Splitting tensile strength of recycled tire steel fiber‐reinforced alkali‐activated slag concrete designed by Taguchi method

Despite offering several benefits in terms of environment and economics, few studies were conducted on the ground granulated blast furnace slag (GGBFS) based alkali‐activated concrete reinforced with recycled tire steel fibers (RTSFs). Therefore, this paper employed the Taguchi method to investigate and optimize the influence of GGBFS content, the alkaline solution to GGBFS content ratio, sodium hydroxide solution concentration, and RTSF volume fraction on splitting tensile strength (STS) of recycled tire steel fiber‐reinforced alkali‐activated slag concrete (RTSFR‐AASC). The microstructural evaluation was done through scanning electron microscopy, and Fourier transform infrared spectroscopy (FTIR) analyses. The results indicated that the fiber content, the most efficient parameter on STS, led to a considerable enhancement in the performance characteristic. Microstructural analysis proved the variations in formation of C–S–H gel. The results of the confirmation experiment on the proposed optimized RTSFR‐AASC mixture with the highest 28‐day STS (7.11 MPa) confirmed the effectiveness of the Taguchi method.


| INTRODUCTION
Concrete, as the second most consumed material, 1 is a major contributor to the emission of greenhouse gases. The manufacturing process of ordinary Portland cement as binding material in concrete is responsible for around 5%-7% of total CO 2 emission. 2 Hence, developing a green alternative to conventional cementitious composites is of the utmost importance. In this regard, the alkali-activated concrete (AAC), which has been occasionally named as geopolymer concrete, can reduce the carbon footprint by producing 0.16 tons of CO 2 . 3 It has also the potential to iron out challenges of limited available landfills for dumping industrial waste materials such as ground granulated blast furnace slag (GGBFS) and fly ash by employing them as an eco-friendly source of secondary raw materials.
Alkali-activated materials are formed through the reaction of aluminosilicate sources, which should be rich in silica (Si 4+ ) and alumina (Al 3+ ), 4 in the presence of the alkaline medium. Alkali-activated concrete has shown excellent properties, for example, higher compressive strength and fire resistance, sustainability, and lower carbon dioxide emission, which have been vividly discussed in several studies. [5][6][7][8][9] However, one of the main technical drawbacks identified in these green materials is the lack of sufficient tensile strength. 10 Several studies have shown that the addition of fiber is a practical solution for this weakness. [11][12][13][14] A total of 60 million tons of different types of fibers are annually employed for producing fiber-reinforced concrete. 15 Wang et al. 16 evaluated the construction performance and the compressive strength of basalt-fiber-reinforced concrete with varying fiber lengths and fiber volume fractions subjected to an axial load. They asserted that adding an appropriate amount of basalt fiber can effectively improve the ultimate bearing capacity of short concrete columns, with maximum and average increases of 28% and 24%, respectively.
Among industrial available fibers, steel fibers are one of the most persuasive ones employed for reinforcing the concrete. Wei et al. 17 investigated the compressive behavior of circular steel tube-confined reinforced ultra-highperformance concrete (UHPC) columns by varying the addition of steel fibers. The failure mode of all the columns was shear failure except for the explosive failure of the reinforced UHPC (RU) columns without fibers. Moreover, the steel fibers addition immensely enhanced the strength and ductility of the RU columns. Steel fibers production is energy-intensive 18 and demands a massive amount of raw materials, 19 which makes its final cost high and, thus, limits its widespread use in the construction industry. 20 Finding an eco-efficient and sustainable alternative to commercially steel fibers has been at the center of the attention of scholars.
In recent decades, the world has been faced with challenges related to the vast amounts of waste tires, which will reach 1.2 billion tons yearly by the end of the 2030s. 21 Burning and also stockpiling waste tires pose a significant threat to the environment and human health. 19,22,23 In this regard, a remarkable turnaround has been observed in US end-of-life tires (ELTs) consumption; 11% of annually produced ELTs consumed by end-use markets in 1990 turned into 75.8% in 2019. 24,25 In 2001, the EU implemented several restrictions on ELTs landfilling and reported a 92% recovery rate in 2018. 26 Tire Stewardship Australia reported that 69% of the 466,000 tons of waste tires generated in Australia in 2018-2019 were recovered for reuse or processing into tire-derived products or in thermal processing. 27 With the help of the extended producer responsibility system, Brazilian have achieved ELTs recovery rate of 99%. 28 Some Asian countries, including Japan and South Korea, have reached 92% and 88% recovery rates, respectively, by developing efficient policies in waste management. 29,30 In general, recycling tires has been encouraged as a sustainable strategy for the safe disposal of ELTs.
Employing recycled tire steel fiber (RTSF), derived from the recycling process of tires, as reinforcement in concrete not only is a promising way to address the issues related to scrap tires but also it could be economically beneficial as its price is much lower than industrial fiber. Several pieces of research studying the engineering properties of cementitious concrete reinforced with RTSF are available in the literature. 21,[31][32][33][34] However, the number of studies investigating the influences of incorporation of RTSF in geopolymers is limited. [35][36][37] In addition to the fiber, other main parameters, including aluminosilicate source, type of alkaline activator, combination and concentration of activator, and alkaline solution to binder ratio, 38 may be influential in the performance of recycled tire steel fiber-reinforced alkali-activated slag concrete (RTSFR-AASC), which should be considered. Although investigating the influence of multiple parameters on performance simultaneously might be possible, it requires a large number of experiments as per full factorial design, taking considerable time and cost. 39 However, the Taguchi method, as an optimization technique that uses orthogonal arrays (OAs), can be employed to evaluate and optimize the influences of parameters on performance with a minimum number of experiments. 40 To the best knowledge of authors, Khalaj et al. 41 employed the Taguchi method to determine the optimum mix proportion of slag-based geopolymer concrete reinforced with industrial steel fiber (ISF) by considering the effects of water curing regime, sodium hydroxide (NaOH) concentration, alkaline activator to cement weight ratio and fiber weight on splitting tensile strength (STS). Aghaie et al. 42 used the Taguchi method to investigate the effects of the curing time, amount of the flat-end fibers, weight ratio of the anhydrous borax to NaOH solution, and weight ratio of the alkaline activator to the slag on geopolymer concrete and reported the highest STS (10.08 ± 0.86 MPa).
It is crystal clear from the above review that the development of GGBFS based AAC reinforced with RTSF has beneficial environmental and economic impacts, and the effect of these steel fibers recovered from scrap tires on the tensile strength of AAC is still inconclusive. On the basis of these considerations, the present investigation is motivated by the production RTSFR-AASC to fill this research gap and develop alkali-activated composites. Therefore, The Taguchi method was employed to arrive at the optimal mix proportion for RTSFR-AASC with regard to STS. Nine experimental runs were evaluated by considering the influence of GGBFS content, the alkaline solution to GGBFS content (Alk/GGBFS) ratio, NaOH solution concentration, and the RTSF volume fraction. The concept of signal to noise (S/N) ratio was performed to evaluate how examined parameters influence performance output and determine the optimum level of parameters. The analysis of variance (ANOVA) was also applied to identify the percentage contribution of each parameter on STS. Microstructure status of mixtures with maximum and minimum STS and fiber-matrix interface were also investigated through scanning electron microscopy (SEM) and Fourier transform infrared spectroscopy (FTIR) analyses, respectively. Finally, verification experiments were conducted on the proposed optimized RTSFR-AASC mixture to assess the efficiency of the Taguchi method for alkali-activated composites.

