.2009. of the corrosion influence on the cyclic behaviour of RC col-
06.005 umns. Eng Struct. 2014;76:112–23. https://doi.org/10.1016/j.
27. El-Sayed AK. Shear capacity assessment of reinforced concrete engstruct.2017.10.020
beams with corroded stirrups. Construct Build Mater. 2017;134: 43. SEAOC. Vision 2000, Performance based seismic engineer-
176–84. https://doi.org/10.1016/j.conbuildmat.2016.12.118 ing of buildings, vols. I and II: conceptual framework. Sacra-
28. Yalciner H, Kumbasaroglu A, Karimi A. Prediction of seismic mento (CA): Structural Engineers Association of California,
performance levels of corroded reinforced concrete columns as 1995.
a function of crack width. Adv Civil Eng Mater. 2019;8:376–97.
https://doi.org/10.1520/ACEM20190035
29. Vecchio FJ, Collins MP. The modified compression-field theory AUTHOR BIOGRAPHIES
for reinforced concrete elements subjected to shear. ACI J.
1986;83:219–31. Alper Celik, Department of Civil
30. Zhou H, Chen S, Du Y, Lin Z, Liang X, Liu J, et al. Field test of
Engineering, Erzincan Binali
a reinforced concrete bridge under marine environmental cor-
rosion. Eng Failure Anal. 2020;115:104669. https://doi.org/10.
Yıldırım University, Erzincan, Tur-
1016/j.engfailanal.2020.104669 key. Email: alper.celik@erzincan.
31. FEMA (Federal Emergency Management Agency), (2007). edu.tr.
“Interim testing protocols for determining the seismic perfor-
mance characteristics of structural and nonstructural compo-
nents” FEMA-461, Washington, DC, 1–113.
32. Liu X, Jiang H, He L. Experimental investigation on seismic
performance of corroded reinforced concrete moment-resisting Hakan Yalciner, Department of
frames. Eng Struct. 2017;153:639–52. https://doi.org/10.1016/j. Civil Engineering, Erzincan Binali
engstruct.2017.10.034 Yıldırım University, Erzincan, Tur-
33. Sezen H, Moehle JP. Shear strength model for lightly
key. Email: Erzincan Binali
reinforced concrete columns. J Struct Eng. 2004;130:1692–703.
https://doi.org/10.1061/(ASCE)0733-9445(2004)130:11(1692)
Yildirim Universitesi.
34. Park R, Paulay T. Reinforced concrete structures. 1st ed. New
Jersey: John Wiley and Sons; 1975.p. 36–45.
35. Li D, Wei R, Xing F, Sui L, Zhou Y, Wang W. Influence of
non-uniform corrosion of steel bars on the seismic behavior
of reinforced concrete columns. Construct Build Mater. Atila Kumbasaroglu, Department
2018;167:20–32. https://doi.org/10.1016/j.conbuildmat.2018. of Civil Engineering, Erzincan
01.149 Binali Yıldırım University, Erzin-
36. Yalciner H, Kumbasaroglu A. Experimental evaluation and can, Turkey.
modeling of corroded reinforced concrete columns. ACI Struct Email: akumbasaroglu@erzincan.
J. 2020;117:20–32. https://doi.org/10.1016/j.conbuildmat.2018.
edu.tr.
01.149
37. Yang SY, Song XB, Jia HX, Chen X, Liu XL. Experimental
research on hysteretic behaviors of corroded reinforced
CELIK ET AL. 19

_
Ahmet Ihsan Turan, Department How to cite this article: Celik A, Yalciner H,
of Civil Engineering, Erzincan Kumbasaroglu A, Turan AI. _ An experimental study
Binali Yıldırım University, Erzin- on seismic performance levels of highly corroded
can, Turkey. Email: ahmet. reinforced concrete columns. Structural Concrete.
turan@erzincan.edu.tr. 2021;1–19. https://doi.org/10.1002/suco.202100065