1

Experimental analysis of the effects of

chloride-induced reinforcement corrosion

in smooth and ribbed rebars on the bond between steel and concrete

Charlotte VAN STEEN1, Michiel YSENBAARDT1, Lucie VANDEWALLE1,

Martine WEVERS2, Els VERSTRYNGE1

1Department of Civil Engineering, KU Leuven, Leuven, Belgium,

charlotte.vansteen@kuleuven.be, michiel.ysenbaardt@student.kuleuven.be,

lucie.vandewalle@kuleuven.be, els.verstrynge@kuleuven.be 2Department of Materials Engineering, KU Leuven, Leuven, Belgium,

martine.wevers@kuleuven.be

Abstract

Corrosion of the reinforcement is one of the most common and most expensive deterioration

mechanism in reinforced concrete (RC) structures. While a lot of research has been performed on the

design of new RC structures, estimating the remaining capacity of existing structures with corroded

reinforcement still requires dedicated attention. One of the important damage modes due to corrosion

is the deterioration of bond between steel and concrete. In this paper, the effect of reinforcement

corrosion on the bond strength was investigated for smooth and ribbed rebars. Bond behaviour was

studied at different target corrosion levels: 0%, 1.5%, 5% and 10% steel mass loss. Reinforcement

bars embedded in concrete were subjected to accelerated corrosion by imposing a constant direct

current while the specimens were partially immersed in a 5% NaCl solution. During the corrosion

process, one of the specimens of each rebar type and every corrosion level was monitored with the

acoustic emission (AE) technique. At target corrosion levels, pull out tests were carried out to study

the bond capacity. Results show that AE effectively detects effects of rebar corrosion. Bond-slip

curves of specimens with smooth and ribbed reinforcement were compared and the effects of different

corrosion levels were analysed. The results were also related to results obtained from the literature.

Keywords

Acoustic emission; bond; reinforced concrete; reinforcement corrosion, chlorides

Introduction

Corrosion of the reinforcement is one of the major deterioration problems in existing

reinforced concrete (RC) structures causing considerable costs for maintenance and repair.

While tools for modelling and design of new RC structures are mature, efficient management

schemes and accurate models for the assessment of existing structures are lacking, leading to

higher repair costs and reduced structural reliability [1]. Dedicated research to quantify the

effects of the degradation processes on RC member’s structural capacity and durability

becomes challenging and timely but is necessary in order to develop these assessment

schemes for existing RC structures. Reinforcement corrosion causes a number of interacting

damage modes: overall/local section reduction of the rebar, tensile stresses in the concrete due

to the expansive nature of corrosion products causing concrete cracking and spalling, a

reduction of ductility of the rebar and degradation of bond within the concrete-reinforcement

interface. Bond deterioration is one of the important damage modes caused by reinforcement

corrosion. Studies have shown that for certain structural elements the loss of bond strength for

unconfined reinforcement is more critical than the loss of cross section [2]. The loss of bond

can go up to 80% while the reduction of the cross section is rather low [3].

2

Bond is the interaction between concrete and steel which results in the composite action

between the two materials. It allows longitudinal forces to be transferred from the

reinforcement to the surrounding concrete. Three aspects lead to bond strength: chemical

adhesion, friction and wedging action. The last one only holds for ribbed rebars. A relative

displacement, which is called slip, between steel and concrete can occur when there is a

difference between steel strains and concrete strains. Extensive research has been performed

by several authors on the bond behaviour of corroded reinforcement, including samples with

smooth or ribbed rebars and with or without lateral confinement (stirrups). The summary of a

literature study on the bond behaviour of corroded specimens is presented below. The bond

strength depends on several parameters such as rebar type (smooth or ribbed) [3], diameter of

the rebar [4, 5], the water-cement ratio [6, 7], the presence of lateral confinement [3, 8] and

the thickness of the concrete cover [5, 6]. Some authors also adapted the well-known

concentric pull out test to a cantilever bond test [9] or corroded the rebars before casting them

[10]. The different experimental results reported in literature for smooth and ribbed rebars

were compared in relation to the obtained corrosion level and are shown in figure 1 and 2. As

different concrete mixtures and sample sizes were used for each experimental campaign, the

normalized bond strength was calculated in order to compare the results. The normalized

bond strength is defined as the bond strength of each sample divided by the bond strength of

the non-corroded sample. Results show a large scatter. It can be concluded from these figures

that a small amount of corrosion (up till 2% mass loss) has a positive influence on the bond

strength. For higher corrosion levels, the bond strength decreases. Figure 2 also shows the

lack of studies on smooth rebars. When developing an assessment tool for existing structures,

more research is needed on this type of rebars as many of the existing structures dating from

the sixties and seventies are constructed with this kind of reinforcement.

