Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Composite Structures 174 (2017) 263–276

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Fiber pullout behavior of HPFRCC: Effects of matrix strength and fiber


type
Doo-Yeol Yoo a, Jung-Jun Park b, Sung-Wook Kim b,⇑
a
Department of Architectural Engineering, Hanyang University, 222 Wangsimni-ro, Seongdong-gu, Seoul 04763, Republic of Korea
b
Structural Engineering Research Institute, Korea Institute of Civil Engineering and Building Technology, 283 Daehwa-dong, Goyangdae-ro, Ilsanseo-gu, Goyang-si,
Gyeonggi-do 10223, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: This study investigated the effects of fiber type and matrix strength on the fiber pullout behavior of high-
Received 25 February 2017 performance fiber-reinforced cementitious composites (HPFRCC). The correlation between single fiber
Revised 14 April 2017 pullout behavior and flexural behavior of HPFRCC was also evaluated. Two different steel fibers, i.e.,
Accepted 25 April 2017
straight and hooked steel fibers, and three different matrix strengths were adopted. Test results indicate
Available online 26 April 2017
that the fiber pullout performance was improved with increasing matrix strength. The hooked fibers
exhibited higher bond strengths and pullout work than the straight fibers, but at large slips, they showed
Keywords:
smaller shear stress at the interface than their counterpart. In addition, the straight fibers were more
High-performance fiber-reinforced
cementitious composites
effective in improving the pullout performance with the matrix strength than the hooked fibers. For
Fiber pullout the straight fibers, the shorter fibers provided higher bond strengths and maximum shear stress at the
Matrix strength interface than the longer fibers. The flexural performance of HPFRCC beams was improved with increas-
Fiber type ing matrix strength. The beams with medium-length straight fibers (lf/df = 19.5/0.2 mm/mm) gave the
Flexure best flexural performance, whereas those with hooked fibers exhibited the worst flexural performance.
Due to several influential factors, the correlation between the single fiber pullout behavior and flexural
behavior of HPFRCC beams was quite low.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction mated the pullout response of straight steel fiber embedded in


ultra-high-performance concrete (UHPC) with various ingredients.
In recent years, (ultra-) high-strength concrete is widely used Sands of SiO2 and ZrSiO4 increased the roughness of the matrix sur-
for building construction due to its several advantages, such as rounding fiber, leading to an improved frictional bond component,
excellent specific strength and durability. However, the increase and when using smaller glass powder and hydrophilic nanosilica,
of strength intrinsically results in an increase of its brittleness, the bond performance was improved. Yoo et al. [4] investigated
and as a result, several methods for improving its ductility have the effect of matrix shrinkage on the pullout behavior of straight
been introduced thus far. Among others, the use of discontinuous steel fiber embedded in UHPC and reported that the bond strengths
fibers has been considered as one of the most efficient methods and pullout work decreased with a decrease in shrinkage, leading
improving the ductility of brittle (ultra-) high-strength cement to a decreased radial confinement pressure. Banthia and Trottier
composites through fiber bridging at crack surfaces [1,2]. Since [5] evaluated the pullout behaviors of three deformed (i.e., hooked,
the pullout capacity of fibers, most significantly affects the fiber crimped, and twin-cone) steel fibers with different inclination
bridging capacity, is influenced by both fiber type and matrix angles embedded in normal, mid-, and high-strength concretes
strength, pullout capacities of various types of fibers embedded (fc0 of 40, 52, and 85 MPa). In their study [5], several important
in (ultra-) high-strength cement matrices with various strengths findings were observed: (1) the aligned deformed steel fibers in
need to be investigated. the pullout direction exhibited higher peak loads and energy
For this reason, many researchers [3–8] have examined the absorption capacity than the inclined fibers, and (2) matrices with
fiber-matrix bond or pullout properties. Wille and Naaman [3] esti- higher strengths resulted in matrix splitting or fiber fracture, incit-
ing a brittle nature. In contrast, Lee et al. [6] reported that the high-
⇑ Corresponding author. est peak pullout loads of straight steel fibers in UHPC occurred
E-mail addresses: dyyoo@hanyang.ac.kr (D.-Y. Yoo), jjpark@kict.re.kr (J.-J. Park),
when the inclination angles were either 30° or 45° due to the com-
swkim@kict.re.kr (S.-W. Kim). bined snubbing and matrix spalling effects. These two divergent

http://dx.doi.org/10.1016/j.compstruct.2017.04.064
0263-8223/Ó 2017 Elsevier Ltd. All rights reserved.
264 D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276

