Assessment of Fiber Corrosion Influence in the Flexural Performance of Steel Fiber-Reinforced Concrete

Abstract

Fiber corrosion impacts on the mechanical performance of steel fiber reinforced concrete (SFRC) have been considered minor. However, this may be true only for ordinary corrosion conditions. For severe corrosion conditions, such as stray currents, the impacts must be investigated. This study addresses the influence of corrosion at different levels, including severe corrosion, on the flexural performance of SFRC. An experimental study focused on a three-point bending test, considering as variables the corrosion level, the fiber content, and the fiber aspect ratio. It was confirmed that corrosion can shift fiber failure from pullout to rupture, and it was found that corrosion can shorten flexural performance by as much as 80%. Therefore, corrosion impacts, in certain conditions, cannot be considered minor; rather, they have to be considered significant.

1. Introduction

The idea of reinforcing brittle matrices with discrete fibers is somehow intuitive and can even be found in wildlife, whose most emblematic example may be the nest of the Hornero (Furnarius rufus), the engineer bird. Considering mankind, the concept dates from nearly 45 centuries ago—when straw fibers were mixed with mud and then molded in rectangular prisms to be sun-dried and produce improved construction blocks—and has been evolving ever since.

Steel fiber-reinforced concrete (SFRC)—a mixture of a concrete matrix with randomly oriented short steel fibers—is probably the most popular application of the concept. This composite has been traditionally applied to flooring along with tunnel and mine linings [1]. Then, its utilization was extended to the compressive layer of bridge decks, precast concrete industry, nuclear waste vessels, water retention, and seismic-resistant structures, amongst others. Recently, SFRC has started to be utilized as complementary reinforcement in prestressed [2] and reinforced concrete [3,4]. All these applications are grounded in the acknowledged ability of SFRC to control cracking development inside the matrix [5], thus improving the flexural and tensile behavior [6,7,8].

The widespread use of SFRC is naturally supported by extensive research that started in the 1960s [9]. The main outcomes of the research carried out so far are codes and recommendations for the testing and design of the mechanical behavior of SFRC [10,11,12,13].

Durability is an important dimension of the performance of any material. According to Granju and Ullah Balouch [14], this subject, concerning SFRC, has been studied since the early 1990s. The existing studies are mostly devoted to the influence of fibers on transport properties [15] and, concerning fiber corrosion, in cracked SFRC. Dsouza et al. [16] reported an increase in transport properties with steel fiber content, while Krishna and Rao [17] observed lower resistance to acid attack with the increase in fiber content. Neves and Gonçalves [18] found no significant modifications in transport properties with fiber addition, which is in accordance with the results of the experiments carried out by Frazão et al., either in ordinary [19] or self-compacting concrete [20], concerning chlorides. The results of Zhang et al. [21] also show no significant influence of steel fibers on chloride diffusion, up to a volume content of 2%.

The first major publication addressing fiber corrosion effects on the mechanical performance of SFRC is from 1987 and reports improvements in the mechanical properties of cracked SFRC subject to corrosion. This has been ratified in later studies. For instance, Granju and Ullah Balouch [14] observed a strength increase for cracked SFRC subject to corrosion, despite the (light) corrosion of the fibers. The improvement is attributed to the rise in friction between fibers and matrix promoted by the accumulated corrosion products. Marcos-Meson et al. [22] concluded in their literature review that corrosion of fibers, bridging cracks with openings of less than 0.2 mm, is limited to staining and does not cause further cracking or spalling.

Several studies [22,23,24] revealed that fiber corrosion is limited to a certain depth, ranging from 0.1 to 5 mm, but can reach as deep as 30 mm if the concrete is mixed with seawater [25]. However, this limited corrosion causes no detachment or spalling [24], but aesthetic damage. Moreover, improvements in the mechanical performance of uncracked SFRC were also reported [22,26,27]. Contradicting the idea of Marcos-Meson et al. [22] that, in SFRC, corrosion damage, either aesthetical or mechanical, tends to stabilize in time, loss of mechanical performance for longer test periods is disclosed in [19,26,28]. Further, for exposure to simulated marine showers, loss of performance at a constant rate is visible in the charts of Mangat and Gurusamy [27].

