Construction and Building Materials 121 (2016) 310–318

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier .com/locate /conbui ldmat

Combined effect of coarse aggregate and fiber on tensile behavior ofultra-high performance concrete

http://dx.doi.org/10.1016/j.conbuildmat.2016.05.0390950-0618/� 2016 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: ljz@cnjsjk.cn (J. Liu).

Jianzhong Liu ⇑, Fangyu Han, Gong Cui, Qianqian Zhang, Jin Lv, Lihui Zhang, Zhiqian YangState Key Laboratory of High Performance Civil Engineering Materials, Jiangsu Research Institute of Building Science, Nanjing 211108, China

h i g h l i g h t s

� Tensile behavior of UHPC incorporating coarse aggregate is investigated.� Coarse aggregate brings an impairment to utilization efficiency of fiber.� Coarse aggregate can be successfully introduced into system of UHPC.� Fiber type has almost no effect on strain hardening behavior of this UHPC.

a r t i c l e i n f o

Article history:Received 5 February 2016Received in revised form 22 April 2016Accepted 4 May 2016Available online 9 June 2016

Keywords:Ultra-high performance concreteCoarse aggregateFiberBonding behaviorTensile behavior

a b s t r a c t

In this study, combined effect of coarse aggregate and fiber properties on tensile behavior of ultra-highperformance concrete (UHPC) was investigated. Four replacement levels of coarse aggregates (0%, 15%,25%, 35% by volume of mortar) and four types of steel fiber (three micro-fibers with a shape differenceand one macro-fiber) were considered. Results showed that replacement level of coarse aggregate hasa critical value of 25% and different fiber types act similarly in regard of compressive strength. Coarseaggregate brought impairment to bonding strength and utilization efficiency of fiber, especially fordeformed ones. Furthermore, coarse aggregate could be successfully introduced into system of UHPCwithout impairing its tensile properties at a favorable replacement level (625%). In addition, phe-nomenon of strain hardening behaviors of UHPC incorporating coarse aggregate could be triggered byfurther increasing fiber dosage to larger than 2.5%, however, it was independent of fiber type due to com-bined effect of coarse aggregate and fiber bridging.

� 2016 Elsevier Ltd. All rights reserved.

1. Introduction

For the design of concrete structures with good long-term per-formance, insight into microstructure of concretes and its relation-ship with performance is of paramount importance [1–2]. As result,ultra-high performance concrete (UHPC) appeared, a new class ofconstruction material with outstanding properties. UHPC has theremarkable mechanical properties and extremely low porosity,which was obtained through dense particle packing. Therefore, itimplies that UHPC performs a high durability, improved resistanceagainst various chemicals as well as higher penetration resistance[3–5]. Presently, there has been increasing interest in the use ofUHPC as a vanguard product for industrial and structuralapplications in the past few years, such as coupling beams inhigh-rise buildings, precast members, infrastructure repairs andspecial facilities like nuclear waste storage containers [6–7].

Basic principles of developing UHPC have been established byRichard and Cheyrezy in 1995 [3]. Normally, UHPC is designed tobe a super plasticized concrete by replacement of traditional coarseaggregates with fine sand [8]. In addition, thermal treatments ofUHPC are mainly preferred in order to reduce curing time andincrease mechanical strengths [9–10]. The well-chosen raw mate-rial and sophisticated technical procedures make it too costly tomeet the demand of large-scale project engineering. To reduce itscost, natural sand has been employed successfully as a substitutefor expensive quartz sand without impairing its mechanicalstrengths [6,11]. Specially, coarse aggregate is recently introducedinto the system of UHPC to further reduce its cost and broaden itsapplication [12–13]. Firstly, coarse aggregate is more economicefficiency than other raw materials used. Secondly, it makes UHPCpossess a better shrinkage performance and lower hydrationtemperature rise [14], which is attributed to the decreased bindercontent and its strong restraint on shrinkage of UHPC. With theinclusion of coarse aggregates, UHPC can also easily possess a highcompressive strength of 180 MPa as well as excellent workability

J. Liu et al. / Construction and Building Materials 121 (2016) 310–318 311

[13]. Unfortunately, the interface around coarse aggregate wouldbecome more weak and more flaws are accompanied [15], result-ing in a poor tensile performance of concrete, especially for thatwith a low water to binder (W/B) ratio, i.e., UHPC [16]. Therefore,for UHPC incorporating coarse aggregate, the optimal content ofcoarse aggregate should be investigated carefully in order to obtainbetter tensile properties.

In addition to the dense microstructure obtained by maximizingpacking density with very fine minerals (silica fume, slag, fly ash,etc.) [8,17–20], superior performance of UHPC is also achieved byenhancing matrix toughness with optimal steel fiber reinforce-ment (smooth, twist, hook, hybrid etc.) [9,21–25]. Effects of steelfiber on the tensile behavior of UHPC without coarse aggregatehave been widely reported in literatures [5,21,24,26–29]. It isaccepted that the tensile behavior of UHPC is controlled by bondbehavior between fiber and matrix, which is attributed to matrixproperties (including particle size and mechanical properties),fiber properties (including geometry, length, diameter, volumecontent, and mechanical properties) and interfacial properties.Some researches show that UHPC produced from macro-fiberswith deformed geometry (i.e. hooked, twist) provides the excellentperformance with respect to post cracking strength, strain capacityand multiple cracking behavior, which is attributed to the high uti-lization efficiency [21,24]. However, utilization efficiency ofsmooth fiber can be improved by designing ultra-high strengthmatrix, which allows development of high tensile stress in thesmooth fiber [26]. When coarse aggregates are introduced intoUHPC, the matrix should be tailored carefully to obtain a highstrength and a better bond condition. Besides, fiber distributioncharacteristic is another critical factor influencing post-crackingbehavior of UHPC [29–31]. In addition to casting method, fiberproperties (i.e. geometry, shape, volume content), fluidity andshape of forms, the existence of coarse aggregates have a strongeffect on fiber distribution [31–32]. Consequently, pull-out behav-ior of varying steel fiber in concrete behaves differently comparedwith that without coarse aggregates. Therefore, to use UHPC incor-porating coarse aggregate in civil infrastructure, combined effect ofcoarse aggregate and fiber on the tensile behavior of UHPC must becarefully investigated. However, to author’s knowledge, there is lit-tle information available in literatures about the combined effect.

