Electrically conductive cementitious composites for marine applications

Conductive cement-based materials have attracted particular attention due to their potential to enable low-cost continuous monitoring of next-generation smart structures. This feature is relevant in highly energetic marine environments, where structural damage due to excessive loads and remote locations are critical to the performance of structures. However, to the best of the authors knowledge the literature exploring this feature in the context of marine environment is inexistent. To fill this gap, an electrically conductive cementitious composite (E3C) was developed for marine applications by introducing waste materials and fibers. Six mixtures were studied, incorporating the same matrix but varying the amount and fiber type, and water content. E3C, which conducts electricity underwater, was successfully developed, demonstrating the capacity to exhibit intelligent properties in marine environments. Self-sensing capability, assessed through electric current measurements during load application, demonstrated sensitivity to mechanical loading. This research presents promising outcomes, reshaping the paradigm of marine structures.


Introduction
Electrically conductive cementitious composites (ECCC) are acknowledged for their ability to exhibit intelligent functionalities, including crack detection, stress/strain monitoring, microstructural sensing, corrosion monitoring, traffic monitoring, self-heating, their usefulness for modeling the penetration of aggressive agents and for monitoring the quality of concrete, for monitoring the hardening and setting processes, as well as for the development of technologies for cathodic anti-corrosion protection .As mentioned, applications are various and the most frequently analyzed property is electrical resistivity.However, although it is an area that is already well explored, none of the studies found in the literature so far were aimed at producing and studying an ECCC that could be applied in submerged marine conditions and exhibit smart features.Marine and offshore structures typically face serious problems of premature degradation due to their remote location, exposure to corrosive environment and varying mechanical loads, and toxic substances are often used to slow their deterioration [27,28].These problems are converted into other problems, such as higher costs of repair operations, the use of more materials during the service life, the generation of waste and the destruction of marine fauna, as well as other social, environmental and economic impacts.Extending the service life of these structures can be achieved by mitigating all the mechanisms that lead to the degradation of the materials involved, for example by using techniques described previously such as self-monitoring and cathodic anticorrosion protection, as well as self-healing through the induction of low voltage electric current.The production of ECCC for marine applications may provide a solution for attaining smart properties underwater, thereby extending the durability and enhancing performance of marine structures.However, for applications in submerged marine structures, and to guarantee material conduction, the material is required to show an electrical resistivity substantially lower than the one observed in seawater, where electrical resistivity typically ranges between 20X:cm and 30X:cm½29: The smart mechanisms mentioned previously require that the electrical conduction is made through the material and not through the seawater.The conventional concrete presents high electrical resistivity (10 9 X.cm) and is commonly seen as a non-conductive material, and to the best of the authors knowledge and considering an extensive literature review , the lowest value of electrical resistivity obtained for ECCC was 65 X.cm, using Electric Arc Furnace Slag (EAFS) aggregate and with 1.29% of oxidized carbon fibers [23].However, this is not conductive enough for marine applications.According to the authors, this mixture was also found to show the best correlation between electrical resistivity and longitudinal strength.
In the present research, the main goal was to develop and characterize the so-called E3C, aimed at submerged marine applications.In this context, it was also necessary to develop a new system for measuring the electrical resistivity of the E3C developed.The methods adopted to measure the electrical properties and, in particular, the procedures for the acquisition and processing of electrical signals, are factors that remain little explored and unclear in most of the studies found in the literature.Considering the more recent studies and after a previous composition design and optimization study, the novel E3C showing electrical resistivity lower than seawater, was studied.The electrical resistivity at 28 d, for longitudinal and transverse directions was measured, as well as compressive and flexural behaviors, in order to understand the interaction between nonlinear mechanical and electrical material behavior, and phase a preliminary assessment of the multiphysics of E3C.Other parameters such as fiber type and fractions, amount of water, water type, and incorporation of EAFS aggregates to improve the conductivity of the E3C were also analyzed.

Materials
In all mixtures, CEM II/A-L 42.5 R cement according to EN 197 standard and fly ash (FA) from coal burning, were chosen as binder material and fine addition, respectively.The particle density of fly ash was 2.35 Mg/m 3 .For all mixtures the conventional coarse and fine aggregates were replaced by Electric Arc Furnace Slag (EAFS 0-4 mm) aggregates, a waste resulting from the steel production using electric arc furnaces.The main characteristics of the EAFS and FA used are presented in Tables 1 and 2 [45], and Figure 1.This aggregate has, approximately, 30% of iron (Table 2) most likely in the form of steel and may provide a good contribution to the electrical properties of the E3C [23].After a previous composition design and optimization study [1][2][3][4][5]24,30,[39][40][41], three types of fibers were used: (i) steel fibers (SF) 20 mm long, 29 mm wide and 1.6 mm thick, with a specific density of 7.85 Mg/m 3 ; (ii) carbon fibers (CF) 6 mm long, 7 mm in diameter, and with a specific density of 1.80 Mg/m 3 ; and (iii) recycled carbon fibers (RCF) with specific density of 1.76 Mg/m 3 and variable lengths, between a minimum of 80 mm and a maximum of 30 mm.SF were selected because of their flat cross-section and larger specific surface, as well as their considerable length, in contrast to CF which have circular cross-section and shorter length.Additionally, in line with eco-friendly considerations, RCF were chosen to offer a more sustainable alternative.Since steel fibers (SF) and carbon fibers (CF and RCF) used tend to significantly reduce workability due to their high specific surface area, a superplasticizer (SP) based on polycarboxylates was used to reduce the need for water and improve the workability and consistency in all mixtures, SIKA ViscoCrete − 3002 HE.

