Академический Документы
Профессиональный Документы
Культура Документы
a r t i c l e i n f o a b s t r a c t
Article history: The role played by a number of fiber properties (like fiber length, content, type, hybridization, tensile
Received 27 September 2016 strength and energy-absorption capability or toughness) is investigated in this paper with reference to
Received in revised form 14 January 2017 shotcrete mixes containing amorphous fibers with random atomic arrangement (well known for their
Accepted 23 January 2017
excellent corrosion resistance) and conventional hooked steel fibers. Several prismatic beams and plates
Available online 31 January 2017
were cast by varying the above-mentioned properties. Compared to steel fiber-reinforced concrete
(SFRC), amorphous metallic fiber-reinforced concrete (AFRC) exhibits a higher flexural strength, but a
Keywords:
lower residual strength in tension after first cracking, a lower energy-absorption capability, and a higher
Fibers
Shotcrete
rebound ratio during the spraying process. The rebound is favored by the fiber ball in the case of amor-
Flexure phous fibers, while a relatively small number of steel fibers gives shotcrete a more cohesive behavior dur-
Toughness ing the spraying process. Amorphous fibers are, therefore, more suitable for quick repairs (for instance, in
Field application sewerage pipes), whenever quick hardening and good short-term mechanical properties in tension are
sought for, together with corrosion resistance.
2017 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tust.2017.01.012
0886-7798/ 2017 Elsevier Ltd. All rights reserved.
86 J.-M. Yang et al. / Tunnelling and Underground Space Technology 64 (2017) 8594
Nomenclature
to their corrosion resistance), to the advantage of shotcrete dura- employed, and they were similar to those used in the laboratory.
bility. Furthermore, amorphous fibers are flexible and pliable The water-to-cement ratio (W/C) was 0.44 and the target slump
between the fingers; therefore, amorphous metallic fiber rein- value was 150 20 mm. A high-strength cement-based mineral
forced shotcrete (AFRS) can be handled with conventional spraying accelerator was employed, providing excellent early age strength
devices. development and a lower rebound of shotcrete (Won et al.,
This study investigates the applicability of AFRS in tunnel appli- 2013). The properties and chemical components of the hardening
cations. To this end, the effects of variables such as the length, ten- accelerator are listed in Table 2.
sile strength, and quantity of amorphous fibers were examined To investigate the flexural performance of FRC, five types of
because of their role in guaranteeing the performance of AFRCs. amorphous fibers in addition to hooked fibers (coming in bundles)
Additionally, the strength and ductility class of AFRC were investi- were used. The properties of the fibers are reported in Tables 3 and
gated according to the European Specification for Sprayed Concrete 4, as declared by the producers. If the cooling rate exceeds a certain
(EFNARC, 1996). Finally, a field application test of AFRS was per- critical value, the random atomic microstructure typical of the liq-
formed in the building site of a real tunnel. uid state remains frozen in the solid state. The atoms solidify in a
non-equilibrium phase and the so-called amorphous alloys are
obtained. Because of the different atomic microstructure compared
2. Experimental program with the so-called crystalline alloys, completely different physi-
cal, chemical, and mechanical characteristics are obtained. Amor-
2.1. Materials and mixture proportions phous alloys have excellent strength, toughness, friction
resistance, and corrosion resistance against moisture and other
In this study, fiber-reinforced concrete (FRC) without an accel- severe environments.