Accordingly, in this paper, combined effect of coarse aggregateand fiber on tensile behavior of UHPC was studied. Three typesof steel micro-fibers with varying geometry and one type of steelmacro-fiber were applied as reinforcement. Tests had been doneto obtain overall tensile stress-strain curves of UHPC. Bondingbehavior between these fibers and matrix was also performed toanalyses combined effect of coarse aggregates and fibers.

Fig. 1. Particle size distribution of cementitious materials.

2. Experimental program

2.1. Raw materials

Portland cement, with a strength class of 52.5 conforming to theChinese Standard GB 175-2007 [33], silica fume, ultra-fine slag andfly ash were used as cementitious materials. Their specific gravitieswere 3.15 g/cm3, 1.87 g/cm3, 2.84 g/cm3 and 2.33 g/cm3

Table 1Chemical composition of cementitious materials.

Composition Al2O3 CaO SiO2 Fe2O3 K2O

Cement 4.94 63.05 19.95 2.92 0.66Silica fume 0.16 0.18 97.31 0.15 0.39Ultra-fine slag 16.8 37.1 31.3 0.43 0.35Fly ash 37.7 4.64 46.6 4.27 0.23

respectively, and their chemical properties, determined by X-rayfluorescence (XRF) (ThermoFisher Scientific ARL QUANTX), areillustrated in Table 1. Moreover, their particle size distributions(PSD) determined by nitrogen sorption isothermal measurement(Coulter ONLNISORP 100 CX) are given in Fig. 1. River sand withan apparent density of 2.63 g/cm3 and grain size below 5 mmwas applied. Crushed basalt with an apparent density of 2.86 g/cm3 and grain size between 5 mm and 20 mm was used as coarseaggregate. A polycarboxylate-based superplasticizer with solidcontent of 40% was adopted as water-reducer. To investigate effectof fiber on the tensile behavior of UHPC incorporating coarseaggregate, three types of steel micro-fibers and one type of steelmacro-fiber were performed. Their properties are given in Table 2and Fig. 2 shows images of each fiber type.

2.2. Mix proportion

Table 3 provides mix proportion of the UHPC series in whichbinder composition is invariable and water to binder ratio is fixedat 0.18 for all mixes. The optimized UHPC binder was made with40% cement, 10% silica fume, 30% ultra-fine slag and 20% fly ash.Single fiber pullout tests were performed on the control specimenwithout coarse aggregate for avoiding the effect of coarse aggre-gate on pullout behavior of fiber. In an attempt to optimize coarseaggregate content with respect to tensile behavior, the coarseaggregate was partially replaced in mortar at four replacementlevels (0%, 15%, 25%, 35% by volume of mortar). It should be notedthat the W/B ratio and superplasticizer dosage of UHPC was ini-tially designed to be invariable for these mixes. However, as thecoarse aggregate replacement level increases, a satisfied fluidity(slump >100 mm) of fresh mixture was needed to realize thewell-dispersion of coarse aggregate and good homogeneity ofmatrix. Therefore, the superlasticizer dosage was adjusted for

MgO Na2O SO3 TiO2 MnO LOI

1.33 0.15 3.83 0.27 – 2.90.82 0.2 0.54 – 0.01 0.249.08 0.38 2.8 0.9 0.34 0.521.41 0.17 1.52 0.78 0.29 2.39

Table 2Properties of various fibers.

Fibertype

Name Diameter(mm)

Length(mm)

Aspectratio(�)

Density(g/cm3)

Tensilestrength(MPa)

Micro Smooth 0.20 13 (13.0)a 65 7.9 2940Spiral 0.20 13 (14.2)a 59 7.9 2860Hooked A 0.20 13 (13.8)a 59 7.9 2940

Macro Hooked B 0.60 30 (32.4)a 50 7.9 1890

a Numbers in parentheses are the nominal length of fiber in the straight form.

Table 3Composition of UHPC mixture by weight ratio

Notation Binder Fineaggregate

Coarseaggregate

Super-plasticizer

Water Fiber

SM2G0a 1 1 0 (0%)b 0.018 0.18 0.133(1.75%)c

SM2G1a 1 1 0.49(15%)b

0.024 0.18 0.155(1.75%)c

SM2G2a 1 1 0.93(25%)b

0.025 0.18 0.170(1.75%)c

SM2G3a 1 1 1.51(35%)b

0.030 0.18 0.194(1.75%)c

SM0G2a 1 1 0.93(25%)b

0.025 0.18 0 (0%)c

SM1G2a 1 1 0.93(25%)b

0.025 0.18 0.097(1.00%)c

SM3G2a 1 1 0.93b

0.025 0.18 0.243c

312 J. Liu et al. / Construction and Building Materials 121 (2016) 310–318

these mixes to achieve a satisfied workability and it has no impair-ment to mechanical strength of these UHPC in the latter age. Thepreferred formulation was further fixed to investigate effect offiber type and dosage.

(25%) (2.50%)

SP2G2a 1 1 0.93(25%)b

0.025 0.18 0.170(1.75%)c

HA2G2a 1 1 0.93(25%)b

0.025 0.18 0.170(1.75%)c

HB2G2a 1 1 0.93(25%)b

0.025 0.18 0.170(1.75%)c

a SM, SP, HA and HB is represent of smooth, spiral, hooked A and hooked B fiber.b Numbers in parentheses are the volume content of coarse aggregate.c Numbers in parentheses are the volume content of fiber.