Preparation of the specimens
Six different E3C mixtures were prepared for assessing the electrical resistivity under controlled conditions, at 98 ± 2% of relative humidity and 18 ± 2 C of temperature.These mixtures were composed by EAFS 0-4 aggregates, CEM II/A-L 42.5 R cement, fly ash, superplasticizer, tap water or seawater, and fibers (SF, CF or RCF).The designations attributed to the different mixtures included the following variables: (i) water type (seawater, SW, or tap water, TW); the amount of water (standard water amount, TW, or extra water amount þ TW) and content and type of fiber used.The studied mixtures were designated as shown Table 3. Table 4 shows the compositions of the mixtures prepared.Additionally, considering that the mixing process has an influence on the electrical resistivity of ECCC [5,30] the preparation process of the specimens (Figure 2) followed these sequential steps: (a) initially, the aggregates, fly ash, and cement were mixed; (b) after, the dry components were blended with water and superplasticizer; and (c) finally, fibers were incorporated and mixed to finalize the procedure until homogeneity was reached.Prismatic specimens with dimensions of 4 Â 4 Â 16 cm 3 were used.Three specimens were produced for each mixture, designated by 1, 2, and 3.After 24 h, all specimens were removed from the molds and left to harden in the same curing conditions, at 98 ± 2% of relative humidity and 18 ± 2 C of temperature.After 7 days, the specimens were prepared for the measurement of the electrical resistivity, as shown in Figure 3(a) and (b).The electrode can be attached or embedded [5,7,24].For attached electrodes such as endplate, tape or wire electrodes, as is the case of this study, a conductive adhesion is required to be applied between the specimen and electrodes to reduce the contact resistance [5,7,24].Considering this, carbon electrically conductive paint and copper electrically conductive adhesive tape were applied to the surface of the specimens, in order to form the electrodes (Figure 2(b)).The conductive paint was applied to reduce the contact resistance between the electrodes (conductive adhesive tape) and the specimen.This method has been shown to deliver reliable measurements with low scatter, while being simple to carry out [5,7,24].
At 28 d after casting each specimen was subjected to the electrical resistivity measurement, and subsequently to the flexure and compression test.One of the halves obtained from the flexure test after failure was used for the compression test, according to EN 1015-11-2009.During the flexure tests the electrical current was continuously measured until failure.For that purpose, the electrodes were positioned in the longitudinal direction of the specimen, with a square shape and dimension of 4 Â 4 cm 2 , as shown in Figure 3(b) and (c).The distance between the electrodes was 16 cm.In the case of the compression test, the electrodes were positioned in the transverse direction with respect to the loading direction.The electrodes with a square shape had as dimensions of 2 Â 2 cm 2 and the distance between the electrodes was 4 cm (Figure 3(d)).

Testing procedures
The electrical current of the E3C was measured using the two-probe method, while subjecting the specimen to a certain potential difference in volts.The measurements were carried out using a DPS3005 digital power supply with direct current (DC) and by imposing a voltage of 10 V.The electrical resistivity was determined according to Equation (1): where U (V) is the potential difference imposed or voltage; I (A) is the measured electrical current (DC); S (cm 2 ) is the electrode surface area and l (cm) is the distance between electrodes.
As mentioned in the literature [4], the curing time can influence the electrical resistivity of conductive cementitious composites, therefore the evolution of electrical resistivity with the curing time was measured.For all specimens the electrical resistivity was measured at 7, 14, 21, and 28 d after casting.
Compressive and flexural behavior were considered as representative of the most frequent loading scenarios under development.Additionally, the strengthening mechanisms provided by the added fibers and their impact on electrical resistivity under different loading conditions, involve antagonistic actuation mechanisms.Therefore, the effect of the compressive and flexural loading on the electrical resistivity of the E3C specimens was assessed through continuous measurements of the electric current during testing, for mixtures M_TW_1SF and M_TW_ 0.5CF.The compressive and flexural tests were carried out according to EN 1015-11-2009 and the precision of used LVDT was 5 mm.A system was developed to acquire electrical current during testing, including a digital DC power supply (DPS3005) with a voltage and current

