A review of five years’ experience on the effect of fibers on the autogenous healing of Ultra High-Performance Fiber-Reinforced Concrete

1. INTRODUCTION

Self-healing materials possess the ability to repair damage autonomously or with minimal external intervention. In construction, the use of self-healing materials is expected to enhance the longevity of structures and reduce maintenance and repair needs (1De BelieN, GruyaertE, Al-TabbaaA, AntonaciP, BaeraC, BajareD, JonkersHM. 2018. A review of self-healing concrete for damage management of structures. Adv. Mater. Interfaces.5(17):1800074. 10.1002/admi.201800074). Reinforced concrete structures often experience small cracks (less than 0.3 mm) due to the material’s low tensile strength, and such cracks are considered normal in many structures, as specified in Eurocode 2. Concrete has a natural self-healing ability, known as autogenous healing, which allows it to seal small cracks through ongoing hydration and carbonation processes (1De BelieN, GruyaertE, Al-TabbaaA, AntonaciP, BaeraC, BajareD, JonkersHM. 2018. A review of self-healing concrete for damage management of structures. Adv. Mater. Interfaces.5(17):1800074. 10.1002/admi.201800074). The healing process has been known for years, especially in concrete water reservoirs and pipes, where crack sealing and leakage stop were detected. The first mention to the phenomena is usually attributed to the French Academy of Sciences (1De BelieN, GruyaertE, Al-TabbaaA, AntonaciP, BaeraC, BajareD, JonkersHM. 2018. A review of self-healing concrete for damage management of structures. Adv. Mater. Interfaces.5(17):1800074. 10.1002/admi.201800074, 2LauerKR. 1956. Autogenous healing of cement paste. In Journal Proceedings Vol. 52, (6): 1083-1098. 10.14359/11661). Well-known technical books also described the phenomena with different names, such as “cicatrization” in (3CánovasMF. 1994. Patología y terapéutica del hormigón armado. Colegio de Ingenieros de Caminos, Canales y Puertos, Servicio de Publicaciones. ISBN 8474932025.), and is always attributed to cracks below 0.2-0.3 mm, and thus, within the allowable crack width values.

Concrete autogenous healing is typically disregarded in the design of typical concrete structures, but it is considered in Eurocode 2 in part 3, EN 1992-3 “Liquid retaining and containment structures”,, which relates to design of watertight concrete structures, usually exposed to one-sided water pressure (such as basements, tunnels and water tanks). EN 1992-3 allows certain crack width values for tightness class 1, depending on the water pressure and thickness of the reinforced concrete wall, with values generally between 0.05 and 0.2 mm, given the requirement that sealing of the cracks would occur in a short-time, which, however, is not always guaranteed.

Recent research has focused on improving autogenous healing (4EdvardsenC. 1999. Water permeability and autogenous healing of cracks in concrete. ACI Materials Journal. 96(4): pp. 448-454. Thomas Telford Publishing. 10.14359/645, 5Ter HeideN, SchlangenE, van BreugelK. 2005. Experimental study of crack healing of early age cracks. Department of Civil Engineering, Technical University of Denmark.), designing specialized products to promote self-healing (autonomous healing) (1De BelieN, GruyaertE, Al-TabbaaA, AntonaciP, BaeraC, BajareD, JonkersHM. 2018. A review of self-healing concrete for damage management of structures. Adv. Mater. Interfaces.5(17):1800074. 10.1002/admi.201800074, 6Roig-FloresM, FormaginiS, SernaP. 2021. Self-healing concrete-what is it good for?Mater. Construcc.71(341):e237. 10.3989/mc.2021.07320), or better understanding these processes. It has been widely reported that the visual closure of cracks does not always guarantee improved durability (7Roig-FloresM, PirritanoF, SernaP, FerraraL. 2016. Effect of crystalline admixtures on the self-healing capability of early-age concrete studied by means of permeability and crack closing tests. Constr. Build. Mater.114:447-457. 10.1016/j.conbuildmat.2016.03.196, 8FerraraL, Van MullemT, AlonsoMC, AntonaciP, BorgRP, CuencaE, De Belie, N. 2018. Experimental characterization of the self-healing capacity of cement based materials and its effects on the material performance: A state of the art report by COST Action SARCOS WG2. Constr. Build. Mater.167:115-142. 10.1016/j.conbuildmat.2018.01.143.), making the study of self-healing in terms of transport properties crucial.