observations indicate that the effect of the inclination angle on the cement, silica fume (SF), blast furnace slag (BFS), and fly ash (FA)
fiber pullout behavior is dependent on the fiber geometry and were incorporated as the cementitious materials, and their chem-
matrix strength. Kim et al. [7] examined the implications of matrix ical compositions and physical properties are summarized in
strengths, i.e., fc0 of 28, 56, and 84 MPa, on the pullout response of Table 2. Silica sand with a grain size between 0.2 mm and
deformed (hooked and twisted) steel fibers. They [7] reported that 0.3 mm was adopted as a fine aggregate, and silica flours with
higher pullout resistance was obtained with an increase in the two different diameters of 4.2 lm and 14.1 lm, containing 98%
matrix strength, and the twisted fibers were more sensitive to SiO2, were used as filler. To improve the tensile or flexural behav-
the matrix strength, meaning the higher enhancements in both ior, coarse aggregate was excluded from the mixture. In the case of
pullout load and energy, as compared with the hooked fibers. Matrix C, only SF was added as 25% of the cement weight; for
Likewise, several studies have been performed to evaluate the Matrices A and B, SF along with BFS or FA were incorporated as
effects of various parameters, i.e., matrix strength, from normal 30% of the cement weight. Since the fluidity of cement mortar
strength to high strength, and fiber type, straight or deformed, was reduced with decreasing W/B ratio, a high amount of high-
on the fiber pullout behaviors. However, to the best of the author’s range water reducing agent, called superplasticizer (SP), was
knowledge, there are no published studies investigating the effects added, and its amount increased with decreasing W/B ratio. In
of matrix strength on pullout behaviors of both straight and order to investigate the effect of fiber type on the fiber pullout
deformed steel fibers embedded in (ultra-) high-strength cementi- and flexural behaviors of HPFRCC, two different types of fibers,
tious composites having matrix strengths greater than 100 MPa. i.e., straight and hooked steel fibers, were used. Three different
Therefore, a study with regard to the pullout behavior of fibers sizes were used for the straight steel fibers (S): S13 with lf/
embedded in (ultra-) high-strength cementitious composites is a df = 13/0.2 mm/mm, S19.5 with lf/df = 19.5/0.2 mm/mm, and S30
pressing need and may provide very useful information for the with lf/df = 30/0.3 mm/mm. One size was used for the hooked steel
practical use of fiber-reinforced (ultra-) high-strength concrete. fibers (H): H30 with lf/df = 30/0.375 mm/mm. The detailed geomet-
Accordingly, in this study, the effects of fiber type and matrix ric and mechanical properties of the used fibers are shown in
strength on the pullout behaviors of various steel fibers in Table 3.
(ultra-) high-strength cementitious composites were evaluated. Similar to the mixing sequence used for UHPC, the following
Two types of steel fibers (i.e., straight and hooked) and three differ- mixing sequence was adopted for HPFRCC used in this study,
ent matrix strengths from 112.2 MPa to 190.2 MPa were consid- because it has low W/B ratios and does not have coarse aggregate.
ered. In addition, the correlation between the single aligned fiber First, cementitious materials, silica sand, and silica flour were pre-
pullout behavior and the flexural behavior of (ultra-) high- mixed for approximately 10 min using a Hobart type mixer with a
strength cementitious composites including steel fibers, denoted capacity of 120 L. Water premixed with SP was incorporated into
as high-performance fiber-reinforced cementitious composites the dry ingredients and mixed for an additional 10 min. Once the
(HPFRCC) hereafter, was examined. mixture had adequate fluidity and viscosity, steel fibers were dis-
persed and mixed for 5 more min.
All of the test specimens were cured at room temperature for
2. Research significance the initial 48 h. The specimens were then demolded and steam
cured with heat (90 ± 2 °C) for 3 days. The specimens were then
Due to an increase of the demand for mega structures, (ultra-) removed from the steam curing room and stored in the laboratory
high-strength concrete has attracted attention from many at room temperature until testing.
researchers and engineers. However, the increase of strength inevi-
tably accompanies brittleness, and as a result, there is a pressing 3.2. Mechanical tests
need for developing high toughness and (ultra-) high-strength con-
crete, denoted as HPFRCC. To improve the ductility of concrete, dis- 3.2.1. Fiber pullout test
continuous fibers can be effectively used. However, even though Half dog-bone shaped pullout specimens were fabricated and
HPFRCC has a wide range of strengths and pullout resistance of tested to evaluate the bond behaviors of straight and hooked steel
fibers is greatly affected by matrix strength, there is no published fibers. For obtaining reliable test data, at least 6 or more specimens
study of the investigation of matrix strength on pullout resistance were used to calculate the average value. Three different embed-
of various fibers embedded in (ultra-) high-strength cement matrix ded lengths of 6.5, 9.75, and 15 mm (half of the fiber lengths) were
with compressive strengths greater than 100 MPa. In addition, only adopted for S13, S19.5, and S30 (or H30), respectively. A single
a few studies examined correlation between the pullout response fiber pullout test was performed based on a previous study [7],
of single fiber and the tensile or flexural behavior of composite ele- and the detailed test setup is shown in Fig. 1. A single fiber was
ments. Accordingly, this study investigated the pullout behaviors embedded in the center of half dog-bone specimen with a cross-
of various steel fibers embedded in (ultra-) high-strength cement sectional area of 25  25 mm2. The specimen was first put into
matrix with compressive strengths of 112.2, 152.5, and the specimen grip; once it was centered, the free end of the fiber
190.2 MPa, and the flexural behaviors of HPFRCC with 2 vol% steel was tightly gripped. A uniaxial tensile load was applied through
fibers were analyzed in relation to the fiber pullout responses. the universal testing machine (UTM) with a maximum load capac-
ity of 250 kN. In order to apply quasi-static loading rate and to
3. Experimental program obtain enough data points up to the peak strength, a loading rate
of 0.018 mm/s was adopted, similar to a previous study [10]. By
3.1. Materials and mixture proportions assuming that the movement of the fiber grip is the same as the
fiber slip and the elastic deformations of the fiber and specimen
Three different matrix types (Matrices A, B, and C) were are small enough, the fiber slip was measured from the vertical dis-
adopted, and their ultimate compressive strengths were placement of the fiber grip using a linear variable differential
112.2 MPa, 152.5 MPa, and 190.2 MPa, respectively. Detailed mix- transformer (LVDT).
ture proportions are shown in Table 1. As it is well known that
the compressive strength of cement-based material is most signif- 3.2.2. Four-point bending test
icantly affected by the water-to-binder (W/B) ratio [9], three differ- The prismatic HPFRCC beams were fabricated and tested as per
ent W/B ratios of 0.3, 0.25, and 0.2 were adopted. Type I Portland ASTM C1609 [11]. To obtain reliable test data, at least 4 (or 5) spec-
D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276 265

Table 1
Mixture proportions.

W/B Relative weight ratios to cement SP (%) fc0 (MPa)


Cement Water SF BFS FA Silica sand Silica flour
Matrix A 0.30 1.00 0.39 0.10 – 0.20 1.20 0.20 (14.1 lm) 0.005 112.2
Matrix B 0.25 1.00 0.325 0.10 0.20 – 1.10 0.30 (14.1 lm) 0.0075 152.5
Matrix C 0.20 1.00 0.25 0.25 – – 1.10 0.30 (4.2 lm) 0.02 190.2

[Note] W/B = water-to-binder ratio, SF = silica fume, BFS = blast furnace slag, FA = fly ash, SP = superplasticizer, and fc0 = compressive strength.

Table 2
Chemical compositions and physical properties of cementitious materials.

Composition % (mass) Cement BFS SF FA


CaO 61.33 42.10 0.38 –
Al2O3 6.40 14.50 0.25 16.60
SiO2 21.01 33.80 96.00 65.30
Fe2O3 3.12 0.01 0.12 5.58
MgO 3.02 5.80 0.10 0.82
SO3 2.30 1.89 – 0.51
Specific surface area (cm2/g) 3413 6240 200,000 3850
Density (g/cm3) 3.15 2.91 2.10 2.13
Ig. loss (%) 1.40 0.05 1.50 3.82

[Note] Cement = Type 1 Portland cement, BFS = blast furnace slag, SF = silica fume, and FA = fly ash.

Table 3
Properties of steel fibers.

Type of fiber df (mm) lf (mm) Aspect ratio (lf/df) Density (g/cm3) fft (MPa) Ef (GPa)
S13 0.200 13.0 65.0 7.9 2788 200
S19.5 0.200 19.5 97.5 7.9 2500 200
S30 0.300 30.0 100.0 7.9 2580 200
H30 0.375 30.0 80.0 7.9 2630 200

[Note] Sxx = straight steel fibers with a length of xx, Hxx = hooked steel fiber with a length of xx, df = fiber diameter, lf = fiber length, fft = tensile strength of fiber, and
Ef = elastic modulus of fiber.

isfy the minimum width requirement of ASTM C1609 that the


beam width or height needs to be larger than 3 times the fiber
length. A clear span length of 300 mm was adopted, and the uniax-
ial load was applied through the UTM with a maximum load capac-
ity of 250 kN. A loading rate of 0.4 mm/min was applied based on
the speed of stroke. The applied load was measured from the load
cell affixed to the crosshead of the UTM, and a steel frame with
LVDTs at both sides was installed to measure the pure mid-span
deflection without support settlement. The detailed test setup for
the four-point flexural test is shown in Fig. 2.

