Strength Design of Ultra-High-Performance Fiber-Reinforced Cementitious Composites Using Local Ecological Admixture

: The ultra-high-performance ﬁber-reinforced cementitious composite (UHPFRC) is a new generation of building material with extremely high mechanical strength and durability, which can be used for ultra-high, thin-wall or long-span construction, that prolongs the service life of construction in severe environments. In this study, UHPFRC was prepared with a high range of local ecological admixture to decrease the material’s cost and the environmental impact. Raw materials’ proportions, water/binder ratio, ﬁber-volume contents, and hybrid-ﬁber ratio were studied on the property improvement of UHPFRC, and an F-test analysis was induced to reveal the important signiﬁcance on compressive strength. The results demonstrated that the compressive strength of 237.8 MPa was achieved with mineral admixture substitution over 40%. The particle-packing density and the binder reactivity both succeeded on the compressive strength. Water/binder ratio determined the hydration degree and the ﬂowability of UHPFRC, which affected compressive strength through hydration products and microstructure. Also, compressive strength was more sensitive with hybrid-ﬁber than ﬁber-volume content. The order of importance for compressive strength was powder proportion > hybrid-ﬁber ratio > ﬁber-volume content > water/binder ratio.


Introduction
With the fast developing of urbanization in decades, high performance and long service life are becoming the primary acquirements for construction materials due to the growing demands and the environmental friendship development [1][2][3][4][5]. Ultra-high-performance fiber-reinforced cementitious composite (UHPFRC) is an advanced construction material with super-high mechanical properties and durability [6][7][8][9][10][11]. The compressive strength of UHPFRC is commonly higher than 150 MPa, and the permeability is suggested to be 5-10 times lower than normal concrete as well [12,13]. So far, UHPFRC plays an increasingly important role in several fields, such as infrastructure construction, super engineering, and structural repairment [14][15][16][17], to improve the performance and prolong the service life. However, a large fraction of high-activity binders and fine powders were necessary in UHPFRC to guarantee the performance [18][19][20][21][22]. For example, cement dosage usually ranged from 900 to 1100 kg/m 3 in UHPFRC [20,23,24]. High cement demand caused a series of problems, including exceeded consumption of natural resources, large CO 2 emissions, high hydration heat, large volume shrinkage, and expensive material costs [11,[25][26][27]. Hence, industrial by-products and solid wastes were attempted to substitute cement or quartz sand to decrease the environmental impacts and materials costs [25,[28][29][30][31][32][33][34], and even some also presented positive effects on shrinkage deformation and corrosion resistance of UHPFRC [35,36]. Ultra-high strength is the most important advantage of UHPFRC that caused both engineering and economic benefits [37]. High strength offered sufficient possible constructions on engineering design and structural evolution [38,39]. UHPFRC was preferred for some super projects for enough structural stability and service life [40], for instance, the river-cross bridge with main span larger than 1 km. High strength UHPFRC also matched better with the high strength steel rebar than normal concrete. In addition, ultra-high strength declined with the usage of construction materials per volume, making buildings lighter, thinner, and lower cost [41]. To achieve the ultra-high strength, high amounts of fine active binders and autoclave curing were commonly induced [42], while the substitution of cement with mineral admixtures was against the strength development. Ali Sadrmomtazi [21] found that with an increase of silica fume replacement, the compressive strength decreased both at 7 and 28 days. According to references, the compressive strength of UHPC was mostly less than 200 MPa when substituted with mineral admixtures and without autoclave curing [20,25,26,29,35]. Unfortunately, using substitutable materials and the non-autoclave method to prepare UHPFRC with compressive strength over 200 MPa is still a challenge.
UHPFRC is a complex and expensive mixture, and its performance far varies with the raw-materials proportion and curing regime. Although several standards were proposed by regions and organizations recent years, it was still hard to achieve a formulary approach for UHPFRC preparation. Moreover, to reduce the cost and environmental impact, ternary or quaternary active industrial by-products and solid wastes were often selected as supplementary cementitious materials (SCMs) to substitute cement or aggregates in UHPFRC [30][31][32][43][44][45][46]. Yang et al. [35] used phosphorous slag to induce a low-carbon design of UHPFRCC, and the compressive strength slightly decreased even with 50 wt% of cement substituted by slag, and the autogenous shrinkage was reduced by 45.6% from the maximum. Li et al. [36] developed quaternary blends, including cement, groundgranulated blast-furnace slag (GGBS), limestone powder (LP), and microsilica, to evaluate the properties of UHPFRC. A sustainability efficiency of 130% was reached, with the decrease of CO 2 about 59%. However, the properties of these ternary/quaternary blends of UHPFRC were far different due to their materials' categories and mix proportions. So far, the performance for UHPFRC products mostly depended on a series of experimental tests rather than theoretical predication. Some researchers studied the design procedures of UHPFRCC [47,48]. Shi et al. [11] considered the cost and CO 2 emission during UHPC preparation and calculated the compressive strength and embodied CO 2 index to design the UHPC with low environment impact. Arora et al. [20] studied the effect of matrix and fiber volume on the mechanical strength and predicted the compressive and flexural strength through the digital-image-correlation method.
In this work, UHPFRC with compressive strength exceeding 200 MPa was designed with a high volume of SCMs and non-pressed curing method, and the strength development factors and the influence efficiency were also studied to guide the proportion design. Fly ash and silica fume as SCMs and local river sand as aggregates were applied to conduct the degree of environmental impact. The effects of the materials' aspects, including binder proportion, water-to-binder ratio, steel-fiber content, and hybrid-fiber ratio, on workability and strength development were investigated. To evaluate the significance of each variable on strength development, an F-test based on ANOVA(Analysis of Variance) was induced. The calculation of F follows from Equation (1) [49]: where SS a is the sum of squared deviations of factors, f a is the freedom degree of factors, SS e is the sum of squared deviation for experimental error, and f e is the freedom degree for SS e . The higher F value indicates that the corresponded factor was more significant for strength development. The results could be used to guide the materials' mix proportions for UHPFRC and helpful for strength adjustment if necessary.