erator and FRS with an accelerator were mixed in the laboratory As a precondition to produce high-quality FRCs, the fibers have
and in the field, respectively. The mixing proportions used in this to be adequately dispersed. In order to evaluate the dispersibility
study are provided in Table 1. In the case of the laboratory FRC of the amorphous fibers, a wash analysis of fresh AFRC was
mix, a water-to-cement ratio (W/C) of 0.426 was adopted. Type 1 tested according to KS F2411 (2010), and its detailed process can
Portland cement was employed. The mass ratio of sand and aggre- be explained as follows. After 50 l of AFRC were mixed with an
gates (sand + gravel) was 0.60. Coarse aggregates, with a specific amorphous fiber volume percentage of 0.5% (=36.0 kg/m3) based
mass of 2.69 kg/m3 and a maximum size of 10 mm, were used. Fine on the lab mix proportion in Table 1, 7 l of fresh AFRC was mea-
aggregates were washed and had a specific mass of 2.59 kg/m3. The sured using a vessel to measure the air content of concrete. The
target slump value was 120 20 mm with an air content of 4 1%. measured AFRC was poured into a 0.09 mm sieve and was washed
Polycarbonate superplasticizer was added to achieve the target with water. The remaining materials in the sieve after washing
slump value for each mix. In the FRS mix, the materials and mix- were dried for 12 h in an oven and then all the amorphous fibers
ture proportions used within the real tunnel construction site were were sorted out with a magnet and weighed. The analytical wash-
ing test was repeated three times (Table 5). The average amor-
phous fiber volume percentage measured from 7 l fresh AFRC
was found to be 0.493% (=35.5 kg/m3) with a standard deviation
of 0.00802%. This result indicates that the dispersion of AFRC is
good. Using an identical mixture, a 100 100 400 mm3 beam
specimen was cast. After curing for one week, the dispersion state
of the amorphous fibers within the hardened beam specimen were
observed via CT (Computed Tomography) scanning at a nearby
hospital. As shown in Fig. 2, the dispersibility of AFRC was gener-
ally good.
2.2. Test setup and procedure to analyze the effects of fiber properties
Table 1
Mix proportions.
Mix W/C (%) S/a (%) Gmax (mm) Unit weight (kg/m3) Fiber (kg/m3) SP (%) AC (%) Slump (mm) Air (%)
W C S G
Lab test 42.6 60 10 200 470 971 660 AMF 29 0.7 0 120 20 41
AMF 36 0.8
AMF 42 0.9
SF 42 0.7
AMF21SF21 0.8
Field test 44.0 60 10 203 461 973 656 AMF 37 1.0 5.0 150 20 41
SF 37 0.8
Note: W/C = water-to-cement ratio, S/a = weight ratio of sand to total aggregate, Gmax = maximum size of gravel, W = water, C = cement, S = sand, G = gravel, AMF = amor-
phous metallic fiber, SF = steel fiber, SP = superplasticizer, and AC = accelerator.
Table 2
Properties and chemical components of the accelerator.
Table 3
Geometrical and mechanical properties of fibers.
Fiber Length Equivalent Width Thickness Slenderness Tensile strength Elastic modulus Density
(mm) diameter (mm) (mm) (mm) (Lf/de) (MPa) (GPa) (g/cm3)
AMF1 30 0.24 1.60 0.029 125 1400 140 7.2
AMF2 20 0.24 1.60 0.029 83 1400 140 7.2
AMF3 15 0.17 1.00 0.024 88 1400 140 7.2
AMF4 30 0.26 1.41 0.037 115 1600 140 7.2
AMF5 30 0.24 1.37 0.032 125 1850 140 7.2
SF1 30 0.50 60 1250 210 7.8
Note: AMF = amorphous metallic fiber, SF = steel fiber, Lf = fiber length, and de = equivalent diameter.
Table 5
Washing analysis test result of fresh AFRC.
Note: Ave. = average flexural strength and St. dev. = standard deviation.
88 J.-M. Yang et al. / Tunnelling and Underground Space Technology 64 (2017) 8594
P
Plate
specimen
Steel plate
(100 x 100 mm)
600 mm
Amorphous 100 mm
metallic fibers
500 mm
600 mm
P
an electronic scale (Fig. 5c). The rebound ratio was thus calculated
by the following equation
Steel
frame Rebound ratio % total weight of rebound material kg
=total weight of shotcrete kg 100 1
Finally, specimens for the compressive and flexural tests were
cast. In order to prepare the compressive strength test specimens,
shotcrete was sprayed into basins that were 800 800 300 mm3
in size; cores (diameter 100 mm and height 200 mm) were
extracted from the shotcrete in the basins, to prepare the cylinders
to be tested in compression. For the tests on beams, the shotcrete
was sprayed directly into prismatic molds (size: 150 150
Prismatic 550 mm3). Later, all the specimens were cured in the lab, at
LVDT T = 23 1 C and RH = 50 5%.
beam
Fig. 3. Four-point flexural test.
