Construction an Building Materials 71 (2084) 510-520, Contents lists available at ScienceDirect Construction and Building Materials ELSEVIER journal homepage: www. evier.com/locate/conbuildmat = Mechanical properties of high performance fiber reinforced cementitious composites Mostafa Fakharifar“', Ahmad Dalvand, Mahdi Arezoumandi®’, Mohammad K. Sharbatdar”, Genda Chen‘, Ali Kheyroddin” “Departmen of Ci Arciectra end Enrmentl Ener, Misour Univers of conrad Technol, 209 Pie Bultin 1304 Xe Ste, Rolle, MO 65408, United ees "Dxparment of Cit Engineering Soran Univesity, Sean, ran “Department of Gi. rctectarl and Environmental Eiri Missouri Universo cece ond Tetnaogy, 212 Butler Cron Hal, M401 Nine Str, ol, MO (5408, United Sates “Deponment of Cl Archetural and Enronmentl Eger, Misia Uniersy of Sence and Teel, 128 Butler Carton Hal, 1401 Nie Sire, ala, MO own HIGHLIGHTS + Compressive and flexural strength of HPFRCC follow normal dstsibution. inear relationship exists between flexural and compressive strength of HPFRCC, steasng iber content improves mechanical properties of HPFRCC; ‘increasing fbr content enhances Ast crack and failuce strength of HPFRCC. ARTICLE INFO ABSTRACT ‘arc ison [Exensive experimental studies on High Performance Fiber Reinforced Cement Composites (HPFRCC) Received 6 February 2014 ‘wing to their remarkable properties have been carried out. Statistical studios Rave Been mainly focused Received in revised form 21 August 2014 fon Fiber Reinforced Concrete (FRC) An extensive study including an experimental/sttistcal approach ‘Accepted 24 August 2014 ae adessing key mechanical properties (compressive and flexural strength) and impact resistance of such high performance composites with inclusion of dillerent volume of fibers has been caried ut on two- hhundfed and forty specimens in this research Results fom this study revealed that compressive and lexural strength as well as impact resistance of LHPFRCC follow the normal distribution. Furthermore, statistical data analyses (both paramettc and non- parametric) showed higher percentage of fibers led in greater values for mechanical properties and iis Impact resistance of HPFRCC. Moreover, based on acquired test results, equations were developed between mechanical properties and impact resistance of HPFRCC materials. (© 2014 Elsevier Li llrightsreserved Kepwors: Compressive tenth Feral stent Inpac esistance Probability dstabution egy abuorpon 1. Introduction High Performance Fiber Reinforced Cementitious Composites (HPFRCC) is a class of fiber cement composites with fine aggre- sates, demonstrating, remarkable properties such as improved Y Covespnding author. Te: #1573 341 4462 fax: +1 575 3414720. Email adareses: mihdcOnstedu (M. Fakhanar, Ahm dalend@ymail ‘om (A. Daan), ma52G@nstedu (M. Arzoumand mshaatdar@semnanccit (ik Stata cheno (Chen) Kiyandinssrnan cr Keyra) ek 11579 416845 on 40 S733 47, 2 Tels #1573 3416972 fu 41573341 4729, hp:ibdiorg)10.10165conbutémat.201408.058 (50-0618) 2014 Eee Ue Al rghis reserve resilience and sustainability. HPFRCC materials also exhibit improved properties compared with normal concrete (NC) andfor Fiber Reinforced Concrete (FRC) in terms of higher ductility, dura- bility and energy dissipation capacity. This material ean be charac- terized by a pseudo-ductile tensile strain hardening behavior with ‘multiple cracking prior to failure [1.2] Tensile behavior of NC, FRC and HPFRCC materials are compared in Fig. ). This figure clearly ‘exhibits three distinct behavior upon cracking. When concrete, ‘mortar or FRC are subjected to tension, brittle degrading behavior at first cracking due to inability to transfer tensile stresses across the crack surface is observed. In contrary, HPFRCC materials tundergo multiple cracking after first cracking, exhibiting aM Fathanor ea Construction and uling Motels 71 (2014) 510-520 tt “Tensile stress Strain hardSaing esion (Mull cracking) sn Single mek and localization ttt Brite failure of NC — a. hl tt a - © tH 7 = Crack Opening 12 Strain or Deformation Crack Opening (Material and sreturaldoctiity) (Surface energy, material ductili) Fig. 1. typical tens stress-strain or deformation relation upto ature of (2) normal caneete NC}: (b) Fier Reinforced Concrete (FRC): and) Migh Peformance Fiber efrced Cements Competes (HPFRCC), adopted fo [2 10), ‘abte1 “ype mechanical apse properties of EC [8 Compresive Fist cackingsuengih_Ulimate tensile suengih Flexural strength Youngs madulis _UNimate tensile stain Density strenath utr. ora) cate. (cra @ (ale 20-95 37 «12 10-30 1834 1 095-23 hardening behavior, ie. strength increase after frst cracking. Only for the HPFRCC materials the post cracking strength, dpc, is higher than the frst cracking strength, In recent years, a new class of HPFRCC materials called engineered cementitious composites (ECC) has emerged, offering ‘promising solutions for structures with longer service life featuring enhanced structural performance. Developed at the University of Michigan, ECC exhibits a typical moderate tensile strength of 4- © MPa and ductility of 3~5% [3], Besides, other types of ECC mate- rials including self-consolidating ECC, early high strength ECC. light ‘weight ECC and green ECC were introduced and studied by various researchers |4~7|, Typical Mechanical and physical properties of [ECC material is presented in Table 1 [8]. Polyvinyl alcohol (PVA) fibers have been widely used in development of ECC materials. However, fiber selection depends on different parameters including: (1) fiber properties (diameter, surface roughness and mechanical behavior), (2) characteristics of matrix (crack resis tance and fiber-matrix interfacial bonding strength), (3) perfor- ‘mance objectives of ECC (desired properties, durability, and sustainability), and (4) cost considerations of ECC materials in filed applications [9 ‘As set forth eatlier and illustrated in Fig. 1, adding fibers into conerete may enhance concrete mechanical properties, including, flexural strength, fracture toughness, thermal shock strength, fat gue strength and impact resistance | 0-15]. Many studies revealed the superior performance of HPFRCC materials based on its ‘observed static mechanical properties. Little research has been ‘undertaken on the loading rate effect on the mechanical behavior ‘of HPFRCC materials. Behavior of fiber, matrix and the interfacial ‘bond between them is dependent on loading (strain) rate [2 Thus, impact tests are valuable tools to characterize completely the behavior of HPFRCC materials under high strain rate loadingsse. IM Fao tl /Contrton and Building Matera 71 (2014) 510-520 Loading rate effect is significantly important on strength and duc- tility of critical regions within structures, such as beam-column joints and coupling beams |16,17) when subjected to earthquake loading. Existing statistical data analyses available in the literature are ‘mainly focused on mechanical properties (compression and flex- lure) and impact strength of FRC materials. Implementation of a reliable large set of experimental tests in addition to statistical data analyses would be an appropriate method to further study and ‘develop the understanding of mechanical properties of these high potential composite materials. Effect of fiber content on statistical parameters and distribution of flexural strength, compressive strength and impact strength of HPERCC is vital and need to be addressed. Energy dissipation of HPFRCC has never been staisti- ‘ally studied previously. To the best of the author's knowledge. a thorough experimentalfstatistical_ study on impact strength, ‘mechanical properties and energy dissipation of HPFRCC materials hhas not been investigated yet. Although, HPFRCC have been studied for their mechanical properties, but not statistically evaluated yet based on adequately large number of compression, flexure and impact test specimens, including different fiber contents. 2. Research significance Research on HPFRCC materials widely indicated their enhanced, properties featuring improved ductility, strength and durability lunder static load condition. Impact tests characterize the perfor- mance of HPFRCC under high strain rates. Effect of fiber content ‘on the impact strength of HPFRCC under dynamic loading is not completely characterized yet. A thorough experimental/statistical study would contribute to better characterization of mechanical properties and impact strength of HPFRCC, including different fiber ‘contents. Thus, a relatively large test matrix was developed by the authors to further investigate the mechanical properties and impact strength of HPFRCC materials. Two hundred and forty spec- imens total, tat is significantly larger than test matrices in similar studies to provide a reliable baseline for statistical analyses, were ‘ast, prepared and tested to achieve this goal. 3. Experimental study and procedures ‘Three volume fraction of polypropylene (PP) fiber equal to 0.5% 0.75% and 1% were considered in this study. Considering the three different fiber content, total specimens in this study would be cat= ‘egorized as: 60 specimens (100 x 100 « 100 mm) for compressive strength (20 specimens per group), 60 specimens (60 « 80 « 320 mm) for flexural strength (20 specimens per group) ‘and 120 specimens (150 x 64 mm) for impact resistance (40 spec- mens per group). The proposed drop-weight test according to the ACI Committee 544 [18], commonly used for impact strength of FRC, has been used for impact resistance test of HPFRCC specimens in the current study, Data obtained from such tests usually have a ‘coefficient of vatiation (COV) of mote than 25%. The variation