Raw Materials and Mix Proportion
The binder materials used in this study were P·I 52.5 Portland cement according to Chinese standard and class F fly ash (FA) and silica fume (SF) from a local powder plant as mineral admixture. The chemical composition and critical physical properties of binders are listed in Table 1. Local river sand with a fineness modulus of 2.4 and maximum grain size of 2.36 mm was utilized as fine aggregate. The particle-size dispersion of raw powder materials is present in Figure 1. High-range polycarboxylate-superplasticizer powder was carried to afford an available flowability of cementitious mixture. Two fibers with different geometry, straight steel fiber (SSF) and end-hooked steel fiber (HSF), were used to reinforce the mechanical properties of UHPFRC. The physical index of fibers is listed in Table 2. To reveal the effect of the materials' proportions on UHPFRC with high strength, powder gradation (M16B1, M16B2, and M16V0), water-to-binder ratio (W/B, M14V1, M15V1, M16V1, and M17V1), fiber content (M16V0, M16V1, and M16V2), and hybrid-mixed ratio (M16V1, M16H1, M16H2, M16H3, M16H4, and M16H5) were varied and the influences on workability and mechanical development were investigated. In experiments, the value for each variable was determined based on our previous work [33,34,50,51] for powder gradation, where PC:FA:SF:Sand = 5:4:1:12, 6:3:1:12, and 6:3:1:10, respectively. W/B ranged from 0.14-0.17, volumes fraction of steel fiber were 1.5%, 2.0%, and 2.5%, and the ratios of SSF:HSF = 1:0, 4:1, 3:1, 2:1, 1:1, and 0:1, respectively. The details of mix proportion are listed in Table 3. and the influences on workability and mechanical development were investigated. In ex-periments, the value for each variable was determined based on our previous work [33,34,50,51] for powder gradation, where PC: FA: SF: Sand = 5:4:1:12, 6:3:1:12, and 6:3:1:10, respectively. W/B ranged from 0.14-0.17, volumes fraction of steel fiber were 1.5%, 2.0%, and 2.5%, and the ratios of SSF: HSF = 1:0, 4:1, 3:1, 2:1, 1:1, and 0:1, respectively. The details of mix proportion are listed in Table 3.    In order to avoid the disruption from sand and steel fibers, some paste samples with the same mix proportion but without sand and steel fiber were also prepared for XRD and TG tests.