4. Experimental results and discussion

In order to obtain highly reliable test results, 10 dog-bone spec-


imens of fiber pullout were fabricated for each variable. During
removal of the forms, the bonding region between the fiber and
matrix in some specimens was damaged because the strength of
the matrix was not fully developed (before steam curing). Thus,
specimens without damage were only used for the pullout tests,
and their test results are summarized in the Appendix. At least 6
or more specimens were used to obtain the average data, and the
detailed procedure on the calculation of the average pullout load
versus slip curves is given in the Appendix.

4.1. Evaluation parameters for fiber pullout performance


Fig. 1. Fiber pullout test setup.
In order to obtain appropriate energy absorption capacity of
fiber-reinforced concrete (FRC) after matrix cracking, the fibers
imens were used for each variable. Beams with a cross-sectional are required to be pulled out without breaking. Therefore, the max-
area of 100  100 mm2 and a length of 400 mm were used to sat- imum tensile stress, rf,max, obtained in the fiber by pullout force
266 D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276

Based on the assumption that the shear stress is uniformly dis-


tributed over the entire embedded length of the fiber, the equiva-
lent bond strength can be calculated based on the fiber pullout
work [12], as follows

2W P
seq ¼ ð5Þ
pdf L2E
where seq is the equivalent bond strength.

4.2. Effect of matrix strength on fiber pullout behavior

The average fiber pullout load (or fiber stress) versus slip curves
are shown in Fig. 3, and several important pullout parameters, i.e.,
pullout work, maximum fiber tensile stress, average bond strength,
and equivalent bond strength, are summarized in Table 4. The pull-
out work is the energy required to completely pullout the fiber,
and if it is assumed that bond strength is constant along the entire
embedment length, the pullout work is used for calculating the
equivalent bond strength, which is closely related to cracking
behavior of HPFRCC [12]. The maximum fiber tensile stress is an
important parameter to predict the fiber pullout mode (pullout
or fracture), and a minimum fiber tensile strength preventing frac-
ture during pullout process can be suggested from this parameter.
Since the average bond strength is based on the peak pullout force,
the pullout resistance of fibers with various geometries is simply
evaluated based on the parameter of average bond strength. The
Fig. 2. Four-point flexural test setup. shape of the pullout load versus slip curve was not significantly
influenced by the matrix strength. However, for the case of S30,
high frequency noises in the descending branch were obtained
needs to be smaller than its tensile strength. The maximum tensile with increasing the matrix strength. This might be caused by the
stress, rf,max, is thus a critical factor for determining the proper increased filling effect of crushed fine particles at the interface
fiber type according to the concrete strength, and it can be calcu- between the fiber and matrix. Once the fiber was fully debonded
lated in terms of the maximum pullout load and area of the fiber, from the matrix, only frictional shear stress was activated at the
as follows interface, and the crushed particles at the interface, generated from
Pmax the debonding of fiber, braked the pullout of fiber. Due to the filling
rf ;max ¼ ð1Þ effect, the pullout resistance of fiber increased in a moment, and
Af
once it reached to the capacity, a sudden drop of pullout load
where rf,max is the maximum fiber tensile stress, Pmax is the maxi- was obtained. This process was repeated in the descending branch
mum pullout load, Af is the area of the fiber (=p  df2/4), and df is the and caused abrasion at the surface of steel fiber, already reported
fiber diameter. by Wille and Naaman [3]. This phenomenon becomes more pro-
The average bond strength between fiber and matrix based on nounced with increasing the size of reinforcement. For instance,
the maximum pullout load and initial embedding length is given no noticeable noise was observed in the specimen S13 regardless
by of the matrix strength, but noticeable and significant noises in
Pmax the pullout load versus slip curve were observed in the round steel
sav ¼ ð2Þ
pdf LE bar with a diameter of 16 mm embedded in UHPC [13]. However,
further research is required to provide more rational and valid rea-
where sav is the average bond strength, and LE is the initial embed- sons for this observation.
ded length of the fiber. The average bond strength (or maximum pullout load), maxi-
In order to evaluate the actual shear stress behavior of the fiber mum fiber stress, equivalent bond strength, and pullout work for
according to the slip, the following formula is used, which is based all test series generally increased with increasing matrix strength,
on the pullout load at any slip divided by the current surface area regardless of the fiber type. For example, the highest average bond
[=p  df  (LE  s)]. strength, maximum fiber stress, equivalent bond strength, and
PðsÞ pullout work of Matrix C with H30 fibers were found to be
sðsÞ ¼ ð3Þ 12.5 MPa, 2000.6 MPa, 7.3 MPa, and 965.5 J, respectively; approxi-
pdf ðLE  sÞ
mately 10.0%, 10.1%, 3.8%, and 3.7% greater than those of Matrix A
where s(s) is the shear stress, P(s) is the pullout force at slip s, and s with the identical fiber (H30). This is because when the matrix
is the free-end slip. strength increases, a denser interfacial zone between the fiber
The fiber pullout work is calculated by the area under the pull- and matrix can be obtained, as shown in Fig. 4 [14,15], which indi-
out load versus slip curves until the pullout load drops to zero and cates the scanning electron microscope-backscattered electron
is given by the following equation image of the fiber-matrix interface in two different matrices. The
Z s¼LE
porosity at the interface is much greater for the high-strength con-
WP ¼ PðsÞds ð4Þ crete than that of Matrix C. In addition, the filling effect of fillers
s¼0 was improved as finer particles were adopted. The filler with a
where WP is the fiber pullout work, and WP⁄ is the fiber pullout smaller diameter of 4.2 lm was incorporated in Matrix C, as com-
energy at unit embedded surface area (see Table 4). pared with that (14.1 lm) of Matrices A and B. Thus, a higher bond
D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276 267

Table 4
Summary of pullout test results of fibers in three different matrices.