in impact strength from the drop-weight test has been reported for the FRC materials [19-24 and not for the HPFRCC yer. All three groups of HPFRCC specimens were cast with identical ‘water-cement ratio of 0.38 with three different volume fiber contents of 0.5%, 0.75% and 1%, which are designated as HP-0:5, HP-0.75 and HP-1, respectively. Numeric value represents the percentage offers used in the corresponding group of specimens. For example, HP-0.5 represents the group of specimens with 0.5% fibers. Compressive strength tests were performed on twenty 100x100 100mm cubic specimens per group prepared according to BS 1881-108:1983 [25]. Flexural strength tests were ‘conducted on twenty 60x 80x 320mm beam specimens per ‘group according to BS EN 12390-5:2009 [26 Although compres- sive and flexural strength test procedures are almost fixed, a brief ‘explanation on drop test procedute undertaken in this research is presented. Drop-weight tests were conducted following the ACT 544 Committee's recommendations. For each concrete mix, ten 150 x 300 mm cylindrical specimens were cast. Then each cylinder was cut into four 64 mm thick cylindrical disks using a diamond blade concrete saw. During impact tests, each cylindrical disk ‘was set centered on a base plate within four positioning lugs, ‘and impacted by repeated blows. Blows were applied through a 4.45 kg, hammer falling continually from a 457 mm height onto a steel ball with the diameter of 63.5 mm, which was centered at the top surface of the disk. The number of blows required to cause the firs visible crack and then failure were recorded. With the aid ‘of a magnifier equipped with built-in flashlight, the number of blows to initiate the first visible crack on the top surface was defined as the first-crack strength, while the number of blows (0 ‘generate failure of the disk was identified as the failure strength, Al tests were performed after curing the specimens for 28 days. 4. Materials and specimens preparation Inehis experimental study, AIM Type Prand coment, rushed sand and fers were wed nthe mk ables 2a 3 present the cement compositions and mechanical properties of P es respectivey high cange water reduce gent twit the commer name of Mapet10" was used to aus the workablty of the ‘concrete mores For batching cement was mixed with aggregate for 1 min The, thewaterandwaterredocer agent wer addedto the mix and mixed for 6-8 min The mature proportions for hee diferent mixes are provided in Ta. Tea est eon- fete was east Imo. cUiC_ form (100 » 100» 100mm) cylinder (150 300mm), and prisms for (0 » RO = 320 mm) for compressive, impact ‘able? Chemical composition and_ physical properties of Composition Peeentge Los a37 Myo. 156 Ko ase GO. a3 cs 51 cs 27 cA 5 CaF na specie wavy sn Specie src en) 3000 Mechanical popes of polypropylene es Tength Diameter Color Tensile Younes ‘ens (oom) (oom) trent (MPa) mods (Pa) _ (hm) ‘abled Mure proportions of once por cuble meter concrete) Wate Cameat Poypropylene Fine fement aio fiers (2) aggregatesM Fathanor ea Construction and uling Motels 71 (2014) 510-520 sis Compressive sens of cable specimens (MPa) Specimen no Compresive rent 30 a) Sra a0 Stes S290 asi 4598 our asat 5501 Sia aah 472 so14 ast 512 4558 4575 a7 sia a6 5029 519 4634 sun saz 898 sais seu 683 ea and flexural tests, respectively Allo the specimens were moist cure at a tempera {ure rage of 3-30°C and a relative humidity range f 85-1005 forthe st 24 ‘Ate demoling, he specimens were stored ina ser-contrled environment with ‘temperaturerangeof 20-25 anda ete hamid range of 35-50 for 28 day, ‘specimens were tested ater curing for 28 das om batting 5. Results and discussion 5.1. Compressive strength ‘The compressive strength tests were cartied out on sixty 100 x 100° 100mm cube specimens, using a digital standard automatic testing machine of 2000kN capacity. Compressive strength test results are presented in Table 5. Results indicate that the addition of fibers into specimens improved. compressive strength and it increases with increasing, PP fiber percentage Fig. 2a presents the histogram of compressive strength fi for 1HP-1.00 specimens. This figure shows thatthe results are normally distributed and fit well withthe superimposed normal distribution curve (it will be explained in more details in Section 5). HP-1 specimen group has the highest mean compressive strength value amongall the specimen groups while exhibiting the highest COV) atthe same time. As itis observed, by increasing fiber percentage in HPFRCC specimens, dispersion of compressive strength data is increasing accordingly a5 well as COV. The COV is lower than a limit of 15% suggested by Swamy and Stavrides [27]. Moreover Day (28) suggested that a COV between 5% and 10% generally rep” resents a reasonable quality contol, asthe acquited results here