Figure 1: Normalized bond strength versus

corrosion level for samples with unconfined

and confined ribbed rebars

Figure 2: Normalized bond strength versus

corrosion level for samples with unconfined

and confined smooth rebars

To allow the development of efficient performance-based approaches for existing structures a

shift towards quantifying the damage level and residual structural capacity is needed. On-site

quantification of structural reliability requires advanced non-destructive techniques (NDT) for

material and damage characterization and quantification. Electrochemical techniques are

widely used to monitor corrosion on-site. Unfortunately, they are dependent on the climatic

3

condition and they might lack to provide precise information. This increases the demand for

calibration with other techniques. A very promising technique to capture not only the

corrosion process itself, but also the initiation and progress of concrete cracking, is the

acoustic emission (AE) technique. Local stress redistributions in the material, such as

cracking, emit high frequency elastic waves that can be recorded by AE sensors on the

concrete surface [11]. The technique has proven its efficiency for localizing damage in metal

structures, such as pressure vessels and pipelines, were direct contact with the metal is

possible [12, 13]. AE monitoring has been successfully applied during rebar corrosion in

concrete and testing of corroded RC samples and components [14-17]. The technique has high

potential for corrosion monitoring in RC structures but still poses some challenges for on-site

use such as filtering in noisy environments, path-dependent distortion of signals by concrete

cracking, linking AE data with structural performance of components and structures and the

development of on-site test schemes. To account for this, dedicated experimental work is

necessary to upscale this technique to the structural level.

This paper will focus on bond behavior of samples with corroded smooth and ribbed

reinforcement bars. The corrosion process was accelerated in the lab and acoustic emission

monitoring was performed continuously. The paper will firstly describe the experimental

setup, secondly discuss the results and finally draw some conclusions.

Experimental approach

Materials and specimen preparation

Two types of specimens were compared during the experimental program: specimens

reinforced with a smooth rebar and specimens reinforced with a ribbed rebar. Four different

corrosion levels were targeted for each rebar type namely 0%, 1.5%, 5% and 10% mass loss.

For every corrosion level, three replicate specimens were casted. In total, 24 specimens were

tested.

Both rebar types had a diameter of 12 mm. The specifications of the steel are given in table 1.

The embedded length was taken 100 mm for all specimens. PVC tubes were used to avoid

bonding of the remaining reinforcement. The first experiments have shown that the embedded

length for the lowest corrosion levels of the samples with ribbed rebars was too large, and

failure of the rebar through rupture was observed at the thread that connected the rebar with

the tensile testing machine. Therefore, for these lowest corrosion rates (0% and 1.5%), the

bond length of the samples with ribbed rebars was reduced to 60 mm (5 times the rebar

diameter). The geometry of the specimens is illustrated in figure 3. Table 3 gives an overview

of the specifications of the different tested specimens.

Table 1: Mechanical properties of reinforcement

Material Yield stress (N/mm2) Tensile strength (N/mm2) Elastic modulus (N/mm2)

Smooth rebar

BE220

358 466 200 040

Ribbed rebar

BE500S/B

454 518 199 150

The rebar was placed horizontally in the center of the wooden mould (150 x 150 x 260 mm)

and was protruding from both sides in order to connect the power supply for the accelerated

corrosion process afterwards. The concrete composition is shown in table 2. For every batch,

also 6 concrete cubes were casted to determine the compressive strength at an age of 28 days

and at the age of pull out testing. Three prisms were made to know the flexural strength at an

age of 28 days. The average compressive strength for the cubes at 28 days was 44.93 MPa

and the average flexural strength was 3.47 MPa.