From the three mechanisms involved in the transport of fluids, permeability, diffusion, and sorption (9NevilleAM. 1995. Properties of concrete (Vol. 4, p. 1995). London: Longman. ISBN: 978-0-273-75580-7), the most relevant mechanism will be different depending on the specific civil work and its contact with water. There are several civil constructions where permeability is the main mechanism with different range of values of water pressure, such as dams (~50-100 meters of water column, mWc), water reservoirs (1-2), concrete pipes (0.5-1 mWc) or even variable mWc in the case of the groundwater level, a situation that can be found in many foundations or constructions under the ground level such as a parking.

Traditionally, concrete permeability has been evaluated in the uncracked state following EN 12390-8. This approach enables the assessment of water penetration depth and whether it reaches the reinforcement. In specimens with cracks, the flow of fluid through the sample under a specific pressure can be evaluated. Testing methods for cracked or damaged samples are quite similar to those used for porous materials in fields like soil mechanics or hydrogeology, where fluid flow is measured using either the constant head or falling head methods (figure 1). From these tests, a permeability coefficient (k) is estimated using Darcy’s law, with slightly different equation depending on the type of test. However, it should be mentioned, that Darcy’s law was formulated for porous mediums and its applicability may be limited in concreted cracked specimens where other phenomena (such as turbulences) may occur. Other classifications on the type of permeability tests can be found in the literature (10DesmettreC, CharronJP. 2011. Novel water permeability device for reinforced concrete under load. Mater. Struct.44:1713-1723. 10.1617/s11527-011-9729-6), such as a differentiation between discrete tests (healing occurs between tests) and continuous permeability tests (where healing occurs simultaneously to the test).

 a)

b)

Figure 1 Different setups for testing water permeability in soils, a) constant head water permeability and b) falling head water permeability.

For the evaluation of self-healing through transport properties, there is a need of inducing a previous damage (usually a crack), unlike with mechanical tests. Usually, water permeability in cracked conditions is measured after inducing these cracks.

Previous authors (4EdvardsenC. 1999. Water permeability and autogenous healing of cracks in concrete. ACI Materials Journal. 96(4): pp. 448-454. Thomas Telford Publishing. 10.14359/645) designed a test apparatus capable of inducing tensile cracks in concrete samples and assessing water permeability under varying pressure levels (ranging from 2.5-20 mWc), and in which healing took place simultaneously with the permeability test. Their findings indicate that specimens subjected to lower water pressures while healing exhibited improved healing rates. Similar setups working at high (~20 mWc) have also been used later on in the literature (7Roig-FloresM, PirritanoF, SernaP, FerraraL. 2016. Effect of crystalline admixtures on the self-healing capability of early-age concrete studied by means of permeability and crack closing tests. Constr. Build. Mater.114:447-457. 10.1016/j.conbuildmat.2016.03.196). In the case of test setups working at low water pressures, other authors (11AldeaCM, ShahSP, KarrA. 1999. Effect of cracking on water and chloride permeability of concrete. J. Mater. Civ. Eng.11(3):181-187. 10.1061/(ASCE)MT.1943-5533.0002050, 12Van Tittelboom, K, De Belie, N, De Muynck, W, Verstraete, W. (2010). Use of bacteria to repair cracks in concreteCem. Concr. Res.40(1):157-166. 10.1016/j.cemconres.2009.08.025) used a method based on sealing PVC ring and applying a water-filled tube system with a water pressure of around 0.3 mWc. In this case, the water permeability is evaluated as the reduction of water height in the pipette. In the end, the test selection should depend the application and objectives.

Ultra-High-Performance Fiber Reinforced Concrete (UHPFRC) is a cementitious material that combines self-compacting behavior with excellent mechanical, and durability properties, and shows high ductility and tensile strength (13HabelK, GauvreauP. 2008. Response of ultra-high performance fiber reinforced concrete (UHPFRC) to impact and static loading. Cem. Concr. Compos.30(10):938-946. 10.1016/j.cemconcomp.2008.09.001.). Its excellent properties are attributed to a dense matrix, low w/c and high content of micro steel fibers (14CharronJP, DenariéE, BrühwilerE. 2007. Permeability of ultra high performance fiber reinforced concretes (UHPFRC) under high stresses. Mater. Struct.40:269-277. 10.1617/s11527-006-9105-0). UHPFRC mixes with high contents of fibers and appropriate mix design may have multiple cracking behavior, thanks to a high stress redistribution that produces several narrow microcracks. The porosity accessible to water in UHPFRCs without cracks at 28 days can be 7 and 4 times lower than a 30 MPa and a 90 MPa concrete, respectively (15ValcuendeM, Lliso-FerrandoJR, Roig-FloresM, Gandía-RomeroJ. M.2021. Porous Structure of ultra-high-performance fibre-reinforced concretes. Materials.14(7):1637. 10.3390/ma14071637, 16SohailMG, KahramanR, Al NuaimiN, GencturkB, AlnahhalW. 2021. Durability characteristics of high and ultra-high performance concretes. J. Build. Eng.33:101669. 10.1016/j.jobe.2020.101669).

