An Experimental Study on Effects of Corrosion and Stirrups Spacing on Bond Behavior of Reinforced Concrete

The current experimental study consists of part of an extensive and ongoing research on bond behavior of RC elements damaged by corrosion, focusing on stirrups spacing effect on bonding. For this, RC specimens with different cases of stirrups spacing were casted. Accelerated corrosion was induced in order to simulate the slow process of nature corrosion on RC specimens and the corrosion damage was estimated in terms of mass loss of steel bars and average width of surface concrete cracking. Subsequently, pull-out tests were carried out to examine the bonding resistance between steel and concrete. The study indicates the great influence of density of stirrups on the percentage mass loss of the embedded reinforcing bar, accompanied by width of surface concrete cracking, as well as on bond strength between steel and concrete. The results of bond stress–slip curves show that the densification of stirrups plays a significant role in bonding, leading to higher bond strength values and delaying the degradation of bond loss as corrosion damage increases. However, it becomes apparent that, although the densification of stirrups (Φ8/60 mm) result in the full anchorage of steel-reinforcing bars, it may be inappropriate, since it can lead to a substantial increase in costs and a rapid rise in corrosion rate, due to potential increase. Furthermore, the recorded values of relative slip at bond strength are between 1 and 3 mm, regardless of corrosion damage or concrete cracking, which depends on the ribs geometry and crushing of concrete in front of them. To conclude, the results of the present manuscript indicate that the increase in transverse reinforcement (stirrups) percentage plays a key role in the durability of reinforced concrete elements and in bond strength maintenance between rebar and concrete.


Introduction
In Reinforced Concrete (RC) elements, the bond mechanism between steel bars and the surrounding concrete plays the key role, in order to ensure that steel reinforcement and concrete react as one material and the tensile forces are safely transferred from steel to concrete. Due to its high bearing capacity and ductility, compared to the low production costs, reinforced concrete is one of the most widely used construction materials in the existing building stock.
However, the corrosion of steel reinforcement consists of the main durability problem of RC structures, affecting not only their appearance but also their mechanical performance and structural integrity. Recent reports have indicated huge economic impacts due to corrosion damage, since a significant part of the annual budget in many countries is spent on the maintenance, repair, and rehabilitation of RC structures [1,2]. For instance, the results of a study conducted by NACE [1] as chloride ions, are penetrated and diffused in concrete through its pores, destroying the abovementioned protection mechanisms (initiation stage of corrosion). When the chloride concentration reaches a critical value, it is considered that the passive layers of the reinforcement are demolished, and then the electrochemical process of corrosion initiates (corrosion onset) [3,4]. The forming oxides on steel surface (rust) occupy a 2 to 6 times greater volume of the attacking mass [5], causing tensile stresses in surrounding concrete and, thereafter, leading to gradual concrete cracking development and spalling of the cover concrete [6], as shown in Figure 1. As a result, the bond mechanism between steel and concrete is significantly affected, and therefore the bearing capacity of RC elements is degraded [7][8][9].
Since steel reinforcement is embedded in concrete and its corrosion damage is difficult to map and quantify, several studies have been conducted with the aim of linking the width of surface cracks with the penetration of the corrosion (or in terms of mass loss) of steel bars; thus, predictive models have been suggested [10][11][12][13][14][15][16][17]. This approach consists of a non-destructive method of corrosion assessment, since crack opening is a useful and representative index of steel bars' damage. The extensive experimental research of Bossio et al. on corroded RC specimens introduced a non-linear model, in which a uniform type of corrosion was assumed, and the limit value of crack width was about 10 mm [17]. However, the outcomes of various experimental studies differ due to different exposure conditions, cover thicknesses, steel bar geometries, positions of steel bars, etc. Hence, there is a need for further research on this framework in order to define the influencing coefficients and establish specific correlations between the width of surface cracking, the degree of corrosion damage, the cover thickness, the density of stirrups, the concrete class, etc. Bonding resistance between steel and concrete is mainly determined by chemical adhesion, friction, and the mechanical interlock between steel bar and surrounding concrete [18]. In general, the anchorage of (deformed) reinforcing bars is due to the confinement level. Several studies have indicated the main factors which affect bonding, such as the mechanical properties of concrete, concrete cover, stirrups spacing, relative rib area of deformed steel bars and steel corrosion. Abosrra et al. [19] and Zandi and Coronelli [20] investigated the influence of corrosion on bond behavior, Since steel reinforcement is embedded in concrete and its corrosion damage is difficult to map and quantify, several studies have been conducted with the aim of linking the width of surface cracks with the penetration of the corrosion (or in terms of mass loss) of steel bars; thus, predictive models have been suggested [10][11][12][13][14][15][16][17]. This approach consists of a non-destructive method of corrosion assessment, since crack opening is a useful and representative index of steel bars' damage. The extensive experimental research of Bossio et al. on corroded RC specimens introduced a non-linear model, in which a uniform type of corrosion was assumed, and the limit value of crack width was about 10 mm [17]. However, the outcomes of various experimental studies differ due to different exposure conditions, cover thicknesses, steel bar geometries, positions of steel bars, etc. Hence, there is a need for further research on this framework in order to define the influencing coefficients and establish specific correlations between the width of surface cracking, the degree of corrosion damage, the cover thickness, the density of stirrups, the concrete class, etc.
Bonding resistance between steel and concrete is mainly determined by chemical adhesion, friction, and the mechanical interlock between steel bar and surrounding concrete [18]. In general, the anchorage of (deformed) reinforcing bars is due to the confinement level. Several studies have indicated the main factors which affect bonding, such as the mechanical properties of concrete, concrete cover, stirrups spacing, relative rib area of deformed steel bars and steel corrosion. Abosrra et al. [19] and Zandi and Coronelli [20] investigated the influence of corrosion on bond behavior, testing specimens Metals 2020, 10, 1327 3 of 14 of different compressive concrete strengths. Apostolopoulos et al. [21], Fischer and Ozbolt [22] and Yalciner et al. [23] carried out accelerated corrosion tests and pull-out to investigate bond loss in the function of different corrosion levels and different concrete cover values. Most experimental studies in the literature have carried out pull-out tests on specimens where the reinforcing bar is set in the center of the cylindrical concrete specimen without transverse reinforcement. Recently, experimental studies by Zandi et al. [24], Lin et al. [25] and Apostolopoulos et al. [21] investigated the impact of both corrosion and stirrups spacing carrying out eccentric pull-out tests on RC elements with the usual design values of concrete cover. However, even today there is a level of uncertainty which depends on the estimation of bond capacity of corroded RC elements. Thus, this experimental study seeks to enhance the current knowledge and provide additional data on the bond behavior of corroded RC elements, taking into account the corrosion damage of the reinforcing bar, as well as the confinement level, through transverse reinforcement (stirrups).

