polymers

Article
Mechanical Behavior of Hybrid Glass/Steel Fiber
Reinforced Epoxy Composites
Amanda K. McBride 1, *, Samuel L. Turek 1 , Arash E. Zaghi 1, * and Kelly A. Burke 2,3,4, *
1 Civil and Environmental Engineering Department, University of Connecticut, 261 Glenbrook Road,
Unit 3037, Storrs, CT 06269-3037, USA; samuel.turek@uconn.edu
2 Chemical and Biomolecular Engineering, University of Connecticut, 191 Auditorium Road, Unit 3222,
Storrs, CT 06269-3222, USA
3 Polymer Program, Institute of Materials Science, University of Connecticut, 97 North Eagleville Road,
Unit 3136, Storrs, CT 06269-3136, USA
4 Biomedical Engineering, University of Connecticut, 260 Glenbrook Road, Unit 3247, Storrs,
CT 06269-3247, USA
* Correspondence: amcbride7392@gmail.com (A.K.M.); arash.esmaili_zaghi@uconn.edu (A.E.Z.);
kelly.burke@uconn.edu (K.A.B.); Tel: +1-732-275-4530 (A.K.M.); +1-860-486-2468 (A.E.Z.);
+1-860-486-3133 (K.A.B.)

Academic Editor: Mohamed Khayet

Received: 21 February 2017; Accepted: 19 April 2017; Published: 23 April 2017

Abstract: While conventional fiber-reinforced polymer composites offer high strength and stiffness,
they lack ductility and the ability to absorb energy before failure. This work investigates hybrid fiber
composites for structural applications comprised of polymer, steel fiber, and glass fibers to address
this shortcoming. Varying volume fractions of thin, ductile steel fibers were introduced into glass
fiber reinforced epoxy composites. Non-hybrid and hybrid composite specimens were prepared and
subjected to monolithic and half-cyclic tensile testing to obtain stress-strain relationships, hysteresis
behavior, and insight into failure mechanisms. Open-hole testing was used to assess the vulnerability
of the composites to stress concentration. Incorporating steel fibers into glass/epoxy composites
offered a significant improvement in energy absorption prior to failure and material re-centering
capabilities. It was found that a lower percentage of steel fibers (8.2%) in the hybrid composite
outperformed those with higher percentages (15.7% and 22.8%) in terms of energy absorption and
re-centering, as the glass reinforcement distributed the plasticity over a larger area. A bilinear
hysteresis model was developed to predict cyclic behavior of the hybrid composite.

Keywords: composite; hybrid; fiber reinforced polymer; mechanical properties; plastic deformation;
energy absorption

1. Introduction
Fiber-reinforced polymer (FRP) composites are commonly comprised of glass or carbon fibers to
provide a high-strength and lightweight material for a variety of industries. However, these fibers are
inherently brittle and have a limited energy absorption capacity prior to failure. This prohibits the
use of FRPs in certain applications, specifically in critical structural elements that may be subjected to
extreme events such as earthquakes, truck collision, or blast, during which energy absorption is crucial.
Researchers have studied various ways to improve composite ductility, including polymer matrix
toughening via silica nanoparticles [1] and the use of more ductile polymeric (e.g., polypropylene) [2]
or natural fibers (e.g., flax) [3]. However, these methods may compromise composite stiffness.
Metal fiber reinforcement has traditionally been used in rubber tires for additional strength in the
form of continuous steel cords [4] and in engineered cementitious composites (ECC) for ductility in