The existing knowledge leads us to think that for mild exposure conditions or short periods, the effect of steel fibers on durability may be negligible, while for severe exposure conditions, longer periods, built-in chlorides, or the presence of stray currents, the corrosion of SFRC may be of concern. In fact, the loss of toughness in corroded SFRC was labeled as dramatic by Kosa et al. [29].

This work aims at deepening our knowledge of the effects of severe corrosion (such as stray current-induced corrosion) on the flexural behavior of SFRC. Besides the corrosion level, the influence of fiber content and fiber aspect ratio on flexural performance indicators is also investigated. Other relevant parameters, such as matrix strength and fiber shape, are not addressed.

2. Materials and Methods

The experimental program comprised one concrete mix, reinforced with different fiber aspect ratios and at different fiber volumes. Fiber aspect ratios (l/d) of 65 and 80 were considered, whereas the adopted fiber contents were 40 and 80 kg/m³, corresponding to fiber volumes of 0.5 and 1%. Mix composition is provided in

Table 1. Concrete composition.

Constituent	Content (kg/m3)
Cement I 42.5R [30]	472
Water	174
Gravel 5/15	744
Gravel 0/5	292
Coarse sand	196
Fine sand	484
Superplasticizer	2.68

, while

Table 2. Fiber characteristics.

Fiber	ZP	RC
Length (mm)	35	30
Diameter (mm)	0.5	0.38
Aspect ratio	65	80
Anchorage type	Hooked end	Hooked end
Steel	Low carbon	High carbon

contains information on steel fibers.

Three combinations of fiber type and fiber content were set, and, therefore, 3 composites were produced (

Table 3. Composites characteristics.

Composite	Fiber Type	Fiber Content (kg/m3)
F80ZP	ZP	80
F40ZP	ZP	40
F40RC	RC	40

).

Fibers were added to concrete after the mixture of the remaining constituents. After mixing, the workability of the composite was assessed through the slump test [31]. Then cubic (150 mm edge) and prismatic (100 × 100 × 500 mm³) specimens were molded and compacted on a vibrating table. Demolding occurred 48 h after casting, and water curing at 20 °C was provided, following the RILEM recommendation [32].

At the age of 60 days, 3 cubes of each composite were subject to compressive strength test [33].

After 6 months of curing, half of the prismatic specimens underwent an accelerated corrosion process. As in [2], a constant imposed current was applied. The procedure was as follows: one week before the acceleration process, the specimens were taken out of water to allow them to dry. Twenty-four hours before turning on the imposed current, the prisms were immersed in a 3.5 wt% NaCl solution. Then, the prisms, lying in a horizontal position, were slightly lifted to level the solution near the top face without covering it and to allow the placing of a titanium mesh under the bottom surface (Figure 1a) to act as an electrode, while at the upper surface another titanium mesh, wrapped in a damp cloth, served as a second electrode. Both meshes were connected to a power supply, making SFRC carry the electrical current when the circuit was on (Figure 1b). The current intensity was set to 1 A and kept for one week for intermediate corrosion (IC) and two weeks for severe corrosion (SC). When the accelerated corrosion process stopped, the prisms were left to dry in a laboratory room at 20 °C and 65% RH (Figure 1c).

The remaining prismatic specimens were kept free of corrosion (NC) until flexural testing.

A three-point bending test, run by displacement, was applied to corroded and uncorroded SFRC beams. Load and displacement were monitored and recorded. The displacement rate was set to 0.01 mm/s, as in [2] and within the range of rates adopted in other flexural tests [20,34]. Vertical displacement was measured by means of LVDT placed at mid-span, in opposite faces, and at 100 mm from mid-span for each side, only in one face. The load measurements were carried out by means of load cells at the load application point and supports (Figure 2).

Load and displacement values were recorded using a data logger, run by specific software.