Experimental Procedure
In this experimental study, the bond behavior of 24 un-corroded and corroded reinforced concrete specimens was investigated. The experimental procedure consists of four main phases: Phase I, the casting and curing of specimens with different amounts of transverse reinforcement; Phase II, the accelerated corrosion of specimens in laboratory for different exposure times in order to achieve different corrosion levels and measure the range of surface cracking on concrete; Phase III, pull-out tests, carried out in order to investigate the bond behavior (bond stress and relative slip between steel and concrete); Phase IV, the breaking of specimens in order to assess the corrosion damage of reinforcement and correlate it with the pull-out results.

Materials and Specimens Casting
It is a fact that in the period 1990-2010, the selected concrete in the majority of structures located in coastal and seismic regions of the Mediterranean (Greece, Italy, Turkey) was class of C20/25, despite the guidelines of EN 206 [26] for the use of higher categories of concrete (e.g., C25/30 or C30/37). Based on the fact that these structures already show symptoms of premature aging, it was considered expedient to develop an extensive, continuous (evolving) experimental study in order to investigate the degradation of bond strength (steel-concrete) in RC structural elements for different levels of artificial corrosion. For this purpose, prismatic reinforced concrete specimens of dimensions 240 × 200 × 310 (mm 3 ) were casted ( Figure 2), in which a main reinforcing steel bar, of technical class B500C and nominal diameter equal to 16 mm, was semi-embedded on the top of the specimens with concrete cover c equal to 25 mm. Furthermore, specimens with different amount of transverse reinforcement were tested, in order to investigate the influence of stirrups spacing on the corrosion damage and bond behavior between main steel bar and concrete. Specifically, four cases were tested: (a) specimens without stirrups, (b) specimens with wide stirrups Φ8/240 mm, (c) specimens with stirrups Φ8/125 mm and (d) specimens with dense stirrups Φ8/60 mm. All stirrups were of nominal diameter equal to 8 mm and four smooth bars were placed at the corners to stabilize them during the casting procedure.
To permit easy identification of test specimens, special labels of three parts were used; the first part of the label refers to the concrete cover thickness, namely c = 25 mm, then the following part of label represents the amount (density) of stirrups-more specifically, N refers to the group of test specimens without stirrups, and S240, S120 and S60 refers to the groups of test specimens with stirrups Φ8 per 240, 120 and 60 mm, respectively. The last digit of each label refers to the corrosion time. Thus, for instance, the 25-S120-4 specimen corresponds to a specimen with cover thickness 25 mm, with wide dense stirrups Φ8 per 120 mm, and to the fourth (in a row) corrosion time. Before mixing the concrete in molds, reinforcing steel bars were carefully cleaned, and their initial mass and total length were documented. With a view to limit the corrosion damage only to the embedded length of main reinforcing bar, and to indicate the influence of corrosion of the main bar on surface concrete cracking and on bond behavior, a 250-mm section of rebar was exposed in corrosion and the rest was protected with a wax layer, and all stirrups were protected with a special anticorrosive coating (epoxy resin). Thereafter, all reinforcement (steel bar and stirrups) were aligned and fastened to the molds to cast the concrete, Figure 3a, compaction of which was conducted with the help of a table vibrator. For the concrete mixture, a Portland cement was used, with water-cement ratio of 0.55 and 20 mm maximum size coarse aggregate. After the casting, all specimens were kept for three days in a wet environment, then the molds were carefully removed and the curing of specimens was taken place for 28 days in a specially designed room under a temperature of 22 °C and continuous wetting, Figure 3b. In parallel, plain concrete cube specimens of 200 mm 3 were casted and the average value of 28-day recorded compressive strength was evaluated fc = 30 MPa.

