Mechanical properties and self-sensing ability of amorphous metallic fiber-reinforced concrete

. The aim of this research work is to develop a corrosion resistant fiber-reinforced concrete for radioactive waste disposal structures. In the case of precast concrete, the use of fibers is a solution to reduce the amount of steel reinforcement while maintaining high mechanical performance and durability. Concrete has a low strain capacity and a limited tensile strength. Generally, reinforcing bars are used to ensure tensile strength. A fiber reinforcement can also help to overcome such a mechanical weakness. For this purpose, an amorphous metallic fiber (AMF), corrosion-resistant and suitable for application in severe environment conditions are used. The fresh and hardened properties of the self-compacting fiber reinforced concrete (SCFRC) are studied with volume fractions of AMF of 0% and 0.28% and with three different aspect ratios (82, 114 and 123). Flexural tensile tests according to European standard EN 14651 are conducted to quantify the contribution of the fiber reinforcement on the residual flexural tensile strength. Since these fibers are electrically conductive, they are also tested to design a smart concrete. For this purpose, electrical resistance of specimens submitted to cyclic flexural loadings are monitored using a Wheatstone bridge.


Introduction
Concrete is a widely used material in construction field for its high compressive strength.The strain capacity of plain concrete is in the range of 0.1 to 0.2 mm/m, which results in brittle failure of the material.Even with low crack opening values, brittle behavior is observed.When a stress is applied to the concrete, very quickly, there is a localization of the deformations and a development of one or several macrocracks.For ordinary concrete, reinforcing bars are used to improve the strain capacity and the tensile strength.However, the presence of steel reinforcement can cause durability concerns due to corrosion.This study is carried out within the framework of the Cigéo project, a French deep geological disposal facility for high-level (HLW) and intermediate-level long-lived (ILW-LL) radioactive waste.The French National Radioactive Waste Management Agency (Andra) is studying alternative to conventional reinforced concrete to reduce corrosion and therefore increase risk control.One of the solutions is to use non-corrosive fibers as concrete reinforcement.Amorphous metallic fibers (AMF) are used to reduce the amount of reinforcement but also to improve crack control in concrete.Since the 1960s, fiber-* bouillar@insa-toulouse.fr reinforced concrete has been the subject of numerous studies to understand its properties.Many FRC applications have emerged in the field of civil engineering, such as bridges, pipelines, or tunnels [1].Depending on the application, there is a wide range of fibers as metallic, vegetable, or synthetic fibers.They can be used to improve post-peak behaviour, increase resistance to fire or impact, and to reduce the risks associated with corrosion.
For a long time, AMF have definitely proved their worth in the reinforcement of radioactive waste containers [2].There is now interest in extending the use of these fibers to other concrete structural components.
Several studies have already been devoted to the contribution of AMF on the mechanical properties of concrete.Some studies such as those of Kim & al. [3], Hameed [4] and Yoo & Banthia [5] show that this type of fiber increases significantly the flexural strength.In contrast, studies of Choi & Ku [6] and Won & al. [7] indicate that the maximum bearing capacity on flexural tensile test are relatively close to those achieved without fiber.Nevertheless, all the studies agree on the fact that these fibers improve the post-peak behavior and that they are especially effective to control the opening of the crack as soon as the latter is initiated.
Additionally, the use of electrically conductive fibers also appears as a solution to improve the "smart material" potential of concrete, that is to say the ability to monitor its level of stress and its state of damage in real time.Electrical conductivity can be defined as the capacity of the material to allow the passage of an electric current.In concrete, there are two main types of electrical conductivity.The first one is the electrolytic conductivity that is induced by the movement of ions in the pore solution [8].In the second one, the electrical conductivity is caused by the movement of free electrons present in the conductive materials (here the fibers).Previous researches worked on this topic and studied self-sensing ability of concrete under compression stresses.For this purpose, different types of fibers such as steel fibers [9]- [10], short carbon fibers [11]- [12], and micro or nano carbon fibers [13]- [14] have been investigated.
Nevertheless, the smart concrete potential has been rarely studied on flexural test even if it is one of the best ways to appreciate the contribution of fibers on the mechanical properties.