| Materials
The aluminosilicate source material in the current study was GGBFS, which was derived from the local steel plant. The chemical analysis performed using X-ray fluorescence spectrometry and physical properties of employed GGBFS are summarized in Table 1.
From the chemical properties, the basicity coefficient, which is the ratio between the total content of basic constituents (CaO + MgO) and the total content of acidic constituents (SiO 2 + Al 2 O 3 ), 43 was calculated as 1.0. Therefore, the utilized GGBFS was classified as a neutral type, 44 suitable as a binder.
As the CaO/SiO 2 ratio was between 0.5 and 2.0, and the Al 2 O 3 /SiO 2 ratio was between 0.1 and 0.6, 4 the chosen GGBFS could be considered suitable to be employed as the binder. As can be seen from Figure 1, which presents the micrograph of aluminosilicate material obtained from the SEM analysis, the particles of GGBFS were angular in shape.
The crushed granite with a maximum size of 12.5 mm and the river sand with fineness modulus of 3.24 were selected as the coarse and fine aggregates, respectively. The grading curves of the aggregates are presented in Figure 2. Blends of NaOH and sodium silicate (Na 2 SiO 3 ) solutions were used as the alkaline activator, and their  Table 2. Polycarboxylate ether-based superplasticizer, which has the best plasticizing effect on geopolymers activated by NaOH and Na 2 SiO 3 , 45 was used to improve the workability of AASC. The adopted RTSFs were provided from a local recycling factory where scrap tires from cars and trucks were collected and shredded. Then their embedded steel fibers were derived by magnetic forces, which were utilized to reinforce the mixtures in this study. Figure 3 shows the extracted RTSFs with the average length and diameter of 16.20 and 0.23 mm, respectively. The chemical analysis of employed RTSF, obtained by Atomic Emission spectrometry analysis, is presented in Table 3.