After flexural testing, fibers were collected from specimens (corroded and non-corroded), and the attached matrix particles were removed. Afterwards, the different sets of fibers were weighed to evaluate the respective mass loss caused by corrosion. Then the relative mass loss was adopted as a corrosion degree indicator. Although in [35,36], the sets were composed of 10 fibers, in this work, a higher number (30) was adopted, thereby reducing the error in mass loss assessment.

3. Results and Discussion

3.1. Workability

The slump values are presented in Figure 3.

All batches of the concrete mix were S3 class [37]. As expected, adding fibers to concrete decreased its workability. The highest relative decrease was observed for composite F80ZP, which has a higher fiber content. Although composites F40ZP and F40RC have the same fiber content, the fibers in composite F40RC have a higher aspect ratio (Table 2). Thus, there is a higher number of fibers in F40RC than in F40ZP, which explains the higher relative workability loss in composite B than in composite C, which agrees with the findings in [38].

3.2. Compressive Strength

Compressive strength was assessed on uncorroded specimens at the age of 60 days. The average values of 3 specimens for each composite were 75.5, 76.5, and 73.6 MPa for composites F80ZP, F40ZP, and F40RC, respectively. The compressive strength of all composites was similar, indicating that fibers do not have a significant influence on compressive strength [9]. Still, it is worth noticing that the decrease in compressive strength with the fiber content was also observed in [4].

3.3. Flexural Response

As mentioned in Section 2, data acquisition was carried out from 3 load cells and 4 LVDT. All specimens developed a first and major crack near mid-span (Figure 4). In the first step, the support reactions and displacements at 100 mm from the supports were used to cross-check the validity of the test and the data to be used in the analysis. Figure 5 shows an example of the crosscheck (corresponding to data from testing uncorroded F40RC).

Similar plots were obtained for all tests. The reaction at the supports corresponds to half of the applied load, and the overlap of symmetrical measurements is remarkable. It is also worth noticing the similarity of displacements on opposite faces.

In a second step, the net deflection at midspan was computed, using the intermediate deflections and beams theory equations.

Finally, the load–displacement curves for each composite (F80ZP, F40ZP, and F40RC) and corrosion condition (uncorroded (NC), intermediate (IC) and severe (SC)) were plotted. Further, flexural performance indicators, summarized in

Table 4. Flexural performance indicators for F80ZP (relative variation between brackets).

Condition	NC	IC	SC
Pcr (kN)	9.2	8.1 (−12%)	3.1 (−66%)
Pmax (kN)	15.4	13.3 (−13%)	5.8 (−63%)
i	17.9	15.3 (−14%)	4.0 (−78%)
I5	2.3	8.8 (+288%)	11.6 (+411%)
Tb (J)	31	21 (−32%)	6.5 (−79%)
fMOR (MPa)	18.5	16.0 (−13%)	6.9 (−63%)
fe (MPa)	4.6	3.2 (−32%)	1.0 (−79%)

,

Table 5. Flexural performance indicators for F40ZP (relative variation between brackets).

Condition	NC	IC	SC
Pcr (kN)	6.1	5.9 (−3%)	3.2 (−48%)
Pmax (kN)	10.3	8.1 (−21%)	3.8 (−63%)
i	6.1	11.2 (+83%)	2.6 (−57%)
I5	8.9	7.0 (−21%)	7.8 (−12%)
Tb (J)	20	12.1 (−39%)	2.8 (−86%)
fMOR (MPa)	12.4	9.7 (−21%)	4.5 (−63%)
fe (MPa)	3.0	1.8 (−39%)	0.4 (−86%)

and

Table 6. Flexural performance indicators for F40RC (relative variation between brackets).

Condition	NC	IC	SC
Pcr (kN)	6.3	5.8 (−8%)	4.0 (−39%)
Pmax (kN)	8.9	10.7 (+20%)	7.2 (−19%)
i	4.3	10.5 (+145%)	21 (+384%)
I5	2.9	7.4 (+157%)	5.2 (+81%)
Tb (J)	21	18 (−16%)	13 (−38%)
fMOR (MPa)	10.7	12.8 (+20%)	8.6 (−19%)
fe (MPa)	3.2	2.7 (−16%)	2.0 (−38%)

, were computed.