4

After curing for 28 days in a curing room (20°C, 95% RH), the specimens were fully

immersed in a 5% NaCl solution for three days. Afterwards, the specimens were placed in the

accelerated corrosion setup (figure 4) in a climatised room (20°C, 60% RH). The accelerated

corrosion process and acoustic emission monitoring started at an age of 31 days.

Table 2: Concrete composition [weight %]

CEM I 42.5 N Sand 0/5 Aggregates 4/14 Water Chlorides W/C

14.52 25.72 52.68 6.8 0.29 0.46

Table 3: Sample specifications

Sample name Target

corrosion level

[%]

Reinforcement

type (D12)

Bond length

[mm]

# samples # samples

monitored

with AE

CR0-S1 – CR0-S3 0 Smooth 100 3 0

CR0-R1 – CR0-R3 0 Ribbed 100 3 0

CR0-R4 – CR0-R6 0 Ribbed 60 3 0

CR1-S1 – CR1-S3 1.5 Smooth 100 3 1

CR1-R1 – CR1-R3 1.5 Ribbed 100 3 1

CR1-R4 – CR1-R6 1.5 Ribbed 60 3 1

CR2-S1 – CR2-S3 5 Smooth 100 3 1

CR2-R1 – CR2-R3 5 Ribbed 100 3 1

CR3-S1 – CR3-S3 10 Smooth 100 3 1

CR3-R1 – CR3-R3 10 Ribbed 100 3 1

Figure 3: Geometry of the specimens ; lateral section of specimens with smooth rebars (all

corrosion levels) and ribbed rebars (5 and 10% mass loss) (top), lateral section of

specimens with ribbed rebars (0 and 1.5% mass loss) (below), cross section all specimens

(right) ; all dimensions in mm

Accelerated corrosion

The corrosion process was accelerated by an imposed direct current. A constant current

density of 100 μA/cm2 was chosen because this was reported to be the maximum current

density occurring in natural conditions [18]. Different corrosion products will be formed

when using a higher current density. Also the internal pressure will be build up too quickly

since the corrosion products will have no time to fill the pores [19]. The rebar was connected

5

to the positive side of the DC regulator and acts as an anode. A stainless steel plate was

connected to the negative side and acts as a cathode. During exposure, the specimen was

partially immersed in a 5% NaCl solution to ensure electrical connectivity and chloride

ingress. The duration of the accelerated corrosion process was estimated with Faraday’s law

(Eq. 1). The actual amount of corrosion was measured afterwards by the weight loss of the

rebar.

∆m=(t x M x I)/(Z x F) (Eq. 1)

with ∆m is mass loss [g],

t is time [s]

I is current [A],

Z is valence with a value of 2

F is Faraday’s constant with a value of 96487 C/mol

Figure 4: setup to accelerate the corrosion process and perform acoustic emission

monitoring

Acoustic Emission (AE) technique

During the accelerated corrosion process, passive AE monitoring was performed continuously

on one of the three specimens of the same type. Two piezoelectric sensors with a flat response

between 100-400 kHz were attached on top of the specimen’s surface with hot melt glue.

Their flat frequency response allows a more reliable acquisition of different sources closer to

their original frequency. The center-to-center distance between the AE sensors was 100 mm

for the specimens with an embedded length of 100 mm and 60 mm for the specimens with an

embedded length of 60 mm. The sensors were connected to pre-amplifiers with 34 dB gain.

These pre-amplifiers were connected to a Vallen AMSY-6 acquisition system with six AE

channels. AE parameters and waveforms were then stored on a PC and the Vallen

VisualAETM software was used to visualize the data in real time. Matlab was applied for

further processing. The threshold was set to 40 dB to avoid falls detections due to background

noise.

Pull Out tests

The pull out tests were carried out on a MFL testing machine with a capacity of 2500 kN. A

specially designed loading frame was used. The rebar was pulled down while on top of the

specimen, an LVDT was attached to the free rebar end to measure the slip. The specimens

were tested in a displacement-controlled regime with a loading speed of 0.3 mm/min for the

entire test. The setup is shown in figure 5.