Polymers 2017, 9, 151; doi:10.3390/polym9040151 www.mdpi.com/journal/polymers

Polymers 2017, 9, 151 2 of 16

the form of short fibers [5]. Advancements in the manufacturing of metal fibers have introduced a
unique class of ultra-thin (<100 µm) stainless steel fibers, possessing both a high stiffness and failure
strain. Recent studies have been performed on the tensile and impact behavior of unidirectional (UD)
and cross-ply polymer composites utilizing these steel fibers [6–11]. These studies have investigated
the effect of brittle and ductile matrices, fiber architecture, and modifying adhesion between fiber and
matrix. This research has demonstrated the potential of these fibers to enhance structural performance
of composites in terms of failure strain and energy dissipation.
Hybrid fiber composites were developed to provide both strong and ductile reinforcement while
reducing material cost and weight. One class of hybrids, known as fiber metal laminates (FML), consist
of alternating layers of thin metal sheets and FRP [12]. FML are commonly layered with aluminum and
glass (e.g., glass laminate aluminum reinforced epoxy (GLARE)) or aluminum and Kevlar (e.g., aramid
fiber reinforced aluminum laminate (ARALL)) and are used as an alternative material for airframe
structures with enhanced fatigue resistance [13,14]. GLARE has been commercially implemented in
the fuselage of the Airbus A380 [15]. Rubio-González and coworkers studied residual strength of
carbon-reinforced aluminum FMLs after fatigue damage through open-hole testing, in which it was
observed that fiber laminates were more fatigue tolerant than the base carbon composite [16].
Mechanical properties and predictive models of nonmetal hybrid fiber-reinforced composites
have been reviewed by Kretsis, and more recently by Swolfs and coworkers, who also reviewed metal
fiber composites [17,18]. The ‘hybrid effect’ was investigated by Hayashi and defined as the apparent
failure strain enhancement of the low elongation fiber (carbon) in a carbon/glass hybrid composite [19].
Marom and coworkers defined the hybrid effect as the deviation of mechanical properties from the Rule
of Mixtures (ROM) [20], expanding the effect to other properties beyond failure strain. In composites
with elastic fibers, longitudinal material properties, such as stiffness and strength, can be estimated
based on the Voigt model or “equal strain” assumption of a viscoelastic material in parallel. This
leads to the ROM, or weighted mean of various properties [21]. That assumption is no longer valid
when incorporating inelastic fibers such as steel; therefore, the prediction of the hybrid composite
properties using the ROM becomes less accurate outside the elastic region. More research is needed to
understand the mechanical behavior in nonmetal-metal reinforced hybrid composites, however, as the
literature is presently limited.
Incorporating metal fibers into conventional composites has previously been studied for
their effect on non-structural properties. For example, Breuer and coworkers investigated a
carbon/steel fiber hybrid composite as a lightweight material for electrical conductivity in aerospace
applications [22]. However, it was realized that composite mechanical properties can also be further
tailored using metal fibers for structural benefits. Satish and coworkers studied the tensile and
compressive behavior of a steel/nylon fiber reinforced polyester composite [23]. The addition of steel
greatly increased the strength and stiffness of the composite, as those properties of steel are superior to
nylon and polyester. However, delamination failure was observed due to a weak interface formed
between the steel and the polyester. Ahmed studied composite multifunctionality by considering the
impact behavior of the hybridization of glass and steel reinforcement, observing that the addition of
metal fibers provides increased energy absorption and lowered the damage area under low velocity
impact [24]. Thysen studied the effect of lay-up and fiber ratios on the tensile strength and failure
strain of glass/steel composites in epoxy and nylon (PA-6) matrices [25]. Without reinforcement, the
nylon matrix was more ductile than the epoxy matrix. However, the failure strain of the reinforced
epoxy composite displayed higher failure strains than the reinforced nylon composite. These results
guided the matrix choice for the experimental study presented in this paper. It was concluded that the
hybrid effect was more prominent with steel on the outside and glass on the inside due to differences
in thermal expansion inducing an initial compressive stress on the glass fibers after manufacturing.
Differences in lay-up were only observed after failure of the glass fibers.
This paper presents the results of an experimental investigation on the mechanical properties,
energy dissipation capacity, and strain re-centering ability of composites comprised of glass and steel
Polymers 2017, 9, 151 3 of 16

fibers in an epoxy matrix. Within the hybrid composites, the effect of the glass-to-steel fiber volume
ratio was studied. Coupons with and without holes were tested under monolithic and half-cyclic
tensile loading to obtain stress-strain relationships, hysteresis behavior, and failure mechanisms.
The validity of the ROM using the ‘equal strain’ assumption was studied for the hybrid specimens.
Failure specimens were examined to characterize the damage progression due to the interaction
of glass and steel fibers. A bilinear hysteresis model was used to confirm the hybrid composite
cyclic behavior. The behavior after damage, energy dissipation during loading, and re-centering
capabilities of the different hybrids are of interest to ascertain the applicability of hybrid composites in
structural engineering.