2.3. Specimen preparation and curing

The specimen preparations were conducted at temperature of20 ± 5 �C and relative humidity of about 65%. A forced single-axismixer with a rotational speed of 45 rpm was used to prepare mix-ture. Cementitious materials and aggregates were first dry-mixedfor about 1 min. After that, water pre-mixed with superplasticizerwas then added gradually and mixed for another 2 min. Then fiberswere dispersed carefully by hand into the mixture and mixed foranother 2 min. The cement mixture with fibers was casted in amold and vibrated for 10 s after workability test. Specimens werecovered with plastic sheets and stored at room temperature for1 day, then moved to a standard curing room with temperatureof 20 ± 2 �C and relative humidity of greater than 95%. Compressivetests were conducted after 7, 28 and 56 days curing and the tensiletests were carried out after 28 days curing.

To assess bonding strength between steel fibers and UHPCmatrix, a type of dog-bone mold with a dimension of235 mm � 25 mm � 25 mm was adopted (as shown in Fig. 3). Aslot with 2 mm width was located in the middle of mold. A PVC

Fig. 2. Images of fibers: (a) smooth micro-fiber, (b) spiral micro-fi

slice with a single fiber passing through the center was insert intothe slot. The embedded lengths of the single fiber on both sides ofslice are both half of fiber length. Specimens were then immersedinto the saturated Ca(OH)2 solution. Pullout tests were also carriedout after 28 days curing.

2.4. Experimental methods and characterization

2.4.1. Fresh properties and compressive strengthFluidity of fresh concrete, including slump and spread values,

was measured with a normal 300 mm slump cone according to

ber, (c) hooked A micro-fiber and (d) hooked B macro fiber.

Fig. 3. Schematic of dog-bone specimen for fiber pull-out tests.

J. Liu et al. / Construction and Building Materials 121 (2016) 310–318 313

ASTMC143. Spread refers to the averagediameterof concrete spreadin this paper. Air content of fresh mixes was determined by an airentrainment meter (Sanyo LS-546) according to ASTM C173. Dueto the incorporation of coarse aggregate, cube specimens of100 mm � 100 mm � 100 mm were used for compressive strengthtesting according to the Chinese StandardGB/T 31387-2015. At leastthree parallel specimens were tested at each age point.

2.4.2. Steel fiber pullout testsBonding strength between yarns and matrix was tested by

using a CMT4103 electromechanical universal tensile machinewith 10 kN load cell. The stroke-control rate was set at 0.5 mm/min. Five parallel samples of each group were tested and averagevalues and standard deviation were presented. The curves ofpull-out load P per unit remaining embedded length versus slipdisplacement Ds were obtained. The remaining embedded lengthis the difference between initial embedded length L and slipdisplacement Ds.

2.4.3. Tensile tests of UHPCTensile tests of UHPC sample were performed on an

electromechanical universal testing machine (MTS-801) with500 kN load cell. The stroke-control rate was set at 0.05 mm/min.The geometry of test specimen and set-up are shown in Fig. 4.According to the previous work [24], tensile specimens with sizesof 100 mm � 100 mm � 500 mm was prepared in order to mini-mize the size effect of fibers, especially for the hooked-B fiber. Car-bon cloth and metal plates were successively glued onto specimensby using epoxy resin to alleviate localized damage and minimizedeformation of matrix at clamps. The gage length between metalplates was 100 mm. In addition, the metal frame was used to mea-sure elongation by using two LVDTs as shown in Fig. 4(a) and aver-aged value from the two LVDTs was used in calculating tensilestrain until peak tensile stress during tensile process. As shown in

Fig. 4. Tensile specimen geometry and te

Fig. 4(b), load was transferred through a pre-embedded steel barwith a diameter of 10 mm, and steel bar was welded with 8 ribsof 5 mm length to prevent its relaxation during testing. To keep loadaxis straight, boundary conditions at both ends of the tensile testset-up were driven by a cardan. The average values and standarddeviation of at least three parallel samples were acquired. Typicalstress-strain curves of different UHPC samples were comparedand first crack strength and maximum bridging stress wereacquired for all samples. The first crack strength rfc is defined asthe applied tensile stress at which a matrix crack spreads through-out the cross section of the sample under tension, and it can beadopted at the point where matrix is completely out of function.The maximum bridging stress rB is defined as the maximum stressthat bridging fibers can transfer across the crack of specimen [34].

3. Results and discussion

3.1. Workability and compressive strengths

Table 4 shows the results of workability and compressivestrengths of all sample series. With the increase of coarse aggre-gates content, UHPC mixture behaves a more poor workability asspread and slump values declined sharply. This is due to thedecreased binder content [13] and the more severe interlockbetween coarse aggregate and steel fibers. When the replacementof coarse aggregates improves to 35%, fresh mixture exhibits a verydry condition with slump of 130 mm. In addition, the decreasedbinder content also results in a low air content of fresh mixtureincorporated with coarse aggregate, where air content of freshpaste can be assumed to maintain unchangeable. As expected,workability of UHPC samples becomes worse with the increase offiber dosage. With incorporation of 1.0%, 1.75% and 2.5% smoothfiber, the flowability gradually decreased by 12.0%, 32.8% and46.0%, and it almost presents a linear tendency [35]. This phe-nomenon should be attributed to the increase in internal surfacearea that produced higher cohesive forces between fibers and con-crete matrix, with the increase of additional fiber content [8].Moreover, steel fibers are randomly distributed in the matrix andact as skeleton, and consequently hinder the flow of fresh concrete[36]. Among these fibers, the mixture with inclusion of hooked-Bfiber presents a preferred workability, followed by the mixturewith spiral fiber, smooth fiber and hooked-A fiber. The poor work-ability accompanied with deformed fibers is resulted from largerfriction induced by the deformed fiber, especially for hooked fiber[22]. Besides the observed difference between hooked fibers is due

st set-up: (a) test set-up, (b) Cardan.