Mix Procedure
Due to low W/B, a reversed mix procedure was adopted to maximize the dispersion effect of polycarboxylate superplasticizer. Powder materials, including binders, sand, and superplasticizer, were first stirred together at low speed for 1 min to mix well. After that, all powders were added into water under low-speed stirring for 5-8 min until a uniform mortar was formed; then high-speed stirring was applied for another 1 min to achieve an available flowability. At last, the steel fibers were mixed and stirred for 2-3 min for thorough dispersion. The fresh UHPFRC mixture was casted and placed under standard conditions (20 ± 2 • C, RH ≥ 95%) for 48 h until demolded, then was steam curried at 85 • C for designed age.

Particle-Size Distribution
The particle-size distribution of binder materials was measured by a laser particle-size analyzer (LPSA), and ethanol was selected as solvent to avoid hydrated reaction. An amount of 5 g powder was measured and ultrasonic dispersed in 100 mL of ethanol for 3 min; then the mixed solution was dropped into the sample chamber of the equipment until enough of the amount formed. For each sample, the test was repeated twice, and we calculated the results for the average.
Since the particle size of sand was mostly beyond the measuring range of the LPSA, the curve of sand-size distribution was fitted from the grading curve.

Flowability of Fresh UHPFRC
The flowability of UHPFRC was evaluated following Chinese standard (GB/T 2419-2005). The fresh matrix was first filled into a cone-shaped mold on a platform of a jolting table. Then the mold was vertically removed and switched on the table in time. After jolting 25 times, two perpendicular diameters were measured, and the mean value was record as the flowability.

Mechanical Strength
The flexural and compressive strengths were measured according to the standard BS EN 196-1. Three prism specimens with the size of 40 mm × 40 mm × 160 mm were prepared for each mix proportion. The flexural strength was determined by three-point bending with the upload rating of 2.4 kN/s, the ultimate strength for each prism specimen was recorded, and the average value was calculated. Then, six broken ends after flexure testing were used for axial compression tests, and the average value of maximum strength was used as the final compressive strength.

Hydration Products
The crystalline structure and the content of hydration products were investigated by X-ray diffraction (XRD) and thermal gravimetric analysis (TG). After steam curing at 85 • C for the designed age, the paste samples were stopped of hydration by alcohol and grinded into powder to pass through the 80 µm sieve. During XRD testing, the sample was scanned from 5 • to 90 • (2θ) at the rate of 0.15 s/step, with the step of 0.02 • (2θ). During TG testing, the sample was ignited from 25 • C to 1000 • C, with the rate of 10 • C/min under nitrogen atmosphere.

Morphology
A scanning electron microscope (SEM) was used to observe the morphology of paste samples by the backscattered electronic (BSE) mode, with the operating voltage of 15 kV. For sample preparation, the paste fragments were first stopped of hydration at the designed age and dried at 40 • C in a vacuum oven for 48 h, then were impregnated by epoxy under vacuum at 40 • C for 48 h until hardened. After that, the samples were polished by sand paper to the roughness of 0.1 µm. Finally, carbon film was coated on the sample surface to avoid charging during the experiment.