6

Figure 5: Test setup pull out

Results and discussion

Corrosion process

During the accelerated corrosion process, non-destructive techniques such as crack

measurements and acoustic emission monitoring were performed. Acoustic emission

monitoring was performed continuously. Every week, the corrosion process was interrupted

to measure the cracks using a crack meter with an accuracy of 0.02 mm. In addition, every

side of the specimen was photographed.

For the higher corrosion levels (5% and 10% mass loss) specimens showed radial cracks as

represented in figure 6. The average crack widths versus the time of corrosion are shown in

figure 7. Each crack was measured at 9 locations and the average was calculated. All

specimens were cracked on the surface that was completely immersed in the salt solution

(surface A) This crack was also noticeable on the sides (surface E and F). One sample was

cracked at the side (surface D). It can clearly be observed in figure 7 that samples with ribbed

rebars started to crack earlier than samples with smooth bars, which is probably due to the

larger effective steel surface of the ribbed rebars in comparison with the smooth rebars.

7

Figure 6: Cracked sample after 90 days of corrosion, crack at surface A (left) which was

also noticeable on surface E (right)

Figure 7: Average crack width versus time with indication of the cracked surface

The cumulative acoustic emission events versus the time are given in figure 8. The moment of

cracking of 3 of the 6 monitored samples, namely CR2-R1, CR3-R1 and CR3-S1, can clearly

be distinguished by the sudden increase in the slope of the cumulative number of AE events.

If the AE events of these 3 samples are compared with the crack measurements (figure 9), a

good resemblance between the two graphs is noticeable in terms of slope increase. A

quantitative relation between the amount of AE events or other AE parameters, such as

energy, and crack width was not sought as quantitative AE data depend on several parameters

such as the placement of the AE sensors and the attenuation and reflection of AE waves due

to the heterogeneous character of concrete. Such analysis will be made in further work after

location and characterisation of the AE signals.

8

Figure 8: Cumulative acoustic emission events versus time

Figure 9: Average crack width and cumulative AE events versus time

Pull out testing

The pull out tests were executed when the time to reach the target corrosion levels according

to Faraday’s law was exceeded. The bond strength was calculated as the ratio of the external

forces on the reinforcement and the surface area of the embedded part of the reinforcement. It

thus expresses an average stress along the bonded length of the reinforcement. The slip was

obtained from the LVDT placed on top of the sample. After testing, the rebars were removed

and cleaned using Clark’s solution according to ASTM G1-76 [20] up to the point that no

corrosion products could be noticed by the microscope. The bars were then weighed to

compare the effective and targeted mass loss. The actual corrosion level was calculated using

Eq. 2. Table 4 shows a comparison between the designed and reached corrosion level. Also

the crack widths before and after pull out testing are given.

9

CL=(g0*l-G)/(g0*l)*100% (Eq. 2)

with g0=weight per unit length of the reinforcement bar

l=embedded length

G=weight of the cleaned rebar (embedded length)

From table 4, it can be noticed that the measured corrosion level for the lowest corrosion rate

(1.5% mass loss) exceeded the target corrosion level. For the higher corrosion levels, the

target corrosion level was mostly not reached. Faraday’s law assumes that all current is

applied to dissolve iron in the electrochemical reaction, which is in reality not the case. It also

does not take into account the concrete permeability and amount of chlorides in the solution.

Table 4: Comparison of the designed and measured corrosion level and crack before and

after pull out testing of the different samples

Sample name Designed

corrosion level

[%]

Measured

corrosion level

[%]

Average crack

width before

testing [mm]

Average crack

width after testing

[mm]

CR1-R1 1.5 2.66 No crack No crack

CR1-R2 1.5 1.25 No crack No crack

CR1-R3 1.5 2.35 No crack No crack

CR2-R2 5 5.55 0.56 2.44

CR2-R3 5 3.64 0.58 3.41

CR3-R1 10 8.18 1.18 3.21

CR3-R2 10 8.97 1.07 4.44

CR3-R3 10 7.56 1.23 Not measured

CR1-S1 1.5 2.95 No crack No crack

CR1-S2 1.5 2.62 No crack No crack

CR1-S3 1.5 2.72 No crack No crack

CR2-S1 5 5.45 0.08 0.24

CR2-S2 5 1.83 No crack No crack

CR2-S3 5 3.52 0.04 0.34

CR3-S1 10 6.07 0.64 1.82

CR3-S2 10 6.55 0.34 0.84

CR3-S3 10 7.41 0.91 2.36

Figure 10 shows bond strength versus slip for the different corrosion levels for samples with

ribbed (left) and smooth (right) rebars. Figure 11 shows the maximum bond strength of each

sample versus the measured corrosion level. In addition, the crack widths before testing due to

the corrosion process are indicated.