3. Experimental results and discussion
tests, a total of four plate specimens were cast (two plates for each 3.1. Effects of fiber properties on the flexural behavior of FRC
variable) with the following dimensions: 600 600 100 mm3. As
shown in Fig. 4, each plate specimen was simply supported on its 3.1.1. Effect of fiber length
four edges with a clear span of 500 mm, while a center point load A comparison of the flexural load versus the mid-span deflec-
was applied at a deformation rate of 1.5 mm/min through a contact tion curves for AFRC beams can be seen in Fig. 6, and several flex-
surface of 100 100 mm2 until reaching a center deflection of ural parameters are summarized in Fig. 7. In Fig. 7, the flexural
25 mm. strength was calculated by means of the formula Pmax.L/bh2, which
is based on the assumption that the material be perfectly rigid-
plastic. Herein, Pmax. is the maximum applied load, L is the clear
2.2.3. Field test span length, b is the beam width, and h is the beam height. Three
The tests on shotcrete containing amorphous fibers were per- prismatic beams were measured for each variable (Fig. 6ac), and
formed in a building site, where the boring of a tunnel was in pro- the average curves were recorded (Fig. 6d). It was evident that
gress. A total of 8 m3 of ready-mixed FRCs (4 m3 for each fiber) the load carrying capacity of AFRC under flexure was improved
containing 37 kg/m3 amorphous and hooked fibers (roughly 0.5% by increasing the length of the fibers. This was consistent with
by volume) were manufactured at a batch plant near the tunnel the findings reported by Yoo et al. (2016a). They noted that the
construction site. The procedures for field testing of shotcrete are flexural strength of ultra-high-performance concrete (UHPC)
summarized in the following. Fresh properties, such as the slump increased with an increase in fiber length given an identical fiber
and air content, of the ready-mixed FRC were first examined. Then, diameter and volume fraction. Image analysis applied to the spec-
FRC was poured into the shotcrete machine and sprayed inside the imens after being tested explains why increasing fiber length (for
tunnel (Fig. 5b). Afterwards, the shotcrete rebound test was per- the same fiber volume fraction and diameter) favors flexural
formed. After spraying 1 m3 of shotcrete onto the walls and ceiling, strength: longer fibers more effectively bridge the cracked
J.-M. Yang et al. / Tunnelling and Underground Space Technology 64 (2017) 8594 89
Lf = 15 mm Lf = 20 mm
(a) (b)
Lf = 30 mm
Lf = 30 mm
Lf = 20 mm
Lf = 15 mm
(c) (d)
3
Fig. 6. Effect of fiber length on the average flexural load versus deflection curves of AFRC at fiber weight of 36 kg/m (Vf = 0.5%) and fiber tensile strength of 1400 MPa; (a)
Lf = 15 mm, (b) Lf = 20 mm, (c) Lf = 30 mm, (d) average data.
surfaces. In the case of amorphous fibers, fiber-to-matrix bond (Lf) of 15 mm exhibited deflection-softening behavior. Therefore,
improves further because these fibers have smaller cross-section both the residual flexural strength (fD600 and fD150) and toughness
and higher slenderness (=length-to-equivalent diameter ratio), (TD150) values increased with an increase in fiber length. The resid-
leading to the increased total bonded surface. ual flexural strengths, fD600 and fD150, were calculated based on the
If it was assumed that the first cracking point corresponded to formula, PDL/bh2, where PD is the applied loads at the deflections of
the limit of proportionality (LOP) and was mostly influenced by L/600 and L/150, respectively, and the toughness (TD150) is the area
the matrix tensile strength as opposed to fiber properties (Yoo under the load versus the deflection curve at the deflection of
et al., 2016c), Fig. 6d reveals that the use of longer amorphous L/150. For example, the highest toughness values were found to
fibers would be more effective in increasing the load carrying be 15.7 N m for beams with 30 mm fiber lengths, approximately
capacity immediately after matrix cracking. Based on this result, 73% higher than that of beams with 15 mm fiber lengths. Conse-
AFRC beams incorporating longer amorphous fibers with lengths quently, it was noted that the use of long amorphous fibers was
(Lf) of 20 and 30 mm exhibited deflection-hardening behavior, more effective toward improving the flexural performance of con-
while beams that included amorphous fibers with a shorter length crete than short ones.
