Uniaxial tensile behavior of aligned steel fibre reinforced cementitious composites

: By applying an external uniform magnetic field to a fresh cement mixture during casting, an aligned steel fibre reinforced cementitious composites (ASFRC) was prepared. This investigation compares the performance of ASFRC with its counterpart - ordinary steel fibre reinforced cementitious composite (SFRC) containing randomly distributed steel fibres. First, the orientation of the steel fibres in ASFRC and SFRC specimens was examined using X - ray computed tomography analysis; this confirmed that the steel fibres were effectively aligned in the ASFRC. Then, uniaxial tensile tests were performed to allow a comparison of the uniaxial tensile stress - strain curves of the ASFRC and SFRC; and to determine the advantages, if any of ASFRC over SFRC in terms of uniaxial tensile strength ( f Utu ), ultimate strain (  Utu ) and energy dissipation ( G f - A ). The uniaxial tensile test results were also used to show that, if the tensile strength of ASFRC is equal to that of SFRC (actually slightly exceeding) using the aligned steel fibre technology, the dosage of steel fibres can be reduced at least 40%. It was also found that the alignment of the steel fibres affects the strain - hardening and multiple cracking behavior of the composites during uniaxial tension testing. Finally, the multiple cracking behavior of the composites was analyzed using a digital image correlation method. These results show that ASFRC exhibits a multiple cracking pattern at a much lower fibre content compared to SFRC.


Introduction
By incorporating steel fibres into the cement mortar matrix, the tensile strength, cracking resistance, impact and fatigue resistance of cementitious composites can be effectively enhanced [1][2][3][4][5][6][7][8]. Hence, steel fibres are usually added into concrete to improve its ductile behavior. However, in recent years, most experimental studies [1][2][3][4][5][6][7][8] focussed on SFRC with randomly dispersed fibres, and the theoretical value of fibre orientation number for dispersed steel fibres in an ordinary SFRC specimen is 0.405 [9]. As in some structural elements (beam, slab, tunnel segment, etc.) in service the direction of the principal stress is constant, the orientation of most steel fibres is therefore not consistent with the direction of the tensile stress during the loading process, and only a part of the fibres act effectively as reinforcement. It has been demonstrated that the orientation and distribution of fibres have a significant influence on the mechanical behavior of SFRC [10][11][12][13]. Obviously, if the orientation of steel fibres can be aligned with the direction of the principal tensile stress of the structural members, the tensile behavior and ductility of the SFRC structural members will be improved significantly.
Several methods have been reported to control the orientation of steel fibres in cementitious composites.
A centrifugal casting technology has been used to prepare aligned steel fibre reinforced concrete poles by Rotondo et al. [14]. A special machine was used to spray steel fibres so that the steel fibres were aligned to the specific axis direction of the structural element [15]. By pouring fresh self-compacting concrete mixed with steel fibres through a chute [16] or a U-shape channel [17,18], the steel fibres tend to be oriented parallel to the flow direction of the matrix. Specimens with different fibre orientations can be obtained by cutting the matrix in different directions, e.g., direction that is parallel or perpendicular to the concrete flow direction. Duque et al. [16] and Abrishambaf et al. [17,18] assessed the influence of fibre orientation on the post-cracking behavior for steel fibre reinforced self-compacting concrete panels (SFRSCC). A U-shape iron core around a copper wire coil was used to align fibres while casting ultra-high performance fibre reinforced cementitious composites (UHPFRC) specimens by Nunes et al. [19,20], the advantage of this set up is that the specimens can be prepared without the influence of the element shape [19][20][21]. Recently, Abrishambaf et al. [22] used a similar device [19,20] to prepare UHPFRC specimens, and the uniaxial tensile test was used to study the influence of fibre orientation on the tensile behavior of UHPFRC. Both the straight smooth fibres and the hooked-end fibres were aligned by placing the sample in an electromagnetic field, formed by a spiral coil; the orientation of fibres was revealed by X-ray images [23]. The deflection of specimens in bending was detected using a digital image correlation (DIC) method as proposed by Michels et al. [23] who also studied the effect of fibre orientation on the flexural behavior. Likewise, Mu et al. [24,25] and Wijffels et al. [26] analyzed the effects of the magnetic field distribution and the time of magnetization on the fibre alignment based on the principles of electromagnetics, respectively. The ASFRC specimens were prepared by a uniform magnetic field, which was formed using energized coils during casting [24,25]. The X-ray CT results of ASFRC specimens showed that the fibre orientation number can exceed 0.9 (w/c=0.42) [24], and the results of splitting, flexural and shear strength tests [24,25] showed that the mechanical performance of ASFRC specimens was evidently superior to SFRC specimens; the reinforcement efficiency of steel fibres was significantly improved by aligning fibres utilizing a magnetic field.
The cracking resistance of SFRC has an important influence on its practical application in engineering, and tensile strength is often used for evaluating cracks in structural members. A uniaxial tensile test is the most reliable and direct method to obtain tensile strength and to study the cracking behavior of SFRC.
Moreover, compared to the splitting and flexural tests, the uniaxial tensile test was used to assess the nonlinear fibre-matrix interfacial response on the post-peak ductile behavior of SPRC [27]. Therefore, the uniaxial tensile test was applied to effectively study the tensile behavior of ASFRC, and contribute to its practical application.
The fibre volume fraction (Vf) and matrix strength have a remarkable effect on the mechanical behavior of the SFRC specimens. With the increase of fibre content or matrix strength, tensile strength and toughness can be significantly improved [28][29][30], the multiple cracking and strain-hardening behavior in the post-cracking stage under uniaxial tension [31][32][33][34][35][36] can be observed. In addition, the orientation of fibres also affects mechanical properties, multiple cracking and strain-hardening were obtained with specimens with sectional dimensions of 30mm×40mm using a griping system [22], and the test results showed that the tensile strength and post-peak toughness can be significantly enhanced by aligning the fibres. However, the effects of fibre content (Vf) and water to cement ratio (w/c) on the tensile behavior of ASFRC specimens were not clear. Furthermore, the strain-hardening characteristics and the evolution of fracture for ASFRC specimens under uniaxial tension as well as affecting other factors are not clarified. Generally, the orientation of fibres will be greatly affected by boundary effects [37,38] if the size of specimen is relatively small, and the tensile strength obtained (often overestimated) is not comparable with the behavior of the structure. So, to facilitate the engineering application of ASFRC, uniaxial tensile tests should be carried out with specimens having relatively larger dimensions.
The aim of this investigation is to study the effect of fibre orientation on the strain-hardening and multiple cracking behavior of steel fibre reinforced cementitious composites in order to better exploit the high performance of cementitious composite materials. Prismatic specimens were prepared, and then uniaxial tensile tests were carried out with SFRC and ASFRC specimens with different fibre volume fractions and matrix strengths. Subsequently, the influence of fibre orientation on the strain-hardening behavior and fracture properties of such composites was studied. Under uniaxial tensile loads, the evolution of fracture for ASFRC specimens was revealed using the DIC method.