Raw-Materials Proportion
Based on the MAA model [43], the optimal packing grading curve was calculated from Equation (2) and is presented in Figure 2 as the target curve. The actual integral grading curves from the series of materials' proportions are also compared in Figure 2. The diversion between the target curve and actual curves is evaluated by the determination coefficients (R 2 ) and listed in Table 4.  The actual particle-packing curves for different powder proportions are also shown in Figure 2. It is seen that the actual curves are not perfectly fitting with the target one. Compared with the target curves, more of a fraction of the large-sized particles over 500 μm existed, and more particles smaller than 10 μm were still preferred. The determination coefficients, which are listed in Table 4, were all beyond 0.9. Notably, the sample M16B2, composed with the proportions of PC:FA:SF = 6:3:1, as well as the binder-to-sand ratio of 1:1.2, presented the most-closed grading curve to the target one, and supposedly had the densest packing proportion.
The flexural and compressive strengths of samples with different material proportion were also tested, and the results are shown in Figure 3.  The actual particle-packing curves for different powder proportions are also shown in Figure 2. It is seen that the actual curves are not perfectly fitting with the target one. Compared with the target curves, more of a fraction of the large-sized particles over 500 µm existed, and more particles smaller than 10 µm were still preferred. The determination coefficients, which are listed in Table 4, were all beyond 0.9. Notably, the sample M16B2, composed with the proportions of PC:FA:SF = 6:3:1, as well as the binder-to-sand ratio of 1:1.2, presented the most-closed grading curve to the target one, and supposedly had the densest packing proportion.
The flexural and compressive strengths of samples with different material proportion were also tested, and the results are shown in Figure 3.
In Figure 3, the strength of the densest packing of M16B2 was higher than that of M16B1, but less than that of sample M16V0, which contained the highest weight percentage of cement. The results implied that the packing density was beneficial to strength development, but the synergistic effect of packing density and cement dosage resulted in the best strength performance.
coefficients, which are listed in Table 4, were all beyond 0.9. Notably, the sample M16B2, composed with the proportions of PC:FA:SF = 6:3:1, as well as the binder-to-sand ratio of 1:1.2, presented the most-closed grading curve to the target one, and supposedly had the densest packing proportion.
The flexural and compressive strengths of samples with different material proportion were also tested, and the results are shown in Figure 3.

Water-to-Binder Ratio
Water was necessary to supply available flowability and promote the hydration reaction of UHPFRC. To investigate the effect of water on strength development, the water-tobinder ratio (W/B) for samples was varied from 0.14 to 0.17, keeping the other parameters constant. Once the W/B was 0.14, the fresh matrix was too dry to cast, and when the W/B increased to 0.17, the flowability of the fresh sample much improved, where even a water film formed on the matrix surface, as seen in Figure 4. The flowability of samples almost linearly increased from 167.5 mm to 280.0 mm with increasing W/B, as seen in Figure 5. It is well accepted that more water leads to better fluidity. However, the viscosity of fresh UHPFRCC also changed significantly with different W/B ratios and will be studied in further work. In Figure 3, the strength of the densest packing of M16B2 was higher than that of M16B1, but less than that of sample M16V0, which contained the highest weight percentage of cement. The results implied that the packing density was beneficial to strength development, but the synergistic effect of packing density and cement dosage resulted in the best strength performance.