Table 4Workability and compressive strengths of UHPC samples.

Notation Slump (mm) Spread (mm) Air content (%) Compressive strength (MPa)

7d 28d 56d

SM2G0 205 495 6.6 96.1 ± 3.5 99.8 ± 4.2 97.9 ± 5.9SM2G1 195 400 3.6 111.5 ± 5.2 123.0 ± 4.9 129.9 ± 3.9SM2G2 168 380 2.9 120.2 ± 6.5 131.1 ± 2.7 138.4 ± 4.1SM2G3 130 220 2.8 112.3 ± 3.9 125.1 ± 1.8 130.9 ± 4.2SM0G2 250 680 3.6 85.6 ± 6.4 96.9 ± 2.8 97.2 ± 3.1SM1G2 220 420 3.8 118.2 ± 6.9 124.7 ± 5.8 135.2 ± 5.1SM3G2 135 240 2.3 126.4 ± 4.7 138.5 ± 5.1 144.3 ± 5.7SP2G2 190 365 3.4 118.5 ± 2.3 121.2 ± 1.5 129.5 ± 3.7HA2G2 122 210 3.6 117.8 ± 2.8 127.1 ± 2.4 131.3 ± 4.2HB2G2 227 470 2.7 116.6 ± 3.4 128.9 ± 3.8 131.9 ± 5.0

Fig. 5. Shear force per unit remaining embedded length versus slip displacement ofpull-out tests for various steel fibers in matrix.

Table 5Frictional bonding strengths of varying steel fibers (MPa).

Fibers Smooth Spiral Hooked-A Hooked-B

Frictional bondingstrength

8.60 ± 0.53 18.37 ± 0.85 24.65 ± 0.42 12.17 ± 0.31

314 J. Liu et al. / Construction and Building Materials 121 (2016) 310–318

to the lowest number of hooked-B fiber involves at the samecontent.

As shown in Table 4, it can be seen that compressive strength ofUHPC samples increases slowly with age. This is because of rela-tively dense structure of UHPC associated with very low W/B ratio,which did not exhibit enough free water for further hydration ofbinder at later ages. Also, it is clear that coarse aggregate has anoptimal point of 25% replacement with respect to compressivestrength of UHPC sample. As coarse aggregate replacement furtherincreases to 35%, a declined compressive strength of UHPC sampleis observed. This phenomenon is in accordance with the resultsreported by Meddah et al. [37]. It occurs because concrete shouldbe examined as a three-phase materials, thereby coarse aggregateshould be proportioned depending on the cement content and itssurface area. Enhanced compressive strength of UHPC is associatedwith the increase of fiber content due to the fact that increasedsteel fibers can bridge more cracks and retard their propagation[36], and it is accepted that relationship between compressivestrength and fiber dosage is confined to a linear tendency [8]. Itshould be noted that the compressive strength of UHPC sampleswith same fiber dosage of 1.75% is almost identical, irrespectiveof fiber type. This confirms that different fiber types act similarlyin regard of compressive strength [38].

3.2. Single steel fiber pullout behavior

Shear force versus slip curves in Fig. 5 presents the debondingcharacteristics of various steel fibers in UHPC matrix. As shown

in Fig. 5, steel fibers are pulled out from matrix and two differentstages in curves can be observed: quickly debonding stage andgently friction one. The latter stage generally tends to be a horizon-tal line except deformed micro-ones, i.e. hooked-A and spiral fiber.Considering the very short region of embedded yarns that adhesivebonding strength exerts on, it is assumed that steel fibers are heldin matrix only by frictional bonding strength with no adhesivebonding strength. The frictional bonding strength depends on aver-age value of the height in the gently stage of curves [39]. Theresults are presented in Table 5. Due to the complex pullout behav-ior, the bonding strength of deformed fibers implies the averageaction strength during whole pullout process for the sake of com-parison with straight fiber and simplicity of calculation. As shownin Table 5, it should be noted that a lower bonding strength of steelfiber in UHPC would be observed with inclusion of coarse aggre-gates, especially for the deformed ones. This is attributed to thefact that existence of coarse aggregate would bring more defectsandmirco-cracks which are impairing to the denser microstructureof matrix [15]. Additionally, the frictional bonding strength ofdeformed fibers are dramatically higher than that of smooth fiber,where the hooked-A micro fiber shows the highest frictional bond-ing strength, followed by spiral and hooked-B macro fiber. Fordeformed ones, mechanical deformation is accompanied with theirpull-out response, thereby resulting in a strengthening shear force[22]. While, for hooked-B macro fiber, plastic deformation of thehooked end is relatively weak due to its large diameter and embed-ment length.

3.3. Effects of coarse aggregate replacements

Fig. 6 shows tensile responses of UHPC samples incorporatingvarying coarse aggregate contents (0%, 15%, 25%, 35% by volumeof mortar), where the strain is only valid up to peak stress of fiberpull-out. Irrespective of coarse aggregate content, all UHPCsbehave similarly and two different stages are clearly identified inthe tensile stress-strain curves as shown in Fig. 6: strain basedelastic part and fiber pullout part. In all cases, together with asteady-state cracking, a sudden drop in load-bearing capacity isfollowed by first crack of UHPC samples, which indicates that astrain hardening and multiple cracking behaviors is not occur inthese samples. This is because the first-crack strength of thesesamples is higher than the corresponding maximum fiber-bridging stress. As reported by Li et al. [34], the presence of

Fig. 7. Dependence of tensile properties of UHPC samples on coarse aggregatecontent.