| Method
The Taguchi method as a process optimization technique was employed in this research to obtain the optimum mix proportion for RTSFR-AASC. The procedure of the Taguchi method is presented in Figure 4 and stated as follows.
Step 1: Determination of control parameters and examined response.
Previous experimental results indicating the effect of different parameters on performances of AAC are listed in Table 4. As can be seen, most of them investigated the properties of fly ash-based geopolymer concrete, and very limited studies evaluated the effects of parameters on STS. Table 5 presents the effects of RTSF on the properties of alkali-activated composites reported in previous studies. The given data would shed light on the research gap that existed in this field; that is, the influence of RTSF on STS of AASC is still ill-defined, and there is also no research examining the impact of mix proportion parameters and RTSF volume fraction simultaneously.
Based on the Tables 4 and 5, in this research four examined parameters, including GGBFS content, Alk/GGBFS ratio, NaOH solution concentration, and RTSF volume fraction, labeled by A, B, C, and D, respectively, were considered. Three levels were chosen for each examined parameter. The selected parameters and their variation levels are illustrated in Table 6. Less resistance to tensile loading has become one of the most technological drawbacks of alkali-activated composites, which limits their structural use. Therefore, the STS was selected as the quality response of RTSFR-AASC.
Step 2: Selection of Taguchi OA. The Taguchi method developed the OAs to decline the number of experiments which must be performed in traditional design like full factorial. The selection of OA depends on the total degrees of freedom (DOF) of the process, which is the sum of DOF of each examined parameter. 59 The number of levels of each parameter minus one gives the individual DOF for each parameter. 59 As four parameters with three levels each were selected, the individual DOF of each parameter was equivalent to two, and the total DOF of the process was T A B L E 2 Chemical properties of sodium silicate (Na 2 SiO 3 ) and sodium hydroxide (NaOH)

Materials
Properties Results Unit Recycled tire steel fiber used for recycled tire steel fiber-reinforced alkali-activated slag concrete equal to 8. Since the OA's DOF should not be smaller than the total DOF, 60 the Taguchi L 9 OA, which had the same DOF (8), could be selected. The experimental runs as per the L 9 OA are presented in Table 7.
Step 3: Conducting experiments and calculating S/N ratios.
Taguchi method uses a statistical measure of performance known as S/N ratio to minimize the influence of uncontrollable parameters (noise), while maximizing the impact of examined parameters (signal). 61 Depending on the required objective characteristic, the S/N ratio can be categorized into three types and computed as follows 62 : (1) Larger is the better (2) Smaller is the better (3) Nominal is the best As this research aims to maximize the STS of RTSFR-AASC mixes, the S/N ratio was calculated according to Equation (1).
Step 4: Performing Analysis of variance. A statistical technique of ANOVA was employed to determine the importance order of examined parameters on response. As the DOF of error was zero due to the selection of over-fitted design, the F-value, which is the ratio of the mean of squared deviations to the mean of squared errors, 60 could not be computed. Therefore, the percentage contribution of each examined parameter was calculated directly by computing the ratio of the sum of squares of each parameter (Equation [4]) to the total sum of squares of all examined parameters (Equation [5]) based on Equation (6). 60,63 Step 6: Conducting confirmation experiment. The confirmation experiment was performed to investigate the validity of the employed method. The optimum value of the S/N ratio can be calculated corresponding to the optimum level of control parameters using Equation (7). 61