Accelerated Corrosion Technique
Due to concrete cover and initial high alkaline level of concrete, the corrosion process in natural environment is slow and takes place over a period of years, even in coastal regions, as there is a period Before mixing the concrete in molds, reinforcing steel bars were carefully cleaned, and their initial mass and total length were documented. With a view to limit the corrosion damage only to the embedded length of main reinforcing bar, and to indicate the influence of corrosion of the main bar on surface concrete cracking and on bond behavior, a 250-mm section of rebar was exposed in corrosion and the rest was protected with a wax layer, and all stirrups were protected with a special anticorrosive coating (epoxy resin). Thereafter, all reinforcement (steel bar and stirrups) were aligned and fastened to the molds to cast the concrete, Figure 3a, compaction of which was conducted with the help of a table vibrator. For the concrete mixture, a Portland cement was used, with water-cement ratio of 0.55 and 20 mm maximum size coarse aggregate. After the casting, all specimens were kept for three days in a wet environment, then the molds were carefully removed and the curing of specimens was taken place for 28 days in a specially designed room under a temperature of 22 • C and continuous wetting, Figure 3b. In parallel, plain concrete cube specimens of 200 mm 3 were casted and the average value of 28-day recorded compressive strength was evaluated f c = 30 MPa. Before mixing the concrete in molds, reinforcing steel bars were carefully cleaned, and their initial mass and total length were documented. With a view to limit the corrosion damage only to the embedded length of main reinforcing bar, and to indicate the influence of corrosion of the main bar on surface concrete cracking and on bond behavior, a 250-mm section of rebar was exposed in corrosion and the rest was protected with a wax layer, and all stirrups were protected with a special anticorrosive coating (epoxy resin). Thereafter, all reinforcement (steel bar and stirrups) were aligned and fastened to the molds to cast the concrete, Figure 3a, compaction of which was conducted with the help of a table vibrator. For the concrete mixture, a Portland cement was used, with water-cement ratio of 0.55 and 20 mm maximum size coarse aggregate. After the casting, all specimens were kept for three days in a wet environment, then the molds were carefully removed and the curing of specimens was taken place for 28 days in a specially designed room under a temperature of 22 °C and continuous wetting, Figure 3b. In parallel, plain concrete cube specimens of 200 mm 3 were casted and the average value of 28-day recorded compressive strength was evaluated fc = 30 MPa.

Accelerated Corrosion Technique
Due to concrete cover and initial high alkaline level of concrete, the corrosion process in natural environment is slow and takes place over a period of years, even in coastal regions, as there is a period