Test set-up
Uniaxial tensile tests were carried out using a servo-hydraulic closed loop testing machine with a tensile capacity of 1000kN as shown in Figure 1 (the actual loading configuration can be seen in Fig. S1 in the Supplementary Material). To load the specimen, a steel plate with a spherical hinge was first glued to each end of the specimen using epoxy resin (the curing cycle of the epoxy resin is a bit longer than 24 h); the specimen was then connected to the loading machine via the spherical hingethis effectively eliminated any applied load eccentricity (similar to the test set-up of Ding et al. [30]). In order to further ensure that the application of the load was concentric, the ends of the specimens were squared off before the steel plates were attached. Two linear variable differential transformers (LVDTs) were fixed on short wooden strips which were installed at the two opposite side surfaces of the specimen, the LVDTs can measure the elongation of the specimen (as shown in Fig. 1). The DIC system was also installed to test full field strain, which can be used to identify the initial cracking and the progressive cracking process until the complete failure of a specimen under uniaxial tension. As it is a non-contact test with full field measurement, the DIC system can precisely monitor the development of cracks [39]. The loading procedure consisted of two stages: the first stage was stress controlledthe loading rate was 0.4MPa/min and the load was increased to about 80% of Pmax (peak load) and this was followed by the second stage which was displacement controlledhere, the loading rate was 0.1mm/min. The specimens were tested 28 days after casting.