| Preparation and testing of specimens
The mixing sequence of the RTSFR-AASC is illustrated in Figure 5. The NaOH solution was prepared in the required concentrations a day before mixing in order to reach the ambient temperature and remove the negative influence of its excessive heat on setting time. 4,64 The NaOH solution with considered concentrations (10, 12, and 14 M) was prepared by mixing NaOH pellets, whose properties is shown in Table 2, with tap water. As the molecular weight of NaOH is 40, the 10 M NaOH solution contains 10 Â 40 g = 400 g of NaOH solids per liter of the solution. The mass of NaOH solids was measured as 155 g per kg of NaOH solution with a concentration of 10 M. After a day, the alkaline solution was first prepared by mixing Na 2 SiO 3 and NaOH solutions around 30 min before mixing with dry materials, suggested by Aliabdo et al. 52 to enhance the properties of geopolymer mixture. The ratio of Na 2 SiO 3 solution to NaOH solution was constant at 1.5. The superplasticizer and extra water were also added into the prepared alkaline liquid and thoroughly mixed to obtain a homogenous liquid component. The dosage of 1% SP, reported by Jithendra and Elavenil 3 as an optimum percentage of SP for aluminosilicate based geopolymer concrete, was considered for this research. Afterward, coarse and fine aggregates, prepared in saturated surface dry condition according to ASTM C128-15 65 and ASTM C127-15, 66 respectively, and GGBFS were mixed in the 20-L pan mixer for 4 min. The premixed liquid component was then added to the mixer and mixed with other ingredients for further 4 min to get uniform nature of the concrete. Finally, fibers were gradually added into the mixture to prevent the balling phenomenon. The detailed mix proportions of the trial mixes are summarized in Table 8. The fresh mixture was poured in cylindrical molds with 200 mm length and 100 mm diameter (200 Â 100) and compacted on a vibration table for the 30 s. Plastic sheets covered the inside surfaces of the molds ( Figure 6[a]) to prevent interaction between steel and fresh alkali-activated matrix. The molds were then stored in the laboratory with a temperature of 20-23 C. The specimens were removed from their molds 24 h after casting and covered with plastic bags for 28 days. The splitting tensile tests were performed on cylindrical specimens using a universal testing machine ( Figure 6[b]) at a constant loading rate of 0.6 kN/s. The STS of RTSFR-AASC was calculated by using Equation (8) according to ASTM C496/C496M-11. 67 In order to investigate the microstructure of fiberreinforced alkali-activated concrete, SEM and FTIR analyses were performed. After completion of STS tests on cylindrical specimens, the fiber-matrix interface in a selected broken sample was observed in SEM images taken by scanning electron microscope: SERON TECH-NOLOGY -AIS2100 model. The functional groups in association with the formed compounds in alkaliactivated mixtures were determined by FTIR analysis so that a mixture of 2 mg of alkali-activated powder and 200 mg of potassium bromide (KBr) was prepared for sample pellets. Then, the FTIR spectra were obtained with the help of Thermo-AVATAR model spectrometer in transmission mode at a rate of six scans per spectrum.

| RESULTS AND DISCUSSION
The results of STS of RTSFR-AASC mixes at 28 days are presented in Figure 7. From this, the highest STS of 7.09 MPa was obtained by TM3 with a GGBFS content of 300 kg/m 3 , Alk/GGBFS ratio of 0.6, NaOH concentration of 14 M and RTSF volume fraction of 1%. In contrast, the TM5 with a GGBFS content of 350 kg/m 3 , Alk/GGBFS ratio of 0.5, NaOH concentration of 14 M and RTSF volume fraction of 0.5% had the lowest STS of 4.73 MPa. The RTSF volume fraction was the major difference between the highest and the lowest mixes which might led to the higher STS. It should be mentioned that in the mixtures reinforced with 1% RTSF, the minimum STS was measured as 6.6 MPa, which was higher than the maximum STS (6.52 MPa) obtaining among the mixtures containing 0.75% RTSF. Moreover, the maximum STS (4.95 MPa) of mixes incorporating 0.5% RTSF was 20% lower than the minimum STS (6.12 MPa) which was achieved in the mixtures reinforced with 0.75% RTSF. Thus, the fiber content might play a prominent role in the performance of STS of RTSFR-AASC.
Khezrloo et al. 68 revealed that the curing time was one of the most influential parameters, and had a considerable impact on STS of slag-based boroaluminosilicate geopolymers. Therefore, to investigate this parameter, the STS test was performed on TM3 and TM5 at 7 and 14 days of T A B L E 5 Effects of recycled tire steel fiber (RTSF) inclusion on some properties of alkali-activated composites  curing from the day of preparation and obtained results are illustrated in Figure 8. As can be found from Figure 8, the STS of mixes improved with curing time. At the early ages (i.e., 7 days), mixtures obtained more than 60% of the 28-day STS; however, as the curing time increased beyond 7 days, the rate of gaining strength decreased. Aghaie et al. 42 reported the STS of the fiber-reinforced boroaluminosilicate geopolymer containing 1% flat-end steel fibers at 7 days (5.47 MPa). They also reported that as the amount of flat-end steel fibers increased from 1% to 7%, the 7-day STS increased around 11% (from 5.47 to 6.14 MPa). Although the achieved STS for TM3 with 1% RTSF (5.12 MPa) in the current study was very close to STS obtained for flat-end steel fiber-reinforced geopolymer, a much higher increasing rate of STS with RTSF content was achieved; as the RTSF volume fraction increased from 0.5% to 1%, the 7-day STS increased 73% (from 2.95 to 5.12 MPa). This difference couldbe related to the shape of fibers. Irregular and deformed shape of RTSFs could increase the anchorage between fibers and aggregates, leading to huge amount of pull-out energy in comparison with flat-end steel fibers.