90 J.-M. Yang et al. / Tunnelling and Underground Space Technology 64 (2017) 8594
7.09 20 7.09 20
fMOR fMOR
7 7
5.99 14.9 15.73 f D600 6.00 15.73 f D600
6 5.41 16 f D150 6 16 f D150
Toughness (Nm)
Toughness (Nm)
T D150 12.23 T D150
Strength (MPa)
Strength (MPa)
5 5
12 12
4 9.1 3.48 4 3.48
3.04
3 8 3 8
2.16
2 1.58 2
4 4
1 0.68 0.58 1 0.54 0.58
0.38
0 0 0 0
15 20 30 29 36
Fiber length (mm) Fiber amount (kg/m3)
Fig. 7. Effect of fiber length on the flexural parameters of AFRC at fiber weight of Fig. 9. Effect of fiber amount on the flexural parameters of AFRC at fiber length of
36 kg/m3 (Vf = 0.5%) and fiber tensile strength of 1400 MPa. 30 mm and fiber tensile strength of 1400 MPa.
Toughness (Nm)
steel fibers) on the flexural behavior of concrete are provided in
Strength (MPa)
25 T D150
Figs. 10 and 11. Higher flexural strengths (fMOR) were obtained 6 5.32 5.39 5.37
4.95 20
for AFRC than that of SFRC, primarily due to the small cross- 4.27
sectional area of the amorphous fibers; nearly 16 times more fibers 4 3.34 15
were incorporated over steel fibers even when the fiber weights 10
(Wf) were identical (Won et al., 2012). Due to its larger bonding 2 1.21
5
area with cement matrix, amorphous fiber exhibited higher maxi-
mum pullout load than the hooked fiber, although it had a lower 0 0
bond strength (Won et al., 2013). For example, the fMOR of AFRC AFRC SFRC Hybrid-FRC
was found to be 8.96 MPa, approximately 66% higher than that of Specimen
Fig. 11. Effect of fiber type and hybridization on the flexural parameters at fiber
weight of 42 kg/m3 (Vf = 0.540.58%).
(a) (b)
Fig. 12. Fiber failure mode (Yoo et al., 2016b); (a) amorphous metallic fiber, (b) hooked-end steel fiber.
improvement of residual flexural strengths (fD600 and fD150) for amorphous fibers exhibited better flexural performance than their
AFRC were obtained with hybrid amorphous and steel fibers. This counterparts, in terms of higher flexural strength and toughness
means that the hybridization of two or more types of fibers could values. For instance, a toughness of 24.5 N m was observed for a
yield synergetic effects with regard to the performance of FRC. high fiber strength (fts) of 1850 MPa, approximately 15% higher
than that of fibers with a lower fts of 1600 MPa. Additionally, AFRC
beams with stronger fibers exhibited a steeper load-deflection
3.1.4. Effect of fiber tensile strength
curve between the LOP and MOR points compared with those with
Fig. 13 shows the strengthening effect of amorphous fibers on
weaker fibers. This is because stronger fibers resisted higher tensile
the flexural behavior of concrete, the relevant parameters of which
loads prior to fracturing than weaker fibers; beams including
are summarized in Fig. 14. Although identical fiber quantities (Wf
stronger fibers exhibit higher post-cracking stiffness and flexural
of 42 kg/m3) were incorporated, specimens with higher strength
strength values than their counterparts (with weaker fibers). Thus,
the use of amorphous fibers with higher strengths is recommended
to improve the flexural performance of AFRC.
fts = 1850 MPa 3.2. Specification of strength and ductility of sprayed FRCs as per
EFNARC
20
6 5.32 than those of AFRC cylinders at all ages; however, both samples
4.32 15 were specified as the same strength class of C48/60 according to
4 EN 206 (EFNARC, 1996). The primary reason for the slightly
10
reduced strength obtained for AFRC could be attributed to a greater
2 1.35 1.21 5 amount of fibers, which increased the porosity (by entrapping
some air) and in the case of AFRC could hardly be adequately
0 0 dispersed. Song and Hwang (2004) also reported a decrease in
1600 1800
compressive strength for high-strength concrete when a high vol-
Fiber strength (MPa)
ume content of steel fibers was incorporated. By contrast, the flex-
Fig. 14. Effect of fiber tensile strength on the flexural parameters at fiber weight of ural strength of AFRC beams was higher than that for SFRC beams
42 kg/m3 (Vf = 0.58%). due to a higher quantity of fibers, which was consistent with the
92 J.-M. Yang et al. / Tunnelling and Underground Space Technology 64 (2017) 8594
Table 6
Compressive strength of sprayed concrete with amorphous metallic and steel fibers.