Accelerated Corrosion Technique
Due to concrete cover and initial high alkaline level of concrete, the corrosion process in natural environment is slow and takes place over a period of years, even in coastal regions, as there is a period of penetration of the aggressive factors in concrete and the subsequent beginning of corrosion must be passed. In order to accelerate the consequences of steel reinforcement corrosion, laboratory experiments on reinforced concrete specimens were carried out by the anodic corrosion method. This method is simple and widely used, since it has the advantage of fixing different levels of the induced current density on specimens and significantly accelerating the corrosion rate of natural procedure.
Based on the abovementioned technique, a power supply was used to induce corrosion by applying a direct electric current to the semi-embedded reinforcing steel bar, which acted as an anode. Each specimen was placed in a tank, semi-filled with 5% sodium chloride (NaCl) solution by the weight of water, in the presence of a stainless-steel bar, which acted as cathode of the circuit, Figure 4a. Initially, before the commencement of corrosion experiments, all specimens were moistened continuously for three days in order to full the pores of concrete with the NaCl electrolyte, and, thereafter, the current was induced to the main reinforcing bar. moistened continuously for three days in order to full the pores of concrete with the NaCl electrolyte, and, thereafter, the current was induced to the main reinforcing bar.
Solution 5% NaCl was selected because it simulates a severe corrosive environment or a coastal environment [27,28] and comes in agreement with existing regulations concerning corrosion tests, such as the B117 ASTM Standard [29]. Additionally, the content of the solution in NaCl accurately represents the case of the Mediterranean countries, where hot climate results in higher salinity of the seawater, in contrast to northern countries, where river estuaries can be met. A typical example is the Arabian basin, where the recorded salinity is even higher. Furthermore, in Mediterranean countries, the salinity of the sea increases during the summer periods, due to the high temperatures [30,31].
With the aim of simulating the corrosion process and achieving different levels of corrosion damage on reinforcing steel bar, the specimens remained in the cells for different exposure times, inducing current of 0.12 A, via the continuous power supply, Figure 4b. Based on Lin et al. study [32], which indicated the bond strength reduction owing to corrosion of stirrups, and with a view to estimate the bond loss, only due to the damage to the main reinforcing bar, all stirrups were protected with a special anticorrosive coating (epoxy resin), in order to sufficiently maintain their corrosion resistance, so that the results of pull-out test refer only to the degradation of longitudinal reinforcing bar and are not additionally affected by stirrups' corrosion. However, as a result, a significant part of the applied current is consumed at the same time to damage the additional protective material. Thus, the current density value of specimens with different cases of stirrups was not the same, resulting in different experimental times being required to cause a similar level of corrosion damage.
Furthermore, in order to investigate the influence of the reinforcing bar's corrosion on bond behavior, the embedded length of 250 mm was free to be corroded and the rest of the steel bar was protected by a wax layer.  Solution 5% NaCl was selected because it simulates a severe corrosive environment or a coastal environment [27,28] and comes in agreement with existing regulations concerning corrosion tests, such as the B117 ASTM Standard [29]. Additionally, the content of the solution in NaCl accurately represents the case of the Mediterranean countries, where hot climate results in higher salinity of the seawater, in contrast to northern countries, where river estuaries can be met. A typical example is the Arabian basin, where the recorded salinity is even higher. Furthermore, in Mediterranean countries, the salinity of the sea increases during the summer periods, due to the high temperatures [30,31].
With the aim of simulating the corrosion process and achieving different levels of corrosion damage on reinforcing steel bar, the specimens remained in the cells for different exposure times, inducing current of 0.12 A, via the continuous power supply, Figure 4b. Based on Lin et al. study [32], which indicated the bond strength reduction owing to corrosion of stirrups, and with a view to estimate the bond loss, only due to the damage to the main reinforcing bar, all stirrups were protected with a special anticorrosive coating (epoxy resin), in order to sufficiently maintain their corrosion resistance, so that the results of pull-out test refer only to the degradation of longitudinal reinforcing bar and are not additionally affected by stirrups' corrosion. However, as a result, a significant part of the applied current is consumed at the same time to damage the additional protective material. Thus, the current density value of specimens with different cases of stirrups was not the same, resulting in different experimental times being required to cause a similar level of corrosion damage.
Furthermore, in order to investigate the influence of the reinforcing bar's corrosion on bond behavior, the embedded length of 250 mm was free to be corroded and the rest of the steel bar was protected by a wax layer.

Mapping of Surface Concrete Cracks
Surface cracking appeared on the concrete of specimens due to the corrosion of steel reinforcement, mainly on the upper side where the main reinforcing bar was placed, as shown in Figure 5a. The average width of these cracks was mapped, measured and calculated in order to estimate the corrosion damage through concrete cracking and to correlate this cracking with bond strength.

Mapping of Surface Concrete Cracks
Surface cracking appeared on the concrete of specimens due to the corrosion of steel reinforcement, mainly on the upper side where the main reinforcing bar was placed, as shown in Figure 5a. The average width of these cracks was mapped, measured and calculated in order to estimate the corrosion damage through concrete cracking and to correlate this cracking with bond strength.
Surface cracks in a clear direction parallel to the longitudinal reinforcing bars were measured along per 5 mm in distinct segments of 20 to 50 mm in length, namely, crack width measurements were taken per 5mm for each individual crack length using a digital caliper. By way of illustration, an individual measuring of crack width on concrete surface is presented in Figure 5b. From these discrete measurements, the average crack width in each reinforced concrete specimen for each corrosion level was finally calculated.

Pull-Out Tests
Based on ASTM C234-91a [33], pull-out tests were carried out after corrosion experiments and mapping of surface cracks. This method is a simple means of bond mechanism testing on the interface between the steel reinforcement and surrounding concrete.
For this purpose, an apparatus was designed, Figure 6a, which allows the direct transfer of the applied force from the steel bar to concrete. The mechanical tests were performed on a generalpurpose servomotor MTS (load capacity of Fmax =250 kN), where its own fitting and the tested reinforced concrete sample were properly fitted, Figure 6b.
An external Linear Variable Differential Transformer (LVDT) was set to the free loaded end of reinforcing bar, to monitor the relative slip between the bar and concrete. All specimens were loaded under displacement-control at a constant displacement rate of 0.4 mm/min. During the mechanical tests, pull-out force and relative slip between steel bar and concrete were automatically recorded using a data logger. Surface cracks in a clear direction parallel to the longitudinal reinforcing bars were measured along per 5 mm in distinct segments of 20 to 50 mm in length, namely, crack width measurements were taken per 5mm for each individual crack length using a digital caliper. By way of illustration, an individual measuring of crack width on concrete surface is presented in Figure 5b. From these discrete measurements, the average crack width in each reinforced concrete specimen for each corrosion level was finally calculated.