2. Materials and Methods

2.1. Research Methodology

An experimental study was conducted to investigate different compositions of a hybrid composite
comprised of glass and steel reinforcing fibers. Non-hybrid composites were prepared as baseline
comparisons for each fiber type. The composites included glass, steel, and three hybrids with
different glass-to-steel fiber ratios. As overall fiber volume fractions were kept approximately constant,
glass-to-steel ratios were varied to investigate the differences in mechanical performance and failure
mechanisms of the different compositions. Table 1 summarizes the compositions that were prepared
and tested. The layer notation is as follows: G = glass fiber and S = steel fiber. As all fibers were
oriented in the 0◦ direction (unidirectional), the fiber direction is omitted from the layer notation;
this was also the direction of all loading. In the non-hybrid composites, repeated layers are shown
using subscripts. For example, [G]5 represents 5 layers of unidirectional (UD) glass fiber fabric and
[SGSGSGS] represents alternating layers of steel and glass. The hybrids (Table 1) were designed to
have anticipated glass-to-steel ratios of 70/30, [SGGGGS], 50/50, [SGSGSGS], and 30/70, [SSSGSSS].
The calculated fiber fractions are presented in Section 2.3.

Table 1. Composition of prepared composites.

Anticipated Fiber Ratio

Composite Type Layer Notation
Glass Steel
UD Glass [G]5 100 0
Hybrid 1 [SGGGGS] 70 30
Hybrid 2 [SGSGSGS] 50 50
Hybrid 3 [SSSGSSS] 30 70
UD Steel [S]8 0 100

Monolithic and half-cyclic tensile testing was conducted. The monolithic tensile testing provided
the load-displacement data to find the ultimate tensile strength, stiffness, and failure strain. Open-hole
testing is the standard mechanical testing for polymer matrix composites and was performed to observe
the effect of a stress concentration on the mechanical properties of the composites. Open-hole testing
also helped dictate failure within the gauge length. The results from the open-hole half-cyclic testing
were used to find the energy dissipated and residual strain throughout damage cycles. This was used
to characterize the strength, energy absorption, and re-centering capabilities of this hybrid material
towards exploring the potential for use as structural composites.

2.2. Raw Materials

2.2.1. Reinforcing Fibers

Figure 1 shows the reinforcement, glass and steel fibers, investigated in this study. Table 2 presents
the properties of these fibers, as provided by the manufacturer The glass reinforcement is a quasi-UD
only present to bind the fiber bundles together and do not contribute to the mechanical properties.
Advantex® is a boron-free, corrosion resistant material that is recommended for use in acidic
environments.
The steel reinforcement studied is a quasi-UD weave consisting of Grade 316 stainless steel
fibers (Figure
Polymers 1b) in the warp direction and polyethylene succinate (PES) cross yarns, provided
2017, 9, 151 4 ofby
16
NV Bekaert SA (Deerlijk, Belgium) [26]. The PES yarns serve to control the spacing between the steel
fibers and maintain the integrity of the weave; the contribution to the mechanical properties of the
woven ® E-CR glass fibers in the warp (0◦ ) direction provided by Owens
fabric isroving consisting
negligible. They of Advantex
were manufactured using a bundle drawing technique [27] and annealed at
>Corning (Toledo,toOH,
800°C (1472°F) USA)
ensure (Figure
high 1a). The fibers
strain-to-failure withoutin the weft (90◦ ) direction
compromising stiffness. are only present to
bind the fiber bundles together and do not contribute to the mechanical properties. Advantex® is a
boron-free, corrosion resistant materialTable 2.that is recommended
Properties for
of reinforcing use in acidic environments.
fibers.
The steel reinforcement studied is a quasi-UD weave consisting of Grade 316 stainless steel fibers
Reinforcing
(Figure 1b) in the warpAerialdirection
Density g/m 2 Fiber Diameter
and polyethylene μm
succinate Fiber cross
(PES) Densityyarns, Young’s
providedModulus
by NV
Fiber
Bekaert (oz/yd[26].
SA (Deerlijk, Belgium) 2 ) The PES yarns (in)serve to control theg/cm
spacing between theGPa
3
steel fibers
andUD
maintain
Glass the integrity of(9.6)
327 the weave; the contribution
10 (0.0004) to the mechanical
2.62 properties of the 82 fabric is
negligible.
UD SteelThey were manufactured
570 (16.8) using a bundle drawing technique
30 (0.0012) at >800 ◦ C
[27] and annealed 193
N.S.
(1472◦ F) to ensure high strain-to-failure without
N.S. = not compromising
specified stiffness.
by manufacturer.