The bond strength for the samples with non-corroded ribbed rebars is clearly higher than the

bond strength of the samples with non-corroded smooth rebars. The ribs of the ribbed rebars

provide more resistance against the pull out force. The bond of the smooth rebars is only

caused by chemical adhesion and friction and they can therefore be pulled out quite easily.

For the specimens with ribbed rebars, the bond strength increases with 75% when the rebar is

slightly corroded (2%). The reinforcement failed due to rupture for these specimens (CR1-R)

explaining the sudden stop of the left graph in figure 10. Due to the expansive nature of the

corrosion products, the rebars are more confined resulting in a higher bond strength and

stiffness. The same explanation is valid for the samples with smooth rebars were the there is a

remarkable bond strength increase by a factor 10.

10

Figure 10: Relationship between bond stress and slip for ribbed rebars (left) and smooth

rebars (right)

Figuur 11: Maximum bond strength versus measured corrosion level for ribbed and

smooth rebars

For higher corrosion levels namely 5% and 10%, the bond strength decreases for the samples

with ribbed rebars as they showed corrosion-induced cracks before testing. These cracks

resulted in a reduced stress build-up inside the sample causing a reduced confinement. During

testing, the specimens split along the already existing corrosion cracks. The crack widths after

testing were measured and are indicated in table 4. The bond strength of the samples with

11

smooth rebars remains the same for 5% corrosion as for the lowest corrosion level. Even

though fine cracks could be observed before testing, the bond strength is still improved

compared with the non-corroded specimens. The roughness of the smooth rebar increases as

they reach higher corrosion levels resulting in more friction and thus a higher bond strength

and decreased slip. When the corrosion level is above 6%, this effect reduces due to the larger

cracks.

In figures 12 and 13, results were compared with the literature study that was presented in the

introduction. For every sample, the normalized bond strength was calculated. For the samples

with ribbed rebars, results are in general in good agreement with what was found in the

literature except for the lowest corrosion level. This might be due to the relatively large

concrete cover thickness in comparison with the diameter of the rebars applied in this

experimental test campaign.

Results of the samples with smooth rebars show overall a higher bond strength than the

results from Fang et al. [3]. As results can only be compared with one other study, more

research on the smooth rebars is needed to confirm the results.

Figuur 12: Normalized bond strength versus

corrosion level for samples with unconfined

and confined ribbed rebars including new

results

Figuur 13: Normalized bond strength versus

corrosion level for samples with unconfined

and confined smooth rebars including new

results

Summary and conclusions

In this paper, the bond-slip curves for samples reinforced with smooth or ribbed rebars were

compared for different corrosion levels. During the corrosion process, one of the three

samples of each type were monitored with the acoustic emission technique. Results show that

AE is able to detect damage due to corrosion and that the moment of cracking can be

determined from the cumulative AE event curves. AE results need further analysis to

distinguish emissions from different AE sources such as the corrosion process itself and

concrete cracking. Samples with ribbed rebars started to crack earlier than samples with

smooth bars as they have a larger effective surface.

Pull out tests were performed at target corrosion levels. Results show an increase of the bond

strength for samples with ribbed and smooth rebars for low corrosion levels (up to 3%).

Corrosion products cause an improved confinement of the rebars due to their expansive

12

nature. For higher corrosion rates, the bond strength of the samples with ribbed rebars

decreases as these samples showed corrosion-induced cracks before testing. Cracking leads to

a reduced tension causing a reduced confinement. For medium corrosion levels of samples

with smooth rebars, the bond strength is almost the same as for the lowest corrosion level

even though these samples showed fine cracks. For higher corrosion levels and larger crack

width, the bond strength decreases.

Future research will focus on the further analysis of AE results and testing samples with

corroded and uncorroded stirrups to address the effect of confinement on the bond strength.

