Materials and Design 117 (2017) 139–149

Contents lists available at ScienceDirect

Materials and Design

journal homepage: www.elsevier.com/locate/matdes

Hybrid-fiber reinforced engineered cementitious composite under

tensile and impact loading
M.A.E.M. Ali a, A.M. Soliman b, M.L. Nehdi a,⁎
a
Department of Civil and Environmental Engineering, The University of Western Ontario, London, Ontario N6A 5B9, Canada
b
Department of Building, Civil and Environmental Engineering, Concordia University, Montreal, Quebec H3G 1M8, Canada

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• New SMA-PVA engineered cementi-

tious composite was developed.
• SMA fibers offer local pre-stress effect
in the novel composite.
• Novel SMA-PVA hybrid composite of-
fers superior performance under im-
pact.
• Composite has potential in design of
critical infrastructure for homeland se-
curity.

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

Article history: The behavior under impact loading of an innovative hybrid fiber-reinforced engineered cementitious composite
Received 31 October 2016 incorporating short randomly dispersed shape memory alloy (HECC-SMAF) and PVA fibers was explored using a
Received in revised form 13 December 2016 drop weight impact test. Test specimens were also heat-treated to investigate possible pre-stressing effects of
Accepted 16 December 2016
SMA fibers on the impact resistance of the ECC. Uniaxial tensile testing on ECC coupon specimens was also con-
Available online 19 December 2016
ducted. A two-parameter Weibull distribution was used to analyze variations in experimental results in terms of
Keywords:
reliability function. Results indicate that SMA fibers significantly enhanced the tensile and impact performance of
Engineered cementitious composite the ECC. Adding fibers beyond a certain dosage led to fiber clustering, thus, no further gain in tensile and impact
Shape memory alloy performance was measured. The impact resistance of HECC-SMAF specimens was further improved after expo-
Impact loading sure to heat treatment. This highlights the significant contribution imparted by the local pre-stressing effect of
Toughness SMA fibers to the impact resistance of the composite. The Weibull distribution was adequate to predict the im-
Heat treatment pact failure strength of ECC, allowing to avert additional costly experiments. This research underscores the poten-
Weibull distribution tial to engineer new cementitious composites with superior tensile properties and impact resistance for the
protection of critical infrastructure in the event of explosive or impact loading.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Concrete is arguably the world's most important construction mate-

rial. Concrete reinforced with steel rebar (RC) has been utilized for
⁎ Corresponding author. many decades in the construction of protective structures to resist im-
E-mail address: mnehdi@uwo.ca (M.L. Nehdi). pact and dynamic loads [1]. Subsequent to the attack on the world

http://dx.doi.org/10.1016/j.matdes.2016.12.047
0264-1275/© 2016 Elsevier Ltd. All rights reserved.
140 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149

Table 1
Mechanical properties of NiTi-SMA and PVA fibers.

Mechanical properties Ultimate tensile strength (MPa) Diameter (mm) Length (mm) Young's modulus (GPa) Elongation (%) Density (kg/m3)

SMA 869 0.635 16 41.0 38 6450

PVA 1620 0.039 8 42.8 6 1300

Table 2
Mixture proportions for the tested ECC specimens.

Mixture ID Cement Fly ash Silica sand w/cm HRWRA PVA (%Vf) SMA (%Vf)

ECC0-0 1 1.2 0.8 0.26 0.012 0 0.0

ECC2-0 1 1.2 0.8 0.26 0.012 2 0.0
ECC2-0.5 1 1.2 0.8 0.26 0.012 2 0.5
ECC2-1 1 1.2 0.8 0.26 0.012 2 1.0
ECC2-1.5 1 1.2 0.8 0.26 0.012 2 1.5

trade center twin towers, research on the resistance of structures to im-
pact loading, explosive waves and penetration of fragments has become
a very active research topic, especially for military and defense facilities.
Under impact loading, the failure of RC elements has often led to partial
or total collapse of buildings, causing substantial economic and human
life losses [2]. Therefore, innovative materials that can enhance the de-
formation capacity, ductility, damage tolerance, and confinement of
structural concrete elements, while minimizing residual crack sizes
and recovering/reducing permanent deformations, are highly sought
after [3]. A potential scheme to achieve such an enhanced performance
is using shape memory alloy (SMA) bars in structural elements either as
total or partial replacement for traditional reinforcing steel [4–6]. How-
ever, using large-diameter SMA bars was found to reduce the hysteretic
area of the SMA and its ability to dissipate energy compared to that of
small-diameter SMA wires. This was attributed to the accumulation of
more distorted martensite crystalline structure and inherent deficien-
cies in larger-diameter bars with respect to smaller-diameter wires
[7]. Moreover, the better spatial distribution of short fibers can facilitate
crack closing in multiple directions as opposed to unidirectional bars [8].
Most pertinent recent research work focused on the impact behavior
of cement-based materials reinforced by mono-fibers [9–11] and/or hy-
brid-fibers with two different types of fibers or more [1,12,13]. For in-
stance, Almusallam et al. [14] investigated the effect of utilizing more
than two different types and sizes of fibers on the impact resistance of
hybrid-FRC slabs. They tested a set of slabs reinforced with steel (SE),
polypropylene (PP) and Kevlar fibers under the effect of high-velocity
projectiles of up to 657 m/s. Results showed that the geometrical
properties of fibers were more influential than their material properties
in improving the impact resistance of the hybrid-fiber reinforced slabs.
The notion behind the growing interest in using hybrid fibers with two
or more fiber types and sizes is that they could create synergistic effects
through their different elastic properties and/or various sizes since
micro-fibers can control the growth of micro-cracks, while large fibers
restrain the growth of larger cracks.
With the advent of engineered cementitious composites (ECC) and
emergence of smart materials such as shape memory alloys (SMA),
new opportunities in tailoring high-impact resistance cement-based
materials have opened up. ECC is a high-performance cement-based
composite, which was developed in the early 1990's [15]. Maalej et al.
[16] explored the behavior of hybrid-fiber ECC panels reinforced with
1.5% polyethylene (PE) and 0.5% steel fibers subjected to a high velocity
projectile. Results demonstrated the potential of ECC to perform as a
protective material in various applications, such as improving shatter
resistance with reduced scabbing and spalling associated with distribut-
ed micro-cracking. Moreover, a series of drop weight impact tests were
conducted by Yang and Li [17] on hybrid-fiber-reinforced ECC panels
and beams incorporating PVA and PE fibers. It was concluded that the
material ductility of ECC specimens led to enhanced impact resistance
through improving the load carrying capacity and energy dissipation
ability of the tested structural elements. Furthermore, Soe et al. [1]
showed that a hybrid-fiber ECC incorporating a combination of 1.75%
PVA and 0.58% steel fibers achieved superior performance, absorbing
more energy than that of the control normal concrete when subjected
to high velocity projectiles.
Although there is significant research on the impact resistance of
ECC as a material, there is dearth of information on the performance
of ECC structural members under the effects of impact loads. Therefore,
in the present study, the impact resistance of innovative hybrid fiber-re-
inforced ECC mixtures incorporating a combination of randomly dis-
persed polyvinyl alcohol (PVA) and nickel titanium shape memory
alloy (NiTi-SMA) short fibers was investigated. The study aims to ulti-
mately coin a composite with superior resistance to impact loading,
and possibly with self-healing capability to create novel alternatives in
designing ductile structural elements capable of mitigating damage
inflicted by impact, explosive and dynamic loading.