Electrical resistivity and curing time
Table 5 shows the average electrical resistivity measured at 28 d, for each mixture tested.Two of the mixtures have reached an electrical resistivity at 28 d lower than the seawater, namely M_TW_0.5CF and M_SW_0.5CF.According to the results, the lowest electrical resistivity at 28 d was reached with mixture M_TW_0.5CF,followed by mixture M_SW_0.5CF(12% difference).The mixture M_ þTW_0.5CF,with a greater amount of water, showed higher electrical resistivity when compared with M_TW_ 0.5CF.On the other hand, the mixture M_TW_0.5CFRunexpectedly showed a significantly higher electrical resistivity, the highest among of all mixtures tested.Regarding mixtures with SF, the electrical resistivity measurements were relatively higher than the ones obtained with mixtures incorporating CF, except when compared to the specimens containing RCF (mixture M_ TW_0.5RCF).The mixture M_SW_1SF achieved almost half the resistivity showed by M_TW_1SF, and such a great difference was not paralleled by the CF based specimens (mixtures M_SW_0.5CF and M_TW_0.5CF).
Figure 4 shows the effect of the curing time on the electrical resistivity evolution for each mixture developed.It is possible to observe that, for all specimens of the mixture M_SW_1SF (Figure 4(a)), the electrical resistivity was essentially maintained over time with a slight tendency to decrease, and the electrical resistivity obtained at 28 d for the different specimens was essentially similar.However, in the case of the mixture M_TW_1SF, the results showed greater scatter and the evolution trend of the electrical resistivity over time was not as clear.In this case, the results indicate that the water type had a greater influence in the measured resistivity, and the mixture with seawater showed more stable and better results in terms of electrical resistivity at 28 d.Contrarily, as shown in Figure 4(b), the electrical resistivity for all mixtures containing CF was approximately constant over the time, with a slight trend for a decrease.Therefore, the water type did not seem to significantly affect the electrical resistivity, in contrast to the behavior observed for the mixtures containing SF.The mixture M_þTW_0.5CFshowed higher electrical resistivity than the mixture M_TW_0.5CF,confirming that also in the case of the use of CF the amount of water significantly impacted the electrical resistivity obtained.In summary, according to the results obtained, the evolution of the electrical resistivity of the mixtures studied over time was influenced mainly by the fiber type, while the electrical resistivity values obtained at 28 d were essentially determined by the fiber type, amount of water and water type.RCF showed to be the less effective option for achieving low electrical resistivity, while CF showed to be the best option, especially if aiming at a  resistivity below 30 X.cm.Therefore, both the type and amount of water significantly affected the electrical resistivity obtained: (i) the mixture M_SW_1SF achieved almost half the electrical resistivity of mixture M_TW_ 1SF, as a result of the use of seawater; (ii) the increase of the amount of water incorporated almost tripled the electrical resistivity of the mixture M_þTW_0.5CF, when compared to the mixture M_TW_0.5CF.The obtained results seem to indicate that the distinct size and geometry of SF and CF lead to different conduction mechanisms [3,40].One may assume that the larger and flat-shaped SF, although at a much lower number, actuate at a larger scale inside the matrix and promote conduction mechanisms at longer distances.In this case, the seawater seemed to favor better electrical conduction between fibers, therefore decreasing the resistivity of the matrix.In contrast, CF are an order of magnitude smaller in diameter, which is within the range of the smaller particles of the mixture and the microstructural features of the cementitious matrix.In this case, different and more effective conduction mechanisms may have been established, with CF acting at a much smaller scale, similar to the one of the cementitious matrix microstructure.Therefore, the possible beneficial contribution of the seawater to increase the composite conductivity was not so evident.Also, because CF are much smaller, they exist in the matrix at a much larger number (5:11 Â 10 9 CF against 1:06 Â 10 7 SF), decreasing the average distance between fibers [36].
Additionally, the excess of water seemed to contribute to reduce the efficiency of the conductive paths, probably due to the alteration of the capillary structure in the matrix and the most likely increase of the average pore size, and therefore reduce the conductivity [15,22,36].As expected, the fresh properties (rheology) were also significantly affected by the increase of the amount of water, and the fresh properties also affect the electrical behavior of the mixtures by influencing the optimal arrangement of the solid skeleton.In this regard, the obtained results agree with previous studies [7,15,22,36].

Fiber reinforcement and mechanical behavior 3.2.1. Compressive behavior
Table 6 and Figure 5 summarize the compression tests results.The average compressive strength obtained for mixture M_TW_1SF was 30.32 MPa, and for mixture M_TW_0.5CF was 30.55 MPa.The coefficient of variation for both mixtures was low.
In general terms, the compositions containing SF and the ones containing CF showed essentially similar performances in compression (Figure 5).Both compositions reached interesting compressive strengths considering the amount of cement added, while showing a generous ability to dissipate energy after the peak force was reached, as well as substantial residual compressive forces.Nevertheless, the compositions containing SF showed a somewhat higher ductility in the post-peak stage of the compressive testing.Therefore, the effect of the use of different fibers, while not visible at the level of the compressive strengths reached by each composition, became somewhat more relevant at the post-peak stage, with a somewhat greater mechanical contribution in the case of the SF.