Materials
The cement used in this work is a CEM III/A 52.5 L CE PM-ES-CP1.The sand (0/4) and the gravel (4/12) are of calcareaous origin.The Betocarb HP, a calcareous filler was also used to adjust the granular distribution.In order to control the rheology of the concrete, the MasterEase 2000 is used.It is a superplasticizer (SP) based on new generation of polymers giving to fresh concrete enhanced rheological properties.The AMF « Fibraflex » come from the Saint-Gobain SEVA Company.They have a straight shape type (see Fig. 1) with a thickness () of 24 or 29 µm, a length (L) of 20 or 30 mm and a width () of 1 or 1.6 mm, all fiber dimensions are given in the Table 1.The industrial manufacturing process of quenching liquid metal consists in pouring a jet of liquid metal on a cooled wheel rotating at high speed.In that way, the metal is cooled very quickly, and the fibers are frozen in an amorphous state.Those fibers are flexible and have a tensile strength of 1700 MPa, a Young's modulus of 140 GPa and a density of 7.2.They are composed of 75% Fe, 5% Cr, 8% P, 10% C and 2% Si.Thanks to presence of Cr, even in small quantities, and to its amorphous state, these fibers show an excellent resistance to corrosion [15]- [16].

Mixing and casting
It is well known that adding fibers to concrete has a significant impact on the fresh state behaviour.Indeed, as well as aggregates, fibers will impact the arrangement of the granular skeleton.Therefore, the higher the fiber content is, the more difficult it is to control rheological behaviour of the concrete.Such a detrimental effect increases with the fiber aspect ratio.
A plain concrete (PC) is cast as control to allow the contribution of the fiber reinforcement to be quantified.To compare different concrete blends, the superplasticizer content was adjusted to obtain equivalent workability.The concrete studied is a self-compacting concrete, classified as SF1 according to NF EN 206/CN [17].It means that the spread at the Abrams cone must be in the range of 550-650 mm.The mixing protocol for fiberreinforced concrete is as follows: for 1 minute, the aggregates are dry-mixed, then the filler and the cement are added, and the whole is mixed for 1 minute.The superplasticizer mixed with water is added progressively for 1 minute, and then mixed for 2 more minutes.Finally, the fibers are added, and mixing continues for 2.5 minutes.
The moulds were filled with fresh concrete by hand without any vibration.After 24 h, they are demolded and placed in a curing room at a temperature of 20°C and at 90% RH.All the tests have been carried out after 28 days of curing.
The mix proportions for each raw material are presented in Table 2.The blends are named as follows: "F20" for their Fibraflex fiber contents of 20 kg/m 3 and then the fibers dimensions are given (see Table 1).

Flexural tensile strength
The European standard NF EN 14651 "Measuring the flexural tensile strength" [19] explains the protocol for measuring the flexural tensile strength of metallic fiber reinforced-concrete.
The tests are carried out using prismatic specimens prepared according to NF EN 12390-1 standard [20], with a height (ℎ) of 150 mm, width () of 150 mm, and a length () of 550 mm.A notch of 5 mm width is realized in the middle of the specimen so that the distance between the notch tip and the top surface of the specimen (ℎ  ) is 125 mm.The span length () is 500 mm.The specimen is subjected to a 3-point bending test and the loading rate is controlled by the crack mouth opening (CMOD).The machine is operated so that CMOD increases at a constant rate of 0.05 mm/min.When the CMOD reaches 0.1 mm, the machine increases the CMOD rate at a constant value of 0.2 mm/min.The test should continue until the CMOD reaches a value of at least 4 mm.The test setup is described in Fig. 2.

Piezoresistivity of concrete submitted to flexural tensile cyclic loading
The principle of the test has been inspired by the study of Ferdiansyah [21], it is to apply different stress levels.The specimens are prismatic of 100×100×500 mm 3 with a notch of 17.5 mm in the span middle.This size of specimen was chosen to ensure the perimeter/area ratio is large enough to have a good sensitivity.Specimens are subjected to 3-point bending test and the loading rate is controlled by the CMOD.The machine is operated so that CMOD increases and decreases at a constant rate of 0,03 mm/min.To carry out the electrical measurements, two fine grooves are made at 25 mm from the middle of the specimen and are 2 mm deep.Down these grooves, a conductive paint is applied to improve the quality of the contact from the copper wires to the concrete.The copper wires are used as electrodes and have a 0,8 mm diameter.They encircle the specimen inside the grooves over the entire section.The choice to use external electrodes (instead of electrodes embedded in concrete) was made to avoid any impact on the orientation and distribution of the fibers inside the concrete.With internal electrodes, the fibers could have been blocked or in direct contact with the electrodes, thus greatly altering the quality and the repeatability of the tests.The test setup is described on the Fig. 3 and Fig. 4.