2. Experimental program

2.1. Materials and mixture proportions

Ordinary portland cement (OPC) compliant with the American Soci-

Fig. 1. Schematic diagram of ECC coupon specimens.
ety for Testing and Materials (ASTM) standard C150 [18] and fly ash
(FA) meeting the ASTM standard C618 [19] were used in the binary
binder for all tested mixtures. Silica sand (SS) with an average particle
size of 200 μm and specific gravity of 2.65 was used as fine aggregate.
A polycarboxylate-based high-range water reducing admixture
 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149 141

Fig. 3. Typical stress-strain curves for ECC specimens.

Fig. 2. Schematic diagram of the general layout of the drop weight impact testing system.

(HRWRA) meeting the requirements of ASTM standard C494 [20] was For the drop weight test, 100 mm × 200 mm [4 in. × 8 in.] cylindrical
used to adjust workability. All ECC mixtures had a constant water-to- specimens were cast and then cut into three similar 100 mm × 50 mm
binder ratio (w/b) of 0.26, fly ash-cement ratio (FA/C) of 1.2 and 1.2% [4 in. × 2 in.] discs. Furthermore, 160 mm × 50 mm × 20 mm
of HRWRA by weight of binder. NiTi-SMA fiber conforming to ASTM [64 in. × 20 in. × 8 in.] coupon specimens were prepared from each mix-
standard F2063 [21] and PVA short fibers were added at different dos- ture for the uniaxial tensile test [23]. After 24 h, all ECC specimens were
ages to the ECC mixtures. Table 1 summarizes various characteristics demolded and cured for 7 days in sealed plastic bags at the laboratory
of the PVA and SMA fibers. temperature (21 ± 2 °C) then left in laboratory condition until testing.
Five different ECC mixtures incorporating various dosages of SMA All reported results represent the average value obtained on triplicate
and/or PVA short-fibers were tested. The first test mixture series was di- specimens.
rected to investigating the effects of using mono PVA fibers at a dosage
of 2% by volume fraction on the tensile and impact resistance of the ECC
matrix. In the second test series, the effects of adding different dosages
of SMA fibers (i.e. 0%, 0.5%, 1%, and 1.5% by volume fraction) on the ten-
sile and impact behavior of hybrid fiber (PVA combined with SMA) ECC
were investigated. Table 2 exhibits the compositions of the different ECC
mixtures. The first number in the abbreviation of mixture labels relates
to the PVA fiber content, while the second number indicates the SMA
fiber content. For example, mixture ECC2-1 refers to an ECC mixture in-
corporating 2% PVA fiber and 1% SMA fiber by volume fraction,
respectively.

2.2. Mixing, casting and curing procedures

Initially, all solid ingredients including the cement, FA and silica sand
were dry-mixed for one minute. Then, the mixing water and HRWRA
were added and mixed with the dry ingredients for another three mi-
nutes until a homogeneous mortar was obtained. The PVA and SMA fi-
bers were gradually added into the mortar over another three minutes
of mixing until the fibers were uniformly distributed based on visual in-
spection. More details about the utilized mixing technique were pre-
sented by Turk and Demirhan [22]. Fig. 4. Percent enhancement in tensile strength of ECC specimens.
142 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149
Fig. 5. Percent enhancement in strain capacity of ECC specimens.
3. Testing method
3.1. Uniaxial tensile test
Three uniaxial tensile coupon specimens, as shown in Fig. 1, were
prepared from each ECC mixture and tested at the age of 90 days to ex-
plore the tensile behavior of the mono- and hybrid-fiber-reinforced ECC
specimens. Tests were carried out using the Instron testing machine
under displacement control and a loading rate of 0.5 mm/min as per
the guidelines of the Japan Society of Civil Engineers (JSCE) [24] for di-
rect tension testing of HPFRC. The specimens were clamped at both
ends and wedge pressure was adjusted to avoid cracks in the clamped
parts. The tensile properties of the ECC coupon specimens including
the ultimate strength and strain capacity were evaluated directly
from a data analysis system connected to the Instron testing
machine. A 50 mm extensometer was attached to the coupon speci-
mens to estimate the strain capacity, which is evaluated by dividing
the extension reading at which the tensile stress on the descending
zone reaches 90% of the ultimate strength, by the extensometer
gauge length. The tensile strength is evaluated via dividing the
applied tensile load by the cross-sectional area of the coupon speci-
men [16].
3.2. Impact test
The drop weight impact test was conducted in compliance with the
guidelines proposed by the American Concrete Institute (ACI) Commit-
tee 544 [25]. Each test specimen was adjusted, at the age of 90 days, in
the testing setup and subjected to impact loading induced by a 4.5-kg
weight dropped from 457 mm height above the specimen, which was
able to produce an impact energy of 20.167 J per hit, as shown in Fig.
2. The numbers of impacts to induce the first visible crack (N1), and fail-
ure (N2), respectively were recorded.
Furthermore, a series of drop weight tests were conducted on similar
ECC specimens after applying heat treatment. The heat treatment pro-
cedure was similar to that suggested by Kim et al. [26]. A heating gun
was fixed at 35 mm from the specimen's front surface for 25 min to mo-
bilize the pre-stressing effect of NiTi-SMA fibers owing to its shape
memory effect. The heat gun raised the specimen's temperature by
about 150 °C, leading to the appearance of multiple micro-cracks at
Table 3
Statistical analysis of test results (N1/N2).
Specimen ID ECC0-0 ECC2-0
Mean 1/1 13/39
Standard deviation 0/0 1/2
Fig. 6. Impact energy sustained by ECC specimens up to failure.
the front face of the tested specimens. The temperature of the speci-
mens' core and surface were measured every minute during the heating
process using embedded thermal sensors and an infra-red (IR) ther-
mometer. The specimens were left to cool down to room temperature,
and then the drop weight test was carried out. The impact energy for
each ECC specimen was calculated according to ASTM standard D5628
[27] using the following equation (Eq. (1)):
IE ¼ Ni :h:w:f ð1Þ
where IE is the impact energy in Joule, Ni is the number of blows, h is the
falling height of the steel mass in mm, w is the mass of the steel hammer
in kg, and f is a constant with a value of 9.806 × 10−3.
4. Results and discussion
4.1. Uniaxial tensile strength
Fig. 3 shows typical stress-strain curves from uniaxial tensile tests of
hybrid-fiber ECC specimens conducted at the age of 90 days. The results
indicate that there is an intrinsic increase in the ultimate tensile
strength of mono-PVA-ECC coupon specimen compared to that of con-
trol fibreless ECC specimens. For instance, the ultimate tensile strength
of the ECC2-0 specimen is 301.7% higher than that of the ECC0-0 control
specimen. Likewise, the strain capacity of the PVA-ECC specimen was
enhanced due to the incorporation of PVA fibers. It can be observed
that the strain capacity at ultimate tensile strength of ECC2-0 is about
hundred and sixty times that of the fibreless ECC control mixture. Fur-
thermore, the PVA-ECC specimen exhibited a strain hardening behavior
with appearance of multiple cracking up to failure. As displayed in Fig. 3,
after first crack, the stress continued to raise with inelastic strain and
started to descend after reaching the peak strength. Conversely, the
fibreless ECC specimen showed fracture localization at the peak
strength due to its brittle nature.
Moreover, it can be observed that SMA fiber addition generally en-
hanced the composite tensile properties. The achieved tensile strength
ranged between 6 MPa and 8 MPa, depending on the SMA fiber content.
Fig. 4 illustrates the percent enhancement in tensile strength capacity of
the hybrid-fiber-reinforced ECC specimens compared to that of ECC
specimens reinforced by 2% PVA fiber alone. The ultimate tensile
ECC2-0.5 ECC2-1 ECC2-1.5
50/594 75/697 63/674
2/4 2/7 4/6
 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149 143