fits in that range of variation, 5.2 Fleaural strength Flexural strength test was performed on sixty 320 x 80 x 60 mm specimens. Schematic ofthe flexural test appa- tatus is shown in Fig. 3. Flexural strength test results from 60 spec- imens for three groups with different fiber contents are presented in Table 6. The highest mean flexural strength values belongs to HP-1 specimens, which is 29% and 21% more than mean flexural strength values of HP-0.5 and HP-0.75 specimens, respectively The frequency histograms of flexural strength, f for HP-1.00 spec- imens are shown in Fig. 2b. This figure shows that flexural strengths follow the normal distsibution, Interestingly enough, the COV for all specimens with different percentages are around 10%, 5.3. Impact resistance Drop-weight test results of 120 disk specimens including forty specimens per group, are given in Table 7. The impact test appara- tus and schematic details are shown in Fig. 4. ‘ : rs Be ° nt ss 3055-60 65 70 Ts (af. Of, ‘ i ‘ _ Ba 17 | A ° °. (ep FC (acu ig. 2 Monat dstibations of MP specimens.su IM Fao tl /Contrton and Building Matera 71 (2014) 510-520 Toaing ap Fosural specimen Fig. 3. Flemral et configuration under four-point ong. ables Fear sengts of beam specimen (MP). Specimen po odkles of ape P05 weave et 4 435 Sor 638 5 4a1 ait 574 6 521 an 639 4 427 351 620 8 538 559 574 10 388 494 ons a 51 5a 633 14 403 5a 485 8 523 nz 731 cov en) 04 805 oa 5.3.1. First crack strength ‘The frequency histogram and fitted normal curve of the flst- crack strength, FC distribution for HP-1.00 are shown in Fig. 2c. First-crack strength of HP-1.00 specimens hardly follows a normal distribution (lower p-values in Table 8). Results implied that with increasing fiber percentage, first-crack strength increases. Mean values for first-crack strength of HP-1 group with 1% fibers was approximately 57% and 18% more than HP-0.5 and HP-0.75 groups, respectively. The COV of first-crack strength for HP-1 group is 16% ‘and 6% more than HP-0.5 and HP-0.75 groups, respectively. 5.3.2 Failure strength Fig. 2d depicts the histogram of failure strength, UC for HP-1.00, impact specimens, with the fitted normal curve superimposed. ‘Similar to frst crack resistance, all three groups hardly follow the ‘normal distribution. According to Table 7, the mean value of the failure strength of HP-1 group was approximately 66% and 19% higher than those of HP-0.5 and HP.0.75 groups, respectively ‘The number of blows to reach the failure strength of specimens ranges from 17 to 105 for the HP-05 group, from 18 to 139 for the HP-0.75 group, and from 25 to 167 for the HP-1 group. Increase of fiber content increased the number of requited blows to failure as well. HP-1 group had the highest COV among all the three ‘groups, revealing higher scattering of data. As shown in Table 7, adding fibers to mixture increases the mean and dispersion of the failure strength. Fig. 5 demonstrates impact specimens at fail- ure including different fiber content featuring radial conerete spit- ting initiated from center and extended radially. 53.3. Bnergy absorption and post crack strength ‘The impact energy per blow was applied by a 4.45 kg hammer ‘dropped repeatedly from 457 mm height on top of a 63.5 mm steel ball Energy absorbed by the concrete disk for frst crack and falure ‘rack strength is shown in Table 7. According to these tables the ‘maximum absorbed energy for first crack and failure strength ‘occurs in HP-1 group. Mean value of energy absorbed by HP-1 _sroup for failure strength was approximately 66% and 19% higher than HP-0.5 and HP-0.75 groups, respectively. The Percentage Increase in the Number of Post initial crack Blows to failure is labeled as the “PINPB" parameter. Mean values of PINPB parameter ‘of HP-1 group is 40% and 12% greater than HP-0.5 and HP-0.75, _groups, respectively. Adding fibers to concrete mixture causes increasing the distance between first-crack strength and failure strength by limiting the intial crack and delaying the ultimate fail- ure, This however maybe regarded as the pseudo-ductility ratio. Fibers provide three-dimensional fibrous reinforcement. which assist a disk specimen in absorbing the impact energy of repeated blows, thus downplaying the impetuousness of the disk specimen ‘against crack progression. Fig. 5 illustrates the efficacy of fiber increase in altering one large single cracking at failure (HP-0.50) to multiple cracks (HP-1,00) 5.4, Statistical data analysis ‘The following section discusses probability distributions of ‘compressive strength, flexural strength, and impact resistance of fiber reinforced concrete. It also explains about relationship between mechanical properties as well as prediction of impact resistance from mechanical properties, 5.4.1. Probability distributions ‘The statistical computer