Previous studies (17NegriniA, Roig-FloresM, Mezquida-AlcarazEJ, FerraraL, SernaP. 2019. Effect of crack pattern on the self-healing capability in traditional, HPC and UHPFRC concretes measured by water and chloride permeability. In MATEC Web of Conferences289: 1-8. 10.1051/matecconf/201928901006) investigated UHPFRC beams loaded with increasing levels of strain in order to produce multiple cracks, representing service and failure conditions. A threshold was detected around the residual strain of 1.3‰, if strains were lower, no differences were detected in the water permeability coefficient of UHPFRC. This value of strain was approximately equivalent to a total crack width of around 130 μm distributed in 100 mm. Similarly, previous research in cracked conditions with 20 mWc water pressure (16SohailMG, KahramanR, Al NuaimiN, GencturkB, AlnahhalW. 2021. Durability characteristics of high and ultra-high performance concretes. J. Build. Eng.33:101669. 10.1016/j.jobe.2020.101669) could not detect water movement through 25 μm cracks, and only extremely low movement for 50 μm cracks. As a result, a modified water permeability test was developed (17NegriniA, Roig-FloresM, Mezquida-AlcarazEJ, FerraraL, SernaP. 2019. Effect of crack pattern on the self-healing capability in traditional, HPC and UHPFRC concretes measured by water and chloride permeability. In MATEC Web of Conferences289: 1-8. 10.1051/matecconf/201928901006, 18DoostkamiH, Roig-FloresM, SernaP. 2021. Self-healing efficiency of Ultra High-Performance Fiber-Reinforced Concrete through permeability to chlorides. Constr. Build. Mater.310:125168. 10.1016/j.conbuildmat.2021.125168), using sodium chloride and its reaction with silver nitrate as an indicator to identify areas where water could penetrate.

The lack of water flow reported in (17NegriniA, Roig-FloresM, Mezquida-AlcarazEJ, FerraraL, SernaP. 2019. Effect of crack pattern on the self-healing capability in traditional, HPC and UHPFRC concretes measured by water and chloride permeability. In MATEC Web of Conferences289: 1-8. 10.1051/matecconf/201928901006, 18DoostkamiH, Roig-FloresM, SernaP. 2021. Self-healing efficiency of Ultra High-Performance Fiber-Reinforced Concrete through permeability to chlorides. Constr. Build. Mater.310:125168. 10.1016/j.conbuildmat.2021.125168) can be a consequence of the cracking methodology in reinforced beams, which makes the crack width to be reduced towards the interior of the beam, typically in a “V” shape. However, other methodologies, such as those proposed in COST Action Sarcos (19Roig-FloresM, DoostkamiH, AlonsoMC, AmmarL, BajareD, Beltran-CobosR, SernaP. (2022). Evaluation of the self-healing efficiency of concrete with a crystalline admixture: interlaboratory analysis from COST SARCOS RRT3 group.) used concrete disks cracked by splitting tests with completely passing cracks in most of the specimens. Therefore, this study aims to compare the autogenous healing capability of UHPFRC with two fiber types and dosages, through crack closure and water permeability. In addition, protection against chloride penetration after healing was evaluated in disks with completely passing cracks after healing.

2. MATERIALS AND METHODS

2.2. Methodology

 ⌅

2.2.1. Pre-cracking and crack width measurement

 ⌅

To assess the self-healing capability of the four mixes, 28-day-old disks with dimensions of Ø100×50 mm were pre-cracked using a splitting test. During the pre-cracking process, the load increased progressively and crack width in the loaded state was measured manually. Once the crack reached the desired size, the specimens were unloaded and the residual crack width was evaluated. The fibers added to the mixes allow controlling the cracking process until the target crack width is achieved. In this work the goal was to create residual cracks ranging between 100 and 450 µm. The width of these cracks was measured the same day after pre-cracking and then again after the samples healed under various conditions. For every healing condition, at least six specimens were cracked.