Fiber type Parameters Matrix A Matrix B Matrix C


3
S13 WP (10 J) 107.74 (22.44) 110.73 (43.41) 114.08 (28.87)
WP* (103 J/mm2) 26.97 (5.20) 29.11 (10.13) 29.54 (7.04)
Pmax (N) 30.68 (3.49) 31.81 (12.80) 35.33 (6.83)
rf,max (MPa) 976.72 (110.96) 1012.64 (407.51) 1124.70 (217.31)
sav (MPa) 7.69 (0.84) 8.36 (3.10) 9.17 (1.77)
seq (MPa) 8.49 (1.55) 9.65 (3.04) 9.61 (2.17)
S19.5 WP (103 J) 195.97 (35.80) 224.24 (40.69) 236.34 (59.86)
WP* (103 J/mm2) 32.91 (5.76) 38.87 (8.09) 39.81 (8.64)
Pmax (N) 42.98 (3.67) 42.76 (9.10) 49.12 (9.51)
rf,max (MPa) 1368.21 (116.89) 1361.18 (289.73) 1563.65 (302.62)
sav (MPa) 7.21 (0.46) 7.43 (1.88) 8.29 (1.33)
seq (MPa) 6.95 (1.23) 8.49 (2.14) 8.44 (1.53)
S30 WP (103 J) 508.24 (130.87) 635.82 (289.46) 656.25 (96.60)
WP* (103 J/mm2) 37.05 (8.68) 46.37 (19.67) 47.42 (6.31)
Pmax (N) 84.61 (28.66) 83.13 (12.01) 86.30 (21.22)
rf,max (MPa) 1196.99 (405.50) 1176.07 (169.85) 1220.87 (300.27)
sav (MPa) 6.16 (1.99) 6.16 (0.87) 6.22 (1.40)
seq (MPa) 5.10 (1.09) 6.38 (2.50) 6.46 (0.79)
H30 WP (103 J) 931.36 (138.64) 927.56 (223.26) 965.54 (89.28)
WP* (103 J/mm2) 52.73 (7.85) 52.67 (12.45) 54.67 (5.05)
Pmax (N) 200.74 (21.84) 213.06 (32.52) 220.96 (18.01)
rf,max (MPa) 1817.53 (197.78) 1929.07 (294.45) 2000.59 (163.09)
sav (MPa) 11.36 (1.24) 12.10 (1.79) 12.50 (1.02)
seq (MPa) 7.03 (1.05) 7.04 (1.63) 7.29 (0.67)

[Note] WP = pullout work, Pmax = maximum pullout load, rf,max = maximum fiber tensile stress by pullout load, sav = average bond strength, and seq = equivalent bond
strength.
*
Pullout energy at unit embedded surface area.

2400
160 Matrix A
Matrix B
140 2000
Matrix C
Fiber stress, (MPa)
Pullout load, P (N)

120
1600
100
80 1200

60 800
40
400
20
0 0
0 2 4 6 8 10 12 14 16
Slip, S (mm)

Fig. 3. Average fiber pullout and slip curves; (a) S13, (b) S19.5, (c) S30, (d) H30.
268 D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276

Fig. 4. Scanning electron microscope-backscattered electron image; (a) high-strength concrete (fc0 = 50.1 MPa), (b) matrix C (fc0 = 211.8 MPa) [14,15].

strength of the fibers was achieved for the matrix with higher com-

Average bond strength ratio


pressive strength, due to the improved stiffness and strength of the 125

based on matrix A (%)


interfacial transition zone. The radial confinement pressure caus- 120 S13
ing the friction bond between the fiber and matrix is dependent S19.5
on the amount of shrinkage of the matrix [4]. It is well-known that 115
the shrinkage magnitude of the matrix is reduced with a higher W/ H30
110 S13
B ratio [16]; thus, smaller radial confinement pressure was also H30
105 S19.5
expected to be generated for the fibers embedded in the matrix S30
S30
with higher W/B ratio (or lower compressive strength). Based on 100
the above two reasons, the fiber pullout performance was
95
improved with an increase in matrix strength. This was verified
150 200
by comparing to previous test results obtained by Kim et al. (Matrix B) (Matrix C)
[7,10,14], who reported that the maximum fiber stresses, rf,max, (a) Matrix strength (MPa)
of hooked steel fiber with a diameter of 0.38 mm and an embed-
ment length of 15 mm in matrices with compressive strengths,
fc0 , of 28, 49, and 83 MPa were obtained as approximately 750,
Pullout work ratio based on

130 S30
1311, and 1500 MPa, respectively. These values were obviously
S30
smaller than the maximum fiber stress (rf,max = 1817 MPa) of 125
S19.5
matrix A (%)

H30 embedded in Matrix A with a higher compressive strength, 120


fc0 , of 112.2 MPa. S19.5
115
Fig. 5 shows the average bond strength and pullout work ratios
based on the values obtained from Matrix A. Notably, the short 110
S13
straight steel fiber (S13) was most effective in improving the aver- 105 S13 H30
age bond strength with an increase in matrix strength. For exam- 100 H30
ple, a nearly 20% higher average bond strength of S13 was
95
obtained when the matrix strength increased from 112.2 MPa to
150 200
190.2 MPa (Matrix A ? C). Alternatively, the average bond (Matrix B) (Matrix C)
strength ratio was reduced with increasing length and diameter (b)
of straight steel fibers. This means that the effectiveness of increas-
Matrix strength (MPa)
ing the matrix strength on the improvement in the average bond
Fig. 5. Ratios of (a) average bond strength and (b) pullout work based on matrix A.
strength of the fiber decreased with increasing fiber size (length
and diameter). Compared to the values of the H30 specimen, the
S19.5 specimen exhibited a lower ratio in Matrix B, whereas it pro-
vided a higher ratio in Matrix C. Thus, the order of effectiveness in increase of pullout work with matrix strength. Consequently, the
increasing average bond strength according to the matrix strength order of efficiency of improving pullout work with the matrix
is given as follows: S13 > S19.5  H30 > S30. However, the pullout strength is given as follows: S30 > S19.5 > S13 > H30.
work ratio based on Matrix A provided different behavior regard- Based on these test results, several important findings were
ing the average bond strength. The S30 specimen exhibited the obtained as follows: (1) the use of straight steel fiber is more effec-
highest improvement in the pullout work by increasing the matrix tive in improving the pullout performance than that of the hooked
strength. For example, approximately 30% higher pullout work of steel fiber with an increase in the matrix strength, (2) the short
S30 was observed when the matrix strength increased from straight steel fiber is effective for improving the bond strength,
112.2 MPa to 190.2 MPa. The effectiveness of increasing the matrix which is related to the post-cracking tensile strength of HPFRCC,
strength on the pullout work was obviously reduced as the size of with the matrix strength, and (3) the long straight steel fiber is effi-
the straight steel fiber decreased. This is the opposite result to the cient in improving the pullout work, which is related to the energy
above observation with regard to the average bond strength. The absorption capacity or post-peak tensile ductility of HPFRCC, with
hooked fibers showed the worst performance regarding the the matrix strength.
D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276 269

