Bond behaviour evaluation between steel reinforcement and self-healing concrete containing non-axenic biomasses

. Although steel reinforcements are used to withstand tensile forces in concrete, cracks are an unavoidable phenomenon. The presence of cracks, in fact, increases the risk for lowering the service life and durability of concrete structures. A critical issue occurs when due to splitting forces, cracks appear in concrete along the tensioned rebars which damage the bonding between the steel and concrete matrix. As a mitigation plan, the cracks should be healed at short notice and the bonding has to be recovered by the potential use of healing agents. This paper aims to investigate the bond behaviour of steel reinforcement in self-healing concrete. Two biomasses were employed as healing agents namely HTN (bacteria-based) and YEAST (fungi-based). The fresh and hardened properties of the normal and self-healing concretes were initially evaluated. The bond properties were investigated by performing pull-out tests on three different states of concrete: uncracked, cracked, and healed. Results revealed that the additions of biomasses did not induce negative effects on the compressive strength of hardened concrete. Moreover, the average bond strength of uncracked concretes containing HTN and YEAST improved by 20% and 8%, respectively, as compared with normal concrete. The introduction of a crack caused a significant reduction in bond strength regardless of the addition of healing agents. Nevertheless, it was found that the bond strength was slightly recovered after healing under water immersion


Introduction
Self-healing concrete continuously gains interest from both researchers and industries as a result of the improved healing mechanism to prolong the service life of the concrete structure and to improve the durability properties. This self-healing technology is developed to improve the limited autogenic ability of traditional concrete that only can heal small cracks of around 100 µm [1]. Many publications [2][3][4][5] have proven the capabilities of various healing agents to either heal or seal the cracks with an improved healing/sealing efficiency. The precipitated or polymerized healing products are able to close the crack and hence avoid the ingress of water and harmful substances that may jeopardize the concrete structure by corroding the steel reinforcement. In the case of corrosion, the bonding between the steel rebar and concrete matrix will slowly weaken by reducing the cross-section of the steel rebar and its mechanical properties. Thus, it should be guaranteed that these cracks can be healed in a relatively short time, before liquids enter the cracks. It has been proven that self-healing strategies can avoid liquid ingress by deposition of healing products or release of water repellent agents. Nevertheless, the question arises * Corresponding author: elke.gruyaert@kuleuven.be whether the self-healing effect is only efficient to artificially close the crack or it may also contribute to the improvement of the bonding between the steel rebar and the cracked concrete. Mousavi et al. [6][7][8] investigated the bonding at rebar-concrete interface with the addition of superabsorbent polymer (SAP). Results showed that the introduction of SAP improved the bond properties and also accelerated the healing process. The particle size of SAP influenced the bonding performance where the large particle size is more effective in enhancing the bond properties than small particle size. The research on the topic of self-healing concrete with respect to the bonding is still scarce and this has been identified by the authors as a research gap. Eventually, this paper aims to investigate the bond properties of self-healing concrete mixed with two types of non-axenic biomass healing agents. The bond strength at the rebar-concrete interface is evaluated in three different states of concretes (uncracked, cracked and healed). Water immersion is selected as the healing condition for all cracked specimens.  