Flexural behavior
Table 7 summarizes the flexural tests results.The specimens of mixture M_TW_1SF reached higher flexural strengths than the specimens of mixture M_TW_0.5CF, with an average flexural strength of 6.40 and 4.68 MPa, respectively.The difference observed can be attributed to the different fiber fractions adopted, the different fiber natures and the different matrix/fiber interface interactions established between CF or SF and cement, which resulted in different fiber reinforcement mechanisms.In general, the values obtained for flexural strength showed less scatter for mixture M_TW_0.5CF,which indicates that there was greater uniformity in the M_TW_0.5CFmixture than in mixture M_TW_1SF.The ratio between the specimen size and the fiber length is also quite distinct, which may cause greater heterogeneity at the scale of the material.In the case of the compositions containing SF, the greater length of the SF leads to a larger representative volume element, and the need to use greater specimens to reach homogeneity.According to the results presented in Table 7, the coefficient of variation obtained for both mixtures (M_ TW_1SF and M_TW_0.5CF)was nevertheless reasonably low, indicating good repeatability and reproducibility.
Figure 6 shows the mid-span force versus deflection experimental responses obtained for each mixture (M_TW_1SF and M_TW_0.5CF).Regarding mixture M_TW_1SF (Figure 6), the three specimens presented very similar behaviors up to about two thirds of the peak force, showing a linear response and essentially similar elastic stiffness.Then, the mid-span force continues to increase but with a gradual loss of stiffness until the peak force is reached.At this stage, the differences between the different specimens become more relevant, most likely due to the differences in fiber distribution at the mid-span cross section.After achieving the peak force, the specimens showed a quick loss of stiffness and, at some point, experienced a sudden force drop.Subsequently, the specimens showed a gradual and sustained softening, until the test was interrupted due to excessive deformation.The specimen M_TW_1SF.3,however, did not experience this sudden force drop, and at the same time reached a lower peak force.On the other hand, the specimen that reached a higher peak force experienced a more pronounced and earlier force drop.The ability to dissipate energy revealed by the different specimens after the peak force is reached was somewhat different, with a trend for more pronounced sudden force drops in the case of the specimens with higher strength.These results can be explained by the distribution of fibers in the cross-section of the specimens.
Mixture homogeneity is very difficult to guarantee, due to the fresh properties and the high percentage of fibers added, considering their large specific surface.However,   in all cases it was possible to observe that the material showed flexural ductility, despite the differences.
Regarding the specimens M_TW_0.5CF(Figure 6), the mid-span force versus deflection experimental responses obtained were very similar.The peak force reached by the different specimens was essentially the same and the overall behaviors showed a similar stiffening and after-peak softening response.The mid-span force versus deflection responses seem to show an essentially elastic behavior almost till the peak force was reached, although in the specimen M_TW_0.5CF.1 this seems even more evident, with the other two specimens showing a slightly earlier gradual stiffness loss.After the peak force was reached, the loss of stiffness was much faster and the responses entered a fast-softening stage, until failure was gradually reached.Although both mixtures showed the ability to dissipate energy after the peak force was reached, this was much lower in the case of the specimens containing CF.However, CF containing specimens still showed some ductility, and failure was not sudden or brittle, as typically observed in unreinforced specimens.
As shown in Figure 6 and Table 8, the mixtures containing SF achieved substantially higher values of the peak forces than the mixtures containing CF.While taking a closer look at the initial stage of the flexural tests results, in general all specimens seem to undergo an initially linear elastic mid-span forcedeflection response.In the case of the specimens containing SF, there seems to occur a slight abrupt force change at about 2.1 kN to 2.8 kN during the hardening branch, which denotes the transition between the pre-cracked and the post-cracked regime.In the specimens containing CF, this transition seems to occur for slightly lower forces, at about 2.1 kN to 2.5 kN (Table 8).After crack initiation, the specimens containing SF still undergo a substantial stiffening, up to forces of between 3.5 and 3.9 kN at midspan displacements of 0.244 mm in average.After reaching the peak forces, specimens containing SF start to slowly soften, until a sudden load drop occurs.At this stage the main flexural crack is progressing, the neutral axis is rising and the thickness of the compressed ligament is reducing.The sudden force drop may be related to a sudden rupture of a group of fibers and the associated sudden progression of the flexural crack, which occurs mainly due to the strong adhesion established between the fibers and the matrix.This strong adhesion tends to lead to brittle failures, as opposed to fiber debonding or slippage which tends to cause ductile failures.This was confirmed at the end of the test, and most of the fibers present at the crack surface of SF specimens were found to be ruptured.Additionally, these sudden force drops seem to occur at higher force levels in the case of the specimens that show higher peak forces, which may be associated to a heterogeneous distribution of the fibers and the accumulation of a larger fiber fraction near the tensioned face.In fact, the specimen that showed a lower peak force did not show this sudden force drop, perhaps because there was a greater accumulation of fibers near the central part of the mid-span cross section, which resulted in lower strains in the fibers and eventually in the reduction of fiber ruptures.
In the case of the specimens containing CF, the stiffening experienced after cracking was not so significant, with the specimens showing a slight force increment up to between 2.4 and 2.5 kN at displacements 0.051 mm in average.After the peak force was reached, softening occurred smoothly but very fast, with specimens containing CF showing much lower energies dissipated during the fracture process when compared to the ones containing SF.The progression of the flexural crack was therefore much faster and not restrained by a significant fiber stiffening effect.Therefore, the integrity of the ligament between the two parts of the specimen was rapidly compromised, and the continuity between the two opposite faces of the specimen at the flexural crack was quickly lost.