strength of ECC2-0.5, ECC2-1 and ECC2-1.5 increased by about 29.8%,

74% and 39.8% compared to that of ECC2-0, respectively. Additionally,
all hybrid-SMA-ECC specimens showed high tensile capacity by up to
8% with multiple cracking. Moreover, the strain capacity corresponding
to peak strength was significantly improved due to incorporating SMA
fibers in the ECC mixtures as shown in Fig. 5.
Additionally, the strain capacity of the hybrid-SMA-fiber-reinforced
ECC specimens reached nearly two orders of magnitude higher than
that of the mono-PVA-ECC specimens. For instance, the strain capacity
at peak strength of ECC2-0.5, ECC2-1 and ECC2-1.5 was enhanced by
about 90%, 146% and 111%, respectively, compared to that of the ECC
specimens incorporating 2% PVA fiber by volume fraction alone. The im-
proved tensile properties of hybrid ECC specimens can be attributed to
increasing the fiber-matrix interfacial bond area (i.e. chemical and fric-
tional) due to increasing the fiber content, consequently, improving the
overall load carrying capacity of the composite. In addition, the de-
creased tensile capacity of ECC2-1.5 specimens can be ascribed to non- Fig. 7. Impact failure energy of ECC specimens with and without heat treatment.
uniformity of fiber dispersion compared to that of ECC2-1 specimens.
Yet, it still achieved higher tensile capacity than that of the ECC speci-
mens incorporating 2% PVA fiber alone. The high fiber content seemed demonstrates the superior ability of SMA fibers to restrain crack propa-
to compromise the consolidation of the ECC and induce additional po- gation in ECC specimens under impact loading.
rosity due to fiber clustering, which was not favorable in terms of im- However, the hybrid-ECC specimen comprising a combination of 2%
pact resistance. Evidence for this is provided below. PVA and 1% SMA fibers exhibited the highest impact resistance com-
pared to all other tested specimens. Using higher dosage of SMA fibers
4.2. Impact resistance in the ECC mixture (e.g. ECC2-1.5) decreased the impact resistance com-
pared to that of ECC2-1. This can be attributed to less effective spatial
4.2.1. At normal temperature distribution of fibers due to agglomeration and possibly the increased
Table 3 summarizes the experimental impact test results for speci- porosity induced in the mixtures incorporating higher than 2% PVA
mens from all tested ECC mixtures at the age of 90 days. All specimens and 1% SMA fibers by volume fraction (Table 4).
were tested up to failure and the average values for impact failure ener-
gy were calculated. Moreover, the statistical analysis of test results ob- 4.2.2. After heat treatment
tained from the experimental drop weight impact test is reported in Fig. 7 displays the results of ECC specimens with and without heat
Table 3. treatment tested up to failure under the effect of repeated impact load-
The impact energy sustained by the ECC specimens up to first crack ing. The impact failure energy sustained by the heated ECC2-0 speci-
and up to failure is reported in Fig. 6. Specimens from the control mix- mens was only 11% lower than that of identical non-heated
ture (ECC0-0) without fiber addition failed after only one hit by the specimens, which indicates some degradation due to heat treatment.
drop weight and split into multiple fragments, which reflects its brittle This can be ascribed to the fact that PVA fibers (which have a glass tran-
behavior under impact loading. Conversely, adding 2% of PVA fiber by sition temperature of 85 °C) lost mechanical strength due to the heating
volume enhanced the composite’s behavior under impact loading. For process which reached 150 °C, particularly at the front surface of speci-
instance, specimens from the mixture ECC2-0 sustained impact loading mens where the heating was more intensive. Hence, it is expected that
up to first crack and failure of about 13 and 39 times that of the control at higher temperature, the degradation will be more significant, in
ECC0-0 specimens, respectively. This reflects the beneficial effect of in- agreement with previous study [26].
corporating PVA fibers in ECC specimens, which led to altering the On the other hand, the impact resistance of the hybrid PVA-SMA
mode of failure of ECC under impact loads from prematurely brittle to fiber-reinforced ECC specimens was enhanced subsequent to the heat
more ductile failure, in agreement with previous study [1]. treatment, despite that the contribution of the PVA fibers was
Generally, SMA fiber addition also improved the impact resistance of
ECC specimens in a similar trend to that achieved by PVA fiber addition.
For instance, 0.5% SMA fiber addition by volume fraction increased the
impact resistance to reach first crack and failure by about 284% and
1423% compared to that of the ECC specimens incorporating 2% PVA
fiber alone, respectively. Furthermore, increasing the SMA fiber addition
level increased the number of impacts to reach first crack and failure of
the specimens. For instance, the impact resistance to first crack and fail-
ure of specimens from mixtures ECC2-1 and ECC2-1.5 increased by
about 477% and 1687%, and 384% and 1628%, compared to that of the
ECC specimens incorporating 2% PVA fiber alone, respectively. This

Table 4
Cumulative mercury intrusion volume of ECC specimens (mL/g).