4.3. Effects of fiber shape and dimension on pullout behavior frictional bond strength at the interface between the fiber and
matrix increases with an increase in matrix strength, the mechan-
4.3.1. Straight fiber vs. hooked fiber ical anchorage effect on the fiber strength was relatively reduced.
In order to evaluate the fiber pullout behaviors according to the From this observation, it is noted that the increase of matrix
fiber shape and dimension, the fiber stress, rf, versus normalized strength is more effective in improving the fiber pullout perfor-
slip, s/LE, curves are compared as shown in Fig. 6. Regardless of mance for the straight fibers than the hooked fibers.
the matrix strength, the hooked steel fibers (H30) exhibited a It is interesting to note that the straight steel fibers provided a
greater fiber stress at the initial normalized slip before the normal- greater fiber stress after the normalized slip of approximately
ized slip, s/LE, of approximately 0.3 mm/mm compared to the 0.3 mm/mm, regardless of the matrix strength. This indicates that
straight steel fibers (S13, S19.5, and S30). This is attributed to the the use of straight steel fiber is more effective in resisting the
mechanical anchorage effect at the hooked end, which is similar external tensile load than that of the hooked fiber when a large
to the results reported by Park et al. [17] for the case of UHPC. At crack width is generated. As the hooked fiber is used, micro-
the beginning of the fiber pullout, adhesion and initial friction were cracks in the surrounding cement matrix are generated by high
activated. Once the adhesion and initial friction were completely localized pressure occurred between the fiber and matrix [18].
activated, the additional component of the bond due to the Thus, a sharper decrease in the fiber stress with increasing slip
mechanical deformation of the fiber was then activated such that was obtained for the hooked fiber once the end was completely
the H30 fiber exhibited higher maximum fiber pullout strength straightened compared to that of a straight fiber. The splitting
than its counterparts (straight fibers). For example, the highest micro-cracks by localized pressure is dependent on the matrix
maximum fiber stress of H30 in Matrix A was found to be strength [18], such that a greater difference between the fiber
1817.5 MPa, which is approximately 86%, 33%, and 52% greater stresses obtained from straight and hooked fibers after a normal-
than those of S13, S19.5, and S30 embedded in Matrix A, respec- ized slip of approximately 0.3 mm/mm was obtained for Matrices
tively. Interestingly, the effectiveness of using hooked fiber (H30) B and C rather than Matrix A. This means that more severe
on improving the fiber strength, rf,max, was mitigated when the micro-cracks occurred in Matrices B and C than Matrix A due to
matrix strength was increased. For example, the ratios of fiber their higher strength (fc0  150 MPa).
strength between H30 and S19.5 were 1.33 and 1.28 for Matrices The H30 fiber provided higher pullout work, average bond
A and C, respectively. This is caused by the finding that, since the strength, and equivalent bond strength than the S30 fiber with

Fig. 6. Fiber stress and normalized slip curves; (a) matrix A, (b) matrix B, (c) matrix C.
270 D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276

an identical length of 30 mm, as summarized in Table 4. Due to its


steep decrease in the pullout load caused by splitting micro-cracks,
the effectiveness of using hooked fiber on improving the equiva-
lent bond strength is much smaller than the average bond strength.
For example, the highest average and equivalent bond strengths of
H30 embedded in Matrix C were 12.5 and 7.29 MPa, respectively,
approximately 100% and 13% greater than those of S30 embedded
in the identical Matrix C. The average and equivalent bond
strengths of H30 in Matrix C were very similar to the results
obtained from a previous study [17] that the average and equiva-
lent bond strengths of hooked steel fibers in UHPC were found to
be 11.7 and 7.5 MPa, respectively. Consequently, for the case of
(ultra-) high-strength concrete, the addition of hooked steel fibers
(instead of straight fibers) is expected to be more effective in
increasing the post-cracking tensile (or flexural) strength from
the higher maximum pullout load than its energy absorption
capacity.
The shear stress, calculated from Eq. (3), versus normalized slip Fig. 8. Picture for deformation at the fiber end [3].
curves are shown in Fig. 7. In similar to the previous studies [3], the
straight steel fibers showed a continuous increase in the shear
stress, s(s), with normalized slip. This is caused by the deformation hooked fiber (H30) exhibited shear stress versus normalized slip
(called ‘‘flattening”) at the fiber end (Fig. 8), generated by the cut- behavior that was different with those of straight steel fibers. A
ting process during fiber production. Therefore, although a straight higher shear stress was obtained for the hooked fiber at small nor-
fiber was adopted, the mechanical component resulting from fiber malized slips (before approximately 0.3 mm/mm) compared to
end deformation was activated along with frictional stress. The that of the straight fibers, mainly attributed to the anchorage effect

Fig. 7. Shear stress and normalized slip curves; (a) matrix A, (b) matrix B, (c) matrix C.
D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276 271

of the hooked end. After the small normalized slip, the smallest son’s ratio effect. With regard to the Poisson’s ratio effect, a higher
shear stress occurred because of the splitting micro-cracks in the pullout force occurred in the fibers with larger diameter and
surrounding matrix, which decreased the fiber confinement. When length, resulting in longer elongation of the fiber, leading to a
the normalized slip values approached 1, the shear stress in the reduction in the radial interfacial pressure due to a decrease in
hooked fiber also steeply increased due to fiber end deformation the fiber diameter. Similar observations were reported by several
by the cutting process. researchers [19] for steel reinforcing bar embedded in UHPC. A
greater reduction in equivalent bond strength with increasing fiber
4.3.2. Effects of fiber diameter and length length and diameter was observed compare to that of the average
As shown in Fig. 6, S13 and S30 produced similar behaviors of bond strength. For example, approximately 26% smaller average
fiber stress versus normalized slip curves, whereas S19.5 exhibited bond strengths were obtained for S30 compared to those of S13.
a higher fiber stress at identical normalized slips. This indicates However, approximately 35% smaller equivalent bond strengths
that the fiber pullout behavior was obviously influenced by the were observed for S30 than S13. In addition, although H30 exhib-
fiber diameter and length. However, it was difficult to draw a trend ited nearly 43% higher average bond strength than S13, its equiva-
of pullout behavior according to the fiber aspect ratio, lf/df. The S30 lent bond strength was approximately 23% lower.
specimen has a fiber aspect ratio of 100, similar to that (lf/df = 97.5) The pullout work, WP, clearly increased with increasing fiber
of S19.5 and much greater than that (lf/df = 60) of S13. However, its length and diameter (Table 4). This was caused by the finding that,
pullout behavior was similar to that of S13, but different than that when the fiber length and diameter increased, the maximum pull-
of S19.5 (much smaller fiber stresses were obtained). Based on the out load and ultimate slip, which is the slip at zero pullout load,
test results, some useful findings were obtained as follows: (1) both increased. A higher pullout energy at the unit embedded sur-
increasing the fiber length (with an identical diameter) was effec- face area, WP⁄, was also obtained with increasing fiber length and
tive in increasing fiber stress, which means that if the maximum diameter due to the increase in maximum pullout load.
fiber stress generated by the pullout force does not exceed the fiber As shown in Fig. 7, a higher shear stress was generated with
strength, its effectiveness increases with increasing fiber length, smaller fiber diameter and length. The order of magnitude of the
and (2) decreasing fiber diameter (at a similar aspect ratio) is effi- shear stress is given as S13 > S19.5 > S30, which was caused by
cient in increasing the fiber stress, meaning that the effectiveness the fact that the nonlinearity of the shear stress distribution along
also increases with decreasing fiber diameter. the embedded length of the fiber increased with increasing fiber
Fig. 9 shows the ratios of the average and equivalent bond length and diameter. In addition, a reduced radial interfacial pres-
strengths based on the S13 specimen. The highest average and sure was generated with an increase in fiber length and diameter
equivalent bond strengths were obtained for S13 compared to because of the Poisson’s ratio effect. All of the fibers exhibited a
S19.5 and S30, regardless of the matrix strength. In addition, the steep increase in shear stress as the normalized slip approached
strengths were reduced with increasing fiber length and diameter. 1 due to the effect of fiber end deformation. However, such high
This was attributed to the following two reasons: (1) nonlinear dis- shear stress obtained at the end of the fiber pullout was insignifi-
tribution of bond stresses along the embedded length and (2) Pois- cant from a practical point of view [3].