90°

0°

12.7 mm 12.7 mm
0.5 in 0.5 in

(a) (b)
Figure
Figure 1.
1. Reinforcement
Reinforcement types:
types: (a)
(a) UD
UD glass
glass fibers;
fibers; (b)
(b) UD
UD steel
steel fibers.
fibers.

2.2.2. Matrix Table 2. Properties of reinforcing fibers.

The two-part thermosetting matrix system used for these studies is EPON 828, a difunctional
bisphenolReinforcing
A/epichlorohydrin Aerial Density Fiber Diameter
derived liquid epoxy resin, Fiber with Density
EPIKURE 3055, Young’s Modulus
an aliphatic amine
Fiber g/m2 (oz/yd2 ) µm (in) g/cm3 GPa
hardener, as a curing agent supplied by Hexion (Columbus, OH, USA). It was mixed with a
UD Glass 327 (9.6) 10 (0.0004) 2.62 82
manufacturer-recommended resin-to-hardener weight ratio of 2:1 to obtain optimal polymer
UD Steel 570 (16.8) 30 (0.0012) N.S. 193
cross-linking. EPON 828 has become a widely used industry resin because of its mechanical
N.S. = not specified by manufacturer.
versatility and high resistance to a broad range of chemicals [28]. EPIKURE 3055 hardener has a low
viscosity with extended pot life, which improves the workability of the matrix. This allows for a
2.2.2. Matrix
faster impregnation of the fibers. For structural applications, thermosets are preferred over
thermoplastic
The two-partresins for superior
thermosetting creepsystem
matrix resistance
used for over a wider
these studiesrange
is EPON of 828,
temperatures. For
a difunctional
comparison, monolithic tensile derived
bisphenol A/epichlorohydrin testing per American
liquid epoxy Society for Testing
resin, with EPIKURE and3055, Materials (ASTM)amine
an aliphatic D638
[29] was performed
hardener, as a curing on the epoxy
agent (no composite
supplied by Hexion fillers) to characterize
(Columbus, the stress-strain
OH, USA). properties
It was mixed with
(elastic modulus (E), ultimate tensile strength (σ ) and failure strain (ε
a manufacturer-recommended resin-to-hardener weight ratio of 2:1 to obtain optimal polymer
ult ult )) of the matrix (E = 2.56
GPa, σult = 56.9EPON
cross-linking. MPa, 828
εult =has
5.06%).
become a widely used industry resin because of its mechanical versatility
and high resistance to a broad range of chemicals [28]. EPIKURE 3055 hardener has a low viscosity with
extended pot life, which improves the workability of the matrix. This allows for a faster impregnation
of the fibers. For structural applications, thermosets are preferred over thermoplastic resins for
superior creep resistance over a wider range of temperatures. For comparison, monolithic tensile
testing per American Society for Testing and Materials (ASTM) D638 [29] was performed on the epoxy
(no composite fillers) to characterize the stress-strain properties (elastic modulus (E), ultimate tensile
strength (σult ) and failure strain (εult )) of the matrix (E = 2.56 GPa, σult = 56.9 MPa, εult = 5.06%).

2.3. Manufacturing of Composite Specimens

The composite specimens were manufactured using a hand lay-up technique. The reinforcement
was oriented on a 25.4 cm × 25.4 cm (10 in × 10 in) square steel plate in a 1.8 mm (0.07 in) thick frame
Polymers 2017, 9, 151 5 of 16