Fig. 4. Piezorestivity measurement setup
The loading scenario is divided into several cycles.The peak value has been estimated before the tests with previous investigations.The 1 st cycle starts from 0 kN and increases to 60% of the predicted peak load.After reaching this value, the CMOD decreases until the load reaches 1 kN.Then, the 2 nd cycle starts, the load increases until the real peak.After that, the CMOD continues to increase while the load is decreasing because at this point, the concrete is damaged.When the load reaches 80% of the peak value, the CMOD starts to be closed until the load reaches 1 kN.Then, during the 3 rd cycle, the load increases until it reaches 60% of the peak, before dropping back to 1 kN.It is the same scenario for the 4 th and 5 th cycle with loads varying respectively from 40% and 20% to 1 kN.The loading scenario is shown in Fig. 5.

Fig. 5. Loading scenario
To monitor the electrical properties of the concrete, a Wheatstone bridge was used.An alternative voltage input of 20 V is applied with a frequency of 1 kHz.The system is made up of 3 potentiometers and 1 unknown resistance.The 4 resistances are connected in a closed loop and the current is applied on two ends.Between the junction of two parallel branches, a galvanometer is connected to measure the difference of voltage.Before proceeding to the test, all the branches of the installation must be balanced.In other words, . Then, the voltage (  ) between the branches is equal to 0. If  4 is considered as the impedance of the concrete (  ) that is unknown, one can deduce its value using equation (1).
The Wheatstone bridge can measure very small changes in electrical properties.A change in impedance will unbalance the bridge and will be observed through the change in voltage on the galvanometer.To proceed to the test, the first step is to balance the bridge by adjusting the three potentiometers and get the voltage   as near as possible to 0 V.The initial value of voltage is recorded as  0 .Throughout the test, the evolution of the voltage is recorded along with the value of the applied load, the deflection, and the crack mouth opening displacement.The voltage measured within the bridge (  ) depends on the different impedances and on the voltage applied by the generator (  ) following the equation (2).
Since  1 ,  2 and  3 are constants, the impedance variation of the concrete (  ) can be followed by measuring the voltage   .One can therefore simply follow the evolution of ΔV defined by the equation (3).