Electrical resistivity and mechanical responses 3.3.1. Compression
Table 9 shows the electrical resistivity results in the transverse direction of the specimens at 28 days after casting, and before carrying out the compression tests.The obtained results were essentially influenced by the fiber type since the matrix composition adopted was the same.The average electrical resistivity obtained for the mixture with SF was 333.30X.cm, with high variability between mixtures.As previously discussed, the SF were more difficult to mix and guarantee a homogeneous distribution, influencing the fresh properties of the material and consequently the electrical resistivity.The high standard deviation values obtained may be due to this lack of homogeneity, likely influenced by high SF length and specimens' dimensions ratio.The representative volume element (RVE) of SF specimens is substantially larger due to the size of steel fibers, and additional scale-size dependent tests could help to clarify this effect.In contrast, in the specimens containing CF, the results showed a much lower scatter.The average electrical resistivity measured was 19.4 X.cm, and the coefficient of variation was low, about 5%.In this case, the type and size of fibers allowed a better distribution in the matrix, influenced by the low CF length and specimens 0 dimensions ratio, as well as a smaller RVE, providing more consistent results.The conductivity and homogeneity of the mixtures revealed to be very sensitive to the fiber size, since the fiber material itself was highly conductive, in both cases.
Figure 7 shows the evolution of the electrical resistivity during the compression tests conducted for specimens of the mixture M_TW_1SF.For reference, the electrical resistivity value obtained before testing is represented for all specimens (Table 9), as the baseline (reference value).The evolution of the electrical resistivity (right y-axis) and the compression force (left y-axis) versus the axial displacement (x-axis) are also presented.In general, the electrical resistivity seems to slightly reduce for increasing force at the initial stage of the test, followed by an increase once the force reaches higher levels.For specimen M_TW_1SF.1, the greatest change of the electrical resistivity was observed between one third and two thirds of peak force.Before that, some variation of the electrical resistivity was observed, but of reduced significance when compared to the baseline.Before the peak force and still during the elastic stage of the experimental response, the electrical resistivity showed sensitivity to force increment, with its rapid increase.After the peak force, the electrical resistivity varied also, due to the coalescence and extension of macro cracks in the E3C that gradually lead to the rupture of conductive paths.Simultaneously, new conductive networks may be created, with the crushing of the material, and therefore the electrical resistivity showed great variations at this stage.For M_TW_1SF.2 the largest variation of the electrical resistivity was observed further ahead in the experimental response, already at about one half of peak force during softening.The same decrease of the electrical resistivity with respect to the reference value was observed before the peak force was reached, as in the case of specimen M_TW_1SF.1.For specimen M_TW_ 1SF.3, similarly to M_TW_1SF.2, the highest variability of electrical resistivity was observed near peak force.However, similarly to the previous two specimens, the electrical resistivity showed the same tendency to decrease below the baseline and increase again before the peak force, although in this case a perturbation seemed to occur at this stage.After failure, the same increase of the electrical resistivity was also observed.Generally, all specimens showed to be sensitive to the force variation, with a tendency to increase abruptly after the peak force was  reached, when the material is cracked and the conductive paths are compromised.
Regarding the specimens with CF, Figure 8 shows the evolution of the electrical resistivity during the compression tests.In this case, all specimens seem to show much smoother variations of the electrical resistivity.Generally, at the onset of testing the electrical resistivity tends to decrease below the reference value and increase again during the elastic stage of the compressive force versus axial displacement response.Then, at about two thirds of the peak force, the electrical resistivity increases at a much faster pace.After the peak force and already at the softening branch of the compressive force versus axial displacement response, the electrical resistivity tended to remain stable for a while, returning to a quickly increasing trend once the force reached between two thirds and one half of the peak force.The variation of the experimental response obtained between specimens was quite low, contrarily to the substantial scatter observed in the specimens of M_TW_1SF.In this case, the results were more consistent between specimens, as well as between variables (electrical resistivity and force).
Figure 9 shows in greater detail the evolution of the electrical resistivity at the initial stages of the compression tests of the mixture M_TW_1SF.In the case of M_TW_ 1SF.1 the electrical resistivity tended to decrease in the first stage and increase afterwards, followed by another decrease and increase cycle, with respect to the reference value.Only after 0.1 mm of axial displacement corresponding, approximately, to one third of the peak force, the electrical resistivity increased drastically well before the peak force was reached, still in the so-called elastic domain.
In the specimen M_TW_1SF.2 the electrical resistivity showed, at the onset of the test, a clearer decreasing trend when compared to the M_TW_1SF.1,which again resulted from the slight reduction of the distance between fibers and EAFS aggregates for increasing compressive stresses, improving the conductive network inside the E3C.After the peak force was reached, an abrupt increase of electrical resistivity was observed, most likely associated to the coalescence of the micro cracks into a limited number of macro cracks.Regarding M_TW_1SF.3 the evolution of electrical resistivity seemed to show the same overall trend as the previous ones, but with more variability.An intermediate rapid peak of electrical resistivity was measured before the peak force was reached and interrupting the overall decrease-increase trend of the electrical resistivity measurements obtained before the peak load.This feature was not clearly observed in the other two specimens, although there seems to exist a slight perturbation also in the other two specimens.The multiple parts of the specimen that will later become separated are being pushed against each other, and the SF existing at the rupture surfaces may intermittently interact with each other to create alternated better and worst conduction paths, creating these disturbances in the main trend of the behavior observed.
Figure 10 shows, in a closer look, the initial stage of the evolution of the electrical resistivity measured during compression testing of the mixture M_TW_0.5CF.The tendency observed for all specimens with CF was the same, with the subtle decrease and subsequent increase of the electrical resistivity measured before the initiation of cracking, and with the more rapid increase of electrical resistivity at about one half to two thirds of the peak force, when micro cracking starts.After the peak force, the coalescence of the micro cracks into fewer and larger cracks tends to accelerate even further the increase of the measured electrical resistivity.When comparing these results with the M_TW_1SF, in the case of the M_TW_ 0.5CF the observed behaviors seem more consistent and showing less scatter.The establishment of alternative conductive paths during the specimen crushing, which may have caused greater variations on the measured electrical resistivity in the case of the SF specimens, was not favored by the much smaller and brittle carbon fibers in the case of the CF specimens.
In general, independently of the fiber type and matrix composition, two main mechanisms seem to govern the electrical-mechanical behaviors observed (Figure 11).The first one is related to the effect caused by compressive stresses, which distribute uniformly through the entire specimens in compression tests, and are restricted to the ligament or the area above the neutral axis during the flexural tests.Compressive stresses tend to reduce the distance between the different conductive fibers or points in the solid skeleton, and in this way, the electrical conductivity tends to increase [7].Tensioned areas of the specimen may also be able to conduct electric current, while in the uncracked state or after cracking while some of the fiber links remain intact, and in this case the fiber type and its properties will influence the result.The second mechanism  is related to the effect that cracking has on the conductivity of the specimens.Cracks, which originate isolated and at the microscopic level, will tend to coalesce and form macro cracks, which will interrupt conduction paths and will reduce conductivity.An additional factor to consider is that all materials tested have shown very high electrical conductivities, therefore the interruption of the main conduction paths, except if significant, showed not to considerably affect the electrical resistivity.This is why the influence of the stress state on the electric current during the pre-cracking stages is not so evident.On the other hand, cracking seems to cause visible perturbations on the conductivity measurements, which become quite relevant because electrical resistivity is very low.The final outcome in terms of electrical-mechanical behavior observed is the result of the combination of the most relevant composition characteristics, such as fiber type used and properties of the matrix, and these two competing factors: slight conductivity increase related to the increase of compressive stresses; conductivity decrease related to the initiation, coalescence and propagation of cracks, which may be smoother and less evident at the micro cracking stage, and more sudden and clearly evident at the macro cracking stage.