The crack width was measured at six points on each disk (three equidistant points on each plane surface). The measuring points are marked yellow (see Figure 2). The results show that 100% of the cracks on the U40 disks were single cracks without (or with minimal) branching, as shown in Figure 2 (left). Conversely, 70% of the measurement points on the U160 disks had several crack branches, i.e. a multi-cracking pattern. Therefore, more than one crack was present at most of the measurement points (Figure 2, right).

Figure 2 Single crack produced in the U40 mix (left) and several branches in U160 mix (right). 

Since crack width was measured in all the individual cracks before and after healing, the parameter of crack closure was evaluated for each individual crack. However, as water and chloride permeability tests were conducted on the entire disk, each disk’s representative initial crack width was calculated. To obtain this representative value, the crack width at each measurement point was determined by summing the widths of all the crack branches at each point. The overall representative crack width for the disk was then calculated as the average of all the width measurements from the six points.

2.2.2. Evaluation of the self-healing efficiency

 ⌅

Two parameters were used for the evaluation of self-healing efficiency: the crack Closure Ratio (CR) and the Healing Ratio (HR). CR is obtained using the crack width values of each individual crack width before and after the healing process (Equation [1]). HR is obtained from the low pressure water permeability tests results, performed on cracked disks following the methodology explained in detail in previous studies (8FerraraL, Van MullemT, AlonsoMC, AntonaciP, BorgRP, CuencaE, De Belie, N. 2018. Experimental characterization of the self-healing capacity of cement based materials and its effects on the material performance: A state of the art report by COST Action SARCOS WG2. Constr. Build. Mater.167:115-142. 10.1016/j.conbuildmat.2018.01.143.). HR was obtained as the ratio of water permeability, before and after the healing process, following Equation [2].

$\mathrm{CR}=1-\frac{\mathrm{Crack\; width\; after\; healing}}{\mathrm{Crack\; width\; before\; healing}}$ [1]  

$\mathrm{HR}=1-\frac{\mathrm{Water\; head\; reduction\; after\; healing}}{\mathrm{Water\; head\; reduction\; before\; healing}}$ [2]  

Afterwards, specimens are exposed to the healing condition: a) W28, 28 days of water immersion at 20°C, and H28, 28 days at 20°C and 95% RH.

2.2.3. Protection against chloride penetration

 ⌅

After the healing period under the two healing conditions and the completion of the final water permeability test, an assessment of chloride penetration through the healed cracks was performed. The method used was a modified chloride permeability test, as proposed in a previous study (18DoostkamiH, Roig-FloresM, SernaP. 2021. Self-healing efficiency of Ultra High-Performance Fiber-Reinforced Concrete through permeability to chlorides. Constr. Build. Mater.310:125168. 10.1016/j.conbuildmat.2021.125168). The results of this method revealed two main penetration zones: penetration through the matrix, labeled P₀ and penetration through the crack, labeled W. To measure P₀, four points were marked at 20 mm intervals along both faces of the disk. To measure W, three points were measured, maintaining a 20 mm distance from the surface of exposure to chlorides, in order to avoid the influence of matrix penetration.

3. RESULTS AND DISCUSSION

3.1. Crack closure

 ⌅

Figure 3 and Figure 4 show Crack Closure Ratio (CR) obtained after healing, depending on the initial crack width of each individual crack for both the U160 mix (blue points), U40 mix (orange points). Figure 3 shows the results healing in water immersion for 28 days (W28) and Figure 4 those healing in humidity chamber for 28 days (H28).

Figure 3 Crack closure ratio vs. initial crack width for every crack of specimens healed in water immersion for 28 days. 

Figure 4 Crack closure ratio vs. initial crack width for every crack of specimens healed in humidity chamber for 28 days. 