| Effect of GGBFS content
The mean S/N ratio of STS of RTSFR-AASC mixtures for GGBFS content are shown in Figure 9. As can be seen from Figure 9, the mean S/N ratio of STS was observed to decrease with the increase of the GGBFS content. F I G U R E 5 Adopted mixing protocol for recycled tire steel fiber-reinforced alkali-activated slag concrete (RTSFR-AASC). GGBFS, ground granulated blast furnace slag; RTSF, recycled tire steel fiber.
This can be attributed to the negative effect of GGBFS on workability, reported in previous studies, 53,56 because of the angular shape of slag ( Figure 1). Reduction in workability led to an increase in porosity and a decrease in compactness within the internal structural of RTSFR-AASC. Thus, the local fractures could easily appear near the pores of micro cracks, and reduce STS. However, when the GGBFS content reached 400 kg/ m 3 , the mean S/N ratio experienced a minor decline. This may be ascribed to the higher total alkaline liquid content in mixes due to a rise in GGBFS content in the Alk/GGBFS ratio. Thus, this excessive amount of liquid content increased the particle mobility of the source materials and compensated the loss of workability.

| Effect of Alk/GGBFS ratio
The mean S/N ratio of STS of RTSFR-AASC mixtures for Alk/GGBFS ratio are presented in Figure 10. Considering the Alk/GGBFS ratio, the mean S/N ratio of STS slightly increased by raising the Alk/GGBFS ratio. This improvement can be attributed to the Si/Al ratio, which was increased by raising the amount of alkaline activator solution. Therefore, the density of Si-O-Si bonds, which are stronger than Si-O-Al and Al-O-Al linkages, increased and led to a rise in STS. 69

| Effect of the NaOH solution concentration
The mean S/N ratio of STS of RTSFR-AASC mixtures for the NaOH solution concentration is illustrated in Figure 11. About the influence of NaOH solution concentration, the mean S/N ratio of STS increased with increasing NaOH solution concentration from 10 to 12 M. This can be ascribed to the increased breakage of Ca-O, Al-O, and Si-O bonds in GGBFS, which consequently contributes to the formation of sodium aluminosilicate and increased STS. 48,70 However, increasing concentration beyond 12 M led to a slight drop. The increase of NaOH solution concentration made the alkaline solution more viscous which made the fresh RTSFR-AASC mixes more cohesive and stickier, leading to the loss of workability and reduction in strength. Aliabdo et al. 51 registered a similar drop in 28-day tensile strength of fly ash-based geopolymer concrete as the NaOH concentration increased above 16 M.