program Minitab 16 [2] was ‘employed to find probability distributions of mechanical proper~ ties as well as impact resistance of fiber reinforced concrete. The Kolmogorov-Smimnov, Ryan-Joiner, and Anderson-Darling meth- ‘ods were used to test normality of the data. The hypotheses for the normal distribution are as follows: Hoy. The compressive strength of fiber reinforced concrete follows the normal distribution, oa. The flexural strength of fiber reinforced concrete follows the normal distribution,1M Fathaor ea /Constraction and uling Motels 7 (2014) 510-520 sis Impact eesistances an predicted aur tenths of specimens Specimen no, HROS0 HPO75 P00 FO US PINT Impoctenergy FC “UCCINPR—mpacrenergy «=~=CUC INP Impact energy (blows) (kN) (Blows) (kN) (Blows) (EN am) 7 a8 4s 9 cg ins a 5 8 108912004556 2 6116682300 0 a 2 a7 ws 75 ” 2 1567 228 " » 2 590 ost 3546 a 6 27509 2 ce so ius 2 mb ee 0 a oF 5 75) 2430 2 7 has 1729 1 ae 189 a 2m 41008 15 rer! ot 7 S158 36 8 m™m |96 16 a 66 95 15s iw oT 2068 © 5 997 19 stg 38 46 731160 20 si 3s m6 5775 a 25 488610 2 ry ms '96 8 ” 6 346 “570 2 nO so 610528 2 3951009 as Bo om 495 4a % 7 15261628 38 % a0 1216285980 3 7107861 36 B67 118013635075 2 % S16 1008 ” % 105 126 216 HB io 302035 28 8 no ot 89535 6 a8 30 aa as 50965175100 8 2» 9161099 oe 0 2 8 oo 4 oe s ~ Nunber of blows for ist rack strat numberof blows for fare strength «Percentage increase in number of post-frst-crck blows Ha. The first crack impact resistance of fiber reinforced concrete follows the normal distribution Hag The ultimate impact resistance of fiber reinforced concrete follows the normal distribution Hy Not Hos 2 “Table 8 shows that p-values for all of the above tests are greater than 0.05. This confirms the null hypothesis atthe 0.5 significance level. In other words, the compressive strength, flexural strength, first crack impact resistance, and ultimate impact resistance of fiber reinforced concrete follow the normal distribution. Iris worth ‘noting that the compressive strength and flexural strength of the LHPFRCC specimens fit better to the normal distribution compared with the first crack impact resistance and ultimate impact resis- tance of the HPERCC specimens (it is more clear in Fig. 2) 5.42. Relationship between mechanical properties and impact resistance It is useful to derive equations to predict the HPFRCC flexural strength from compressive strength test results. Fig. 6 shows that there is linear relationship between flexural and compressive strength of HPFRCC for all of the fiber percentages. Eqs. (1)-(3) show these relationships. Interestingly, the HP-0.75 and HP-1.00, showed the same slope, but for HP-0.50 it was around 45% steeper than both the HP-0.75 and HP-1.00 mixes. In other words, adding (050% fiber was more effective than both 0.75% and 1.00% in terms of improving flexural strength (Fig. 6) J.=0A19f, ~3.75. (HP-0.50) a J.=0.3f,-1.57 (HP-0.75) @ J, =0.13f, 0.99 (HP-1.00) 3)Stee! pipe with 64 mm Sts ap Base fre 200 mi Stool with 6 (Matha onstrctin and Bing Mater 71 (2014) 510-520 ©) Fig 4. Fling hammer test etal for impact strength (a) Impact test configuration: schematic (let) and apparatus (ight) an (b) details f impact Bll and tel cap: schematic (het and aparates gh where fs the compressive strength of concrete (MPa) and fis the flexural strength of concrete (MPa). Both the first crack and ultimate impact resistance of specimens, can be predicted by using the compressive strength of specimens. ‘The linear relationship (Eqs. (4)-(9)) observed between first crack) tltimate impact resistance and compressive strength for HPFRCC specimens are included as well. As it can be seen from Fig. 7, in terms of first crack resistance, HP-1.00 and HP-0.75 specimens had the highest and lowest slope, respectively: however, the ultimate impact resistance of HP-0.50 had the highest slope that followed by HP-0.75 and HP-1.00. The higher slope means more effectiveness of percentage of fiber to resist impact force. FC-6.77f,-27333 (HP-050) @ UC=8:77f,-36042 (HP-0.50) o FC 647f,- 272.51 (HP-0.75) 6 UC=8.66f,—36655 (HP-0.75) o FC 7.69f, 344.90. (HP-1.00) ® UC-6.08f,-274.86 (HP-1.00) ® Where FC is the First crack resistance and UC i the ultimate impact resistance. Values forthe normality eto at Tet HPS Hem Hoo “ompresve senate a 0705 0986 oon ws ais, >ao Fesuralserath » 0959 0960 955 5 ato rst crock restance 8 pao timate impect resistance is sas 8 ato ‘adeson- Dating * Koimagorov-Smirov «yan Another valuable result can be predicting the ultimate impact resistance from the first crack resistance of specimens. Similar to the aforementioned equations, the ultimate impact resistanceM Fathanor ea Construction and uling Motels 71 (2014) 510-520 si (a) HP-OS (b) HP-0.75 (©) HP-1.00 Fg 5. Fllre mode of impact disk specimens foal groups. 