5. Correlation between the single aligned fiber pullout behavior


and flexural behavior of composites

5.1. Effect of matrix strength

The main purpose of the fiber pullout tests was to presume the
tensile or flexural behavior of HPFRCC. Therefore, the effect of the
matrix strength on the flexural behaviors of HPFRCCs containing
2% (by volume) S13 fibers was investigated. The average flexural
load versus deflection (or flexural stress versus normalized deflec-
tion) curves are shown in Fig. 10. The flexural stress, rF, was calcu-
lated using the following equation: FL/bh2, where F is the applied
load, L is the clear span length, b is the beam width, and h is the
beam height. The average curve was obtained based on linear
interpolation with an equal space of the deflection axis of
0.0005 mm [20]. All of the specimens showed deflection-
hardening behavior, leading to a higher load carrying capacity after
matrix cracking. Therefore, the flexural strengths of all tested spec-
imens were closely related to the fiber bridging capacity rather
than the matrix cracking strength. The flexural strength increased
with increasing matrix strength. This is consistent with the find-
ings from Kim et al. [21] that direct tensile strength of HPFRCC
including deformed steel fibers increased with increasing the
matrix strength. For example, the highest average flexural strength
was 38.6 MPa for Matrix C, approximately 42% and 22% greater
than those of Matrices A and B, respectively. Based on the pullout
test results, the highest maximum pullout load was obtained as
35.3 N for Matrix C and was approximately 15% and 11% greater
than those of Matrices A and B, respectively. Although the flexural
strength, rF,max, was mainly influenced by the fiber pullout capac-
Fig. 9. Ratios of (a) average bond strength and (b) equivalent bond strength based ity, its improvement in the matrix strength was greater than that of
on S13.
272 D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276

increase in toughness was not identical to that of the pullout work


increase.
Fig. 12 shows the comparative cracking behaviors of HPFRCC
beams. All of the tested beams exhibited multiple micro-cracks,
followed by crack localization (Fig. 12(b)). The multiple micro-
cracking behavior is only able to be obtained when deflection-
hardening behavior is achieved. The average number of cracks
was clearly reduced with increasing matrix strength, while the
average crack spacing increased, which was attributed to the
increased brittleness and cracking strength of the matrix. For
example, the highest number of cracks was found to be 47.3 for
Matrix A, which was approximately 18% and 25% greater than
Matrices B and C, respectively.

5.2. Effect of fiber type

To evaluate the implication of fiber type on the flexural behav-


iors of HPFRCC, a number of HPFRCC beams were fabricated and
tested. In accordance with a previous study [23], the fiber align-
ment in the tensile load direction was disturbed when long steel
Fig. 10. Effect of matrix strength on average flexural load and deflection curves of fibers were adopted because of its increased fiber–fiber interaction
HPFRCC beams with 2 vol.% S13.
at a high volume fraction of 2%, as compared to the short or
medium-length steel fibers. The change in fiber orientation by
fiber-fiber integration at high volume fractions was also reported
the maximum pullout load. This might be affected by several
by Martinie and Roussel [24]. Thus, in order to provide similar fiber
parameters such as the fiber inclination effect and the existence
orientation to all of the beams containing various types and sizes of
of fibers in the matrix. However, we could effectively presume that
steel fibers, a lower fiber volume fraction of 1.5% was applied. The
the increase in matrix strength results in an improvement in the
flexural behaviors of HPFRCCs including 1.5% (by volume) fibers
flexural strength of HPFRCC beams based on the increase in maxi-
(S13, S19.5, S30, and H30) and exhibiting fc0 of 200 MPa are shown
mum fiber pullout load (or average bond strength).
in Fig. 13, which is the average flexural load versus deflection (or
The implication of matrix strength on the energy absorption
flexural stress versus normalized deflection) curves. It is interest-
capacity of HPFRCC beams was also investigated based on the
ing to note that the HPFRCC beams with S19.5 exhibited the high-
toughness value, T, which is the area under flexural load versus
est flexural strength of approximately 38.4 MPa, followed by those
deflection curve up to a certain deflection point. According to
ASTM C1609 [11], the toughness of FRC shall be evaluated at the
deflection point of L/150. Three other deflection points of L/600,
L/75, and L/50 were also considered because HPFRCC beams exhib-
ited deflection-hardening behavior and excellent load carrying
capacity even at a large deflection, similar to those adopted in a
previous study [22]. The deflection point of L/600 is included
within the deflection-hardening zone between the first cracking
point and peak point. The calculated toughness values are shown
in Fig. 11. The post-cracking toughness values were apparently
related to the fiber pullout work because the toughness is the
absorbed energy through the fiber pullout mechanism. Thus, the
toughness increased with increasing matrix strength regardless
of the deflection point, which is consistent with the findings from
the fiber pullout test results: the pullout work increases with
increasing matrix strength. The highest toughness at the deflection
point of L/150 was found to be 218.1 kNmm for Matrix C, approx-
imately 45% and 31% greater than Matrices A and B, respectively.
However, due to the parameters that mentioned above, the