| Effect of RTSF volume fraction
The mean S/N ratio of STS of RTSFR-AASC mixtures for RTSF volume fraction are reported in Figure 12.  As observed from Figure 12, the mean of S/N ratio of STS significantly increased when the RTSF volume fraction increased. This enhancement can be mainly attributed to the high efficiency of fibers in arresting the cracks, 31 which resulted in a higher splitting tensile performance.
It should be mentioned that the improvement in the mean S/N ratio of STS due to the increase of RTSF volume fraction from 0.5% to 0.75% was slightly higher than that of 0.75% to 1%. This can be ascribed to a large  (Figure 13), leading to weaker interfacial transition zones. 71 Therefore, local fracture can be more easily formed near the fibers, 21 causing a loss in the efficiency of fiber. Figure 14 shows the fracture mechanism in STS for RTSFR-AASC. As the STS of the alkali-activated matrix was reached, a longitudinal crack (Figure 14[a]) initially appeared at the center of the specimen. After the initiation of the main crack, the tensile stresses were transferred across the failure's cross section, 32 in which the fibers restrained the extension of cracks in concrete. Once the load increased and the main crack propagated along the direction of the load transfer, 33 secondary cracks ( Figure 14[b]) formed in specimen, developed in depth and parallel to the main crack. 72 Through the bridiging mechanisim, shown in Figure 14(c), when the applied stressess overcome fiber-matrix bond strength, the fibers pulled out from the alkali-activated matrix. Meanwhile, a small number of fiber rupture might took place in twisty fibers which demanded higher fiber-pull out energy. 73 Thus, the dominant failure mode of RTSFR-AASC was a pull-out failure accompanied with a small number of fiber ruptured.
Previous study 41 indicated that the mean STS of slagbased geopolymer concrete containing 1% ISF was around 5.15 MPa, which was 32% lower than the mean F I G U R E 1 1 Mean signal to noise (S/N) ratio of splitting tensile strength of recycled tire steel fiber-reinforced alkaliactivated slag concrete for the sodium hydroxide (NaOH) solution concentration F I G U R E 1 2 Mean signal to noise (S/N) ratio of splitting tensile strength of recycled tire steel fiber-reinforced alkali-activated slag concrete for recycled tire steel fiber (RTSF) volume fraction F I G U R E 1 3 Both sides of failure surface of failed specimen STS (6.84 MPa) recorded in this study for the mixtures (TM3, TM4, and TM8) reinforced with 1% RTSF. This is attributed to the interfacial bonding between fibers and matrix.
The RTSFs were a hybrid of short and long fibers which had different mechanism in bridging the cracks. The short steel fiber can restrict the micro cracks. As the load increases and the micro cracks extend and merge into the macro ones with higher crack width, the long fibers can restrain the macro cracks in an efficient manner. 74 Thus, the bridging effect of short and long RTSFs, each in different stage of loading, results in a higher resistance to the loading. Moreover, based on the results reported in Reference [32], RTSF possess rougher surface with deeper grooves than ISF. Thus, a stronger bond can be formed at the interfacial of RTSF and matrix. As a result of these two facts, the greater enhancement in STS was attained in AAC reinforced with RTSF compared to those AAC in which ISF were used.

| SEM analysis
To investigate more, the SEM micrographs of RTSF surface texture, the failure surface of the matrix, fibermatrix interface and the surface of the RTSF after pull out were taken so that the interfacial bond between RTSF and matrix could be assessed. Figure 15(a) shows SEM picture of RTSF, revealing that residual rubber impurities were still attached to the RTSF surface. From Figure 15 (b), it can be seen that the corrugation at the surface of the fiber was not limited to the direction of axis which would lead to a higher potential for a frictional bond with the matrix. 12 The pores are shown by red circles in Figure 15(c), which could be caused by the micro bubbles formed while curing. 75 As can be seen in Figure 15(c,d), RTSF adhered well to the alkali-activated matrix to such an extent that the surface of it was stuck by some residues of hardened matrix. This may be attributed to the hydrophilic characteristics of the RTSF that let the fresh alkali-activated paste cover the fiber and make a strong contact with it. 10 F I G U R E 1 4 Fracture surface of recycled tire steel fiberreinforced alkali-activated slag concrete specimen (TM5) under loading. RTSF, recycled tire steel fiber.