Fig. 8. Tensile test results of UHPC samples with varying fiber dosage.

Fig. 6. Tensile test results of UHPC samples with different coarse aggregatesreplacements.

J. Liu et al. / Construction and Building Materials 121 (2016) 310–318 315

multiple cracking is determined by first-crack strength, which inturn is dependent on crack size. Furthermore, according to thefounding provided by Wille et al. [26], a value of energy absorptioncapacity g prior to tension softening gP 50 kJ/m3 is suggested todescribe strain-hardening behavior. The energy absorption capac-ity g is mainly dominated by matrix toughness and fiber-bridgingeffect, which is related to the fiber properties (including fiberlength, radius, volume content, elastic modulus and distribution),matrix properties (including tensile behavior and the correspond-ing elastic modulus) and interface properties (including bondingbehavior and strength) [40]. In post-cracking curves, it can be seenthat a lower fiber-bridging stress is observed for UHPC incorporat-ing coarse aggregate and it is almost independent of coarse aggre-gate content. This verified that coarse aggregate would bringimpairment to the microstructure of UHPC and result in a poorfiber bonding stress. However, in addition to the microstructureof fiber–matrix interfacial zone, fiber-bridging stress is also con-tributed by dispersion of fibers, which is influenced by the exis-tence and amount of coarse aggregate. With the addition ofcoarse aggregate, a severe interlock between fiber and coarseaggregate may be obtained and uniform dispersion of fibersbecomes hard to reach. Consequently, resulting in a more complexmicrostructure of UHPC sample.

Fig. 7 summarizes values of first-crack strength and maximumbridging stress obtained for theUHPC incorporating different coarseaggregate content. As shown in Fig. 7, with incorporation of 15%,25% and 35% coarse aggregate, the first-crack strength of UHPC sam-ples gradually declines by 4.7%, 5.0% and 6.1%, respectively. Thismay be due to increased flaws associatedwith the increase of coarseaggregate content. However, difference between first-crackstrengths is relatively small. Additionally, it can be seen that maxi-mum fiber-bridging stresses show a high scattering. Although theeffect of coarse aggregate on maximum fiber-bridging stress is notclear, coarse aggregate content should be carefully tailored. Becausea large replacement level will lead to a severe overlay of steel fiberand a lower bonding strength, this will bring a serious impairmentto fibers pullout behavior. Hence, it can be summarized that coarseaggregate can be successfully introduced into the systemof UHPC tofurther reduce its cost without almost impairing to its tensile prop-erties, regarding tensile properties of UHPC.

Becauseof its satisfactory tensile properties and relative lowcost,the binder system containing 25% coarse aggregate was selected asoptimized formulation of UHPC for the following studies.

3.4. Effects of fiber dosage

It is accepted that another key parameter affecting tensilebehavior of UHPC sample is fiber dosage. In this section, with thecoarse aggregate replacement level fixed at 25%, four smooth fiberdosage were adopted as shown in Table 3 (0, 1.0%, 1.75%, 2.5%) tostudy the influence of fiber dosage on tensile behavior of UHPCincorporating coarse aggregate. Their tensile responses are pre-sented in Fig. 8 and tensile properties are summarized in Fig. 9and Table 6.

As a comparison, only an elastic part is presented in tensilecurve of the unreinforced UHPC sample and its first crack strengthreaches around 6.65 MPa, which is apparently lower than that ofUHPC samples with introduction of steel fiber. As shown in Figs. 8and 9, both first-crack strength and fiber-bridging stress improveas fiber dosage increases. To evaluate utilization of steel fiber, a uti-lization factor g, which is defined as the ratio of tensile propertiesto the ultimate tensile strength of UHPC sample [39], is introducedas Eq. (1). The results are given in Table 6.

Fig. 9. Dependence of tensile properties of UHPC samples on steel fiber dosage.

Fig. 10. Tensile test results of UHPC samples with different fibers.

Table 6Tensile properties of UHPC samples with varying fiber dosage and correspondingutilization efficiency of fiber.

Notation Vf

(%)rf

(MPa)rfc (MPa) gfc

(�)rB (MPa) gB

(�)rB/rfc

(�)

SM0G2 0 2940 6.64 ± 0.22 – – – –SM1G2 1.00 2940 7.26 ± 0.11 0.34 5.85 ± 0.19 0.39 0.806SM2G2 1.75 2940 7.34 ± 0.17 0.22 6.41 ± 0.31 0.24 0.874SM3G2 2.50 2940 8.28 ± 0.18 0.19 8.10 ± 0.21 0.22 0.979

316 J. Liu et al. / Construction and Building Materials 121 (2016) 310–318

g ¼ rt

rm þ aVfrfð1Þ

where rt is tensile properties of UHPC sample, i.e. rfc and rB, rm istensile strength of matrix, it is equal zero for calculating the utiliza-tion factor during fiber pullout stage, a is an orientation factor andfixed at 0.5 for a three-dimensional orientation [21], Vf is the vol-ume fraction of steel fiber, rf is the ultimate tensile strength of steelfiber.

It can be seen from Table 6 that utilization efficiency of fibersbecomes worsen with the increase of fiber dosage, whatever at firstcrack and peak stress in post-cracking curves. This result is dis-agreed with the finding reported by Wille et al. [21], where utiliza-tion efficiency of fibers in UHPC sample without coarse aggregate isalmost unchangeable as fiber dosage increases from 1.5% to 2.5%.As mentioned above, with the fiber dosage increases, interlockbetween fiber and coarse aggregate and overlay of steel fibersbecome severe, which consequently leads to a lower utilizationefficiency of fiber. Furthermore, it can be seen that the utilizationefficiency of fiber at peak stress in postcracking curves is slightlyhigher than that at first crack, which indicates that steel fiber ismore effective at enhancing maximum fiber-bridging stress thanfirst-crack strength. This is attributed to that the first-crackstrength is mainly contributed by matrix toughness, whereas con-tribution of fiber reinforcing effect is relatively weak [34].