Acknowledgements

The financial support of the Research Fund KU Leuven is gratefully acknowledged.

References

[1] Lundgren, K., Zandi, K., Nilsson, U. (2015). A model for the anchorage of corroded

reinforcement: validation and application. Concrete - Innovation and Design fib Symposium.

(eds.). Copenhagen, Denmark.

[2] Auyeung, Y., Balaguru, P., Chung, L. (2000). Bond Behavior of Corroded Reinforcement

Bars. ACI Materials Journal, 97(2), 214-221.

[3] Fang, C., Lundgren, K., Chen, L., Zhu, C. (2004). Corrosion influence on bond in

reinforced concrete. Cement and Concrete Research, 34, 2159-2167.

[4] Cabrera, J. G. (1996). Deterioration of Concrete Due to Reinforcement Steel Corrosion.

Cement & Concrete Composites, 18, 47-59.

[5] Al-Sulaimani, G. J., Kaleemullah, M., Basunbul, I. A., Rasheeduzzafart (1990). Influence

of Corrosion and Cracking on Bond Behavior and Strength of Reinforced Concrete Members.

ACI Structural Journal, 87(2), 220-231.

[6] Yalciner, H., Sensoy, S. (2012). An experimental study on the bond strength between

reinforcement bars and concrete as a function of concrete cover, strength and corrosion level.

Cement and Concrete Research, 42, 643-655.

[7] Lee, H.-S., Noguchi, T., Tomosawa, F. (2002). Evaluation of the bond properties between

concrete and reinforcement as a function of the degree of reinforcement corrosion. Cement

and Concrete Research, 32, 1313-1318.

[8] Zhao, Y., Lin, H., Wu, K., Jin, W. (2013). Bond behaviour of normal/recycled concrete

and corroded steel bars. Construction and Building Materials, 48, 348-359.

[9] Almusallam, A. A., Al-Gahtani, A. S., Aziz, A. R., Rasheeduzzafart (1996). Effect of

reinforcement corrosion on bond strength. Construction and Building Materials, 10(2), 123-

129.

[10] Chung, L., Jay Kim, J.-H., Yi, S.-T. (2008). Bond strength prediction for reinforced

concrete members with highly corroded reinforcing bars. Cement & Concrete Composites, 30,

603-611.

[11] Wevers, M. (1997). Listening to the sound of materials: Acoustic emission for the

analysis of material behaviour. NDT & E International, 30(2), 99-106.

[12] Jirarungsatian, C., Prateepasen, A. (2010). Pitting and uniform corrosion source

recognition using acoustic emission parameters. Corrosion Science, 52, 187-197.

[13] Fregonese, M., Idrissi, H., et al. (2001). Initiation and propagation steps in pitting

corrosion of austenitic stainless steels: monitoring by acoustic emission. Corrosion Science,

43, 627-641.

[14] Yoon, D. J., Weiss, W. J., Shah, S. P. (2000). Assessing damage in corroded reinforced

concrete using acoustic emission. ASCE Journal of Engineering Mechanics, 126(3), 273-283.

[15] Patil, S., Goyal, S., Karkare, B. (2015). Acoustic emission-based mathematical procedure

for quantification of rebar corrosion in reinforced concrete. Current Science, 109(5), 943-948.

13

[16] Ohtsu, M., Tomoda, Y. (2007). Phenomenological Model of Corrosion Process in

Reinforced Concrete Identified by Acoustic Emission. ACI Materials Journal, 105(2), 194-

199.

[17] Van Steen, C., Wevers, M., Verstrynge, E. (2017). Detection of chloride-induced

corrosion-induced damage at the reinforcement-concrete interface with X-ray computed

tomography and acoustic emission. XIV DBMC. Ghent, Belgium.

[18] Andrade, C., Alonso, C. (1996). Corrosion rate monitoring in the laboratory and on-site.

Construction and Building Materials, 10(5), 315-328.

[19] Caré, S., Raharinaivo, A. (2007). Influence of impressed current on the initiation of

damage in reinforced mortar due to corrosion of embedded steel. Cement and Concrete

Research, 37, 1598-1612.

[20] ASTM (1990). ASTM G1 Standard Practice for Preparing, Cleaning, and Evaluation

Corrosion Test Specimens.