Note: Ave. = average compressive strength and St. dev. = standard deviation.
Table 7 exhibited the flexural stresses at the deflections between 0.5 and
Flexural strength of sprayed concrete with amorphous metallic and steel fibers (after 2 mm higher than class 3 and class 0, they belonged to class 3
28 days).
and class 0, respectively.
Flexural strength (MPa) Strength class
Ave. St. dev.
3.2.3. Effect of fiber type on the energy absorption class by EFNARC
AFRC 7.66 1.358 C44/55
SFRC 5.24 0.332 C44/55 plate test
The average load and deflection curves up to the deflection
Note: Ave. = average flexural strength and St. dev. = standard deviation.
point of 25 mm can be seen in Fig. 16a. Compared to the uniaxial
flexural test results, the SFRC plates exhibited higher maximum
findings reported by Won et al. (2012) and Nayar and Gettu (2015). loads than the AFRC plates. It has been reported by Kim et al.
However, they were specified as the same strength class of C44/55 (2013) that the highest tensile strength obtained at the bottom
in accordance with EN 206 (EFNARC, 1996). Consequently, it was surface of plain concrete plates is generally higher than that
concluded that although different fiber types (amorphous fibers obtained at the bottom surface of the uniaxial concrete beams. In
vs. hooked fibers) resulted in a variation of strengths in concrete, addition, the first cracking strength of SFRC beams is similar to that
they did not influence the strength class specified by the European of plain concrete beams since it is closely related to matrix crack-
standard EN 206 for sprayed concrete. ing rather than the fiber bridging (Yoo et al., 2016c). From these
two reasons, the first cracking strength of AFRC plates was
expected to be higher than that of AFRC beams. The amorphous
3.2.2. Effect of fiber type on the residual strengths by EFNARC beam fibers included in the plates, aligned in the direction of the tensile
test
The average flexural stress versus deflection curves of AFRC and
SFRC can be seen in Fig. 15. In order to evaluate their residual SFRC
strength class, a classification of residual strength is also presented
AFRC
in Fig. 15 according to the EFNARC (1996). As reported in Sec-
tion 3.1, the AFRC beams exhibited higher flexural strengths and
a steeper decrease in the post-peak load carrying capacity, as com-
pared to the SFRC beams. The residual strengths of the SFRC beams
at all deflection points were slightly lower than class 4, but were
obviously higher than class 3. The AFRC beams exhibited higher
residual strengths at a deflection of 0.5 mm than class 4, whereas
the residual strengths at deflections of 2 and 4 mm were lower
than class 1 and were slightly higher than class 0 due to a steep
decrease in load carrying capacity after the peak point. According
to EFNARC (1996), the beams should maintain a flexural stress
on, or above, a certain boundary of residual strength class at the
deflections between 0.5 and 2 mm. Since the SFRC and AFRC beams (a)
SFRC
SFRC AFRC
AFRC
Class 4
Class 3
Class 2
Class 1
Class 0
(b)
Fig. 15. Effect of fiber type on the average residual stress versus deflection curves of Fig. 16. Effect of fiber type on; (a) average load versus deflection curve, (b) energy
FRC beams. absorption capacity of FRC plates.