In addition, it is interesting to note that the gap between first-crack strength and maximum fiber-bridging stress is graduallybridged with fiber dosage increases as shown in Fig. 9 and Table 6.Ratio of maximum fiber-bridging stress to first-crack strength canbe used to characterize strain hardening behavior [24,34]. Thisratio is strongly dependent on fiber dosage for UHPC incorporatingcoarse aggregate as shown in Table 6. If the fiber dosage furtherincreases, maximum fiber-bridging stress is undoubtedly stronger

than first-crack strength and this ratio should be larger than 1.0,thereby a phenomenon of strain hardening and multiple crackingbehavior of UHPC samples incorporating coarse aggregate couldbe also triggered. This indicates that fiber dosage will be a criticalfactor that enhancing strain hardening and multiple crackingbehaviors for UHPC sample incorporating coarse aggregates. More-over, due to the low utilization efficiency of fiber in this paper, analternative method of realizing strain hardening and multiplecracking behaviors would be to further improve bond behavior inorder to increase the fiber utilization efficiency, which would thenresult in a higher fiber-bridging stress for a given low fiber volumefraction such as surface treatment or fiber type.

3.5. Effects of fiber type

As mentioned above, another key parameter affecting tensilebehavior of UHPC samples is fiber type. In this section, four differentsteel fibers with dosage fixed at 1.75% were adopted as shown inFig. 2 to investigate influence of fiber type on tensile behavior ofUHPC samples incorporating coarse aggregates. The tensileresponses are given in Fig. 10 and tensile properties are summarizedin Table 7.

As shown in Fig. 10, the overall shape of tensile-strain curves ofall samples are identical, irrespective of fiber type. However, fibertype has a strong effect on first crack strength and fiber-bridgingstress, especially in the post-cracking curve. For micro-fibers, theUHPC with deformed fiber shows a higher tensile properties, com-pared with that with smooth fiber. Among these, UHPC withhooked-A fiber has the highest tensile properties. Differencesobserved in tensile properties are attributed to different bondingproperties associated with fiber geometry. As shown in Table 5,the frictional bonding strength of hooked-A fiber is highest and itis almost three times higher than that of smooth one. While UHPCwith macro-fiber, i.e. hooked-B fiber, exhibits the lowest tensileproperties and its first crack strength is only slightly higher thanthat without fiber. This is attributed to two aspects: one is thelower frictional bonding strength as shown in Table 5, the otheris the lowest fiber number introduced into UHPC, compared withother micro-fibers at same dosage. However, due to its long anchorlength, the macro-fiber can significantly improve strain capacity ofUHPC samples, where the tensile strain at maximum bridgingstress is largest as shown in Fig. 10. This phenomenon is alsoobserved by Wille et al. [21], Park et al. [24], and Wu et al. [36].Thus, UHPC tensile performances with respect to post cracking

Table 7Tensile properties of UHPC with different fibers and corresponding utilizationefficiency of fiber.

Notation Vf

(%)rf

(MPa)rfc (MPa) gfc

(�)rB (MPa) gB

(�)rB/rfc

(�)

SM2G2 1.75 2940 7.34 ± 0.17 0.22 6.41 ± 0.31 0.24 0.874SP2G2 1.75 2860 7.95 ± 0.21 0.25 6.97 ± 0.24 0.28 0.877HA2G2 1.75 2940 8.53 ± 0.22 0.26 7.32 ± 0.26 0.29 0.859HB2G2 1.75 1890 6.70 ± 0.16 0.29 5.80 ± 0.19 0.35 0.866

J. Liu et al. / Construction and Building Materials 121 (2016) 310–318 317

strength, strain capacity and multiple micro-cracking behavior canbe successfully tailored by hybrid fiber system [24,41].

As shown in Table 7, ratio of maximum fiber-bridging stress tofirst-crack strength for UHPC with different fibers presents a verysmall fluctuation, which is located between 0.86 and 0.87. This phe-nomenon is not agreedwith UHPCwithout coarse aggregate [24], asPark et al. reported that this ratio is strongly dependent on fibertype. Difference observed may be attributed to the combined effectof fiber bridging and coarse aggregate, where existence of coarseaggregate has a strong effect on fiber distribution and interfacemicrostructure between fiber and matrix [31–32]. Because of thelimit distribution space, overlay fiber become severe for deformedones and deformedfibers bringmore flaws around the coarse aggre-gate. Although single fiber pullout behavior is significantly influ-enced by fiber type as shown in Fig. 5, pullout behavior of overallfibers bridging across matrix crack plane for UHPC with introduc-tion of coarse aggregate cannot simply integrate over the contribu-tion of individual fibers like UHPC without coarse aggregate.

It also can be seen from Table 7 that the utilization efficiency ofdeformed fibers is higher than that of smooth fiber. This can beattributed to the additional pullout force induced by mechanicaldeformation of deformed fiber. Furthermore, fiber geometry withrespect to length and diameter has a more stronger effect on theutilization efficiency, where the utilization efficiency of hooked-Bfiber shows an increase of 16% and 25% at first crack strengthand maximum fiber-bridging stress, compared to that of hooked-A fiber. Therefore, coupled with micro-fiber, macro-fiber providesthe feasibility of improving bonding behavior through increasingfiber utilization, which may result in a higher tensile properties.The result can be confirmed by the bonding strength result andthe favorable results of blending fiber reinforcement UHPC [24,41].