Mixture ID Specimen 1 Specimen 2 Specimen 3 Average

ECC0-0 0.07384 0.07906 0.07910 0.07733

ECC2-0 0.09391 0.09135 0.09112 0.09213
ECC2-0.5 0.09539 0.09088 0.10028 0.09552
ECC2-1 0.09306 0.09635 0.10467 0.09803
Fig. 8. Failure patterns of ECC coupon specimens: a) ECC matrix, b) PVA-ECC, and c) HECC-
ECC2-1.5 0.11066 0.10914 0.10447 0.10809
SMAF.
144 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149

Fig. 9. Failure patterns of ECC disc specimens: a) ECC0-0, b) ECC2-0, c) ECC2-0.5, d) ECC2-1, e) ECC2-1.5, and f) SMA fiber bridging.

simultaneously degraded as discussed above. For instance, the impact increased as the SMA fiber addition increased. For example, the impact
failure energy for specimens containing 0.5% SMA and 2% PVA fibers failure energy of the heat treated ECC 2-1 and ECC2-1.5 specimens in-
was increased by about 12.46% compared to that of the non-heated creased by about 34.15% and 29.53% compared to that of identical
specimens. Also, the impact failure energy for the heated specimens non-heated specimens, respectively. This can be attributed to the

Fig. 10. Effect of heat treatment on ECC specimens: a) surface cracking, and b) PVA deterioration.
 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149
Fig. 11. Weibull distribution of first crack impact energy of ECC specimens.
Fig. 12. Weibull distribution of impact energy at failure of ECC specimens.
beneficial pre-stressing effect in ECC specimens owing to the shape
memory alloy fibers, which restrained crack growth, a behavior also re-
ported by Kim et al. [26]. The local pre-stressing effect of SMA fibers as
evidenced by the enhanced impact resistance subsequent to heat treat-
ment opens interesting research directions for its application in protec-
tive systems in critical facilities for national security.
4.3. Mode of failure
The mode of failure of tested specimens under uniaxial tensile and
impact loading depends essentially on the matrix strength and fiber-
matrix interfacial bond strength. Under tensile loading, the wedge pres-
sure was adjusted to provide uniform friction and pressure, assuring
that the failure will occur within the failure zone (mid-zone) of speci-
mens (away from wedges). Fig. 8a, b and c show the different failure
patterns observed in fibreless ECC, mono-PVA-ECC and HECC-SMAF
specimens, respectively.
Furthermore, Fig. 9 illustrates the different failure modes under
impact loading. The failure mode of the control specimens consisting
of the fibreless ECC matrix was brittle (Fig. 9a). The specimens were
broken suddenly under the effect of the first hit and shattered into
multiple fragments. Conversely, the mono-fiber ECC specimens (in-
corporating 2% PVA fiber) exhibited a different mode of failure
(Fig. 9b). The PVA fiber reinforcement transformed the mode of fail-
ure from brittle to ductile through sustaining multiple impacts
145
before failure. Moreover, the crack pattern changed from one main
crack in the case of the control specimens, to multiple fine cracks in
the PVA fiber-reinforced ECC2-0 specimens. The area under the
impactor was shriveled, displaying a small indent diameter and
depth in the PVA-ECC specimens. This is consistent with findings of
previous investigations [1,16,28].
A similar trend was displayed by the hybrid fiber-reinforced ECC
specimens incorporating a combination of PVA and SMA fibers, as
shown in Fig. 9c, d and e. The number of cracks induced by impact load-
ing increased with the increase of the SMA fiber content. For instance, as
displayed in Fig. 9c, d and e, the number of cracks was increased by
about 66%, 133% and 167% in the ECC2-0.5, ECC2-1 and ECC2-1.5 speci-
mens compared to that of ECC2-0. Also, the shriveling area, indent di-
ameter and its depth increased with increasing SMA fiber content
likely owing to the SMA fiber crack bridging capability, which held the
specimen together across growing cracks and performed as a crack
growth restraint (Fig. 9f).
Fig. 10 illustrates the effect of the heat treatment on the ECC spec-
imens. It can be observed that all specimens showed multiple fine
cracks specifically at the front surface, which was facing the heat
gun as shown in Fig. 10a. Also, the PVA fiber was deteriorated, partic-
ularly at the front surface exposed directly to the heating, thus incur-
ring damage to its mechanical properties as displayed in Fig. 10b.
Despite this damage to the PVA fiber and the deterioration that the
matrix incurred due to heat treatment, the addition of SMA fibers
contributed to superior performance of the heat-treated specimens
under impact loading.
4.4. Statistical analysis of test results
4.4.1. Weibull distribution
Different probabilistic models have been used over the last few de-
cades to statistically analyze the fatigue and impact test data of concrete.
In particular, the two-parameter Weibull distribution was widely used
in many research works for describing the fatigue life and impact
performance of concrete [29–31]. The Weibull distribution function is
characterized by a probability density function as follows (Eq. (2):
α
nα−1 −ðnuÞα
f ðnÞ ¼ e ð2Þ
u u
where α represents the shape parameter (i.e. Weibull slope), u is the
scale parameter, and n is the specific value of the random variable N
(i.e. N1 and N2 in the present study). Integrating Eq. (2) gives the follow-
ing (Eq. (3)):
n α
F N ðnÞ ¼ 1−e−ðuÞ
ð3Þ
where FN (n) is the cumulative distribution function. According to
Saghafi et al. [32], the probability of survivorship function is represented
by (Eq. (4)):
n α
LN ¼ 1−F N ðnÞ ¼ e−ðuÞ ð4Þ
Taking the natural logarithm twice on both sides of Eq. (5), it can be
rewritten as follows (Eq. (5)):
  
1
ln ln ¼ αlnðnÞ−αlnðuÞ ð5Þ
LN
Eq. (5) represents a linear relationship between ln [ln (1 / LN)] and ln
(n). In order to illustrate Eq. (5) graphically, the empirical survivorship
function LN for impact test data is obtained from the following relation
[31] (Eq. (6)):
i−0:3
LN ¼ 1− ð6Þ
k þ 0:4
146 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149

Table 5
Shape, scale parameters and coefficient of determination of ECC specimens.