| FTIR analysis
FTIR analysis was conducted in order to observe the chemical bonds in alkali-activated mixtures. Figure 16 displays the FTIR spectra of the TM3 and TM5 mixtures with maximum and minimum STS at the age of 28 days, respectively. The asymmetric bending vibration of Si-O-Si at 451 cm À1 and Al-O bond at 668 cm À1 in alkali-activated mixtures were ascribed to the unreacted precursor materials. 56 The broad band ranging from 966 to 1003 cm À1 was attributed to the asymmetric stretching vibration of Si-O-T (T = tetrahedral Si or Al) bonds, indicating the geopolymer reaction process and formation of C-S-H gel. 56,[75][76][77] The exact shifting of wavenumbers and the span of shifting rely on the Si/Al ratio of the precursor materials and the reaction conditions. 77 When Si is replaced with Al in the Si-O-T bond, its angle reduces, and due to the smaller bond force constant of Si-O-Al in comparison with that of the Si-O-Si bond, a lower wavenumber is achieved for the main band of the alkaliactivated system. 56,77 A lower wavenumber as a result of the formation of the strong Si-O-Al bonds implies the strengthening of this bond, leading to higher mechanical strength in early ages. 75 TM3 mixture enjoyed a higher amount of alkaline solution, lower amount of binder, and lower amount of extra water (i.e., free form of water) than the TM5 mixture, which could be the major reason for the formation of Al-rich geopolymer gel, shifting Si-O-T bond to lower wavenumbers and, finally, higher STS.
The band varying from 1404 to 1413 cm À1 was related to the stretching vibration of CO 3 2À , indicating carbonated alkalis remained in pores. 35,56,77 The peak centered around 1650 cm À1 and the broad band varying from 3505 to 3556 cm À1 obtained from alkali-activated mixtures were associated with the bending vibration of the H-O-H group and stretching vibration of -OH group, respectively, which were related to the water absorbed in the rings of geopolymeric products. 35,56,76

| Analytical assessment
Based on the experimental results, the average STS of RTSFR-AASC was related to the influential parameters, that is, GGBFS content, the Alk/GGBFS ratio, NaOH solution concentration, and RTSF volume fraction. The equations were derived by employing the linear regression approach and presented in Figure 17. As can be found in Figure 17, a strong relationship was found between the average STS of RTSFR-AASC and GGBFS content, the Alk/GGBFS ratio, and RTSF volume fraction with the coefficient of determination (R 2 ) higher than 0.779. The rate of RTSFR-AASC gaining strength can be evaluated by the gradient of the obtained equations. The maximum rate of improving STS was observed when rising RTSF content. All other parameters had negligible influence on strength gaining rate. Derived empirical equations would help form realistic opinions of the changes in RTSFR-AASC behavior considering various parameters without conducting costly and time-consuming experimental procedures.

| Optimization of examined parameters of RTSFR-AASC
The optimum level of each experimental parameter is determined based on the highest mean of S/N ratio. Thus, according to the Figures 9-12, optimal STS value could be obtained at A 1 B 3 C 2 D 3 combination with a GGBFS content of 300 kg/m 3 , alkaline solution to GGBFS F I G U R E 1 6 Fourier transform infrared spectroscopy spectra of TM3 and TM5 mixtures at 28 days content ratio of 0.6, NaOH solution concentration of 12 M and RTSF volume fraction of 1%.

| Analysis of variance
The ANOVA results of the S/N ratio for STS of RTSFR-AASC mixes, computed by the Minitab software based on Equations (4) and (5), and the percentage contribution of each parameter, calculated as per Equation (6), are presented in Table 9. From this table, the volume fraction of RTSF was the most significant parameter affecting the STS with a percentage of participation of over 97%. Other examined parameters had negligible effects on strength with less than 3% participation.
Contour plots ( Figure 18) drawn with the help of Minitab software, were used to investigate the variation of response (STS), considering the effect of RTSF as the most influential parameter than the other ones. The contour plot for the mean S/N ratio for varying RTSF volume fraction and GGBFS content is illustrated in Figure 18(a). It revealed that increasing the RTSF volume fraction in lower levels of GGBFS content resulted in a higher mean S/N ratio. On the other hand, at higher levels of Alk/GGBFS ratio and NaOH solution concentration (Figure 18[b,c]), when the RTSF volume fraction increased, the maximum mean of S/N ratio in the range of 16-16.8 was obtained.