4. Conclusions

In this study, tensile behavior of UHPC incorporating coarseaggregates was investigated. Three types of steel micro-fibers,including smooth, spiral and hooked, and one type of steelmacro-fiber of hooked were applied as reinforcement. Bondingbehavior between these fibers and matrix was also performed.Combined effect of coarse aggregate and fiber properties, includingfiber dosage and type, on the overall tensile stress-strain curvesand tensile properties of UHPC was analyzed. The following con-clusions could be drawn.

(1) With inclusion of coarse aggregate, the UHPC mixturebehaves a more poor workability due to the decreased bin-der content. With respect to compressive strength, thereplacement level of coarse aggregate has a critical valueof 25% and different fibers act similarly.

(2) Coarse aggregate brings impairment to the bonding strengthof steel fiber in the UHPC matrix incorporating coarse aggre-gate, especially for deformed ones. Among these, thehooked-A micro fiber shows the highest frictional bondingstrength, followed by spiral and hooked-B macro fiber,whereas the smooth fiber shows the lowest frictional bond-ing strength.

(3) At a favorable replacement level (625%), coarse aggregatecan be successfully introduced into the system of UHPC tofurther reduce its cost without almost impairing to its ten-sile properties, regarding first-crack strength and maximumfiber-bridging stress.

(4) Unlike UHPC without coarse aggregate, a lower utilizationefficiency of fiber is observed as fiber dosage increases from1.0–2.5%, due to the fact that interlock between fiber andcoarse aggregate and overlay of steel fibers become severe.In addition, the phenomenon of strain hardening and multi-ple cracking behaviors of UHPC incorporating coarse aggre-gate can be also triggered by further increasing fiberdosage to larger than 2.5%, where it can bridge the gapbetween first-crack strength and maximum fiber-bridgingstress of UHPC incorporating coarse aggregate.

(5) Compared with smooth fiber, UHPC with deformed fibershows better tensile properties except macro-deformedfiber. Different from UHPC without coarse aggregate, fibertype has almost no effect on strain hardening behavior,due to combined effect of fiber bridging and coarse aggre-gate. Utilization efficiency of fiber is strongly dependent onthe fiber geometry, especially for length and diameter.

Acknowledgements

The authors gratefully acknowledge financial support from KeyProject of National Nature Science Foundation of China (Grant No.51438003), the Ministry of Science and Technology of China ‘973Project’ (Grant No. 2009CB623203) and National Nature ScienceFoundation Project of China (Grant No. 51508244).

References

[1] L. Basheer, J. Kropp, D.J. Cleland, Assessment of the durability of concrete fromits permeation properties: a review, Constr. Build. Mater. 15 (2001) 93–103.

[2] P.C. Aıtcin, The durability characteristics of high performance concrete: areview, Cem. Concr. Compos. 25 (2003) 409–420.

[3] P. Richard, M. Cheyrezy, Composition of reactive powder concretes, Cem ConcrRes. 25 (1995) 1501–1511.

[4] L. Sorelli, G. Constantinides, F. Ulm, F. Toutlemonde, The nano-mechanicalsignature of Ultra High Performance Concrete by statistical nanoindentationtechniques, Cem Concr Res. 38 (2008) 1447–1456.

[5] D.L. Nguyen, G.S. Ryu, K.T. Koh, D.J. Kim, Size and geometry dependent tensilebehavior of ultra-high-performance fiber-reinforced concrete, Compos. B 58(2014) 279–292.

[6] S.L. Yang, S.G. Millard, M.N. Soutsos, S.J. Barnett, T.T. Le, Influence of aggregateand curing regime on the mechanical properties of ultra-high performancefibre reinforced concrete (UHPFRC), Constr. Build. Mater. 23 (2009) 2291–2298.

[7] M. Schmidt, E. Fehling, Ultra high performance concrete research developmentand application in Europe, in: Seventh International Symposium on theUtilization of High Strength/High Performance Concrete, 2005, pp. 51–78.Washington, USA.

[8] R. Yu, P. Spiesz, H.J.H. Brouwers, Mix design and properties assessment ofUltra-High Performance Fibre Reinforced Concrete (UHPFRC), Cem. Concr. Res.56 (2014) 29–39.

[9] K. Wille, A.E. Naaman, S. EI-Tawil, G.J. Parra-Montesinos, Ultra-highperformance concrete and fiber reinforced concrete: achieving strength andductility without heat curing, Mater. Struct. 45 (2012) 309–324.

[10] Y.W. Chan, S.H. Chu, Effect of silica fume on steel fiber bond characteristics inreactive powder concrete, Cem. Concr. Res. 34 (2004) 1167–1172.

[11] Z. Rong, W. Sun, Experimental and numerical investigation on the dynamictensile behavior of ultra-high performance cement based composites, Constr.Build. Mater. 31 (2012) 168–173.

[12] J. Yang, G.F. Peng, Y.X. Gao, et al., Characteristics of mechanical properties anddurability of ultra-high performance concrete incorporation coarse aggregate,in: Proceedings of 3rd International Symposium on UHPC and Nanotechnologyfor High Performance Construction Materials, 2012, pp. 257–264.

[13] C. Wang, C.H. Yang, F. Liu, et al., Preparation of ultra-high performanceconcrete with common technology and materials, Cem. Concr. Compos. 34(2012) 538–544.

[14] J. Ma, M. Orgass, F. Dehn, et al., Comparative investigation on ultra-highperformance concrete with and without coarse aggregates, in: Proceedings ofthe International Symposium on Ultra High Performance Concrete, 2004, pp.221–228. Kassel, Germany.

318 J. Liu et al. / Construction and Building Materials 121 (2016) 310–318

[15] C. Tasdemir, M.A. Tasdemir, F.D. Lydon, et al., Effects of silica fume andaggregate size on the brittleness of concrete, Cem. Concr. Res. 26 (1996) 63–68.