A displacement ductility index, i, was computed, following the definition of [39] (Figure 6), along with the ASTM [11] toughness index I₅ (Figure 7a) and the JSCE [40] flexural toughness T_b (Figure 7b).

Besides these indexes, peak load P_max, modulus of rupture f_MOR (Equation (1)), and equivalent flexural strength f_e (Equation (2)) were also considered.

$$f_{\mathit{MOR}}=\frac{3P_{\mathit{max}}L}{b{h}^2},$$

(1)

where P_max is the peak load, L is the distance between supports, and b and h are the width and height of the specimen’s cross section, respectively.

$$f_e=\frac{150T_b}{b{h}^2},$$

(2)

where T_b is JSCE flexural toughness, and b and h are the width and height of the specimen’s cross-section, respectively.

The results of these performance indicators are presented in Table 4.

Figure 8 depicts the flexural behavior of the different composites at the same corrosion level. The composite with a higher fiber content (F80ZP) exhibited the maximum peak load, which is in accordance with the conclusions of the statistical analysis of the flexural strength of SFRC, i.e., the most influential parameter in SFRC flexural strength is the fiber content. Composites F40ZP and F40RC, with identical fiber content, showed similar peak loads. The equivalent flexural strengths of non-corroded composites with a fiber content of 40 kg/m³ are comparable to those of Soutsos et al. [41], while the modulus of rupture range is similar to those reported in [27,42], and fits the prediction obtained through the model presented in [43].

The results in non-corroded composites are in agreement with the finding of Shafighfard et al. [7], i.e., fiber volume is the dominant parameter on the peak load.

Under moderate corrosion, the flexural performance rank follows the reinforcing index, RI, rank.

$$\mathrm{RI}=V_f\frac{l}{d},$$

(3)

where V_f is the fiber content (vol./vol.), and l and d are the fiber length and diameter, respectively.

Concerning severe corrosion conditions, the best-performing composite was F40RC, while there is a notorious loss of bending stiffness for F80ZP and F40 ZP.

Figure 9 depicts the flexural behavior of the same composite at different corrosion levels.

The loss of flexural performance with increasing corrosion level is remarkable for composites with ZP fibers (F80ZP and F40ZP), while for RC fibers (F40RC), it is less pronounced. Nevertheless, the performance loss is masked by the behavior of F40RC between cracking and the maximum load for the uncorroded condition. This is attributed to the fact that for light corrosion, the stresses built up by the corrosion products, along with the irregularities created at the fiber surface, enhance the friction between the fiber and matrix [14,26], increasing the gap between P_cr and P_max.

Apart from this, the loss of flexural performance increases with the level of corrosion.

All corroded specimens showed a lower peak load than their uncorroded companions. Both the peak load and the cracking load showed a negative association with the corrosion level, i.e., they decreased with the increase of the corrosion level. The same is valid for JSCE toughness. For f_MOR and f_e, the relative variation is equivalent to that of P_max and T_b, respectively, as there is a linear relationship between them.

The losses on T_b for intermediate corrosion are comparable to those obtained by Zhang et al. [44] in SFRC mixed with seawater, while the losses in flexural strength are comparable to those found in [45]. For severe corrosion, the relative losses in flexural toughness and strength are in line with the results reported in [26] and with those reported by Frazão et al. [20] for diametral compression, in SFRC subject to severe corrosion also through imposed current.

The variation in the displacement ductility and toughness indexes with the corrosion level was not consistent. Shafighfard et al. [7] reported that, concerning fiber reinforcement, the displacement ductility index is mostly ruled by the fiber volume, which is in agreement with the results for the non-corroded composites. The same authors defend that the most influential factor on the displacement ductility index is the peak load. As peak load decreases with increasing corrosion level and the corrosion impact in the pre-peak response is higher than in the post-peak response, the displacement ductility index turns out to be a bad indicator of corrosion influence on flexural performance. The influence of corrosion in the pre-peak response is shaped in Figure 9, with noticeable lower flexural stiffness of the composites under severe corrosion. This can be attributed to both the microcracks caused by the volume increase due to corrosion products, the corrosion products acting as voids, given their non-cohesive nature, and the decrease in effective fiber volume.