Mixes and specimen preparation
Several groups of prism specimens with dimensions of 100 mm×100 mm×400mm were cast. The ends of the specimen were ground parallel with a machine such that the actual height of the specimens was 300mm after cutting. Straight smooth steel fibres with a diameter of 0.50mm and a length of 30mm were used in the mortar matrix. The Ordinary Portland Cement (OPC) of grade P.O 42.5 and silica sand with a fineness modulus of 2.6 were used to prepare the mortar investigated in this study. In order to facilitate the alignment of the steel fibres, the coarse aggregates were excluded, and polycarboxylate superplasticizer was added to the water to ensure that the mortar had sufficient fluidity and an excellent workability. The procedure for mixing different mortar mixes and the details of preparing ASFRC were previously described in a publication [24]. During the preparation of ASFRC specimens, the steel fibres and fresh mortar were mixed evenly and poured into plastic moulds, which were placed on the vibrating table with a uniform magnetic field device (basically, it is a square solenoid coil looped around a hollow chamber, which creates a uniform magnetic chamber in which the ASFRC specimens were placed). During vibration, steel fibres in the fresh mixture rotated in the direction of the magnetic field produced by the electromagnetic field. In this study, SFRC specimens (Vf =2.0%) with a water to cement ratio of 0.42, 0.36 and 0.32 were prepared (six types of mixtures are presented in Table 1); three specimens for each mix proportion were cast.

Orientation of aligned fibres in ASFRC and SFRC
The orientation of the steel fibres in the mortar samples was first visually investigated by examining the cracked section of the specimens after the tensile tests, the failed specimens are shown in Figure S2 in the Supplementary Material. As expected, it can be identified from the cracked ASFRC specimens that most of the steel fibres were oriented parallel to the direction of the principal stress. Also note, that more steel fibres cross the cracked section of ASFRC specimens compared to SFRC specimens (the average number of fibres of ASFRC specimens Nf-A with Vf =0.8%, 1.2% and 2.0% was 239, 337 and 612, respectively; the average number of fibres of SFRC specimens Nf-R with Vf =0.8%, 1.2% and 2.0% was 193, 271 and 495, respectively). The orientation of the steel fibres in the specimens can also be directly observed once the failed specimens were separated into their two parts, as shown in Figure 2. The steel fibres within the cracked section of the ASFRC specimen are effectively aligned parallel to the direction of the principal tensile stress, whilst those in the cross section of the SFRC specimen were randomly oriented. Besides visually comparing the differences in orientation of steel fibres in the ASFRC and SFRC specimens, X-ray CT tests were also performed. The X-ray CT test is non-destructive and a three-dimensional assessment; the fibre orientation number can be determined from the X-ray CT analysis, as described by Mu et al. [24].

Results
The effects of fibre orientation on the uniaxial tensile stress-strain curves, tensile parameters

SFRC ASFRC SFRC ASFRC
and multiple cracking behavior were quantitatively analyzed for the experimental test results of SFRC and ASFRC.

The effect of fibre orientation on the uniaxial tensile stress-strain relationship
The tensile stress-strain curves of ASFRC are significantly higher than that of ordinary SFRC (as shown in Fig. 4, the segment from the end of the elastic part to the peak load point of the curve is defined as the hardening branch, and that after the peak load point is defined as the softening branch). This suggests that the ASFRC composite is an improvement over the traditional SFRC composite, certainly in terms of tensile behavior and ductility. Although the tensile strength is improved, the hardening phase is not pronounced if the matrix strength is relatively small (as shown in Fig. 4(a) (c)). The matrix has not cracked in the initial stage of tension (elastic stage); the effect of the fibre orientation at this stage therefore is small (see Fig. 4); although as expected the SFRC specimens cracked at a lower stress than the ASFRC specimens. After matrix cracking, the stress-strain curves of ASFRC specimens are characterized by a longer hardening branch under uniaxial tension, especially with a higher matrix strength, as shown in Figures 4(b) and (c). Normally, three main phases can be identified in the stress-strain curves: elastic, hardening and softening. A new cementitious composite with a higher tensile performance and ductility can be designed through the alignment of the steel fibres by using a magnetic field. The tensile strength fUtu and ultimate strain Utu (corresponding to the tensile strength fUtu) were significantly enhanced by the alignment of the steel fibres. The uniaxial tensile results for each series of SFRC and ASFRC specimens are given in Table 2 (the "ASFRC0.8-1" denotes the No.1 ASFRC specimen with Vf =0.8%), where it can clearly be seen that at increasing matrix strength, the improvement of fUtu for the ASFRC specimens is higher than that of the SFRC specimens. Compared with the SFRC specimens, the average tensile strength of ASFRC specimens increases by 26.96%, 44.09% and 57.14% corresponding to w/c-ratios of 0.42, 0.36 and 0.32, respectively. Obviously, with the increase in matrix strength, the reinforcement efficiency of fibres with regard to fUtu becomes higher. This characteristic is due to fact that the enhancement of the steel fibres highly depends on the mechanical interaction of the bond, friction and mechanical interlocking between the fibre and matrix [27]; the bond between fibres and matrix increases at increasing strength of the matrix. As described in Section 3, the change in the orientation of steel fibers causes the difference in the number of fibers in the different sections, more steel fibres crossed the cracked section of ASFRC specimens compared to SFRC specimens. In addition, the orientation of fibres influences the damage of the matrix. As shown in Figure 5 and based on acoustic emission signal analysis [40], the matrix of SFRC specimens with random fibres is damaged more severely than that of ASFRC specimens. The total resistance force is greater in the ASFRC specimen compared to SFRC specimens with the same fibre volume fraction. Note: "/" represents failed specimen Fig. 5. Change in the embedded length of a steel fiber in SFRC and ASFRC [40] The reinforcement due to the action of the steel fibres on the toughness behavior can be determined from the ultimate strain Utu. As shown in Table 2, the mean values of Utu were similar for the SFRC specimens with different w/c-ratios. However, Utu of the ASFRC specimens varied greatly for the different w/c-ratios compared to the SFRC specimens. The increase in Utu for the ASFRC specimens is more pronounced compared to the matrix strength increase. Due to the alignment of the steel fibres, the toughness of the composites was significantly enhanced (strain-hardening occurred).