Specimen ID α u R2
N2 N1 N2 N1 N2 N1

ECC2-0 24.146 12.505 −161.7 −70.08 0.9322 0.9963

ECC2-0.5 138.44 31.038 −1300 −214.9 0.9505 0.9869
ECC2-1 102.41 46.759 −978.6 −342.6 0.9973 0.9874
ECC2-1.5 107.96 17.272 −1028 −123.98 0.9998 0.9778

where i represents the failure order number and k denotes the number additional impact testing. Similar conclusions were drawn by others [8,
of data points. A linear regression analysis was applied to the ln [ln (1 / 34] for steel fiber-reinforced concrete. However, further tests are re-
LN)] and ln (impact energy) values as shown in Figs. 11 and 12. The linear quired to generalize this concept for ECC composite with different
trend is established by drawing the best fit line between data points characterizations.
using the method of least squares. The slope of the line provides an es-
timate of the shape parameter (α) and the scale parameter (u), which
4.4.2. Regression analysis of drop weight test results
can be obtained by calculating the value at which the line intersects
Fig. 13 illustrates the correlation between the number of blows
the ln [ln (1 / LN)] axis. The shape parameter (α), scale parameter (u)
which caused the first visible crack of the fiber-reinforced ECC speci-
and the coefficient of determination (R2) for the ECC specimens are
mens and the number of blows which caused its failure. This correlation
shown in Table 5.
was used to develop prediction equations for the number of blows
Tables 6 and 7 display the impact energy values for ECC specimens at
which can cause failure of ECC specimens under impact loading as
the first visible crack and failure stage, respectively based on reliability
shown in Table 8. Since the R2 values displayed in Table 8 for ECC spec-
analysis. The first crack impact energy of ECC2-0, ECC2-0.5, ECC2-1,
imens are equal to or higher than 0.89, which exceeds the limit (0.7)
and ECC2-1.5 specimens was approximately equal to or higher than
suggested by Rahmani et al. [33], these equations can be used with rea-
187.927, 876.453, 1377.93, and 1004.0088 with R2 of 0.9963, 0.9869,
sonable accuracy to represent the correlation between the number of
0.9874, and 0.9778, respectively. Likewise, the impact energy at failure
blows causing the first visible crack and the ultimate failure of the ECC
of ECC2-0, ECC2-0.5, ECC2-1, and ECC2-1.5 was approximately equal
specimens produced with and without SMA fibers.
to or higher than 667.926, 11614.5, 13502.0, and 13087.796 with R2 of
0.9322, 0.9505, 0.9973, and 0.9998, respectively. According to Rahmani
et al. [33], a coefficient of determination of 0.7 or higher is sufficient for a 4.4.3. Analysis of fiber dispersion
reasonable reliability model. Since all impact test data had R2 were equal Fiber distribution and free areas significantly affect the mechanical
to or higher than 0.93, a two-parameter Weibull distribution can be performance of fiber-reinforced composites [35]. It was observed that
used to establish the statistical distribution of impact test data for ECC by achieving better fiber distribution with decreased fiber-free area,
incorporating SMA fibers. These developed reliability curves can be con- the first crack stress, multiple cracking response and fracture toughness
sidered as a useful tool to quickly estimate the impact resistance of ECC of the composite could be improved [36]. Enhanced fiber spatial distri-
at cracking and failure, without conducting costly and time-consuming bution with minimal fiber clumping can enhance the fiber crack bridg-
ing capability and transferring of load to other parts of the composite.
For instance, Akkaya et al. [35] concluded that fiber-reinforced concrete
sections with higher fiber clumping percentages would crack before
Table 6 others with better dispersed fibers.
Weibull distribution for first crack impact energy of ECC specimens. In the present study, image processing analysis for cracked surfaces of ECC
Reliability level ECC2-0 ECC2-0.5 ECC2-1 ECC2-1.5 specimens was carried out as shown in Fig. 14. The cracked surface was finely
0.99 187.927 876.453 1377.93 1004.0088
clipped with diamond cut and then polished. It can be observed that ECC2-0.5
0.90 226.776 945.382 1448.94 1150.3270 specimens showed uniform distribution of SMA fibers with large free spaces
0.80 240.801 968.518 1472.39 1201.4080 between fibers along the cross-section. Likewise, ECC2-1 specimens displayed
0.70 250.004 983.264 1487.23 1234.4780 favorable SMA fiber dispersion with smaller fiber-free areas compared to that
0.60 257.289 994.710 1498.70 1260.4200
of the ECC2-0.5 specimens. Furthermore, ECC2-1.5 showed non-uniformity of
0.50 263.646 1004.540 1508.51 1282.8900
0.40 269.597 1013.610 1517.54 1303.7890
0.30 275.548 1022.570 1526.43 1324.5640
0.20 282.019 1032.180 1535.94 1347.0120
0.10 290.213 1044.160 1547.75 1375.2350
0.01 306.753 1067.740 1570.86 1431.5470

Table 7
Weibull distribution for impact energy at failure of ECC specimens.