Table 1 shows the concrete mix designs developed in this study. Two healing agents produced by AVECOM nv, namely HTN and YEAST, were used. These agents were classified as non-axenic agents, meaning the liquid agent contains different species of microorganism, not strictly limited to a single species. The HTN was composed of a consortium of bacteria that was dependent on CO2, H2 and acetate. HTN had a good resistance to a high pH environment and this type of biomass may produce a small amount of nitrite. HTN was classified as heterotrophic nitrifiers and the microbial-induced calcite precipitation (MICP) can be achieved through the denitrification process. On the other hand, YEAST involved a consortium of fungi. This agent supplied the urease enzyme that was useful for the ureolysis to favour calcium carbonate precipitation. However, since the agents are in development stage, the exact mechanisms for the microbially induced precipitation is being elucidated. Both healing agents were available in the form of solution and they were used as a full replacement of mixing water. The pH values of these agents were 7 and 9 for YEAST and HTN, respectively. In HTN agent, the solution contained bacteria-based microorganisms with the pure dosage (or active dry content) of 0.05% by weight of cement (given the mix design), while in YEAST agent, the pure dosage (or active dry content) of fungi-based microorganisms was 0.45% by weight of cement (given the mix design). Previous studies by the producer revealed that the addition of biomasses into the cementitious mortar enhanced the healing performance and improved the durability. All in all, it was decided to upscale these agents for use at the concrete level. The fresh properties of mixtures were initially assessed by means of slump, air content and fresh density tests. The mechanical properties were evaluated by performing the compression test on hardened cube specimens (150×150×150 mm) at 28 days. In order to assess the bond properties of the concrete, the pull-out specimens were fabricated as shown in Figure 1. The cylindrical PVC tube was used as a mould to cast the fresh concrete and was simultaneously used as a confinement to control the crack. As a note, the concrete was not demoulded from the tube, thus it was still attached during the cracking or pull-out test. The short pull-out test was aimed in this study, thus a small 50mm long PVC tube was inserted over the steel rebar of Ø10 mm inside the mould to create a debonded section. For each mix design, sixty pull-out specimens were cast and later the specimens were divided into three different states: uncracked, cracked and healed. The uncracked concretes were tested by pull-out tests at 49 days. Some specimens were cracked by a Brazilian splitting test at 49 days and they were tested by pull-out tests at the same day. The other specimens were cracked at 21 days, and then healed under water (22±1°C) for 28 and 112 days. As a note, all specimens were cured under full water immersion. To prevent the corrosion due to the direct contact between steel rebar and water, the protruding steel rebar on the pull-out specimens was covered with aluminium tape. The pull-out test was performed by means of an Instron instrument as shown in Figure 2. The pull-out specimen was initially inserted into the support plate and reaction plate. At the bottom of the reaction plate, a load cell/transducer (Kraftaufnehmer C6A, 500 kN, 2 mV/V) was mounted onto the steel rebar and was tightened by a rebar anchorage. Two LVDTs were placed: one on top of the concrete surface (LVDT 1) and another one on the bottom of the support plate (LVDT 2). The instrument was programmed with the frame displacement rate of 0.05 mm/s. The pull-out test mechanism was based on the fact that with the increase of the load, the setup will induce tensile stresses to the bottom side of the rebar and tensile stresses will be transferred to the concrete specimen by bond stresses along the embedded region The force transducer will measure the load applied to the rebar. The test was stopped when reaching a slip of between 2.5 and 5.0 mm. The bond stress-slip responses were obtained by taking into account the passive displacement or slip from the LVDT 1.   Statistical analysis by ANOVA showed that actually none of the strength differences were statistically significant (level of significance = 0.05, p-values = 0.12 for HTN, 0.99 for YEAST). As a matter of fact, these agents were not developed to increase the strength, but rather aimed to enhance the healing efficiency. As shown in Figure 4, the hardened densities of all mixtures were identical.