2.3. Manufacturing of Composite Specimens

Polymers 2017, 9, 151 5 of 16
The composite specimens were manufactured using a hand lay-up technique. The
reinforcement was oriented on a 25.4 cm × 25.4 cm (10 in × 10 in) square steel plate in a 1.8 mm (0.07
and saturated
in) thick frame with
andepoxy (Figure
saturated 2a). epoxy
with The composite plates
(Figure 2a). Thewere cured using
composite thewere
plates compression molding
cured using the
method. Curing was executed at 93 ◦ C (200 ◦ F) for 2 h as per the manufacturer’s recommendation
compression molding method. Curing was executed at 93 °C (200 °F) for 2 h as per the
and under a pressure
manufacturer’s of 7 bar (100and
recommendation psi) under
to allow excess resin
a pressure of 7tobar
bleed
(100 out while
psi) reaching
to allow the resin
excess desired
to
thickness. The plates (Figure 2b) were then cooled for approximately 30
bleed out while reaching the desired thickness. The plates (Figure 2b) were then cooled for min under atmospheric
pressure. Care 30
approximately wasmin
taken to minimize
under atmospheric voids throughout
pressure. Care the
wasmanufacturing
taken to minimize process.
voidsAfter curing,
throughout
the
the plates were cut process.
manufacturing into coupons
Afterwith dimensions
curing, the platesas shown
were in Figure
cut into coupons2c (254
withmm length ×as
dimensions 25.4 mm
shown
width × 1.78 mm thickness). G10 fiberglass beveled end tabs were applied using
in Figure 2c (254 mm length × 25.4 mm width × 1.78 mm thickness). G10 fiberglass beveled end tabs Loctite 4014 instant
adhesive (Henkel
were applied usingAdhesives, Westlake,
Loctite 4014 instant OH, USA).
adhesive The end
(Henkel tabs were Westlake,
Adhesives, necessary OH,
to avoid
USA).premature
The end
failure due to stress concentrations at the testing grips. For a number of each of the specimens
tabs were necessary to avoid premature failure due to stress concentrations at the testing grips. For a specified
in Section
number of2.1,
each a hole
of thewas drilled inspecified
specimens the center with a diameter
in Section 2.1, a holeequal
was to 1/6 ofinthe
drilled thecoupon width,
center with a
per ASTM specifications. Prior to testing, specimens were inspected to ensure there
diameter equal to 1/6 of the coupon width, per ASTM specifications. Prior to testing, specimens were was no damage
surrounding the hole.
inspected to ensure there was no damage surrounding the hole.

(a) (b)

(c)
Figure 2.2.Composite
Figure Composite preparation:
preparation: (a) Fiber
(a) Fiber hand hand
lay-up;lay-up; (b)composite
(b) cured cured composite plates; (c)
plates; (c) coupon coupon
dimensions
dimensions with
with end tabs. end tabs.

The layup compositions of the hybrids were selected to ensure all composites were symmetric
The layup compositions of the hybrids were selected to ensure all composites were symmetric so as
so as not to introduce any bending-extension coupling. However, the effect of fiber layup was not
not to introduce any bending-extension coupling. However, the effect of fiber layup was not explicitly
explicitly studied. The composites were designed so that the total fiber volume fractions were
studied. The composites were designed so that the total fiber volume fractions were comparable.
comparable. Table 3 presents a summary of the actual fiber volume fractions of the composites
Table 3 presents a summary of the actual fiber volume fractions of the composites manufactured in
manufactured in this study. The volume fractions (vf) were determined using Equation (1) [21],
this study. The volume fractions (vf ) were determined using Equation (1) [21], based on the composite
based on the composite thickness (t), fiber density (ρ), and fabric areal density (Aw), and number of
thickness (t), fiber density ($), and fabric areal density (Aw ), and number of fiber layers (n) using the
fiber layers (n) using the following equation:
following equation:
vf =vf(n= ×
(n A
× wA)/($
w)/(ρ×× t).
t). (1)
(1)
Polymers 2017, 9, 151 6 of 16

Table 3. Prepared composite fiber volume fractions.