Disks from the U160 mix showed distributed cracking with a higher number of thinner cracks, therefore, the CR parameter was analyzed in a higher number of cracks in the U160 mix. U160 mix had perfect CR in the W28 conditions for most cracks below 360 μm. However, some cracks remained open with CR=0. When U160 disks heal in the H28 conditions, the CR result is dispersed and below 63% in most of the cases, with only two cracks with perfect CR (100%). In the case of the U40 mix healed in W28, it showed several cracks with perfect healing, some cracks with disperse healing between 6 and 60% healing and some cracks that remained completely open CR=0. In the case of U40, healing in H28, the cloud of values obtained is between 0 and 84%, and none of the cracks had perfect healing. Comparing these results with the literature (20Lo MonteF, RepesaL, SnoeckD, DoostkamiH, Roig-FloresM, JacksonS, Blanco AlvarezA, NasnerM, Paul BorgR, SchröflC, GiménezM, AlonsoMC, Serna, P, De BelieN, FerraraL. 2024. Multi-performance experimental assessment of autogenous and crystalline admixture-stimulated self-healing in UHPFRCCs: Validation and reliability analysis through an inter-laboratory study. Cem. Concr. Compos.145:105315. 10.1016/j.cemconcomp.2023.105315), the mix U120 had most cracks with a lower initial width, around 100 μm, and they obtained a wide dispersion of crack closure ratios, with most of the values above 60%. Other research analyzing UHPFRC with very thin cracks (21CuencaE, D'AmbrosioL, LizunovD, TretjakovA, Volobujeva O, FerraraL. 2021. Mechanical properties and self-healing capacity of Ultra High Performance Fibre Reinforced Concrete with alumina nano-fibres: Tailoring Ultra High Durability Concrete for aggressive exposure scenarios. Cem. Concr. Compos.118:103956. 10.1016/j.cemconcomp.2021.103956), below 20 μm, had dispersed values between 60 and 90%, for specimens healing in humidity chamber for 28 days and over 70% for specimens healing in geothermal water.

3.2. Water permeability

 ⌅

Figure 5 shows HR depending on the initial crack width and the healing condition. The water permeability test was conducted on the entire disk, therefore, a representative crack width was calculated for each disk as the initial crack width value. The results show that U160 had improved healing when healed in W28, or disks with a representative crack width of less than 150 μm. However, disks with wider cracks displayed more variable results, and some disks obtained HR values of zero. In the case of the mix U40, the results are below 50% even for disks with representative cracks below 100 μm when healing in W28 and below 10% when healing in H28.

Figure 5 Crack healing ratio for every disk vs. the average initial crack width of the disk. 

Most published studies focus on healing through one parameter, but information on how the different healing indices compare offers meaningful insights into the long-term durability of structures. Figure 6 compares global crack closure of each specimen (average of all the cracks) with its healing ratio. Specimens healed in water immersion are generally aligned in the bisecting line (CR=HR) or they are U160 specimens with high CR ratio and low HR ratio. Specimens healed in humidity chamber are mainly gathered in the area of almost zero CR and almost zero HR. This observation indicates is consistent with findings from previous studies (18DoostkamiH, Roig-FloresM, SernaP. 2021. Self-healing efficiency of Ultra High-Performance Fiber-Reinforced Concrete through permeability to chlorides. Constr. Build. Mater.310:125168. 10.1016/j.conbuildmat.2021.125168), which suggest a direct correlation between crack closure and healing ratios when healing occurs under water immersion conditions. In addition, the lower values of global crack closure when compared with the individual crack closure of single cracks, indicate that healing may occur at specific locations of an element without improving its overall behavior. Other works on UHPFRC (21CuencaE, D'AmbrosioL, LizunovD, TretjakovA, Volobujeva O, FerraraL. 2021. Mechanical properties and self-healing capacity of Ultra High Performance Fibre Reinforced Concrete with alumina nano-fibres: Tailoring Ultra High Durability Concrete for aggressive exposure scenarios. Cem. Concr. Compos.118:103956. 10.1016/j.cemconcomp.2021.103956) also showed good correlation comparing these two types of self-healing with most of the results placed on the bisecting line.

Figure 6 Healing ratio vs. crack closure ratio depending on the fiber content and healing condition. 

3.3. Chloride penetration in healed cracks

 ⌅

Even though the crack closure and healing ratios for U160 healed in water immersion (W28) were almost 100%, the chloride penetration patterns indicated that chlorides still managed to penetrate through the cracks to some extent, regardless of whether they exhibited single or multiple cracking patterns. Figure 7 shows the cracking and chloride penetration patterns revealed by silver nitrate pigmentation.

Figure 7 Example of chloride penetration and crack patterns of disks healed in H28, U40 mix (left) and U160 mix (right). 

Chloride penetration was quantified through the average P₀ and W values for each disk under all mixes and healing conditions.