Reliability level ECC2-0 ECC2-0.5 ECC2-1 ECC2-1.5

0.99 667.926 11,614.5 13,502.0 13,087.796

0.90 736.193 11,813.3 13,815.4 13,375.778
0.80 759.432 11,877.5 13,917.0 13,469.077
0.70 774.328 11,917.8 13,980.9 13,527.718
0.60 785.933 11,948.8 14,030.0 13,572.802
0.50 795.930 11,975.2 14,071.9 13,611.228
0.40 805.183 11,999.3 14,110.3 13,646.460
0.30 814.340 12,023.0 14,147.9 13,681.018
0.20 824.189 12,048.3 14,188.1 13,717.850
0.10 836.504 12,079.5 14,237.8 13,763.433
0.01 860.866 12,140.1 14,334.5 13,852.084
Fig. 13. Relationship between N2 and N1 of ECC specimens.
 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149
Table 8
Relationship between N2 and N1 of ECC specimens.
Reliability level Equation
ECC2-0 N2 = 1.5 N1 + 19.833
ECC2-0.5 N2 = 2.5 N1 + 469.5
ECC2-1 N2 = 4.2857 N1 + 377
ECC2-1.5 N2 = 1.6892 N1 + 567.35
SMA fiber dispersion with existence of fiber clustering and appearance of
fiber-free areas.
Additionally, statistical analysis of fiber distribution was conducted
using the method of statistical point approaches. According to the
work of Stoyan et al. [37], and Repley [38], two statistical functions
known as K-function and F-function can be utilized to statically analyze
the SMA fiber dispersion in ECC specimens. The K-function represents
the tendency of fibers to clump, while the F-function measures the
fiber-free area. After applying the impact load, the cracked cross-section
orthogonal to the loading direction was divided into central, edge, and
corner zones. The center area was defined as all points laying at least
1/8 of the specimen smallest dimension from any edges. Other regions
were not investigated due to limitation of the analysis process for
edge and corner zones [35]. The K-function can be calculated as follows
(Eq. (7)):
" #
number of fibers within r for a given fiber
K ðr Þ ¼ ∑  ðAR Þ
fibers∈ AR ðtotal number of fibers in AR Þ
2 ECC specimen reinforced by 1.5% SMA fibers showed fiber clumping
where r is the distance from fiber, ranging between zero and R (in this
study R = 20 mm (half the smallest dimension of central area), AR is
the observed region (central area). The evaluated K-function was then
compared to complete the spatial randomness K-function, also known
as a Poisson's process. The random K-function can be calculated as fol-
lows (Eq. (8)):
K ðr Þ ¼ π r 2
Fiber-reinforced specimens with K-function of less than the random dis-
tribution pattern tend to have fibers that avoid each other (better distribut-
ed), while specimens with a K-function higher than the random
distribution pattern denotes a fibers clumping trend (Fig. 15). The percentage
of fibers clumping in a specimen with respect to the random dispersion can
be evaluated as follows (Eq. (9)):
 