Flexure
Table 10 shows the electrical resistivity measured in the longitudinal direction of the prismatic specimens at 28 d after casting, before conducting the flexural tests.For mixture M_TW_1SF the results obtained show some variability, as previously shown for the transverse direction, although the variability is not so pronounced.Additionally, when transverse and longitudinal measurements are compared, it is noticeable that the electrical resistivity is significantly lower in the longitudinal direction, which may be the consequence of preferential orientation of the streel fibers along the longitudinal direction.In contrast, the electrical resistivity results for mixture M_TW_0.5CFwere very similar, transverse and longitudinal, as well as the variability of results was very low.Among other factors, the thinner and smaller CF may be mostly the result of a lower trend for CF to align in the longitudinal direction.The obtained values of the electrical resistivity for the mixture with CF were encouraging, very low when compared to literature and other similar studies [5,9].In average, the mixtures M_TW_1SF and M_TW_0.5CFachieved 154.2 X.cm and 19.8 X.cm, respectively.Similarly, when the standard deviation is considered, also CF mixture showed less scatter and greater consistency of results than the SF mixture (Table 10).
Figure 12 shows the evolution of electrical resistivity during the flexural tests for specimens of the M_TW_ 1SF.As before, the reference value of the electrical resistivity is presented as a baseline, obtained from the longitudinal measurements in the prismatic specimens (Table 10).In general, the electrical resistivity seems to show an initial decrease with respect to the baseline value, similar to the one observed previously for the compression tests.This may be justified by the formation of the compressed area within the specimen due to the flexural stresses, which to a certain extent experiences the same behavior that the specimens under compression have shown previously.After that, the electrical resistivity tends to increase, exceeding the baseline value.While the flexural force approaches the peak value and even exceeds it, the electrical resistivity shows a gradual and steady increase.For M_TW_1SF.1 and M_TW_1SF.2 the abrupt variation of electrical resistivity happened when an abrupt force drop was observed, most likely associated to the formation and fast extension of the flexural crack.In contrast, in the case of M_TW_1SF.3 the behavior showed an abrupt variation of electrical resistivity only at the end of the test.During the initial part of the test and while only elastic deformation occurs, a small variation of the electrical resistivity was observed in all specimens, showing that the material was sensitive to small forces.
Regarding M_TW_0.5CF(Figure 13), the obtained results were essentially similar.Nevertheless, the results were more consistent and showed lower scatter between the different specimens of the same mixture.In the specimen M_TW_0.5CF.2, initially the electrical resistivity showed a fast increase which is not coherent with the observations in the remaining specimens or even the other composites.Its value exceeded the baseline.Still for low forces and during the loading stage, the electrical resistivity started to decrease and right after it increased, while the peak force was approached, resembling the results observed for the other specimens.The anomaly observed at the onset of the test may be explained by a small damage in the electrodes, which may have caused the slight increase regarding baseline value.Since the electrical resistivity values measured are very low, all possible perturbations, including changes in the position of the electric wires used to carry out the measurements, or external  stresses inadvertently applied to the electrodes, may result in small but visible variations of the electrical resistivity measurements during the test.
As before, also the specimens made with CF showed sensitivity to loads in the elastic range.After the peak force was reached the variation of electrical resistivity was more abrupt for M_TW_0.5CF.1 and M_TW_0.5CF.2.For M_TW_0.5CF.3, the electrical resistivity value seems to show the same trend as the other two specimens, with the initial reduction and subsequent increase of the electrical resistivity during the pre-cracking stage.However, a second reduction of the electrical resistivity seems to occur, already after crack initiation.Since the force at this stage is still increasing, most likely the compression forces at the ligament are still increasing, and as a result the electrical resistivity may be decreasing despite the flexural crack is forming.Although not so evident, also in the other two specimens this same type of behavior is visible.
Overall, in both cases (M_TW_1SF and M_TW_ 0.5CF) the results revealed an abrupt increase of electrical resistivity when the peak load was reached.In the elastic domain, the electrical resistivity varied in all cases, showing two types of behaviors: increase and decrease of the electrical resistivity in relation to the reference value.
The complexity of the processes involved in the flexural tests are increased because there is a greater interference between the competing effects of compressive stresses causing the increase of the conductivity, and the tensile stresses causing the decrease of the conductivity, as well as the type and properties of the fibers, and their interaction with the matrix during fracture [7].
In order to better understand the potential of these mixtures to be used as sensors at low load levels, Figure 14 shows the evolution of the electrical resistivity with the increment of the mid-span displacement for M_TW_1SF specimens, in a closer look at the initial stages of the tests.Generally, the electrical resistivity seems to show a decrease with respect to the baseline, as well as the subsequent increase while approaching the first cracking force.At this stage, the initial decrease may be associated to the increase of the compressive stresses at the top part of the specimen, in agreement to the previous discussions.At a later stage, with the onset of diffuse micro cracking at the tensioned region, the decreasing trend of the electrical resistivity inverts.In the case of specimen M_TW_1SF.3, the variation of electrical resistivity was very subtle until the cracking force was reached, but to some extent, the same behavior may be observed, as in the other specimens.The most significant increments of the electrical resistivity start to occur somewhere between the cracking force and the peak force, with the transition from a diffusely micro cracked area into a few macro cracks that coalesce and start to propagate.The increase of the electrical resistivity is not monotonic, some oscillations occur, and these may be due to the alternate breakage and reestablishment of conductive paths in the tensioned area, where EAFS aggregates and fibers are constantly rearranging due to the propagation of the crack, and loosing or reestablishing contact.At some point, the increase of the electrical resistivity stops, already at the softening stage of the flexural tests.Probably, at this stage, the flexural crack or cracks are fully formed and stabilized, and more stable paths for electrical conduction may be temporarily established leading to lower levels of electrical resistivity.In the case of specimen M_TW_1SF.3,this effect was clearly visible and the electrical resistivity even returned to the reference value or baseline.At a later stage, the fibers that remain active start to break and the electrical resistivity increases again, until the complete failure of the specimen is attained.Figure 15 shows the evolution of electrical resistivity during the flexural tests of the mixtures produced with CF.In general, the same behaviors were observed.In the case of specimens M_TW_0.5CF.1 and M_TW_0.5CF.3, electrical resistivity showed the typical decrease relatively to the reference value and subsequent increase, reaching the lowest value at approximately one third of peak force, and the highest value shortly before the cracking force was reached.After crack initiation there seems to be a tendency to a new reduction of the electrical conductivity, probably because the force continues to increase, but since the flexural crack is already formed, the neutral axis rises and the thickness of the compressed ligament reduces significantly.As a result, compressive stresses at the ligament increase rapidly, and since the material is highly conductive the electrical resistivity may even decrease due to the added compression stresses and the consequent decrease of the distance between conductive elements in the matrix.In the case of specimen M_TW_0.5CF.2,  initially the electrical resistivity increased in relation to the reference value, which could denote some accidental movement or excess of tension in the wires or the electrode during the onset of the test.But afterwards it tended to decrease, similarly to the other two specimens of the same mixture.After the peak force was reached there was a tendency to the stabilization of the electrical resistivity measured at reasonably low levels.This may result from the fact that, despite the existence of the fully formed flexural crack, the crack opening is small and most of the fibers remain intact while bridging its two opposite faces.At the same time, the ligament in the compressed area of the specimen, which is highly conductive, remains stable while no further progression of the flexural crack occurs.Eventually, the fibers in the tensioned area start to break and the compressed ligament starts to accumulate excessive damage, leading to the rapid increase of the electrical resistivity until complete failure of the specimen is reached.Figure 16 shows a summary of this correlation between the mid-force force and electrical resistivity versus deflection, presenting the expected behavior and the obtained general behavior for all specimens assessed, considering both mechanisms involved in flexural tests namely, compression and tensile stress effects.