Pull-Out Tests
Based on ASTM C234-91a [33], pull-out tests were carried out after corrosion experiments and mapping of surface cracks. This method is a simple means of bond mechanism testing on the interface between the steel reinforcement and surrounding concrete.
For this purpose, an apparatus was designed, Figure 6a, which allows the direct transfer of the applied force from the steel bar to concrete. The mechanical tests were performed on a general-purpose servomotor MTS (load capacity of F max = 250 kN), where its own fitting and the tested reinforced concrete sample were properly fitted, Figure 6b.
An external Linear Variable Differential Transformer (LVDT) was set to the free loaded end of reinforcing bar, to monitor the relative slip between the bar and concrete. All specimens were loaded under displacement-control at a constant displacement rate of 0.4 mm/min. During the mechanical tests, pull-out force and relative slip between steel bar and concrete were automatically recorded using a data logger. In order to quantify the bonding resistance between steel and concrete, average bond strength was estimated, taking the nominal diameter of the reinforcement bar into account and considering the bonding stresses as constant and uniform along the embedded length. This simplification leads to lower values than the actually developing ones, especially as the embedded length increases. Thus, the average bond strength between steel and concrete fcb for the embedded length was expressed by Equation (1) where Fmax is the maximum pull-out load, D = 16 mm the nominal diameter of the main reinforcing bar, and Lemb = 250 mm the embedded length. Furthermore, the maximum pull-out load, Fmax, the bond loss ratio fcb,cor/fcb,uncor and the relative slip values corresponding to the maximum pull-out load (bond strength), smax, were recorded.

Results and Discussion
After concreting and curing period of all specimens, no surface crack on concrete was noticed. Hence, all recorded surface crack widths were deemed due to the corrosion of steel reinforcement. The results of the current experimental work are summarized in Table 1 for each corrosion time. In In order to quantify the bonding resistance between steel and concrete, average bond strength was estimated, taking the nominal diameter of the reinforcement bar into account and considering the bonding stresses as constant and uniform along the embedded length. This simplification leads to lower values than the actually developing ones, especially as the embedded length increases. Thus, the average bond strength between steel and concrete f cb for the embedded length was expressed by Equation (1), as shown in Figure 7 f cb = F max π · D · L emb (1) where F max is the maximum pull-out load, D = 16 mm the nominal diameter of the main reinforcing bar, and L emb = 250 mm the embedded length. In order to quantify the bonding resistance between steel and concrete, average bond strength was estimated, taking the nominal diameter of the reinforcement bar into account and considering the bonding stresses as constant and uniform along the embedded length. This simplification leads to lower values than the actually developing ones, especially as the embedded length increases. Thus, the average bond strength between steel and concrete fcb for the embedded length was expressed by Equation (1), as shown in Figure 7 where Fmax is the maximum pull-out load, D = 16 mm the nominal diameter of the main reinforcing bar, and Lemb = 250 mm the embedded length. Furthermore, the maximum pull-out load, Fmax, the bond loss ratio fcb,cor/fcb,uncor and the relative slip values corresponding to the maximum pull-out load (bond strength), smax, were recorded.

Results and Discussion
After concreting and curing period of all specimens, no surface crack on concrete was noticed. Hence, all recorded surface crack widths were deemed due to the corrosion of steel reinforcement. The results of the current experimental work are summarized in Table 1 for each corrosion time. In Furthermore, the maximum pull-out load, F max , the bond loss ratio f cb,cor /f cb,uncor and the relative slip values corresponding to the maximum pull-out load (bond strength), s max , were recorded.