8 fa) [FC = 6.77f, -273.33 (-090)] . = [Ucar 042 e090) g 0 5 wl 2 7 é yet ae g 3 mes 2 7 ae © rum F.(tPay - ‘ —-nras| (a) HP-0.50 vm 40 a0 30 60 70 80 [FC = 6.47f, - 272.51 w'-09] cay [c= Nsar, 36655 0-009) © i Fe 6 Fowl rent. conpresve sent of pce. 8 & — : 3 = 4 --K and the fst crack resistance ate related to each other linearly by e Eqs. (10)-(12). "SD 0 0. uc—1.27FC- 101 (HP.050) (ao) fare) uc=1.27FC +168 (HP-075) ay () HP-0.75 Uc 1.23FC +527 (HP-1.00) 02) “Ee Fig. 8 shows that HP-0.50 and HP-.75 specimens have identical = 160 = = slope, whereas HP-1.00 specimen shows the lowest slope. That ©“ [FC=Gonr,-27486 0] A means adding 1% fiber, decreased the gap between first crack S120 and ultimate impact resistance compared to 050% and 0.75% fiber Egg cantent. E 4 543. Parametric and nonparametric tests Statistical tests (bath parametric and nonparametric) were used 0 o to evaluate whether there is any statistically significant difference Fora) between the mechanical properties of HPFRCC materials with df= iourine ferent fiber contents. . 5.43.1. Parametric test. The paired t-test is a statistical technique used fo compare two population means. This test assumes that the differences between pairs are normally distributed. If this assumption is violated, the paired t-test may not be the most pow- erful test. The hypothesis for the paited t-test is as Follows: ‘The compressive strength of the HP-0.75 mix is higher than the HP-0.50 mix [f(HP-0.75) > f.(HP-0:50)| aa The compressive strength ofthe HP-.00 mix is higher than the HP-0.75 mi [f(HP-100) > f.(HP-0.75) Fig 7. The fst crackultinate impact resistance vs. compressive strength of specimens. Hy. The compressive strength of the HP-1.00 mix is higher than. the HP-050 mix f,(HP-1.00) ~ f,(HP-0:50)). Hq. The flexural strength of the HP-0.75 mix is higher than the HP“0.50 mix if HP-0.75) > f,{HP-0:50) Hs. The flexural strength of the HP-1.00 mix is higher than the HP-0.75 mix [f,(HP-1.00) > f,(HP-0.75),sis (Matha onstrctin and Bing Materia 71 (2014) 510-520 2a 1s0 & 100 so] a i [uC= ITC LOTR O95) a 30 To 150 2 Few (a) HP-050 210 & ww so] « TCT TRE ODA) ‘ w 100 150 200 FCW) () HP-075 209 1s = 1m s0| | TRC STROH) » o 100 150 210 Fo@ (©) HP-1.00 Fie BT imate impact esac it rack esa. Statistical data analysis esas pale) Number of replications requ to hep the ero under specified init at 90% and 95t evel of confidence. bor) HPOSO HRO7S Poo CE one Ce Level af conden 20 od en oa 25 a5 58 os 0 oa a 1 3s yo Bd 53 <0 2 3 2 2 eo So 1 1 1 1 1 1 95% Lee of comience “0 Ce He. The flexural strength of the HP-1.00 mix is higher than the HP-0:50 mix Jf HP-1.00) > f,(HP-0.50), Hox. The frst crack resistance of the HP-0.75 mixis higher than the HP-0.50 mix [FC(HP-0.75) > FC(HP-0.50)) Hg. The first crack resistance of the HP-1.00 mix is higher than the HP-0.75 mix [FC(HP-1.00) > FC\HP-0.75)) Ho. The first crack resistance of the HP-1,00 mix is higher than the HP-0:50 mix [FC(HP-1.00) > FC\HP-0.50)) Horo. The ultimate impact resistance of the HP-0.75 mix is higher than the HP-0.50 mix [UC(HP-0.75) > UC\HP-0.50)) Hon. The ultimate impact resistance of the HP-1,00 mix is higher than the HP-0.75 mix [UC(HP-1.00) > UC(HP-0.75)) Fav 075)» POS) av) > OTS) Tar)» F030) ‘Neal 7s oss oa Nonparametric 1000 1.00 1000 900.075)» J,04-.050) 0.90)» f44.075) 0.00)» f-050) oralty 0210 oa “029 Parametric 1.00 | 1.000 3 Nonpramete 100 1.000 . Fo(HP-075) >» FeHP.050) Fc(HP-1.00)-FeHe-075) FciHP-100) -FEHP-0.50) oraliy 052s one Nonparametric 007 ‘00 oraliy 038 0525 Nonparametric 000 ‘000 ~ Sine p-value ses thn 005, te pareve and nonparametric tts cannot be performedM Fathanor ea Construction and uling Motels 71 (2014) 510-520 sig Hoja. The ultimate impact resistance of the HP-1.00 mix is higher than the HP-0.50 mix [UC{HP-1.00) > UC\HP-0.50)) He. Not Hes: ‘The statistical computer program SAS 9.2 [30] was employed to perform these statistical tests. The Anderson-Darling test showed the data ~ the differences between the mechanical properties of specimens with different percentage of fiber ~ follows a normal distribution (see Table 9). Therefore, the paired t-tests could be performed, The result of the paired ¢-test presented in Table 9 that shows the p-values were greater than 0.05 forall of the hypothe- ses. This confirms the null hypotheses atthe 0.05 significance level. Inother words, increasing the fiber content improved the mechan- ical properties of HPFRCC, 5.43.2. Nonparametric test. Unlike the parametric tests, nonpara- metric tests are referred to as distribution-free tests. These tests have the advantage of requiring no assumption of normality, and they usually compare medians rather than means. The Wilcoxon signed-rank test is usually identified as a nonparametric alterna- tive to the paired t-test. The hypothesis for this test is the same a those for the paired