| Verification test
The verification experiment was carried out to compare the STS of the A 1 B 3 C 2 D 3 mixture (named as TM10), which was established as an optimum combination of examined parameters in this paper based on the Taguchi method and TM3 mixture, which was considered as the initial optimum mix with the highest STS among nine suggested mixtures (Figure 7) based on L 9 OA. Also, the optimized mixture A 1 B 3 C 2 with no RTSF was considered the control mix. Thus, the STS was conducted at TM10 and control mix. The data of the confirmation experiment are presented in Table 10.
The predicted S/N ratio of optimized mixture (17.08), computed based on Equation (7), was close to obtained S/N ratio (17.05) based on confirmation test results. The improvement on S/N ratio indicated that new combination of optimal parameters performed better than the TM3 mixture with respect to the STS. The STS of optimized RTSFR-AASC mix was highest among all strengths which was obtained for nine initial mixtures ( Figure 7). As well as this, as can be seen from Table 10, the control mix obtained the STS around 3.30 MPa, showing the considerable influence of RTSF in RTSFR-AASC. A previous study reported that the STS of slag-based boroaluminosilicate geopolymers reinforced with 7% of flat-end steel fiber (10.8 MPa) was 25% more than the STS of the plane specimen (8.10 MPa); however, in this study, the STS of TM10 which was reinforced with 1% RTSF was more than twice the STS of the plane mix (control mix).

| CONCLUSION
As mentioned in literature, ACC is a promising green construction technology compare to the well-known traditional Portland cement concrete and RTSF is an ecoefficient alternative to the industrial type. Thus, in this study special attention was paid to determine optimum mix proportions for RTSFR-AASC by employing Taguchi method. In the application of this method, GGBFS content, Alk/GGBFS ratio, the NaOH solution concentration, and RTSF volume fraction were selected as examined parameters with response of 28-day STS. Based on experimental results, following conclusions can be drawn: 1. The STS of RTSFR-AASC increased with increasing RTSF volume fraction and decreased slightly as the GGBFS content increased. STS was also found to improve with the increase in Alk/GGBFS ratio and NaOH solution concentration until a certain limit (12 M). 2. According to the mean S/N ratio, the RTSFR-AASC with a GGBFS content of 300 kg/m 3 , Alk/GGBFS ratio of 0.6, NaOH solution concentration of 12 M, and RTSF volume fraction of 1%, was suggested as optimal mixture regarding STS. 3. The results of ANOVA revealed that the fiber content was the most influential parameter with the greatest impact (97.08%) on STS, followed by GGBFS content (2.77%), NaOH solution concentration (0.15%), Alk/GGBFS ratio (0.01%). 4. The shifting of the peak related to the Si-O-T bond in the FTIR spectra of mixtures with maximum and minimum STS confirmed the influence of different parameters on geopolymer reaction process and the formation F I G U R E 1 8 Contour plots of (a) signal to noise (S/N) ratio versus ground granulated blast furnace slag (GGBFS) content, recycled tire steel fiber (RTSF) volume fraction, (b) S/N ratio versus alkaline solution to ground granulated blast furnace slag content (Alk/GGBFS) ratio, RTSF volume fraction, and (c) S/N ratio versus sodium hydroxide (NaOH) solution concentration, RTSF volume fraction of C-S-H gel. The SEM images also revealed the appropriate adhesion of the alkali-activated mixture to RTSF. 5. The results of the verification experiment revealed that the optimal combinations of the examined parameters attained by Taguchi method performed well in STS.
The information presented in this study takes a step forward in development of GGBFS based alkali-activated slag concrete reinforced with RTSF. It is worth mentioning that using GGBFS, a by-product of iron and steel making industries, and reusing RTSF, abstracted from ELTs, are encouraged to produce RTSFR-AASC because of their environmental benefits.

Notations
In Equations (1)-(3), n presents the number of samples of each mixture, y i shows the experimental data of ith sample of each mixture, y is the average of the experimental data and s 2 y is the variance of y. In Equation (4), where n is the number of experiments at level jth of parameter ith, S/N ji represents the mean of S/N ratio at level jth of parameter ith and S/N mean is the overall mean of the S/N.
In Equation (5), where S/N i shows the S/N ratio of ith mixture and S/N mean is the overall mean of the S/N.
In Equation (6), PC i shows the percentage contribution of ith parameter, SS i represents the sum of squares of the ith parameter and SS t shows the total sum of squares of all examined parameters.
In Equation (7), η presents predicted S/N ratio of the optimized mix, p shows the number of examined parameters η m is the total average of the S/N ratio and the average of S/N ratio at the optimal level for the ith parameter is given by η i .
In Equation (8), STS is shown by T, P is the maximum applied load indicated by the testing machine, l presents length of specimen, and diameter of specimen is illustrated by d.

ACKNOWLEDGEMENTS
Open access publishing facilitated by The University of Western Australia, as part of the Wiley -The University of Western Australia agreement via the Council of Australian University Librarians.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.