Bonding properties
The bond-slip behaviour of steel reinforcement and concrete was obtained from the pull-out tests. In this paper, the ultimate bond strengths were mainly investigated. Figure 5 shows the ultimate bond strengths of the 49-days-old uncracked (UNCR) concretes. The average ultimate bond stress (τu) of REF concrete was registered at 14.2 MPa with employing five repetitions. The inclusions of HTN and YEAST seemed to improve the bond properties by increasing the average ultimate bond strength by 20.4 and 7.6%, respectively, but the differences were not statistically significant (p-values based on ANOVA were 0.55 for HTN and 0.98 for YEAST). This result showed a similar trend as for compressive strength where a (non-significant) improvement was achieved by HTN, followed by YEAST with a minor increment. In order to investigate the effect of cracking, some specimens were cracked at 49 days with three different crack sizes: 200-300, 300-400 and 400-500 µm. The pull-out tests on cracked (CR) specimens showed that there was a significant reduction of bond strength due to the presence of the crack. The strength reduction due to cracking was found to be in the range of 60-80%. Nevertheless, it was also found that the size of the crack did not really influence the obtained results as the bond stresses were statistically not significantly different for all cracked specimens at all crack widths. This indicates that the occurrence of the crack, which crosses the steel reinforcement, induces a critical effect on the bonding. Next, the investigation was continued on the healed (HL) concretes where the remaining specimens were cracked at 21 days with three different size ranges of crack width and they were healed under water for 28 days. The crack closing effect was visible on the healed concretes as shown in Figure 6 with a precipitation of healing products on the crack mouth. Figure 6 also shows the typical formation of healing products coming out from the crack. After 28 days of healing, there was a formation of white minerals at the crack mouth observed at all concretes. Moreover, after 112 days, the healing products formed abundantly. On REF specimens, the crack was fully closed by the healing products that seemed more denser than the products observed at 28 days. On HTN and YEAST specimens, the healing products overfilled the crack and there were formations of long stalactites. It seems that a massive production of minerals coming out of the crack occurred due to the inclusion of biomasses. The pull-out test results on healed specimens were subsequently conducted and the results are summarized in Figure 5. The bond strengths of healed concrete were still relatively low and under 10 MPa and were relatively comparable with the cracked series. This may be attributed due to insufficient healing at a short period (28 days), thus the improvement of bonding is not apparent. Another series of HL concretes was subjected to 112 days of healing and the results on their bond strengths are also summarized in Figure 5. Apparently, the bond strength of healed concrete was still low even after a long healing period and was comparable with the specimens after a short period. Thus, it can be confirmed that a long healing time does not seem to provide a significant improvement on the bond strength. This may be attributed to the fact that the healing products are relatively fragile and these products could not provide a strong connection in the gap between the concrete and embedded steel rebar. Nevertheless, the phenomenon of crack closure is still evident as shown in Figure 6.
To quantify the efficiency of healing agents on the bonding performance at the rebar-concrete interface after healing, a bond-healing improvement (BHI) is defined as follows: ) × 100% (1) In Equation (1), the mean values are considered for ultimate bond stresses for UNCR, CR and HL series. Figure 7 depicts the bond-healing improvement based on different mixtures and various crack widths.

Conclusions
This paper aims to investigate the influences of healing agent on the fresh and hardened properties, and mainly on the bond properties between the steel reinforcement and the self-healing concrete matrix. Two biomass healing agents were used, namely HTN which was based on a bacteria consortium and YEAST which was based on a fungi consortium. The pull-out specimens were prepared with three different states: uncracked, cracked and healed with the aim to investigate the effect of cracking on the bond strength and the effect of healing agents with regard to the improvement of the bonding-healing ability. The following conclusions can be drawn: 1. The workability of fresh concrete was reduced by the addition of HTN, while in contrast, the introduction of YEAST slightly increased the slump. 2. In terms of mechanical properties, the inclusion of HTN slightly increased the compressive strength by 5.8%, while the addition of YEAST had similar mean strength as reference concrete (REF).
3. An improvement of bond strength between the steel reinforcement and uncracked concrete was achieved by HTN with 20.4% increment and YEAST with 7.6% increment (although differences were not statistically significant). 4. The presence of cracks in the range of 200-500 µm showed a critical effect on the bond performance with a significant reduction of ultimate bond strength up to 80%. 5. There was a limited improvement of the bond strength after healing for all concretes. Especially in case of large cracks between 300-500 µm, the addition of HTN and YEAST gave a slightly better improvement than the reference concrete after a long healing time at 112 days.
This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 860006.