[16] T. Akçaoglu, M. Tokyay, T. Çelik, Effect of coarse aggregate size and matrixquality on ITZ and failure behavior of concrete under uniaxial compression,Cem. Concr. Compos. 26 (6) (2004) 633–638.

[17] H. Yazıcı, M.Y. Yardımcı, S. Aydın, A.S. Karabulut, Mechanical properties ofreactive powder concrete containing mineral admixtures under differentcuring regimes, Constr. Build. Mater. 23 (2009) 1223–1231.

[18] H. Yazıcı, M.Y. Yardımcı, Y. Yigiter, S. Aydın, S. Türkel, Mechanical properties ofreactive powder concrete containing high volumes of ground granulated blastfurnace slag, Cem. Concr. Compos. 32 (2010) 639–648.

[19] N. Randl, T. Steiner, S. Ofner, E. Baumgartner, T. Meszoly, Development ofUHPC mixtures from an ecological point of view, Constr. Build. Mater. 67(2014) 373–378.

[20] H. Yazıcı, H. Yigiter, A.S. Karabulut, B. Baradan, Utilization of fly ash andground granulated blast furnace slag as an alternative silica source in reactivepowder concrete, Fuel 87 (2008) 2401–2407.

[21] K. Wille, D.J. Kim, A.E. Naaman, Strain-hardening UHP-FRC with low fibercontents, Mater. Struct. 44 (2011) 583–598.

[22] J. Feng, W.W. Sun, X.M. Wang, X.Y. Shi, Mechanical analyses of hooked fiberpullout performance in ultra-high-performance concrete, Constr. Build. Mater.69 (2014) 403–410.

[23] A. Beglarigale, H. Yazici, Pull-out behavior of steel fiber embedded in flowableRPC and ordinary mortar, Constr. Build. Mater. 75 (2015) 255–265.

[24] S.H. Park, D.J. Kim, G.S. Ryu, K.T. Koh, Tensile behavior of ultra highperformance hybrid fiber reinforced concrete, Cem. Concr. Compos. 34(2012) 172–184.

[25] D.J. Kim, S.H. Park, G.S. Ryu, K.T. Koh, Comparative flexural behavior of hybridultra high performance fiber reinforced concrete with different macro fibers,Constr. Build. Mater. 25 (2011) 4144–4155.

[26] K. Wille, S. El-Tawil, A.E. Naaman, Properties of strain hardening ultra highperformance fiber reinforced concrete (UHP-FRC) under direct tensile loading,Cem. Concr. Compos. 48 (2014) 53–66.

[27] N.T. Tran, T.K. Tran, D.J. Kim, High rate response of ultra-high-performancefiber-reinforced concretes under direct tension, Cem. Concr. Res. 69 (2015)72–87.

[28] S. Pyo, K. Wille, S. El-Tawil, et al., Strain rate dependent properties of ultra highperformance fiber reinforced concrete (UHP-FRC) under tension, Cem. Concr.Compos. 56 (2015) 15–24.

[29] S.T. Kang, J.K. Kim, The relation between fiber orientation and tensile behaviorin an Ultra High Performance Fiber Reinforced Cementitious Composites(UHPFRCC), Cem. Concr. Res. 41 (2011) 1001–1014.

[30] D.Y. Yoo, S.T. Kang, Y.S. Yoon, Effect of fiber length and placement method onflexural behavior, tension-softening curve, and fiber distributioncharacteristics of UHPFRC, Constr. Build. Mater. 64 (2014) 67–81.

[31] D.Y. Yoo, S.T. Kang, J.H. Lee, et al., Effect of shrinkage reducing admixture ontensile and flexural behaviors of UHPFRC considering fiber distributioncharacteristics, Cem. Concr. Res. 54 (2013) 180–190.

[32] J. Liu, C. Li, J. Liu, et al., Study on 3D spatial distribution of steel fibers in fiberreinforced cementitious composites through micro-CT technique, Constr.Build. Mater. 48 (2013) 656–661.

[33] GB175-2007, Common Portland Cements, 2007 (in Chinese).[34] V.C. Li, K.Y. Leung, Steady-state and multiple cracking of short random fiber

composites, ASCE J. Eng. Mech. 11 (1992) 2246–2264.[35] S. Grünewald, J.C. Walraven, Parameter-study on the influence of steel fibers

and coarse aggregate content on the fresh properties of self-compactingconcrete, Cem. Concr. Res. 31 (2001) 1793–1798.

[36] Z. Wu, C. Shi, W. He, et al., Effects of steel fiber content and shape onmechanical properties of ultra high performance concrete, Constr. Build.Mater. 103 (2016) 8–14.

[37] M.S. Meddah, S. Zitouni, S. Belâabes, Effect of content and particle sizedistribution of coarse aggregate on the compressive strength of concrete,Constr. Build. Mater. 24 (2010) 505–512.

[38] P. Soroushian, Z. Bayasi, Fiber type effects on the performance of steel fiberreinforced concrete, ACI Mater. J. 88 (1991).

[39] F.Y. Han, H.S. Chen, K.F. Jiang, et al., Influences of geometric patterns of 3Dspacer fabric on tensile behavior of concrete canvas, Constr. Build. Mater. 65(2014) 620–629.

[40] F.Y. Han, H.S. Chen, W.L. Zhang, et al., Influence of 3D spacer fabric on dryingshrinkage of concrete canvas, J. Ind. Text. (2014), http://dx.doi.org/10.1177/1528083714562087.

[41] W. Yao, J. Li, K. Wu, Mechanical properties of hybrid fiber-reinforced concreteat low fiber volume fraction, Cem. Concr. Res. 33 (2003) 27–30.

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具