Water-to-Binder Ratio
Water was necessary to supply available flowability and promote the hydration reaction of UHPFRC. To investigate the effect of water on strength development, the waterto-binder ratio (W/B) for samples was varied from 0.14 to 0.17, keeping the other parameters constant. Once the W/B was 0.14, the fresh matrix was too dry to cast, and when the W/B increased to 0.17, the flowability of the fresh sample much improved, where even a water film formed on the matrix surface, as seen in Figure 4. The flowability of samples almost linearly increased from 167.5 mm to 280.0 mm with increasing W/B, as seen in Figure 5. It is well accepted that more water leads to better fluidity. However, the viscosity of fresh UHPFRCC also changed significantly with different W/B ratios and will be studied in further work.     The strength development with different W/B ratios is revealed in Figure 6. The compressive strengths of samples were all over 180 MPa, with a maximum of 216 MPa. It is usually known that lower W/B ratios are positive for strength development. However, in this research, with the W/B increasing, the compressive strength was first increased to the maximum at W/B = 0.16, then decreased again, and the flexural strength was decreased about 12.6% with the incremental W/B ratio.
UHPFRC can be defined as the hardened cementitious composites formed by unreacted binder and aggregates bonded with hydration gel, where the gel content is important to immobilize the skeleton. Since cement was not fully reacted in UHPFRC, the dosage of hydration gel was increased with the raised W/B ratio, leading to a better bonding property. On the other hand, the strength of gel was lower than aggregates and unreacted binder, so the extensive gel caused worse strength performance. Proper gel content is vital for strength improvement. To examine this inference, the hydration degree and the hydration-product content were analyzed by TG and SEM. The TG curves for samples are presented in Figure 7. There were several mass-loss steps that corresponded to the thermal reaction of the composites with water vapor or CO2 released during the heating process. The weight loss below 105 °C could be attributed to the evaporation of free water and interlayer water from C-S-H. The mass loss between 105 °C and 410 °C was due to the dehydration of hydration products, including C-S-H, ettringite, and AFm. The reactions at 410-640 °C and 640-800 °C were due to the decompositions of Ca(OH)2 and CaCO3, respectively [52]. The step at the temperatures between 800 °C and 900 °C was possibly caused by the phase transformation from αH' C2S to αL' C2S. UHPFRC can be defined as the hardened cementitious composites formed by unreacted binder and aggregates bonded with hydration gel, where the gel content is important to immobilize the skeleton. Since cement was not fully reacted in UHPFRC, the dosage of hydration gel was increased with the raised W/B ratio, leading to a better bonding property. On the other hand, the strength of gel was lower than aggregates and unreacted binder, so the extensive gel caused worse strength performance. Proper gel content is vital for strength improvement. To examine this inference, the hydration degree and the hydration-product content were analyzed by TG and SEM.
The TG curves for samples are presented in Figure 7. There were several mass-loss steps that corresponded to the thermal reaction of the composites with water vapor or CO 2 released during the heating process. The weight loss below 105 • C could be attributed to the evaporation of free water and interlayer water from C-S-H. The mass loss between 105 • C and 410 • C was due to the dehydration of hydration products, including C-S-H, ettringite, and AFm. The reactions at 410-640 • C and 640-800 • C were due to the decompositions of Ca(OH) 2 and CaCO 3 , respectively [52]. The step at the temperatures between 800 • C and 900 • C was possibly caused by the phase transformation from α H ' C 2 S to α L ' C 2 S.  Since the CaCO3 in samples was mostly generated from the carbonation of Ca(OH)2, the content of Ca(OH)2 and non-evaporated water can be calculated from Equations (3) and (4): (3) where WC and Wn are the contents of Ca(OH)2 and non-evaporated water in the sample, respectively. Ldh, Ldx, and Ldc are the relative mass losses on the TG curve due to the dehydration of C-S-H and ettringite, the dehydroxylation of Ca(OH)2, and the decarbonation of CaCO3, respectively. According to Equations (2) and (3), the content variation of Ca(OH)2 and the relative hydration degree compared to M14V1 were calculated and are shown in Figure 8. The CH content was increased to W/B = 0.15, then decreased, and the relative hydration degree reached the maximum when W/B = 0.16.  Since the CaCO 3 in samples was mostly generated from the carbonation of Ca(OH) 2 , the content of Ca(OH) 2 and non-evaporated water can be calculated from Equations (3) and (4): where W C and W n are the contents of Ca(OH) 2 and non-evaporated water in the sample, respectively. Ldh, Ldx, and Ldc are the relative mass losses on the TG curve due to the dehydration of C-S-H and ettringite, the dehydroxylation of Ca(OH) 2 , and the decarbonation of CaCO 3 , respectively. According to Equations (2) and (3), the content variation of Ca(OH) 2 and the relative hydration degree compared to M14V1 were calculated and are shown in Figure 8. The CH content was increased to W/B = 0.15, then decreased, and the relative hydration degree reached the maximum when W/B = 0.16.  Since the CaCO3 in samples was mostly generated from the carbonation of Ca(OH)2, the content of Ca(OH)2 and non-evaporated water can be calculated from Equations (3) and (4): (3) where WC and Wn are the contents of Ca(OH)2 and non-evaporated water in the sample, respectively. Ldh, Ldx, and Ldc are the relative mass losses on the TG curve due to the dehydration of C-S-H and ettringite, the dehydroxylation of Ca(OH)2, and the decarbonation of CaCO3, respectively. According to Equations (2) and (3), the content variation of Ca(OH)2 and the relative hydration degree compared to M14V1 were calculated and are shown in Figure 8. The CH content was increased to W/B = 0.15, then decreased, and the relative hydration degree reached the maximum when W/B = 0.16. Figure 8. The Ca(OH)2 content and relative hydration degree with W/B from 0.14 to 0.17. Figure 8. The Ca(OH) 2 content and relative hydration degree with W/B from 0.14 to 0.17.
It is well understood that more water promoted the hydration reaction of cement particles. However, the relative hydration degree decreased when W/B = 0.17 in this research.
Combined with the flowability results, it is assumed that the mortar at W/B = 0.17 was too moist so that part of water evaporated from the surface quickly after casting, causing a lower hydration degree. This result was also confirmed according to the strength test.
Moreover, compared between samples M15V1 and M16V1, although the hydration degree of M16V1 was higher, the content of CH was less than that of M15V1. This revealed that the pozzolanic reaction in M16V1 was more intensive and consumed more CH, and more secondary hydration gel was generated, which is also good for compressive strength.
The BSE images for UHPFRC with different W/B are also shown in Figure 9. The bright irregular particles were unreacted cement, and the round particles were FA. The marked hydration products in Figure 9 were the following: 1-inner C-S-H, 2-C-S-H from FA hydrated, 3-out C-S-H. From Figure 9, it is clear that unreacted cement and FA particles were surrounded by uniform C-S-H gel. However, with the W/B increased, more broken cement particles with blurry boundaries were found in Figure 9d, which implies that the hydration degree increased. It is well understood that more water promoted the hydration reaction of cement particles. However, the relative hydration degree decreased when W/B = 0.17 in this research. Combined with the flowability results, it is assumed that the mortar at W/B = 0.17 was too moist so that part of water evaporated from the surface quickly after casting, causing a lower hydration degree. This result was also confirmed according to the strength test.
Moreover, compared between samples M15V1 and M16V1, although the hydration degree of M16V1 was higher, the content of CH was less than that of M15V1. This revealed that the pozzolanic reaction in M16V1 was more intensive and consumed more CH, and more secondary hydration gel was generated, which is also good for compressive strength.
The BSE images for UHPFRC with different W/B are also shown in Figure 9. The bright irregular particles were unreacted cement, and the round particles were FA. The marked hydration products in Figure 9 were the following: 1-inner C-S-H, 2-C-S-H from FA hydrated, 3-out C-S-H. From Figure 9, it is clear that unreacted cement and FA particles were surrounded by uniform C-S-H gel. However, with the W/B increased, more broken cement particles with blurry boundaries were found in Figure 9d, which implies that the hydration degree increased.