Flexural tensile strength test
For each batch, 3 specimens have been tested.The results from flexural tensile strength tests are given in Table 3.The load-deflection curves are shown in Fig. 6 and Fig. 7.
The plain concrete mix (PC) has a brittle behaviour since after the peak load the bearing capacity decreases drastically and is totally lost at a CMOD of about 0.4 mm.The three blends with fibers show similar trend before peak load.First, the load increases linearly up to the peak, the material is considered then as undamaged.Just before the peak, the curve becomes non-linear, henceforth, microcracks have formed.They grow and get localized in one or more macro-crack(s).The post peak is characterized by a sudden decrease in residual bearing capacity that stabilizes around 60% of the peak load.
Then, for the F20-20L6 and F20-20E0 mixes, a hardening behavior is noticed.This second stage could be explained by the fact that the crack could begin to have a large enough opening for the fibers to be stressed.As a result, they would transfer the stress on both sides of the cracks, thus increasing the load-bearing capacity of the concrete.The F20-30L6 mixes does not show the same trend at this moment, even if the curve becomes slight, it does not show any second peak.It shows that at the same content (20 kg/m 3 ) F30L6 are less effective for stress transfer through the crack than F20L6 and F20E0.After this hardening phase, a softening phase leads slowly to the loss of the residual bearing capacity.The NF EN 14651+A1 [19] standard gives the equation (4), allowing the limit of proportionality (LOP) to be calculated.This value is equal to the maximum tensile stress before reaching a CMOD of 0.05 mm, considering the material undamaged.It should however be specified that the standard ignores the notch effect and gives a simplified stress distribution by ignoring the stress concentration.
Where:   is the maximum load reached before 0.05 mm of CMOD, l is the span length,  is the width of the specimen and ℎ  is the distance between the tip of notch and the top of the specimen.
The results show that the addition of fibers does not significantly impact the value of the maximum bearing capacity, and thus, the maximum tensile stress.In the case of plain concrete, the LOP is 7,44 MPa, and for the F20-20L6, F20-20E0 and F20-30L6 mixes, the values are 7.93 MPa, 7.36 MPa and 7.66 MPa respectively.
According to the relation ( 5), the standard [17] also allows residual post peak tensile strength to be calculated for given CMOD values.
Where:   is the residual flexural tensile strength for a given   .  is calculated for CMOD0 = 0.25 mm ; CMOD1 = 0.5 mm ; CMOD2 = 1.5 mm ; CMOD3 = 2.5 mm and CMOD4 = 3.5 mm and results are shown in Fig. 8.As expected, the plain concrete has a brittle behaviour since residual strength drops very rapidly to zero before reaching CMOD value of 0.5 mm.However, for fiberreinforced mixes, the concrete has a much better strain capacity.The F20-20L6 and F20-20E0 mixes have higher post-peak residual strength than F20-30L6.It can be assumed that the 20 mm fibers are long enough for strong anchorage in the matrix and stress transfer across the crack.As a consequence, for given content, their higher number than that of 30 mm fibers makes it possible to be more effective in restraining crack opening conferring to the material improved residual post peak strength.
The strain capacity of a specimen can be quantified by the total energy dissipated (W) to a deformation level.This energy is defined here as the area under the loaddeflection curve.The higher this energy is, the more the concrete will have a ductile behavior.Fig. 9 shows W for each one the 4 mixes until a deflection value of 4 mm is reached.As expected, the plain concrete has a low strain capacity, the energy dissipated up to the failure is only 0,79 kN.mm which is almost 10 times lower than that of F20-20L6 and of F20-20E0 and more than 5 times lower than that of F20-30L6.Before reaching 0.4 mm deflection, the fiber-reinforced mixes (F20-30L6, F20-20L6 and F20-20E0) have dissipated 75%, 54% and 65%, respectively, of the energy dissipated up to 4 mm deflection confirming that AMF are effective in controlling small crack openings.Table 3. Summary of flexural test results (values after "±" are the standard deviations)

3.2.
Piezoresistivity of concrete submitted to flexural cyclic loading

Load vs. CMOD
The load-CMOD curves of the four mixes are given in Fig. 10 and Fig. 11.The 5 cycles of loading scenario are the ones presented in a previous section (see Fig. 5): the first one reaches 60% of the peak load, the second reaches the peak before decreasing down to 80%, and the following ones go to 60%, 40% and 20% of the peak.For  all mixes, during the first cycle, the load increases linearly with the increase of the CMOD up to 60% of the peak load.During the unloading phase, the load comes back to the initial value at 1 kN before increasing again following the same slope.After reaching 80% of the peak load, the "CMOD vs load" curve continues to increase in a nonlinear way up to the peak load because of the microcracks initiation in vicinity the notch tip.After the load has reached 80% of the maximum load, the CMOD continues to increase as the micro-cracks grow and multiply [22].After the peak load, they coalesce and localize to form a main macrocrack that start from the notch tip, the place where the stress was initially concentrated.The load then decreases as the crack propagates and CMOD increases.During the unloading phases, the CMOD partially closes without returning to the initial value because the material is irreversibly damaged.The slope of the curve during the loadingunloading phase indicates the loss of stiffness of the samples during the test.The first loading phase when the material is considered as undamaged give the initial slope of the curve that indicates the initial state of the concrete.As the CMOD increases and the material gets more and more damaged, during the loading/unloading phases the slope of the load-CMOD curve decreased, indicating the loss of stiffness.3) and CMOD over the time for PC, F20-30L6, F20-20L6 and F20-20E0 specimens.For all mixes, ΔV varies slightly during the first cycle.As previously stated, during the first loading cycle, the concrete can be considered as undamaged, so there is no major change within the microstructure of the concrete.The self-sensing sensitivity of the plain concrete occurs quite late in the process.In fact, the electrical properties start to evolve significantly after the 3 rd cycle.Before that, even if the concrete is already damaged, it does not give any information with monitoring of electrical properties.This is in contradiction with the results from a previous work by Ferdiansyah [21] where the author noticed that even before the peak load, the electrical voltage varies with the loading level.This difference can be explained by several reasons.On the one hand the electrodes used by Ferdiansyah were embedded in the concrete during the casting allowing more sensitivity and more accurate measurements.On the other hand, the loading rate was set at 2 µm/min, which is 15 times lower than the one used in the tests presented here.The cracks are thus formed more quickly, and the variation of resistivity is not easy to monitor.After the 3 rd cycle, the CMOD continues to increase and the ΔV curve follows the trend of the CMOD one.For the plain concrete, when cracks propagate, the resistivity of the concrete evolves because the interconnectivity of the concrete porosity in no longer assured.Hence, the conductivity, which is in that case mainly governed by the transportation of ions in the pore solution, is reduced.Regarding fiber reinforced mixes, during the first loading cycle and for the same reasons stated above, the variation of electrical voltage is not significant.The concrete is considered as undamaged and there is no major change within the concrete microstructure.After the peak load, the concrete is then damaged but the CMOD is still low, and the fibers only start to be stressed.As the CMOD increases, fibers are getting stressed, leading to alteration of the fiber-matrix interface.The fibers gradually approach their tensile strength or their debonding from the matrix.It should be noted that thanks to their large specific area, AMFs have a high bond with the matrix so elongation to rupture of the fiber instead of debonding is the most recurrent mode of failure.However, in the case of a fiber whose end is close to the crack, failure can be caused by a debonding and a pull out of the fiber.Whatever the cause of such failure, the latter induces change of the electrical resistance.After reaching 0.2 mm of CMOD, ΔV evolves linearly with the CMOD.When cracks are closed, the ΔV decreases and when they are again opened, ΔV increases.As illustrated in Fig. 16, plain concrete seems to have the highest sensitivity for ΔV variation.At a given CMOD value, the PC mix has reached a higher ΔV compared to the fiber-based mixes.A crossing of Fig. 7 and Fig. 16, confirms that the PC exhibits a lower residual post cracking strength but is more sensitive for ΔV variation.One can notice that for a given CMOD value, the highest sensitivity ranking is in this order PC, F20-20E0, F20-20L6 and finally F20-30L6.