Layer Notation Glass Fiber Fraction Steel Fiber Fraction Total Fiber Volume Fraction
[G]5 34.7 ± 0.1% - 34.7 ± 0.1%
[SGGGGS] 28.2 ± 0.1% 8.2% 36.4 ± 0.1%
[SGSGSGS] 20.3 ± 0.2% 15.7 ± 0.2% 36.0 ± 0.4%
[SSSGSSS] 6.5% 22.8 ± 0.1% 29.3 ± 0.2%
[S]8 - 30.9 ± 0.5% 30.9 ± 0.5%

2.4. Experimental Methodology

Composite coupons were first tested under monolithic tensile loading to obtain the ultimate
tensile strength (σult ) and failure strain (εult ) of different material compositions. The testing was
conducted on a hydraulic MTS 810 test machine (Eden Prairie, MN, USA) and performed according to
ASTM D3039 [30]. The displacement was applied at 2 mm/min (0.08 in/min), the load was recorded
using an 89 kN (20 kip) load cell, and the longitudinal strain was measured using a strain-gauge based
extensometer with a 10.16 cm (4 in) gauge length and a 350 Ω resistance. All data output was collected
via an HBM data acquisition system. All tests were conducted under standard laboratory conditions at
room temperature. Four coupons of each composite type were tested under monolithic tension and
failure inside the gauge length, away from the grips, was successfully achieved for all specimens.
To investigate energy dissipation and re-centering capabilities of this hybrid material, monolithic
(2 coupons) and half-cyclic tensile (2 coupons) tests were then performed on open-hole specimens for
each composite type. The open hole simulates realistic stress concentrations that can be introduced
in the material and may arise from bolt holes for structural connections or general accumulated
damage, for example. First, monolithic open-hole tensile (OHT) testing was performed on each coupon
according to ASTM D5766 [31] to obtain the open-hole ultimate tensile strength. Per the standard,
this is calculated using the gross cross-sectional area, disregarding the presence of the hole. The true
strength (σtrue ) was calculated adjusting for reduced area. Next, open-hole half-cyclic (OHC) tensile
testing was performed. The loading protocol began and returned to a benchmark of 445 N (100 lbs) as
the applied load was increased by 0.1 σult for each cycle at a constant load rate of 111 N/s (25 lb/s)
until failure.

3. Results and Discussion

All experimental results are presented in Table 4. For monolithic testing, average values are
reported with standard deviation. For open-hole tension and cyclic testing, average values are reported.

Table 4. Ultimate tensile strength, failure strain, and true strength results of no-hole and open-hole
tensile tests.

No-Hole Tension Open-Hole Tension Open-Hole Cyclic

Composite σ ult , MPa σ ult , MPa σ true 1 , σ true / σult , MPa
εult , % εult , % εult , %
(ksi) (ksi) MPa (ksi) σ ult,no-hole (ksi)
667 ± 18.4 474 569 455
[G]5 2.50 ± 0.12% 2.46% 0.852 2.55%
(96.7 ± 2.67) (68.7) (82.5) (66.0)
642 ± 35.5 530 598 521
[SGGGGS] 2.90 ± 0.30% 3.68% 0.932 3.41%
(93.1 ± 5.15) (76.9) (86.7) (75.6)
469 ± 29.7 333 422 348
[SGSGSGS] 2.71 ± 0.13% 2.25% 0.900 2.83%
(68.0 ± 4.31) (48.3) (61.2) (50.5)
276 ± 11.2 218 260 225
[SSSGSSS] 2.71 ± 0.12% 2.25% 0.941 2.34%
(40.0 ± 1.62) (31.6) (37.7) (32.6)
208 ± 6.67 161 194 169
[S]8 12.0 ± 0.01% 4.50% 0.932 3.32%
(30.2 ± 0.97) (23.4) (28.1) (24.5)
1 True strength was calculated adjusting for the reduced cross-sectional area due to the hole.
Polymers 2017, 9, 151 7 of 16

208 ± 6.67 161 194 169

[S]8 12.0 ± 0.01% 4.50% 0.932 3.32%
(30.2 ± 0.97) (23.4) (28.1) (24.5)
Polymers 2017, 9, 151 7 of 16
1 True strength was calculated adjusting for the reduced cross-sectional area due to the hole.