test. The Wilcoxon signed rank test assumes that the distribution of the difference of pairs is syramet- ‘cal. This assumption can be checked; ifthe distribution is normal, it is also symmetrical. As mentioned earlier, the data follows the normal distribution and the Wilcoxon signed ranks test can be used, As i can be seen in Table 9, similar to the parametric study all the p-values for the Wilcoxon signed rank were greater than (0.05. Like the parametric test, the nonparametric test confirms the null hypotheses at the 0.05 significance level. interestingly, the p-values for both the parametric and nonparametric tests are very close to each other. Overall, results of the statistical data anal- yyses showed that the HPFRCC mix with higher fiber content pos- sesses higher mechanical properties, 5.44, Number of replications ‘The COV ofthe test results, presented in Table 7 ean be used 10 determine the minimum number of tests “n*, required for guaran- ‘teeing that the percentage error “e” in the measured average value is below a specified limit. This number of tests can be calculated using Eq, (13) (27| (cove e ‘where COVis the coefficient of vatiation and ris the value of Student ‘distribution fo the specified level of confidence. The value of Stu- dent distribution “r" not only depends on the level of confidence, bbut also itis dependent on the degree of freedom, which is related {to the numberof samples. For degrees of freedom of more than 120, the value of "" approaches 1.645 and 1.282 at 95% and 90% level of confidence, respectively (31,32), Table 10 presents the minimum ‘number of replications required to keep the error under specified limits at the 90% level of confidence. Table 10 shows that, if the error is retained lower than 10% for first-crack strength, the mini- ‘mum number of requited replications would be 26, 32 and 36 at LHP-0.50, HP-0.75 and HP-1,00 groups, respectively, at 90% level of confidence. Also, for HP-0.50, HP-0.75 and HP-1 groups for failure strength at 90% level of confidence, i the error is retained lower than 10%, the minimum number of replications required are 30, 32 and 33, respectively. Table 10 also demonstrates the number of tests required to keep the error under a specific limit at the 95% level of confidence. According to this table, ifthe etror is retained lower than 10% the minimum numbers of replications for HP-0.50, HP-0.75 and HP-1.00 group are 43, 52 and 59 for the (13) first-crack strength and 49, 53 and 95 for the failure strength, respectively. It is clear in Table 10 that adding fibers to concrete ‘mixture increases the umber of tests Fequired at each level of 6, Conclusions and findings In this study, effects of polypropylene fibers on mechanical properties of HPFRCC specimens with fine aggregates were investi- gated experimentally and statistically, based on 240 implemented tests. Based on the acquired test data of 240 specimens divided in tree groups (60 cubic specimens for compressive strength, 60 bbeam specimens for flexural strength, and 120 disk specimens {or impact resistance), the following conclusions can be drawn: ‘The compressive and flexural strength, first crack and ultimate impact resistance follow the normal distributions. + In all cases, the mean and the COV both increase with the Increase in percentage of fibers. The more data dispersion in concrete specimens with higher percentage fibers is likely attributed to the pronounced effect of fiber-concrete interfacial bond. ‘The COV for the impact resistance from drop-weight tests is highest among the three mechanical properties. Ths is likely attributed to additional uncertainties involved in impact test- ing. such as specimen surface roughness and loading position + Linear relationship was observed between flexural and com- pressive strength of HPFRCC. + The HP-0.50 specimens have the highest rate of gaining flexural strength. + The first crack and ultimate impact resistance can be predicted by using compressive strength with high coefficient of determi- nation (F) + HP-1,00 and HP-0.50 specimens showed the best performance in terms of the fist crack and the ultimate impact resistance, respectively. ‘©The first crack and ultimate impact resistance are related to each other linearly. + HP-1.00 specimen showed inferior performance after first crack compared to both HP-0:50 and HP-0.75 specimens that per- formed almost identical + Statistical data analysis (both parametric and nonparametric) showed that increasing fiber content improves the mechanical properties of HPFRCC, ‘Adding fibers into concrete increases the number of blow Counts required to cause both the first visible crack and the fail- ture of test specimens, and the number of blow counts from the first crack to failuc. + The mean energy absorption ofthe specimens with 1% fibers is, 