J.-M. Yang et al. / Tunnelling and Underground Space Technology 64 (2017) 8594 93
load, were therefore subjected to higher pullout loads immediately observed with the naked eye. The surroundings of the fiber ball
after matrix cracking than those in the beams; thus, the AFRC possessed an insufficient shotcrete matrix, resulting in reduced
plates were limited to the increase of load carrying capacity, as adhesion and a higher rebound ratio for the case with amorphous
compared to the beams, due to the early breakage of amorphous fibers. Therefore, to reduce the rebound ratio of AFRS, the fiber dis-
fibers. On the other hand, the steel fibers were all pulled out with- persion should be carefully controlled or chemical admixtures
out breakage. Additionally, it was well-known that the deflection- need to be included that increase the adhesion of the matrix. Addi-
hardening behavior of FRC was more easily obtained under the tionally, some amorphous fibers, especially near the circumference
biaxial stress state using plate specimens than under the uniaxial of the nozzle, were randomly scattered without adhesion to the
stress state using beam specimens when fibers were pulled out shotcrete matrix. This may be attributed to the lower density
without breakage (di Prisco et al., 2013). Consequently, a signifi- and flexible characteristics of the amorphous fibers.
cant increase in the load carrying capacity after matrix cracking The compressive strength of the core cylinders was found to be
was observed for the SFRC plates, as shown in Fig. 16a, compared 21.4 MPa (standard deviation of 2.82 MPa) for the AFRS and
to the case of SFRC beams, and much higher maximum load was 23.0 MPa (standard deviation of 2.63 MPa) for the steel fiber rein-
obtained for the SFRC plates than for the AFRC plates. forced shotcrete (SFRS), respectively. This result was consistent
The cumulative energy absorbed by the SFRC plates was similar with the laboratory test results in Section 3.2.1, in which higher
to that of the AFRC plates until reaching a deflection of approxi- compressive strengths were obtained for SFRC as opposed to AFRC.
mately 3 mm, in which a much higher energy may be absorbed The average flexural load and deflection curves of shotcrete pris-
in the SFRC plates as opposed to the AFRC plates, as shown in matic beams with cross-sectional dimensions of 150 150 mm2
Fig. 16b. This was mainly due to the different fiber failure modes and a clear span length of 450 mm can be seen in Fig. 18, in addi-
(breakage for amorphous fibers and pullout for steel fibers). The tion to important flexural parameters. As expected, the AFRS
EFNARC (1996) specified sprayed concrete as one of three types exhibited higher flexural strengths with a steeper decrease in the
in terms of their toughness classification. When the energy absorp- post-peak load carrying capacity, resulting in lower toughness
tion of concrete plates for deflection up to 25 mm was higher than (TD150) values compared to those of SFRS. For example, the tough-
500, 700, and 1000 J, they belong to class a, b, and c, respectively. ness of shotcrete with steel fibers was almost 2 times higher than
Thus, the SFRC plates were specified as class c, whereas the AFRC that of shotcrete with amorphous fibers.
plates may not be specified because their absorbed energy at a The results of the tests performed in this study suggest that
deflection of 25 mm was lower than 500 J. using amorphous fibers in the shotcrete to be applied to tunnel
The typical failure modes of AFRC and SFRC plates can be seen in surfaces be limited, because of several drawbacks of the material,
Fig. 17. In the case of AFRC plates, cracks were localized in two per- such as fairly low residual strength (i.e., after first cracking) and
pendicular directions. On the other hand, SFRC plates exhibited
multiple random cracks due to deflection-hardening behavior.
Multiple random cracks were also observed by Salehian et al. fMOR = 5.49 MPa
(2014) for the SFRC plates, and multiple cracking behaviors were f D600 = 3.15 MPa
generally observed when hardening behavior was obtained (di AFR-shotcrete
f D150 = 0.51 MPa
Prisco et al., 2013). Since the tensile stress generated by the point T D150 = 40.74 Nm
load at the bottom surface of the plate was identical in the circum-
ferential direction, cracks were randomly generated from weak
points in the SFRC plates. fMOR = 4.07 MPa
Based on the above EFNARC beam and plate test results, it is SFR-shotcrete f D600 = 4.01 MPa
noted that the application of hooked steel fibers to tunnel lining f D150 = 3.24 MPa
shotcrete is more appropriate than that of amorphous fibers. T D150 = 84.22 Nm
Fig. 17. Typical failure modes; (a) AFRC plate, (b) SFRC plate.
94 J.-M. Yang et al. / Tunnelling and Underground Space Technology 64 (2017) 8594