3.1. No-hole
No-hole Composites
Composites

Tensile Properties
3.1.1. Tensile
stress-strain behavior
The stress-strain behaviorof ofall
allcomposites
compositesisispresented
presentedininFigure
Figure 3. 3.
Following
Following thethe
elastic limit,
elastic all
limit,
composites
all continue
composites carrying
continue increasing
carrying stresses stresses
increasing until the until
ultimate
thestress is reached.
ultimate stressThe unidirectional
is reached. The
(UD) glass composite,
unidirectional (UD) glass[G]composite,
5 , displayed [G]the largest ultimate
5, displayed strength
the largest at 667
ultimate MPa (96.7
strength at 667ksi)
MPaand(96.7
the
lowest
ksi) andfailure strainfailure
the lowest of 2.50%. TheofUD
strain steelThe
2.50%. fiberUDcomposite, [S]composite,
steel fiber 8 , had a significantly
[S]8, had a larger failure
significantly
strain of
larger 12.0%,strain
failure but the
of lowest
12.0%, ultimate
but the strength. The tensile
lowest ultimate tests showed
strength. a clear
The tensile nonlinear
tests showedresponse
a clear
for the hybrid
nonlinear composites.
response for the hybridThe composites.
[SGGGGS] hybrid had a strength
The [SGGGGS] hybrid had of 642 MPa (93.1
a strength ksi).
of 642 MPa As(93.1
the
amount
ksi). of glass
As the amountdecreased,
of glass the ultimatethe
decreased, strength
ultimate also decreased.
strength The [SGSGSGS]
also decreased. and [SSSGSSS]
The [SGSGSGS] and
specimens had
[SSSGSSS] ultimate had
specimens strengths of 469strengths
ultimate MPa (68.0ofksi)469andMPa
276 MPa
(68.0(40.0
ksi)ksi),
andrespectively.
276 MPa The (40.0hybrid
ksi),
failure strainsThe
respectively. were 2.90%,
hybrid 2.71%,
failure and 2.71%,
strains in order
were 2.90%, of increasing
2.71%, and 2.71%, steelinfiber
orderpercentage,
of increasing andsteel
did
not show
fiber a significant
percentage, difference
and did not show between the hybrid
a significant compositions.
difference betweenThese resultscompositions.
the hybrid suggest that adding
These
steel increased
results the failure
suggest that addingstrain of glass/epoxy
steel increased composites.
the failure strain ofThe failure patterns
glass/epoxy are discussed
composites. The failurein
Section 3.1.2
patterns to obtain aingreater
are discussed Sectionunderstanding
3.1.2 to obtain of the stress-strain
a greater behavior.
understanding of the stress-strain behavior.

Figure
Figure 3.
3. Monolithic
Monolithic tensile
tensile stress-strain
stress-strain relationships
relationships of specimens with no holes.

Figures
Figures 4–6
4–6 depict
depict theoretical
theoretical tensile
tensile stress-strain
stress-strain relationships
relationships of of curves
curves calculated
calculated fromfrom theory
theory
using the rule of mixtures (ROM), which sums the stresses of individual
using the rule of mixtures (ROM), which sums the stresses of individual constituents based on constituents based
volume on
volume
fraction offraction of the
the hybrid hybrid composites.
composites. The figure Thealso figure
presents also
thepresents the actual experimental
actual experimental curve. The glass curve.
and
The glass and steel curves (Figures 4–6) were obtained by subtracting the contribution
steel curves (Figures 4–6) were obtained by subtracting the contribution of the epoxy matrix, thus of the epoxy
the
matrix, thus the
epoxy matrix epoxy as
is shown matrix
its ownis shown as its own
entity. These figuresentity.
showThese figures
that the ROMshow that the
was valid ROM
in the wasregion,
elastic valid
in
as the
seenelastic region,
by the as seen
prediction ofbythethe prediction
stiffness in theof linear
the stiffness
regionin the linear
(Figures 4–6region (Figures
insets). Slight 4–6 insets).
differences
Slight differences between curves can be attributed to limitations in the
between curves can be attributed to limitations in the accuracy of the volume fraction measurement accuracy of the volume
fraction
or minormeasurement
defects, suchoras minor
fiberdefects, such as fiber
misorientation and misorientation
resin voids. Both andCallens
resin voids. BothThysen
[6] and Callens[25]
[6]
and Thysen [25] discuss a lower experimental stiffness than theoretically predicted
discuss a lower experimental stiffness than theoretically predicted due to fiber misorientation in the due to fiber
misorientation
manufacturingin the manufacturing
process. After the steel process.
beginsAfter the at
to yield steel begins to yield
approximately at approximately
0.2%, the rule of mixtures0.2%,
the rule over-predicts
slightly of mixtures slightly over-predicts
the stress-strain the stress-strain
relationship. relationship.
The experimental Theis experimental
curve curve is
consistently 92%–95%
consistently 92%–95% of the theoretical curve in the plastic region. Additionally,
of the theoretical curve in the plastic region. Additionally, the failure strain of the experimental the failure strain of
the experimental hybrid is consistently greater than the theoretical (as calculated
hybrid is consistently greater than the theoretical (as calculated from the ROM) value for all hybrids. from the ROM)
value for all hybrids.
This confirmed Thiswas
that there confirmed that there
a synergistic effectwason a synergistic
failure strain effect
due toon failure
fiber strain due to fiber
hybridization.
hybridization.
Polymers 2017, 9, 151 8 of 16
Polymers 2017, 9, 151 8 of 16
Polymers 2017, 9, 151 8 of 16
Polymers 2017, 9, 151 8 of 16