666% and 19% higher than that ofthe specimens with 0.5% and 0.75% fibers, respectively. The mean post-crack blow count of the specimens with 1% fibers is 40% and 12% greater than that of the specimens with 0.5% and 0.75%, respectively. Acknowledgements Financial support for this study was partly provided by the Department of Civil, Architectural, and Environmental Engineering, at Missouri University of Science and Technology, US. and by the US. National Science Foundation under Award No. CMMI-1030399, The results, discussion, and opinions reflected in this paper are ‘those ofthe authors only and do not necessarily reflect the official views or policies of the funding institutions20 (Matha onstrctin and Bing Mater 71 (2014) 510-520 References. 11) ive. wu ie Constons for pscdo strain-hardening in fiber reinforce brite Imax composts] Ap ech Rey 192438) 390-8 [2] Naaman AE, Reinhard MW. Choratterization of igh performance fb Fenfrce cement cmposites, HPFRCC-2 1986. p. 1-28 |p) seer G. Wang, LC. Design of engineered Eementtions composes for Processing and wertabltyrequlements ne Seventh Internacional ‘Shmpostum on brite matrx Composites 2003. 29-36. 141 Kong Hike 5. UVC. Development of a slfcompactng engineered cementitious composite employing eletosteie dispersion sabzation J {Cem Coner compos 2003:250);301-8. 1) WeSC Heh ens nice cementws compote Ac fo} Wang 5 I Ve. Maes design of lighewelght FVA-EC. In; Naaman AE, Reinhardt HW. editors, Proc HPFRCC. Rl: 2003. 379-90, 171 Lepech MD, L Ve. Roberson RE. Reottan GA. Desian of tle engineered ‘ements composts for improved sustainably. AC Mater] 207 Is) LV Engineered cemenciious composts (Ecc) matena, struct, 20d ‘uralty performance An Abr Mt 48109: Univesity of Michigan: 2007 Ip) Ize kien kt YW Predicion of EC tensile strss-stain curves based 03 mated ber bing relations considering fiber dstibution characteristics. {Comput Cone Int} 201071. Wo] seer 6, Vc. structura composts with EC. Xl Y, Mabin SA ets Proceedings othe ASCES USC 2000p. 1001-8, Inn] Yan Sun Chen The elect sca fame apd steel ber on the dye Imechanieal perfomance of bighstreth “concrete. Cem Cone Res Ion.a423 6 [12] Bana’ N, Sappakitipakorn M. Toughness enhancement in steel Aer ‘Solace content Aer hydration Cot es na] Altezainy A, Sroushian Mira F Mechaicl properties of palypropylene fiber reinforced concrete and the eet of pozzalanc materials Com Coner (Comps 1996:182)35-92, [na] estsy FS Harwich K. Compressive strength and deformation of ste! re vnlored concrete under high rate of sein ne Cer Campos Lighight oner 1885:71)21-8. [n5}Toahim 8 he kar MB. fects om mechanical properties of ndusrasnd Steal iees aan tO normal weight concrete. Proc ERg.2011:18 116] Para: Montesinos G) High-performance Aber enforced cement composites: Sh aternative for sami design of stestures ACI ret] 20051023) 107) Fakhantor Mt sharbatdar MK LinZDalvand A ivand-Pour A Chen G Seismic performance and gota ductility of KC tame rehabiated wie retoted Joins by CF Laminates. arin Eng Eng Vib 2014:131 59-73, [ne] AC Commitee 544 Measurement of properties of tibrreitored concrete, ‘AC Mater | 1988:85(6 583-99, luo) Song Ps Wa J, Hwang. Shew BC. Asesment of stats vations in Impact resistance of gh stengrh concrete and high strength steel Ae Fefrced concrete} Com Concr Res 2005135:395-2 (20) Barat shou Atal F tens Andree Ktastcal variations in mpoet resisanee of polypronvine “bre-reinreed concrete} Impact ng ‘ovw:s 1907-20. [2t) Songs, uJ: Hvangs, shew Bc. Satis analysis of impact strength and strength celal of see-polypropyene hb fbe-enfrced concrete} Constr Bul Mater 2005;19°1-3 [22] Nataraja MC Daan N: Gupta AP. Statistical varlatons a impact estance of steel fber-cinforced conerete subjected to drop wea test Cem Cone Res 1900,25089-95, (2s) fanman T Kan shekarch MSA A Statistical and experimental analysis fon the bevior of ber relafrced concretessubeced to drop weight fet nat ld Mater 201257360. [241 Swany RN Jj AH Impact resistance of stelle enforced lightweight concrete Int} Cem Compo Uihtwegt Cone 1982.4) 200-20, 125) Bris standaras nsittion, 6 1861-108:1985. Method fr making test cubes rom fea cancet: 1985 (26) Bits Staaas, Isto, BS. EN 12390-52009, Testing, tardened oncete-fexual stegth est specimens. Min Keyes, UK: 2008 [271 Swamy RN. Savides,H. Some staustcal considerations of steel Are Fefrced composites Cem Concr es 1975:92):201-15 {2s} Day HOW. Concrete mi design, quity contol and speciation. 2nd ‘Landon! ERFM Spon 1985 (29) Minitab 17 Stas! Sofware. [Compute sofware, State Calls, Pa: Minitab In. sr mina cons, 2010" 30) Sao" 91. Procedures gulde Copy 2011. Cary. NC USA: SAS nstut ne [31] Bo CEP, Hunter WG, Hunter J. Statistics for experimenters. USA: Wiley: 378, 132] Moore DS, McCabe GP. intrdetion to the practice of statisti. New "York WH. Freeman 8 Company: 1589.