Fig. 12. Comparative cracking behaviors of HPFRCC beams with 2 vol.% S13; (a)
Fig. 11. Effect of matrix strength on toughness of HPFRCC beams with 2 vol.% S13. number of cracks and average crack spacing, (b) crack patterns.
D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276 273

tial cracking was achieved in S19.5, followed by S30 and S13. The
H30 specimen provided the smallest post-cracking stiffness. Thus,
it is concluded that the toughness at the initial deflection point,
within the deflection-hardening zone, is greatly influenced by the
post-cracking stiffness. This is consistent with the finding that
the toughness at the deflection point of L/600 increased with
increasing matrix strength at the identical amount of fibers (S13)
because of the increase in post-cracking stiffness (Figs. 10 and
11). After the peak point, the HPFRCC beams containing S19.5
and S30 fibers showed much higher toughness than those with
S13 and H30 fibers due to their higher load carrying capacity. At
the deflection point of L/150, the HPFRCC beams with S19.5
showed the highest toughness, while the HPFRCC beams with
S30 provided the highest toughness at the deflection point of
L/50. This is attributed to the fact that S30 fiber resulted in better
post-peak ductility due to the longer length. Consequently, the use
of S19.5 or S30 fibers in the HPFRCC mixture with a fc0 of 200 MPa is
considered more effective in improving the energy absorption
capacity compared to that of S13 or H30 fibers.
Fig. 13. Effect of fiber type on average flexural load and deflection curves of HPFRCC HPFRCC beams containing H30 fibers produced the worst flexu-
beams with fc0 of 200 MPa. ral performance including the lowest flexural strength, post-peak
ductility, and energy absorption capacity (toughness), although it
provided the highest average bond strength, sav. This observation
with S30, S13, and H30. This is inconsistent with findings from pre-
can be explained as follows. (1) Poor fiber dispersion was obtained
vious fiber pullout tests. The highest flexural strength obtained in
in H30, as shown in Fig. 15(a), because bundle-type hooked fibers
the present study was nearly 32% higher than that of UHPC beams
were included. Each hooked fiber should be separated from the
including hybrid 1 vol% twisted and 1 vol% short straight steel
fiber bundles during the mixing process. However, insufficient sep-
fibers [25]. The S13 provided the higher average and equivalent
aration of fibers was observed for HPFRCC, caused by the absence
bond strengths compared to those of S19.5 and S30. However,
of a coarse aggregate, the inclusion of a large amount of the fibers,
the beams with S13 showed much smaller flexural strength than
the high fluidity of the matrix, etc. (2) Deteriorated pullout capac-
their counterpart with longer straight fibers (S19.5 and S30),
ity of the hooked fiber occurred due to fiber congestion, leading to
mainly attributed to the increased possibility of existing fibers at
a decrease in bonding surface area by a fiber bundle effect [8]. In
the crack surface. This explanation was demonstrated in a previous
addition, the splitting micro-cracks generated in the surrounding
study performed by Yoo et al. [26] in which the actual number of
matrix affected the pullout performance of adjacent hooked fibers.
S13 fibers per unit area at the crack surface was 34.0 ea/cm2 at a
vf of 2% and when it was placed at the corner, which was very sim-
ilar to the number (35.8 ea/cm2) of S19.5 fibers at the identical vol-
ume fraction and placement method. Even though the number of
fibers added to the mixture was reduced with increasing fiber
length (S13 ? S19.5), the actual number of detected fibers at the
localized crack was similar. For these reasons, the HPFRCC beams
containing S19.5 and S30 fibers provided much greater flexural
strength compared to that with S13 fibers because of greater bond-
ing areas between the fiber and matrix.
Fig. 14 exhibits the comparative energy absorption capacities of
HPFRCC beams with various steel fibers. It can be noted that the
greatest toughness at the deflection point of L/600, included within
the deflection-hardening zone, was 38.3 kNmm for the case of
S19.5, whereas the smallest toughness was 34.7 kNmm for H30.
The S13 and S30 specimens exhibited similar intermediate values,
which is consistent with the trend of post-cracking stiffness, as
shown in Fig. 13. The steepest increase in the flexural load after ini-

Fig. 15. Reasons of poor flexural performance of HPFRCC beams with H30 fibers: (a)
Fig. 14. Effect of fiber type on toughness of HPFRCC beams with fc0 of 200 MPa. image for cracked surface, (b) image for hooked fiber pullout [27].
274 D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276

(3) Matrix cracking at near the fiber end hook was obtained espe- 3) The maximum fiber stress, rf,max, increased with increasing
cially when the fibers were inclined, as shown in Fig. 15(b) [27]. A fiber length and decreasing fiber diameter. In contrast, the
large anchorage force by the end hook led to cracks in the matrix short fiber (S13) resulted in higher average and equivalent
because of its low tensile strength and high brittleness. For this bond strengths and maximum shear stress at the interface
reason, a large number of non-straightened hooked fibers were as compared to the longer fibers (S19.5 and S30).
observed in Fig. 15(a). 4) The flexural strength and energy absorption capacity of
HPFRCC beams increased with increasing matrix strength,
6. Conclusions identical to the fiber pullout performance. However, the best
flexural performance was obtained for the beams with S19.5
In this study, the implications of fiber type and matrix strength fibers, whereas the worst performance was obtained in the
on the fiber pullout behavior of HPFRCC were investigated. Addi- beams with H30, which are different with the trend of fiber
tionally, the flexural behaviors of HPFRCC beams with different pullout behavior. This implies that the correlation between
matrix strengths and fiber types were estimated to cross- the single aligned fiber pullout behavior and flexural behav-
correlate with fiber pullout behavior. From the aforementioned ior of composites is fairly low, due to several influential
discussions, the following conclusions were drawn: parameters.

1) Higher matrix strength resulted in the improved fiber pull-


out capacities including bond strength and pullout work. Acknowledgements
The use of straight steel fibers was more efficient in improv-
ing the pullout performance with increasing matrix This work was supported by the Korea Agency for Infrastructure
strength, compared to that of hooked steel fibers. With an Technology Advancement (KAIA) grant funded by the Ministry of
increase in the matrix strength, the short straight fiber was Land, Infrastructure and Transport (Grant 13IFIP-C113546-01).
effective in improving the average bond strength, whereas
the long straight fiber was efficient in improving the pullout Appendix A.
work.
2) The hooked fiber produced higher pullout work and average Fiber pullout behaviors of all test series are shown in Figs. A1–
and equivalent bond strengths than the straight fiber with A3. In order to obtain the average curves, a total of 10 specimens
an identical length of 30 mm. At the initial slip (normalized were fabricated and at least 6 or more specimens were tested for
slip < 0.3), the hooked fiber exhibited much higher fiber each variable. Because the individual curves showed similar
stress and shear stress at the interface than the straight shapes, a simple averaging procedure was adopted as follows.
fibers, whereas the straight fibers provided higher fiber The free-end slip was first assumed to increase in 0.01 mm incre-
and shear stresses after that (normalized slip  0.3). ments, and then, the pullout load was calculated by linear interpo-

Fig. A1. Pullout behavior of single fibers embedded in matrix A (fc0 = 112.2 MPa): (a) S13, (b) S19.5, (c) S30, (d) H30.
D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276 275

Fig. A2. Pullout behavior of single fibers embedded in matrix B (fc0 = 152.5 MPa): (a) S13, (b) S19.5, (c) S30, (d) H30.