Results and Discussion
After concreting and curing period of all specimens, no surface crack on concrete was noticed. Hence, all recorded surface crack widths were deemed due to the corrosion of steel reinforcement. The results of the current experimental work are summarized in Table 1 for each corrosion time.
In particular, the assessment of corrosion damage is depicted through percentage mass loss of the main steel bar, by measuring of weight after pull-out tests, and through the average surface cracking width of concrete. Moreover, derived from pull-out tests for each corrosion level, the ultimate pull-out load, bond strength, bond loss ratio and relative slip at bond strength are noted. The bond loss ratio f cb,cor /f cb,uncor refers to proper normalization of the loss of bond strength between corroded reinforcement and concrete, and subsequently the reduction in the load-carrying capacity of the reinforced concrete element. The results of the accelerated corrosion procedure showed that an increase in corrosion time leads to more intense concrete cracking. As previous studies reported, crack width does not obey any specific rule with respect to mass loss of steel reinforcement, as it may, to some extent, depend on the class and porosity of concrete, steel bar geometry, concrete cover and ratio of transverse reinforcement [17,21]. In former studies, the non-linear relationship between the crack width on concrete surface and the uniform corrosion penetration has been presented via the corresponding models [11,12,15,16]. However, it is well known that corrosion damage on a steel reinforcement embedded in concrete is characterized by strong non-uniformity, which is reflected by pits on the surface of reinforcing bars and local cross-sectional reductions. Figure 8 shows the significant reduction in the cross-section of a corroded reinforcing bar, located only in a small length of it, while the rest of its length has a slow corrosion rate. The mass loss index, used to estimate corrosion damage, corresponds to the mass loss of material along its entire tested length, which is exposed to corrosion. Thus, mass loss gives an overview of corrosion damage, without attributing it to local phenomena. Although it can be explained from the above-mentioned that mass loss index is widely followed by the scientific community, it lags behind the concept of cross-sectional loss of reinforcing bars, and it cannot always correlate the corrosion damage of steel reinforcement with the surface cracking of concrete or the bond loss between steel and concrete. Nonetheless, surface crack width tends to be an appropriate index for the bond strength evaluation, as there a correlation has been indicated between these two parameters. More specifically, the current research agrees with the experimental results, which have shown that the bond loss of reinforced concrete due to corrosion is in line with the increase in surface cracking. In this framework, predictive models of bond loss have been introduced by different researchers [21,22,25,[34][35][36][37][38]. Apostolopoulos and Koulouris [21] proposed exponential predictive curves of bond loss in tandem with surface cracking width, by non-linear regression analysis, based on experimental results ( Figure  9), by drawing attention to the influence of stirrups spacing. The functions of predictive models are shown in Equation (7) e    w cor A c cb uncor cb f f (2) where A is a parameter that depends on the concrete cover thickness (c) and the amount of transverse reinforcement (stirrups spacing or absence of stirrups). The values of parameter A for each predictive model, and the corresponding values of the R 2 coefficient, are presented in Table 2. These models indicated the significant effect of the presence and density of stirrups on the degradation rate of bond loss in RC elements. In particular, the densification of stirrups in RC specimens benefit bond behavior by delaying the reduction in bond strength due to steel corrosion. In the case of specimens without stirrups, the bearing capacity dramatically decreases at a percentage of 85% (for mass loss equal to 7.73%), whereas specimens with dense stirrups (Φ8/60 mm) show a slight drop of 18% (for mass loss equal to 8.68%), respectively. Hence, the presence of dense stirrups spacing has a significantly positive influence on the low degradation rate of the initial (uncorroded) bearing capacity of RC specimens. Nonetheless, surface crack width tends to be an appropriate index for the bond strength evaluation, as there a correlation has been indicated between these two parameters. More specifically, the current research agrees with the experimental results, which have shown that the bond loss of reinforced concrete due to corrosion is in line with the increase in surface cracking. In this framework, predictive models of bond loss have been introduced by different researchers [21,22,25,[34][35][36][37][38]. Apostolopoulos and Koulouris [21] proposed exponential predictive curves of bond loss in tandem with surface cracking width, by non-linear regression analysis, based on experimental results (Figure 9), by drawing attention to the influence of stirrups spacing. The functions of predictive models are shown in Equation (7) f cor where A is a parameter that depends on the concrete cover thickness (c) and the amount of transverse reinforcement (stirrups spacing or absence of stirrups). The values of parameter A for each predictive model, and the corresponding values of the R 2 coefficient, are presented in Table 2.
Metals 2020, 10, x FOR PEER REVIEW 10 of 14 Figure 9. Exponential predictive models of bond loss in function with average crack width [21].