Fiber Content and Fiber Geometry
The fracture of UHPFRC specimens during strength testing was caused by the unstable propagation of cracks, while fibers exhibit outstanding resistance on crack formation and propagation due to the bridge effect. In our previous work, the pull-out energy of fibers with different geometry and the flexural strength of hybrid-reinforced UHPFRC were discussed in [34]. The ability of fibers for crack control was dominated by fiber geometry, fiber content, and fiber hybrid. The strength development of UHPFRCC varied with fiber-volume content and hybrid-steel-fiber ratio, which are presented in Figures 10 and 11, respectively. around (46.5 ± 2) MPa, as shown in Figure 11.
The results implied that the compressive strength was not sensitive for steel-fiber content, and when the fiber exceeded the amount used, the strength decreased due to more initial defects. While HSF was beneficial for compressive-strength development, the strength increased with more HSF in the available range. However, the flexural strength performed opposite to the compressive strength. The flexural strength increased with the raised usage of steel fiber, and the increased range was about 25% when the steel-fibervolume fraction increased from 2% to 3%. As the fiber volume remained constant, the flexural strength vibrated in a narrow range.  around (46.5 ± 2) MPa, as shown in Figure 11.
The results implied that the compressive strength was not sensitive for steel-fiber content, and when the fiber exceeded the amount used, the strength decreased due to more initial defects. While HSF was beneficial for compressive-strength development, the strength increased with more HSF in the available range. However, the flexural strength performed opposite to the compressive strength. The flexural strength increased with the raised usage of steel fiber, and the increased range was about 25% when the steel-fibervolume fraction increased from 2% to 3%. As the fiber volume remained constant, the flexural strength vibrated in a narrow range.  In Figure 10, the compressive strength and flexural strength, respectively, of UHPFRC were 201. 3 and 44.4, 202.4 and 48.5, and 192.3 and 55.6 MPa, with the steel-fiber volume fractions of 2%, 2.5%, and 3%, respectively. Moreover, as the HSF/SSF ratio increased from 0% to 100%, the compressive strength first increased from 203.7 MPa to a maximum of 237.8 MPa, then decreased to 206.8 MPa again, while flexural strength changed slightly around (46.5 ± 2) MPa, as shown in Figure 11.
The results implied that the compressive strength was not sensitive for steel-fiber content, and when the fiber exceeded the amount used, the strength decreased due to more initial defects. While HSF was beneficial for compressive-strength development, the strength increased with more HSF in the available range. However, the flexural strength performed opposite to the compressive strength. The flexural strength increased with the raised usage of steel fiber, and the increased range was about 25% when the steel-fibervolume fraction increased from 2% to 3%. As the fiber volume remained constant, the flexural strength vibrated in a narrow range.
The essential fact of compressive or flexural fracture is the through-crack formed under stress, and the fiber improves strength by resisting the crack extension due to the bridge effect. Under flexural stress, a large crack formed in the bending area, and the crack-resistance property of steel-fiber was more dependent on the friction during the pull-out process. The higher fiber-volume fraction increased with the resisted fiber amount.
However, in compressive cases, the crack width was much smaller than the former, so the anchoring effect worked significantly on crack resistance. Since HSF performed with higher adhesion strength, the compressive strength increased with the raised ratio of HSF/SSF. However, exceeded fiber content or HSF ratio also brought distribution problems, which were harmful for strength development.