Fig. A3. Pullout behavior of single fibers embedded in matrix C (fc0 = 190.2 MPa): (a) S13, (b) S19.5, (c) S30, (d) H30.
276 D.-Y. Yoo et al. / Composite Structures 174 (2017) 263–276

lation of the test data obtained. Once it was applied for all test [14] Kim JJ, Kim DJ, Kang ST, Lee JH. Influence of sand to coarse aggregate ratio on
the interfacial bond strength of steel fibers in concrete for nuclear power plant.
specimens, the load data was then averaged at identical deflec-
Nucl Eng Des 2012;252:1–10.
tions. Therefore, the data points were equally spaced on the x- [15] Yoo DY, Banthia N, Kang ST, Yoon YS. Size effect in ultra-high-performance
axis, which is slip. concrete beams. Eng Fract Mech 2016;157:86–106.
[16] Hossain AB, Weiss J. Assessing residual stress development and stress
relaxation in restrained concrete ring specimens. Cem Concr Compos
References 2004;26(5):531–40.
[17] Park SH, Ryu GS, Koh KT, Kim DJ. Effects of shrinkage reducing agent on pullout
[1] Banthia N, Gupta R. Hybrid fiber reinforced concrete (HyFRC): fiber synergy in resistance of high-strength steel fibers embedded in ultra-high-performance
high strength matrices. Mater Struct 2004;37(10):707–16. concrete. Cem Concr Compos 2014;49:59–69.
[2] Yoo DY, Yoon YS, Banthia N. Flexural response of steel-fiber-reinforced [18] Wille K, Xu M, El-Tawil S, Naaman AE. Dynamic impact factors of strain
concrete beams: effects of strength, fiber content, and strain-rate. Cem Concr hardening UHP-FRC under direct tensile loading at low strain rates. Mater
Compos 2015;64:84–92. Struct 2016;49(4):1351–65.
[3] Wille K, Naaman AE. Effect of ultra-high-performance concrete on pullout [19] Ahmad FS, Foret G, Le Roy R. Bond between carbon fibre-reinforced polymer
behavior of high-strength brass-coated straight steel fibers. ACI Mater J (CFRP) bars and ultra high performance fibre reinforced concrete (UHPFRC):
2013;110(4):451–62. experimental study. Constr Build Mater 2011;25(2):479–85.
[4] Yoo DY, Kang ST, Lee JH, Yoon YS. Effect of shrinkage reducing admixture on [20] Yoo DY, Zi G, Kang ST, Yoon YS. Biaxial flexural behavior of ultra-high-
tensile and flexural behaviors of UHPFRC considering fiber distribution performance fiber-reinforced concrete with different fiber lengths and
characteristics. Cem Concr Res 2013;54:180–90. placement methods. Cem Concr Compos 2015;63:51–66.
[5] Banthia N, Trottier JF. Concrete reinforced with deformed steel fibers, part I: [21] Kim DJ, Wille K, Naaman AE, El-Tawil S. Strength dependent tensile behavior of
bond-slip mechanisms. ACI Mater J 1994;91(5):435–46. strain hardening fiber reinforced concrete. In: proceedings of RILEM
[6] Lee Y, Kang ST, Kim JK. Pullout behavior of inclined steel fiber in an ultra-high international workshop on high performance fiber reinforced cement
strength cementitious matrix. Constr Build Mater 2010;24(10):2030–41. composites-HPFRCC 6, Ann Arbor, USA, June 20–22, 2011.
[7] Kim DJ, El-Tawil S, Naaman AE. Effect of matrix strength on pullout behavior of [22] Yoo DY, Kang ST, Yoon YS. Enhancing the flexural performance of ultra-high-
high-strength deformed steel fibers. ACI Spec Publ 2010;2010(272):135–50. performance concrete using long steel fibers. Compos Struct
[8] Li VC, Wang Y, Backer S. Effect of inclining angle, bundling and surface 2016;147:220–30.
treatment on synthetic fibre pull-out from a cement matrix. Composites [23] Park JJ, Yoo DY, Park GJ, Kim SW. Feasibility of reducing the fiber content in
1990;21(2):132–40. ultra-high-performance fiber-reinforced concrete under flexure. Materials
[9] Yoo DY, Banthia N, Yang JM, Yoon YS. Mechanical properties of corrosion-free 2017;10(2):118–31.
and sustainable amorphous metallic fiber-reinforced concrete. ACI Mater J [24] Martinie L, Roussel N. Simple tools for fiber orientation prediction in industrial
2016;113(5):633–43. practice. Cem Concr Res 2011;41(10):993–1000.
[10] Kim DJ, El-Tawil S, Naaman AE. Loading rate effect on pullout behavior of [25] Nguyen DL, Kim DJ, Ryu GS, Koh KT. Size effect on flexural behavior of ultra-
deformed steel fibers. ACI Mater J 2008;105(6):576–84. high-performance hybrid fiber-reinforced concrete. Compos Part B 2013;45
[11] ASTM C 1609/C 1609M. Standard test method for flexural performance of (1):1104–16.
fiber-reinforced concrete (using beam with third-point loading). West [26] Yoo DY, Kang ST, Yoon YS. Effect of fiber length and placement method on
Conshohocken, PA: ASTM International; 2012. p. 1–9. flexural behavior, tension-softening curve, and fiber distribution
[12] Kim DJ, El-Tawil S, Naaman AE. Correlation between single fiber pullout characteristics of UHPFRC. Constr Build Mater 2014;64:67–81.
behavior and tensile response of FRC composites with high strength steel fiber. [27] Flanders L, Rushing T, Landis E. Energy dissipation mechanisms in the fracture
In: Proceedings of the Fifth International Workshop of High Performance Fiber of fiber reinforced ultra high performance concrete. In: 4th International
Reinforced Cementitious Composites, RILEM Proceedings, Pro. 53, S.A.R.L., Symposium on Ultra-High Performance Concrete and High Performance
Cachan, France, 2007, pp. 67–76. Construction Materials (HiPerMat2016), Kassel, Germany, 2016, pp. 1–8.
[13] Yoo DY, Park JJ, Kim SW, Yoon YS. Influence of reinforcing bar type on
autogenous shrinkage stress and bond behavior of ultra high performance
fiber reinforced concrete. Cem Concr Compos 2014;48:150–61.

You might also like