Energy dissipation
Improving energy dissipation (i.e., impact resistance and ductility) is an important aspect for the behavior of a structure [41][42][43]. Strain-hardening and strain-softening behaviors are characterized by the energy dissipated per unit volume gf-V (in the stage of strain-hardening) and energy dissipated per crack surface area Gf-A (in the post-peak stage/strain-softening) [22,44], respectively. The Gf-A of the cementitious materials can be increased significantly by aligning the steel fibres, as shown in Table 2 (the integral range for Gf-A was from Utu to 11%. Also note in Fig. 4 that there was no obvious hardening branch in the curves of the SFRC specimens; the gf-V of the SFRC could not be calculated). Compared with the SFRC specimens, the Gf-A of the ASFRC specimens (Vf =2.0% and w/c =0.32) increased by 56.87%, which is approximately equal to the increase in magnitude of fUtu (57.14%). Obviously, with the increase in matrix strength, the ascending amplitude of Gf-A for the ASFRC specimens is higher than that of the SFRC specimens; this can be seen in Figure 6. In this study, where the matrix of the SFRC and ASFRC specimens is without aggregates, the properties of the interface between the fibre and the matrix was significantly affected by the w/c. Two

4.2.2.Multiple cracking behavior of ASFRC as determined by the DIC method
The strain fields of ASFRC specimens at different moments during testing are presented in Figure 7 (Pmax represents the peak load, " " represents loading [i.e., ascending branch of stress-strain curve] and " " corresponds to measurements of the descending branch). For SFRC specimens (the results can be seen in Fig.   S4 in the Supplementary Material), a single micro-crack gradually evolved into a macro-crack with an increase in the tensile load. After Pmax was surpassed, due to the debonding and pull-out of the fibres, the rapid macro-crack propagation led to the complete failure of the specimens. In comparison with the ASFRC specimens, the whole process of single crack development is faster in SFRC specimens, and the tensile strain of the specimen is smaller. As shown in Figure 7, the first micro-crack appears between 70 to 80% of Pmax for ASFRC specimens with w/c=0.32 and Vf =2.0%. Then, multiple cracks are formed in the matrix as the load was increased to about 90% of Pmax (during the hardening stage). In the softening stage, with the development of the macro cracks, the specimen gradually loses its bearing capacity. The multiple cracking behavior for the ASFRC specimens with a lower fibre content was not observed during tension.
The orientation number of the steel fibres in the ASFRC specimens (w/c=0.32, Vf =2.0%) is around 0.90, as determined by the X-CT images analysis. Obviously, the number of fibres within the cracked section at complete failure of the ASFRC specimens (Nf-A) is higher than that for the SFRC specimens (Nf-R) due to higher fibre orientation number (the average Nf-R is 495 for the SFRC specimens, the average Nf-A is 612 for the ASFRC specimens [Vf =2.0%], which is a 23.6% increase). This will lead to higher and additional stresses being transferred to the surrounding matrix. If the additional tensile stresses transferred to the surrounding matrix exceed the cracking strength of the matrix, then further multiple cracks are formed. As more cracks are generated, the resistance of the steel fibres bridging these cracks contributes to the post-cracking toughness of the ASFRC composites which far exceeds the tensile capacity found in composites failing with the opening of only a single crack. Hence, the Utu of ASFRC is significantly higher than that of SFRC. It is worth mentioning that the micro-cracks were only revealed by the DIC equipment (they could not be identified from a visual inspection).
The formation of multiple cracks of SFRC specimens was often accompanied by strain-hardening (i.e., good ductility performance), especially for UHPFRC material, where the uniaxial tensile test results of UHPFRC with 2.0% hybrid fibres (1.0% macro-fibres and 1.0% micro-fibres) also produced multiple cracking [33]. The multiple cracking also was observed during the execution of uniaxial tensile test with high-performance fibre reinforced cementitious composites (HPFRCC) with 2% Torex fibres [36]. Similar results can be found in the uniaxial tensile tests for SFRC with large hooked-end fibre contents (Vf =5.0%) [45] and for UHPFRC with over 2% of fibres [22,32,34,35]. Generally, these specimens with multiple cracks also show superior tensile behavior and post-cracking ductility [46]. Compared with SFRC, multiple cracks can be formed in ASFRC at a lower fibre content, and the post-cracking toughness and tensile behavior are significantly improved.