As shown in Figure 10, extending this investigation and focusing on the values of maximum pull-out force and, subsequently, on bond strength, and not bond loss, the role of stirrups spacing is more highlighted, as bonding between steel and concrete degrades due to corrosion and the developed maximum pull-out force drops. The usage of dense transverse reinforcement contributes to bond behavior, not only reducing the bond loss rate, but leading to greater values of maximum  These models indicated the significant effect of the presence and density of stirrups on the degradation rate of bond loss in RC elements. In particular, the densification of stirrups in RC specimens benefit bond behavior by delaying the reduction in bond strength due to steel corrosion. In the case of specimens without stirrups, the bearing capacity dramatically decreases at a percentage of 85% (for mass loss equal to 7.73%), whereas specimens with dense stirrups (Φ8/60 mm) show a slight drop of 18% (for mass loss equal to 8.68%), respectively. Hence, the presence of dense stirrups spacing has a significantly positive influence on the low degradation rate of the initial (uncorroded) bearing capacity of RC specimens.
As shown in Figure 10, extending this investigation and focusing on the values of maximum pull-out force and, subsequently, on bond strength, and not bond loss, the role of stirrups spacing is more highlighted, as bonding between steel and concrete degrades due to corrosion and the developed maximum pull-out force drops. The usage of dense transverse reinforcement contributes to bond behavior, not only reducing the bond loss rate, but leading to greater values of maximum pull-out force due to confinement.
Metals 2020, 10, x FOR PEER REVIEW 10 of 14 Figure 9. Exponential predictive models of bond loss in function with average crack width [21].
As shown in Figure 10, extending this investigation and focusing on the values of maximum pull-out force and, subsequently, on bond strength, and not bond loss, the role of stirrups spacing is more highlighted, as bonding between steel and concrete degrades due to corrosion and the developed maximum pull-out force drops. The usage of dense transverse reinforcement contributes to bond behavior, not only reducing the bond loss rate, but leading to greater values of maximum pull-out force due to confinement. In an effort to estimate the influence of stirrups spacing on non-corroded condition, the presence of wide stirrups (Φ8/240 mm) presents a maximum bond strength equal to 7.04 MPa, whereas quite dense (Φ8/120 mm) and dense stirrups (Φ8/60 mm) are equal to 9.10 and 9.53 MPa, respectively. The percentages attributing the increase in bond strength against specimens without stirrups are 35.9%, 75.6% and 84%, respectively. It is noteworthy that quite dense stirrups spacing (Φ8/120 mm) results in sufficient bond strength levels, whereas further densification leads to a minor increase in bond strength, which nevertheless delays the bond loss. Extrapolating the abovementioned on real RC structures, stirrups' densification above a certain threshold, is considered inappropriate since it could lead, on the one hand, to a substantial increase in costs and on the other hand to a rapid rise in corrosion rate, due to the potential increase.
Moreover, bond stress-relative slip curves are exported from pull-out tests for each corrosion level and for each category of stirrups spacing, respectively. The typical curves of uncorroded and corroded specimens are shown in Figure 11. In an effort to estimate the influence of stirrups spacing on non-corroded condition, the presence of wide stirrups (Φ8/240 mm) presents a maximum bond strength equal to 7.04 MPa, whereas quite dense (Φ8/120 mm) and dense stirrups (Φ8/60 mm) are equal to 9.10 and 9.53 MPa, respectively. The percentages attributing the increase in bond strength against specimens without stirrups are 35.9%, 75.6% and 84%, respectively. It is noteworthy that quite dense stirrups spacing (Φ8/120 mm) results in sufficient bond strength levels, whereas further densification leads to a minor increase in bond strength, which nevertheless delays the bond loss. Extrapolating the abovementioned on real RC structures, stirrups' densification above a certain threshold, is considered inappropriate since it could lead, on the one hand, to a substantial increase in costs and on the other hand to a rapid rise in corrosion rate, due to the potential increase.
Moreover, bond stress-relative slip curves are exported from pull-out tests for each corrosion level and for each category of stirrups spacing, respectively. The typical curves of uncorroded and corroded specimens are shown in Figure 11. The harmful influence of the corrosion phenomenon on the bond behavior between steel and concrete can be seen examining the bond stress-slip curves of Figure 11 and in a particular group of specimens without stirrups. Uncorroded specimens-where no surface cracking had been initially observed-showed a quasi-linear relationship between bond stresses and relative slip until the point of bond strength's development. During this phase, cracks occurred parallel to the axis of steel bar, and gradually were developed due to the radial stresses, which were transferred from steel to concrete. After the development of bond strength, a sharp decline in bond stresses and complete bond loss followed. In the case of corroded specimens (without stirrups), intense surface concrete cracking was recorded, due to steel corrosion, even in low mass loss levels. As a result of this cracking, bond behavior degraded dramatically, as confirmed by the corresponding curves of Figure 11, since the steel-concrete interface is damaged by the corrosion oxides and the confinement level is deteriorated by the cracks in concrete cover.
On the other hand, as shown by the groups of specimens with stirrups, surface cracking and the degree of bond loss is strongly correlated with stirrups spacing. In particular, specimens with quite wide stirrups (Φ8/240 mm) showed that transverse reinforcement has a positive impact on bond behavior in uncorroded conditions, as greater values of bond stress were indicated. However, when corrosion occurs, accompanied by surface concrete cracking, bond resistance degrades, as was reflected in responding curves, where bond strength is reduced significantly, and subsequently low values of residual bond stress were recorded.