C analysis. ECC2-0.5 and ECC2-1.5 specimens had F-function similar to the ran-
%clumping ¼ −1  100
π dom F-function, which reflects their random fiber dispersion performance.
where C is a constant term in Cr2 (Fig. 15). Although, the K-function alone is
not completely representative for the fiber clumping behavior [39], the use of
the F-function together with the K-function can be used in capturing unusual
Fig. 14. SMA fiber dispersion of: a) ECC2-0.5, b) ECC2-1, and c) ECC2-1.5.
147
performance [36]. The F-function can be estimated by planning a regular grid
of points on AR and then finding the nearest fiber to each grid point. The F-
R2 function is evaluated as follows (Eq. (10)):
0.9643
0.8929 number of grid points in AR whose nearest fiber is lesser than r away
F ðr Þ ¼
0.9967 total number of grid points in AR
0.9686
ð10Þ
Similarly, the F-function is compared to the F-function under the
complete spatial randomness condition and can be evaluated as follows
(Eq. (11):
F ðr Þ ¼ 1−e−πλr
2
ð11Þ
where λ is the fiber density in the process, which can be estimated through
dividing the cross-sectional area of all fibers by the area of the cracked section.
The fiber-free area is represented by evaluating the term 1 − F(r). The exis-
tence of fiber-free area can be observed via higher values of 1 − F(r) and
vice versa. Fiber-reinforced specimens with F-function higher than the com-
plete spatial randomness F-function refer to a uniform distribution perfor-
mance. Conversely, specimens with lower F-function than that of the
random F-function denote a clumping behavior.
Fig. 16 illustrates the evaluated K-function values of different ECC
specimens compared to that calculated under the complete spatial ran-
domness condition (random K-function). It can be observed that the
ð7Þ
performance with a trend higher than that of the random K-function.
The percentage of fiber clumping of ECC2-1.5 is about 22.28% with re-
spect to the K-function under the spatial randomness condition. Con-
versely, the ECC2-0.5 and ECC2-1 specimens displayed better SMA
fiber distribution with K-values lower than that of the random K-func-
tion. This is consistent with aforementioned observations acquired by
visual inspection. While ECC2-0.5 and ECC2-1 specimens showed regu-
lar SMA fiber distribution, ECC2-1 specimens displayed better SMA fiber
ð8Þ distribution than that of ECC2-0.5 specimens, especially at higher r dis-
tances. This can be observed through the lower K-function values of
ECC2-1 specimens at r N 8 mm.
On the other hand, Fig. 17 displays the calculated F-function values of dif-
ferent ECC specimens compared to that of the random F-function. The ECC
specimens incorporating 1% SMA fibers achieved the highest F-function
trend compared to that of ECC2-0.5, ECC2-1.5 specimens, and F-function
under spatial randomness condition, which indicates regularity of SMA
fiber distribution. This is consistent with results acquired earlier by K-function
ð9Þ
However, this trend was not valid for the ECC2-1.5 specimens at higher r dis-
tances. The F-function of ECC2-1.5 specimens was much lower than the ran-
dom F-function (typically at r N 12 mm), which indicates likelihood of fibers
clumping.
148 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149
Fig. 15. Typical K-function of fiber dispersion: a) clustered, b) random, and c) uniform.
In addition, Fig. 18 portrays the percentage of fiber-free area of different
ECC specimens with respect to r distances. At lower r distances, all ECC spec-
imens showed availability of fiber-free area. The ECC2-1 specimens achieved
the lowest fiber-free area, which reflects uniformity of SMA fiber dispersion
with relatively small spaces between fibers. In contrast, ECC2-0.5 specimens
indicated the highest percentage of fiber-free area, followed by ECC2-1.5
specimens. Similar performance was observed at medium r distances. How-
ever, at higher r distances (typically r N 10 mm), fiber-free area in ECC2-0.5
and ECC2-1 specimens no longer existed, indicating more uniform distribu-
tion of fibers. Conversely, free areas between SMA fibers in the ECC2-1.5 spec-
imens still existed at higher r distances. This is ascribed to the tendency of
SMA fibers to cluster at high fiber dosage. The analysis of fiber distribution
above provides insight into how various fiber dosages affected the tensile
and impact behavior of specimens incorporating various proportions of hy-
brid fibers.
4.4.4. Analysis of variance of test results
Recently, experimental test data of cement-based materials was statically
analyzed using different probabilistic models. Particularly, the analysis of var-
iance (ANOVA) has been among the most widely used [40–42]. To investigate
whether an experimental variable (e.g. SMA fiber addition and/or effect of
heat treatment on ECC performance) is statistically significant, an Fo value is
estimated and compared to an F standard (critical) value of an F-distribution
density function obtained from statistical tables based on the significance
level (α1) and the degrees of freedom of error obtained from an experiment
using the number of variables and observations. Exceeding the F critical value
denotes that the tested variable affects the mean of the results [43].
Fig. 16. K-function comparison of different ECC specimens and random dispersion.
Fig. 17. F-function comparison of different ECC specimens and random dispersion.
ANOVA at a significance level α1 = 0.05, displayed that variation in the
addition level of SMA fiber had a significant effect on the mean value of the
tensile strength and impact resistance of the ECC composite. The determined
Fo values for the tensile strength and impact test results were 25.21 and 77.51,
respectively, which is higher than the corresponding critical F value of 4.07
(F0.05,3,8). This reflects the significant effect of incorporating SMA fibers in
the production of ECC composites which is consistent with results obtained
earlier. Due to the overlap observed in impact test results due strength degra-
dation of ECC2-1.5 compared to that of ECC2-1 (Fig. 6) owing to fiber cluster-
ing and that of ECC2-0 due to heat treatment (Fig. 7), further ANOVA had
been applied especially for these results. The determined Fo value estimated
for impact resistance of HECC-SMAF composite reinforced by N2% PVA and
1% SMA fibers was 7.286 which is higher than the corresponding critical F
value of 5.79 (F0.05,2,5). This reflects the significant effect of using higher dosage
of SMA fibers in the ECC mixture (ECC2-1.5) leading to decreased impact re-
sistance compared to that of ECC2-1. This is consistence with aforementioned
observations. Conversely, heat treatment indicated an insignificant effect on
the impact resistance of the PVA-ECC composite (ECC2-0). The estimated Fo
value for the impact resistance results of the heated and non-heated ECC2-
0 specimen was 3.675 which is lower than the corresponding critical F
value (F0.05,2,5).
5. Conclusions
The performance of engineered cementitious composite specimens
incorporating hybrid PVA-SMA fiber reinforcement under tensile and im-
pact loading was evaluated by conducting direct uniaxial tensile and drop
Fig. 18. Percentage of fiber-free areas of different ECC specimens with respect to r distance.
 M.A.E.M. Ali et al. / Materials and Design 117 (2017) 139–149
weight impact tests. Based on the experimental results the following con-
clusions are drawn:
• Generally, the tensile capacity of ECC composites was improved due to
SMA fiber addition. ECC specimens incorporating 2% PVA and 1% SMA
fibers achieved highest tensile capacity at the same testing age.
• The impact resistance of mono- and hybrid-fiber ECC specimens was
found to be 30 to 700 times higher than that of the plain fibreless
ECC control specimens.
• The addition of PVA and SMA fibers into the ECC matrix changed the
failure mode of ECC specimens under the effect of impact loading
from brittle to ductile.
• The performance of ECC specimens subjected to impact loading was
improved due to SMA fiber addition. However, the ECC mixture incor-
porating 2% PVA and 1% SMA fibers by volume fraction achieved the
highest impact resistance compared to that of all other specimens.
Fiber dosages beyond this level seemed to compromise fiber disper-
sion and create fiber clustering, while disrupting the matrix and caus-
ing higher porosity, which resulted in reduced performance under
impact loading.
• Heat treatment damaged the PVA fibers and generally compromised
the impact resistance of ECC specimens incorporating PVA fibers
alone, leading to the appearance of multiple fine cracks at the front