Discussion
Results of ASFRC uniaxial tensile tests with w/c=0. 36 and Vf =1.2% are shown in Fig. 8 (ASFRC specimens with Vf =0.8%, 1.6% and energy dissipation (Gf-A) with different Vf also can be seen in Figure S5 and Figure S6 in the Supplementary Material, respectively). From Figure S5(a) in the Supplementary Material, it can be seen that even when the fibres are aligned, if the fibre content is low (Vf =0.8%), there will be insufficient fibres in the cross-section of the cracks to influence the toughness of the ASFRC specimen. As Vf increases, strain-hardening becomes more significant. However, there is no similar strain-hardening branch in the curve for the SFRC specimens (w/c=0.36) with Vf =2.0% (see Fig. 8), which is further evidence that the post-cracking tensile behavior and toughness can be significantly enhanced by the alignment of the steel fibres, especially for specimens at a relatively high fibre content.
From the tests results of the ASFRC specimens (w/c=0.36) with different Vf and those of the SFRC specimens (w/c=0.36) with Vf =2.0% (as shown in Fig. S5 and Fig. 8 considering the orientation of the aligned fibres. Moreover, the energy dissipation capacity of the ASFRC specimens far exceeds that of the SFRC specimens (see Fig. 8), which is a significant practical benefit. It should be noted that the performance of ASFRC in other directions is less than that of SFRC.

Conclusions
The ASFRC specimens were prepared using a uniform magnetic field formed from energized coils during casting, and the effect on the alignment of steel fibres was examined using an X-ray CT system. The influence of fibre orientation on the strain-hardening and fracture properties of SFRC specimens were investigated. The process of multiple crack propagation in the ASFRC specimens subjected to uniaxial tension was assessed using the DIC method. Based on the analysis of the test results, the main conclusions of this study are as follows: (1) Compared with SFRC, ASFRC performs better in terms of uniaxial tensile strength (fUtu), post-cracking toughness and energy dissipation (Gf-A). With an increase in the matrix strength and fibre content, this improvement becomes more obvious. Based on these results, the alignment of steel fibres using a magnetic field technology appears to be an effective method to produce high performance cementitious composite materials.
(2) The uniaxial tensile behavior of ASFRC specimens (w/c=0.36) with Vf =1.2% is equivalent to that of SFRC specimens (w/c=0.36) with Vf =2.0%; this results in at least 40% saving in fibre content if only considering the orientation of aligned fibre. Moreover, the ASFRC specimens exhibit strain-hardening after first crack formation during the uniaxial tension tests (for Vf > 0.8%) and an significant improvement in the post-peak ductile behavior is obtained (again this improvement increases with an increase in fibre content).
(3) Compared with ordinary SFRC with dispersed fibres, the multiple cracks for ASFRC (w/c=0.32) can be formed at a low fibre content (Vf=2.0%) due to the alignment of fibres perpendicular to the crack surface, the post-cracking toughness and tensile behavior is significantly improved for cementitious materials. The initial cracking load of the ASFRC specimens under tension is around 70% Pmax; multiple cracking takes place in the matrix up until the tensile load of around 90% Pmax.