A greater contribution of stirrups to bond behavior was noticed on specimens with quite dense stirrups spacing (Φ8/120 mm), where even higher bond strength values are indicated, and while bond behavior degrades due to corrosion damage, does not demonstrate a massive drop, mainly due to confinement. In addition, in the case of dense stirrups (Φ8/60 mm), the full use of bond behavior, even after corrosion damage, occurs, with residual bond stress recorded after the peak in bond strength. It is also noteworthy that specimens of this category finally ended due to failure under tension. Thus, densification of stirrups makes full use of the bearing capacity of steel reinforcing bars and, subsequently, of RC elements. From the aforementioned, it becomes obvious that the use of dense stirrups (Φ8/60 mm) leads to full bonding between steel and concrete, in uncorroded and corroded conditions, respectively, marking both high bond strength and high residual bond stress after the ultimate pull-out force.
To conclude, it should be noticed that in all groups of specimens with stirrups, the relative slip at bond strength is between 1 and 3 mm, regardless of corrosion damage or concrete cracking. This The harmful influence of the corrosion phenomenon on the bond behavior between steel and concrete can be seen examining the bond stress-slip curves of Figure 11 and in a particular group of specimens without stirrups. Uncorroded specimens-where no surface cracking had been initially observed-showed a quasi-linear relationship between bond stresses and relative slip until the point of bond strength's development. During this phase, cracks occurred parallel to the axis of steel bar, and gradually were developed due to the radial stresses, which were transferred from steel to concrete. After the development of bond strength, a sharp decline in bond stresses and complete bond loss followed. In the case of corroded specimens (without stirrups), intense surface concrete cracking was recorded, due to steel corrosion, even in low mass loss levels. As a result of this cracking, bond behavior degraded dramatically, as confirmed by the corresponding curves of Figure 11, since the steel-concrete interface is damaged by the corrosion oxides and the confinement level is deteriorated by the cracks in concrete cover.
On the other hand, as shown by the groups of specimens with stirrups, surface cracking and the degree of bond loss is strongly correlated with stirrups spacing. In particular, specimens with quite wide stirrups (Φ8/240 mm) showed that transverse reinforcement has a positive impact on bond behavior in uncorroded conditions, as greater values of bond stress were indicated. However, when corrosion occurs, accompanied by surface concrete cracking, bond resistance degrades, as was reflected in responding curves, where bond strength is reduced significantly, and subsequently low values of residual bond stress were recorded.
A greater contribution of stirrups to bond behavior was noticed on specimens with quite dense stirrups spacing (Φ8/120 mm), where even higher bond strength values are indicated, and while bond behavior degrades due to corrosion damage, does not demonstrate a massive drop, mainly due to confinement. In addition, in the case of dense stirrups (Φ8/60 mm), the full use of bond behavior, even after corrosion damage, occurs, with residual bond stress recorded after the peak in bond strength. It is also noteworthy that specimens of this category finally ended due to failure under tension. Thus, densification of stirrups makes full use of the bearing capacity of steel reinforcing bars and, subsequently, of RC elements. From the aforementioned, it becomes obvious that the use of dense stirrups (Φ8/60 mm) leads to full bonding between steel and concrete, in uncorroded and corroded conditions, respectively, marking both high bond strength and high residual bond stress after the ultimate pull-out force.
To conclude, it should be noticed that in all groups of specimens with stirrups, the relative slip at bond strength is between 1 and 3 mm, regardless of corrosion damage or concrete cracking. This is explained by the ribs geometry of steel bars, since it is commonly accepted that slip is mostly due to the crushing of concrete in front of the ribs. As mentioned before, the corrosion of steel reinforcing bars is accompanied with intense pits on the surface, causing a reduction in the total loss of ribs. It is also noteworthy that corrosion damage on the steel surface is mainly depicted only on the outer half of the steel bar's circumference, adjacent to the free surface of specimens, as Figure 12 shows.
Metals 2020, 10, x FOR PEER REVIEW 12 of 14 is explained by the ribs geometry of steel bars, since it is commonly accepted that slip is mostly due to the crushing of concrete in front of the ribs. As mentioned before, the corrosion of steel reinforcing bars is accompanied with intense pits on the surface, causing a reduction in the total loss of ribs. It is also noteworthy that corrosion damage on the steel surface is mainly depicted only on the outer half of the steel bar's circumference, adjacent to the free surface of specimens, as Figure 12 shows.

Conclusions
In the current manuscript, an extensive experimental study was conducted on RC elements, where the influence of steel bar's corrosion damage and stirrups spacing on bond behavior between steel and concrete was investigated, in function with the average width of surface concrete cracking. From the present experimental study, the following outcomes were obtained: 1. Corrosion of the main steel bar causes surface concrete cracking, the width of which is closely related to the existence and densification of stirrups. The increased density of stirrups, on the one hand, contributes to the reduction in the range of surface cracking and on the other hand, to the reduction in the bond loss rate in corrosion conditions; 2. In non-corroded conditions, the use of wide stirrups (Φ8/240 mm) enhances the bond strength, in uncorroded conditions, by about 36% in comparison with a group of specimens without stirrups. However, in corrosion conditions, an 8.5% mass loss of reinforcing bar leads to about a 60% reduction in bond strength compared to those of uncorroded specimens; 3. Specimens with quite dense stirrups (Φ8/120 mm) indicated higher bond strength values in noncorroded specimens, and bond strength degrades by about 32% due to an 8.5% mass loss of the reinforcing bar; 4. Dense stirrups spacing, specimens with Φ8/60 mm, ensure a high level of bond behavior, either in terms of bond strength or of residual bond stress, both in uncorroded and corroded conditions.