Regarding the parameter of matrix penetration (P₀), the results indicated that the U160 mix had 0 mm or not detectable penetration, regardless of the healing condition. This lack of penetration was also observed in uncracked reference disks. These findings are consistent with previous literature publications (18DoostkamiH, Roig-FloresM, SernaP. 2021. Self-healing efficiency of Ultra High-Performance Fiber-Reinforced Concrete through permeability to chlorides. Constr. Build. Mater.310:125168. 10.1016/j.conbuildmat.2021.125168), which reported complete resistance to matrix penetration in UHPFRC containing 160 or even 130 kg/m³ of short steel fibers after healing in water immersion condition. On the other hand, the P₀ values for the U40 mix were higher, despite having the same UHPFRC matrix composition. The results of penetration when healing in water immersion were an avg. of 0.92 mm (std. dev. 1.14 mm), while those healed in the humidity chamber had an avg. P₀ of 2.48 mm (std. dev. 2.52 mm).

Figure 8 presents the results of W, the depth of chloride penetration through the cracks. The U160 disks healed in W28 showed full resistance to chloride penetration, for cracks < 300 μm. In contrast, the W values for U40 disks healed in W28 varied from 4-6 mm for initial crack widths between 100-300 μm. For U160 and U40 disks cured in H28, both mixes showed similar W values ranging from 0-2 mm.

Figure 8 Chloride penetration depth through the crack (W) vs. initial crack width, depending on the mix and healing condition. 

Regarding previous results (18DoostkamiH, Roig-FloresM, SernaP. 2021. Self-healing efficiency of Ultra High-Performance Fiber-Reinforced Concrete through permeability to chlorides. Constr. Build. Mater.310:125168. 10.1016/j.conbuildmat.2021.125168), UHPFRC samples with a same mix composition with 160 or even 130 kg/m³ of short steel fibers, healed in water immersion for 28 days exhibited penetration through the crack (W) between 4 and 6 mm. However, that study also observed some results may have been influenced by contamination introduced by the wet-cutting methods used during the preparation of the samples. Similarly, (20Lo MonteF, RepesaL, SnoeckD, DoostkamiH, Roig-FloresM, JacksonS, Blanco AlvarezA, NasnerM, Paul BorgR, SchröflC, GiménezM, AlonsoMC, Serna, P, De BelieN, FerraraL. 2024. Multi-performance experimental assessment of autogenous and crystalline admixture-stimulated self-healing in UHPFRCCs: Validation and reliability analysis through an inter-laboratory study. Cem. Concr. Compos.145:105315. 10.1016/j.cemconcomp.2023.105315) showed penetration values through the matrix below 6 mm, in a UHPFRC mix with 120 kg/m³.

When comparing the W and P₀ values, Figure 9, U40 samples healed in water immersion show values close to the bisecting line, indicating a similar value of the two parameters. U40 samples healed under H28 showed more dispersed W and P₀ values. U160 samples healed under the W28 condition demonstrated complete resistance to penetration of chlorides through both the matrix and the crack path, with values placed in the origin of the graph. However, U160 disks healed in the H28 condition showed crack penetration (W) that did not correspond to matrix penetration, which may be caused by internal multi-cracking and a concentration of micro-cracks.

Figure 9 Chloride penetration depth through the crack (W) vs. chloride penetration through the matrix (P), depending on the mix and healing condition. 

4. CONCLUSIONS

The results obtained in this study highlight the importance of the presence of fibers, not only for improving the mechanical properties of UHPFRC, but also to improve the durability and self-healing behavior of the material.

On the one hand, the UHPFRC mix with 160 kg/m³ of fibers demonstrated a complete closure of cracks, particularly for initial cracks measuring below 350 μm when healing immersed in water for 28 days. This contrasts with the U40 mix, which displayed a limited crack healing capability, with most cracks closing less than 60%. Under high humidity conditions, both mixes showed reduced crack closure capability, highlighting the need for direct water contact to promote autogenous healing reactions in UHPFRC.

On the other hand, the healing ratio, evaluated by the recovery of impermeability showed that the only mix able to reach complete healing was that with 160 kg/m³ of fibers, and only when healing in water immersion. In addition, this mix also displayed complete resistance to chloride penetration, even for initial cracks up to 300 μm. In contrast, the UHPFRC mixes with lower contents of fiber showed some extent of penetration, around 4 mm in all the cases.

Additionally, it should be noted that crack closure measures might overestimate the self-healing based on water permeability. At the same time, water permeability tests could overestimate the level of protection achieved against chloride penetration.