Conclusions
The research conducted and the obtained results allowed to identify particularly interesting behaviors of E3C compositions for marine applications, which may be summarized as follows: Alterations in the E3C matrix, including the introduction of EAFS aggregates and different conductive fibers, promoted a significant enhancement of the electrical properties of the E3C, in agreement with other authors [23]; The lowest electrical resistivity value was obtained with the mixture containing CF, of 19.40 X.cm and 19.80 X.cm for transverse and longitudinal directions, respectively.These are below the typical electrical resistivity found in seawater and are therefore promising for applications in marine structures.Electrical resistivity of the developed E3C was significantly affected by the type and length of fibers, fiber content, water type and amount of water in agreement with the literature [3,5,7,23,30,33,[39][40][41][42].CF showed better results than SF.However, RCF showed the highest electrical resistivity among all the studied mixtures.Regarding the evolution of the electrical resistivity with the curing time, mixtures incorporating CF showed more stable behaviors when compared to mixtures incorporating SF.Possibly, the mixing process, the lower level of homogenization and different conductivity mechanisms created by SF can explain these results, which require further dedicated studies.In general, the electrical resistivity tended to decrease with the curing time.Overall, all mixtures showed similar mechanical responses in compression.The mixtures containing SF showed a somewhat greater level of residual stresses after peak when compared to the ones containing CF.However, the peak force achieved for all specimens was identical, indicating a negligible contribution of the fibers to the peak force reached.Regarding the flexural behavior, the mixtures studied presented different behaviors.In the elastic domain the specimens of the different mixtures showed similar stiffness, although higher forces were achieved in the mixtures with SF.The post-peak behavior was also very different for the developed mixtures.The mixture with SF showed greater ductility when compared to the mixtures with CF.Electrical resistivity of the materials developed showed to be sensitive to the different forces and damage levels imposed.Generally, the most significant change in the electrical resistivity was observed before the peak force was reached.In this sense, good indications were obtained about the possibility to use this electrical property of the material as a variable for the monitoring of the structural integrity of submerged or partially submerged marine constructions.Sustainable formulations with promising electrical conductivity values have been achieved, envisaging their application in marine environments as multifunctional structural materials and systems for diverse applications.