Conclusion
This contribution focused on the effect of AMF reinforcement on the mechanical properties of concrete.Results showed that AMFs are efficient to improve significantly the residual flexural post peak strength and the strain capacity of concrete resulting in an improved flexural toughness.These fibers are most effective at low CMOD values.In the case of flexural test according to the NF EN 14651, they showed the best efficiency between a CMOD reached at the peak load and a CMOD of 0.5 mm (see Fig. 6).As they are also resistant to corrosion, they are particularly for applications where containment is a priority, including in an aggressive environment.By restraining the crack opening, these fibers allow a sustainability to be achieved [23].
The piezoresistive behaviour under flexural test has been examined in this study.Results showed that both plain and fiber-reinforced concrete have the ability of monitoring their damage state according to electrical measurements once they reached an advanced damage state.Before the peak load, ΔV the voltage variation in the Wheatstone bridge is not significant.At this loading level, only micro-cracks have been initiated and they do not cause any major change in the electrical properties of the concrete.After the peak load, micro-cracks coalesce and localized forming one or more macro-cracks.For fiberreinforced concrete, ΔV evolution is mainly due to the change of intrinsic electrical resistance of the fibers, i.e., the evolution of the fiber-matrix bond and the progressive

Declaration
Conflict of interest: All the authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest is the subject matter or materials discussed in this manuscript.

Fig. 9 .
Fig. 9. Dissipated energy (W) during the flexural test up to a deflection of 4 mm.

Fig. 16 .
Fig. 16.ΔV vs. CMOD curves /doi.org/10.1051/matecconf/202236402004ICCRRR 2022 rupture of the fibers with the crack opening.Regarding plain concrete, the ΔV evolution is mainly governed by the limitation of the ionic conduction due to discontinuity in the material.Author contributions: All the authors contributed to the study conception and design.Experimental preparations, tests, and data processing were made by T. Bouillard.The draft version of the paper has been written by T. Bouillard and A. Turatsinze, JP.Balayssac, A. Toumi, O. Helson and X. Bourbon corrected it and commented it.All the authors read and approved the final manuscript.

Table 1 .
Dimensions of the fibers.

Table 2 .
Details of the mix proportions.(SP for superplasticizer, S for sand, G for gravel and C for cement)