The Effect Efficiency Index F
From the above results and discussions, the UHPFRC with local SCMs achieved 237.8 MPa after 7 days of steam curing. However, the compressive strength varied significantly with different material proportions. To reveal the dominant factor for strength design, the importance index of different material aspects was evaluated by the F-value.
The significance of powder packing, W/B ratio, steel-fiber content, and hybrid-fiber ratio on strength development was evaluated by the efficiency index F, and the value for each aspect was calculated according to Equation (1) and is listed in Table 5. It is generally accepted that W/B ratio is the key factor for the ultimate strength of UHPFRC, and the lower W/B ratio refers to the higher strength. However, from this research, the compressive strength was more sensitive with powder packing than the fiber geometry. The higher strength of UHPFRC was contributed from the less initial defect and strong bonding strength between components. The dense packing powders helped to construct a compact skeleton for UHPFRC, and less C-S-H gel was needed to bind particles and fill pores. In practical application, the particle-size distribution of the component and the component proportion should be carefully selected for ultra-high-strength UHPFRC design. The W/B ratio affected the rheological properties of fresh mortar and the hydration degree, as well as the pore volume and pore structure, whose influence on strength was indefinite and should be cautiously experimented on in practical applications.

Conclusions
In this research, FA, SF, and river sand from local areas were used as substitution materials to design ecological UHPFRC, and the effect of materials for strength development was investigated. The importance of factors on strength was evaluated by the F-value of the ANOVA method. The following conclusions can be drawn: (1) With a high range of substitution materials, the compressive strength of UHPFRC was beyond 180 MPa, and the maximum strength reached 237.8 MPa after 7 days of steam curing. (2) The powder proportion played the key role in strength development. The strength increased with the synergetic effect of compact packing and high activity of binders. (3) The fiber content was more efficient for flexural strength. Additionally, the compressive strength was more sensitively with the ratio of HSF, which performed well with high adhesion strength. However, the massive usage of fiber and HSF brought distribution problems that decreased the compressive strength. (4) The W/B ratio ranged from 0.14 to 0.17 and was hesitated for use with compressive strength in practical applications. High W/B ratio was preferred to improve the flowability and increase the hydration degree but also induced more pores and cracks when water was consumed.
(5) The significance of different materials' aspects on compressive strength was evaluated by the F-value with the ANOVA method, and powder proportion was most important in the strength design of UHPFRC.