Figure 3 (
c) and (d) show both setups adopted for the measurement of the electrical current while conducting the flexure and compression tests on prismatic and cubic specimens, respectively.

Figure 11 .
Figure 11.Schematics showing the typical experimental responses in terms of compressive force and electrical resistivity versus deflection of conductive cementitious composites during compressive testing (a) schematics showing the typical compressive force versus deflection experimental response; (b) schematics relating the evolution of the electrical resistivity measured and the compressive response, as obtained by other authors (expected) [7], and the experimentally obtained in this study (obtained behavior).

Figure 16 .
Figure 16.Schematics of the typical experimental responses in terms of mid-span force and electrical resistivity versus deflection of conductive cementitious composites during flexural testing (a) schematics showing the typical mid-span force versus deflection experimental response; (b) schematics relating the evolution of the electrical resistivity measured and the flexural response, as obtained by other authors (expected) [7], and the experimentally obtained in this study (obtained behavior).
a Rietveld method.

Table 3 .
Designation of the mixtures developed.

Table 4 .
Composition of the mixtures developed.

Table 5 .
Average electrical resistivity measured at 28 d for each mixture tested.

Table 6 .
Compressive strength test results.

Table 8 .
Main parameters of the flexural responses.

Table 9 .
Electrical resistivity at 28 d measured before testing, in transverse direction.

Table 10 .
Electrical resistivity at 28 d, measured before testing in the longitudinal direction.