The residual load bearing capacity at a midspan deflection of l/150, already used by Sakthivel and Aravind [46], showed to be consistent with the corrosion level as well as with the reinforcing index.

3.4. Corrosion Degree

Aiming at quantifying the corrosion level, an indicator of corrosion degree, as detailed in Section 2, was used. The corrosion degree (CD) was obtained through the following equation:

$$CD(\%)=\frac{m_u-m_c}{m_u}\times 100$$

(4)

where m_u is the mass of 30 uncorroded fibers and m_s is the mass of 30 corroded fibers.

The results are summarized in

Table 7. Fiber corrosion degree.

Condition	IC	SC
F80ZP	2.07%	2.65%
F40ZP	4.76%	4.04%
F40RC	4.91%	3.02%

.

For intermediate corrosion levels, the variation of CD is consistent with the fact that for the same current intensity, the higher the content of steel, the lower the relative corrosion-induced mass loss. For the same steel content, the higher the specific surface, or the lower the diameter, is, the higher the relative mass loss.

The CD for severe corrosion is not higher than CD for intermediate corrosion, except for F80ZP, while it would be expected to be higher considering the visual aspect of the specimens, the longer period of impressed current, and even the lower flexural performance. The cause of these apparently contradictory results lies in the fact that for severe corrosion, it was difficult to integrate fibers at the most stressed regions, i.e., outer regions. So, fiber had to be collected from inner regions, where the corrosion process was less intense. This did not happen for F80ZP, as the CD was lower. Jang and Yoo [36] found that for a corrosion degree above 2%, the fiber tends to fail rather by fiber rupture than by fiber pullout. This phenomenon will be more pronounced with the decrease in fiber diameter, which is in agreement with the fact that the CD in F40RC is lower than in F40ZP.

4. Conclusions

The literature indicates that corrosion in SFRC is less serious than in conventionally reinforced concrete. Also, the results of short-term experiences have shown limited corrosion-induced damage to SFRC. This causes the issue of stray current-induced corrosion, which has been overlooked, together with other particular conditions that can cause severe corrosion.

The influence of corrosion on the flexural behavior of SFRC was studied. It was found that:

• Severe corrosion may cause a loss of maximum bearing capacity of up to 65%, and up to 81% in flexural toughness or equivalent flexural strength.
• In general, the highest losses were found in the composite with a higher fiber content. Composites with RC fibers (smaller diameter) showed the lowest losses in maximum bearing capacity and flexural toughness.
• It was confirmed that beyond a certain corrosion threshold, the fiber failure shifts from pullout to rupture.
• The displacement ductility and toughness indexes did not show an association with the corrosion level.
• The attained corrosion levels, even those considered intermediate, caused a decrease in flexural performance, as evaluated per cracking load, peak load, toughness, modulus of rupture, and equivalent flexural strength indicators.

Even though, in other investigations, an improvement in flexural performance for light corrosion is reported, this study proved that fiber corrosion may have a quite significant impact on the flexural behavior of SFRC; thus, it is a factor that cannot be neglected, even if a low quantitative level of fiber corrosion is attained. The flexural performance improvement reported in the literature for mild corrosion is attributed to the fact that the smaller volume of generated corrosion products was enough to build up fiber confinement but insufficient to cause microcracking around corroded fiber, as happened to a certain degree with thinner fibers in this work.

For future research, it is considered interesting to:

• Assess the implications of equivalent corrosion levels in the mechanical properties of concrete reinforced with conventional steel bars having the same volume of steel, allowing a fair comparison between the impacts of steel corrosion in SFRC and conventional reinforced concrete.
• Cause mild corrosion in the composites used in this research, using the corrosion techniques found in the literature, to broaden the quantification of corrosion influence in the flexural performance of steel fiber-reinforced concrete for a wider range of corrosion severity.
• Consider other than hooked fiber shapes and different fiber tensile strengths (namely fibers recycled from decommissioned steel elements), to investigate whether, under a certain corrosion degree, fiber failure occurs by fiber rupture or pull-out.