surface of specimens directly subjected to the heating.
• Conversely, the heat treatment process activated the SMA fibers to
exert a local pre-stressing effect in the ECC composite, consequently
improving its impact resistance. This is despite the damage caused
by the heating to the coexisting PVA fibers.
• The Weibull distribution function proved capable of adequately
representing the impact test data with a linear correlation between
the number of impact which causes the first visible crack and that
which initiated the ultimate failure for all ECC specimens.
• Fiber dispersion analysis reveals that the ECC2-1 specimen shows bet-
ter fiber distribution compared to that of the ECC2-0.5 and ECC2-1.5
specimens, which is consistent with its better tensile and impact per-
formance.
• The performance of the hybrid PVA-SMA fiber-reinforced ECC under
impact loading points to the need for further research and indicates
the possibility of developing novel composites with superior impact
performance to protect infrastructure of national security importance
against explosive, dynamic and impact loading.
References
[1] K.T. Soe, Y.X. Zhang, L.C. Zhang, Impact resistance of hybrid-fiber engineered cemen-
titious composite panels, Composite Structures 104 (2013) 320–330.
[2] F. Alemdar, H. Sezen, Shear behaviour of exterior reinforced concrete beam-column
joints, Struct. Eng. Mech. 35 (1) (2010) 123–126.
[3] G.J. Parra-Montesinos, S.W. Peterfreund, S.-H. Chao, Highly damage-tolerant beam-
column joints through use of high-performance fiber-reinforced cement compos-
ites, ACI Struct. J. 102 (3) (2005) 487–495.
[4] M. Nehdi, A. Shahria, M. Youssef, Development of corrosion-free concrete beam-col-
umn joint with adequate seismic energy dissipation, Eng. Struct. 32 (9) (2010)
2518–2528.
[5] A.H.M.M. Billah, M.S. Alam, Seismic performance of concrete columns reinforced
with hybrid shape memory alloy (SMA) and fiber reinforced polymer (FRP) bars,
Constr. Build. Mater. 28 (1) (2012) 730–742.
[6] E. Choi, J.W. Hu, J. Lee, B. Cho, Recovery stress of shape memory alloy wires induced
by hydration heat of concrete in reinforced concrete beams, J. Intell. Mater. Syst.
Struct. 26 (1) (2015) 29–37.
[7] A. Zafar, B. Andrawes, Fabrication and cyclic behavior of highly ductile superelastic
shape memory composites, J. Mater. Civ. Eng. 26 (4) (2014) 622–632.
[8] G. Murali, A.S. Santhi, G. Mohan, Impact resistance and strength reliability of fiber
reinforced concrete using two parameter Weibull distribution, ARPN Journal of En-
gineering and Applied Sciences 9 (4) (2014) 554–559.
[9] P.S. Song, S. Hwang, B.C. Sheu, Strength properties of nylon- and polypropylene-
fiber-reinforced concretes, Cem. Concr. Res. 35 (8) (2005) 1546–1550.
[10] W. Li, J. Xu, Mechanical properties of basalt fiber reinforced geopolymeric concrete
under impact loading, Mater. Sci. Eng. A 505 (2009) 178–186.
149
[11] D.-Y. Yoo, N. Banthia, S.-W. Kim, Y.-S. Yoon, Response of ultra-high-performance
fiber-reinforced concrete beams with continuous steel reinforcement subjected to
low-velocity impact loading, Compos. Struct. 126 (2015) 233–245.
[12] M. Nehdi, J. Duquette, Fibre synergy in hybrid fibre-reinforced self-consolidating
concrete, ACI Mater. J. 101 (6) (2004) 508–517.
[13] A.A. Nia, M. Hedayatian, M. Nili, V.A. Sabet, An experimental and numerical study on
how steel and polypropylene fibers affect the impact resistance in fiber-reinforced
concrete, International Journal of Impact Engineering 46 (2012) 62–73.
[14] T.H. Almusallam, A.A. Abadel, Y.A. Al-Salloum, N.A. Siddiqui, H. Abbas, Effectiveness
of hybrid-fibers in improving the impact resistance of RC slabs, International Journal
of Impact Engineering 81 (2015) 61–73.
[15] V.C. Li, On engineered cementitious composites (ECC): a review of the material and
its applications, J. Adv. Concr. Technol. 1 (3) (2003) 215–230.
[16] M. Maalej, S.T. Quek, J. Zhang, Behaviour of hybrid-fiber engineered cementitious
composites subjected to dynamic tensile loading and projectile impact, J. Mater.
Civ. Eng. 17 (2) (2005) 143–152.
[17] E.-H. Yang, V.C. Li, Tailoring engineered cementitious composites for impact resis-
tance, Cem. Concr. Res. 42 (2012) 1066–1071.
[18] ASTM Standard, Standard Specification for Portland Cement, ASTM Standard C150/
C150M, ASTM International, West Conshohocken, USA, 2015.
[19] ASTM Standard, Standard Specification for Coal Fly Ash and Raw or Calcined Natural
Pozzolan for Use in Concrete, ASTM Standard C618, ASTM International, West
Conshohocken, USA, 2015.
[20] ASTM Standard, Standard Specification for Chemical Admixtures for Concrete, ASTM
Standard C494/C494M, ASTM International, West Conshohocken, USA, 2015.
[21] ASTM Standard, Standard Specification for Wrought Nickel-Titanium Shape Memo-
ry Alloys for Medical Devices and Surgical Implants, ASTM Standard F2063, ASTM
International, West Conshohocken, USA, 2012.
[22] K. Turk, S. Demirhan, The mechanical properties of engineered cementitious com-
posites containing limestone powder replaced by microsilica sand, Can. J. Civ. Eng.
40 (2) (2013) 151–157.
[23] Y. Zhu, Y. Yang, Y. Yao, Use of slag to improve mechanical properties of engineered
cementitious composites (ECCs) with high volumes of fly ash, Constr. Build. Mater.
36 (2012) 1076–1081.
[24] JSCE, Recommendations for Design and Construction of High Performance Fiber Re-
inforced Cement Composites With Multiple Fine Cracks, Japan Society of Civil Engi-
neers, 2008.
[25] ACI Standard, Measurement of Properties of Fiber Reinforced Concrete, ACI Standard
544.2R-89, Farmington Hills, MI 48331-3439 USA, Reapproved, 2009.
[26] M.K. Kim, D.J. Kim, Y.-S. Chung, E. Choi, Direct tensile behavior of shape-memory-
alloy fiber-reinforced cement composites, Constr. Build. Mater. 102 (2016)
462–470.
[27] ASTM Standard, Standard Test Method for Impact Resistance of Flat, Rigid Plastic
Specimens by Means of a Falling Dart (Tup or Falling Mass), ASTM Standard
D5628, ASTM International, West Conshohocken, USA, 2015.
[28] J. Zhang, M. Maalej, S.T. Quek, Performance of hybrid-fiber ECC blast/shelter panels
subjected to drop weight impact, J. Mater. Civ. Eng. 19 (10) (2007) 855–863.
[29] H. Li, M. Zhang, J. Ou, Flexural fatigue performance of concrete containing nanopar-
ticles for pavement, Int. J. Fatigue 29 (7) (2007) 1292–1301.
[30] S. Raif, A. Irfan, Statistical analysis of bending fatigue life data using Weibull distribu-
tion in glass-fiber reinforced polyester composites, Mater. Des. 29 (6) (2008)
1170–1181.
[31] B. Raman, C. Rakesh, Fatigue-life distributions and failure probability for glass-fiber
reinforced polymeric composites, Compos. Sci. Technol. 69 (9) (2009) 1381–1387.
[32] A. Saghafi, A.R. Mirhabibi, G.H. Yari, Improved linear regression method for estimat-
ing Weibull parameters, Theor. Appl. Fract. Mech. 52 (3) (2009) 180–182.
[33] T. Rahmani, B. Kiani, M. Shekarchi, A. Safari, Statistical and experimental analysis on
the behavior of fiber reinforced concretes subjected to drop weight test, Constr.
Build. Mater. 37 (2012) 360–369.
[34] C. Xiang-yu, D. Yi-ning, C. Azevedo, Combined effect of steel fibres and steel rebars
on impact resistance of high performance concrete, Journal of Central South of Uni-
versity of Technology 18 (2011) 1677–1684.
[35] Y. Akkaya, S.P. Shah, B. Ankenman, Effect of fiber dispersion on multiple cracking of
cement composites, J. Eng. Mech. 127 (4) (2001) 311–316.
[36] Y. Akkaya, J. Picka, S.P. Shah, Spatial distribution of aligned short fibers in cement
composites, J. Mater. Civ. Eng. 12 (3) (2000) 272–279.
[37] D. Stoyan, W.S. Kendall, J. Mecke, Stochastic Geometry and Its Applications, second
ed. John Wiley & Sons, Chichester, 1995.
[38] B.D. Repley, Statistical Inference for Spatial Processes, first ed. Cambridge University
Press, 1991.
[39] A.J. Baddeley, B.W. Silverman, A cautionary example on the use of second-order
methods for analyzing point patterns, Biometrics 40 (4) (1984) 1089–1093.
[40] S. Ahmad, S.A. Alghamdi, A statistical approach to optimizing concrete mixture de-
sign, The Scientific World Journal 2014 (2014).
[41] A.M. Soliman, M.L. Nehdi, Effect of natural wollastonite microfibers on early age be-
havior of UHPC, J. Mater. Civ. Eng. 24 (7) (2012) 816–824.
[42] D. Nieves-Mendoza, C. Gaona-Tiburcio, H.L. Hervert-Zamora, J.R. Tobias, P. Castro-
Borges, O.R. Colas, P. Zambrano Robledo, A. Martínez-Villafañe, F. Almeraya-
Calderón, Statistical analysis of factors influencing corrosion in concrete structures,
Int. J. Electrochem. Sci. 7 (2012) 5495–5509.
[43] D.C. Montgomery, Design and Analysis of Experiments, eighth ed. John Wiley and
Sons, Hoboken, New Jersey, USA, 2012.