Figure4. 4. [SGGGGS]hybrid
[SGGGGS] hybrid composite curve.
curve. (Inset) The
Thelinear
linearregion
regionexpanded from thethe
encircled
Figure
Figure 4. [SGGGGS] hybridcomposite
composite curve. (Inset)
(Inset) The linear region expanded
expanded from
from encircled
the encircled
region.
Figure
region. 4. [SGGGGS] hybrid composite curve. (Inset) The linear region expanded from the encircled
region.
region.

Figure 5. [SGSGSGS] hybrid composite curve. (Inset) The linear region expanded from the encircled
Figure5. 5. [SGSGSGS]hybrid
[SGSGSGS] hybrid composite curve.
curve. (Inset) The
Thelinear
linearregion
regionexpanded from thethe
encircled
Figure
region.
Figure 5. [SGSGSGS] hybridcomposite
composite curve. (Inset)
(Inset) The linear region expanded
expanded from
from encircled
the encircled
region.
region.
region.

Figure 6. [SSSGSSS] hybrid composite curve. (Inset) The linear region expanded from the encircled
Figure 6. [SSSGSSS] hybrid composite curve. (Inset) The linear region expanded from the encircled
region.
Figure
Figure 6. 6. [SSSGSSS]hybrid
[SSSGSSS] hybridcomposite
composite curve.
curve. (Inset)
(Inset) The
The linear
linearregion
regionexpanded
expandedfrom the
from encircled
the encircled
region.
region.
region.
Polymers
Polymers 2017,9,
Polymers2017,
2017, 9,9,151
151
151 999of
of16
of 16
16

3.1.2.Failure
3.1.2. FailureMechanisms
Mechanisms
3.1.2. Failure Mechanisms
The failure
The failure of of specimens
specimens was was visualized
visualized using using aa digital
digital camera
camera (Olympus
(Olympus Stylus Stylus Tough
Tough TG-4 TG-4
The
16.0 MP failure
MP Compact
Compact fromof specimens
from Olympus was
Olympus America visualized
America (Center using
(Center Valley, a digital
Valley, PA, camera
PA, USA))
USA)) to (Olympus
to characterizeStylus
characterize the Tough
the response TG-4
response of of
16.0
16.0
the MP Compactspecimens
composite from Olympus to theAmerica
monolithic(Center Valley, PA, USA))
stress-strain testing. toFor
characterize
each the response
composite of the
tested, all
the composite specimens to the monolithic stress-strain testing. For each composite tested, all
composite specimens
samples displayed
displayed to theand
similar monolithic
reproduciblestress-strain testing.
patterns.For
failure patterns. The each composite
general failuretested,
mechanism all samples
of the
the
samples similar and reproducible failure The general failure mechanism of
displayed
UD glass similar
composites, and reproducible
[G] , can be failure
described patterns.
as a The
“brooming” general failure
effect as mechanism
described by of the
Harik UD
et al.glass
[32],
UD glass composites, [G]5, can be described as a “brooming” effect as described by Harik et al. [32],
5
composites,
and shown
shown in [G] 5 , can be
in Figure
Figure 7a.described as
This isis aa resulta “brooming”
result of scattered
scattered effect
fiberasbreakage
describedand
breakage by debonding
Harik et al. from [32], and shown
the matrix.
matrix.
and 7a. This of fiber and debonding from the
in
AsFigure
shown 7a.inThis
the is a result of scattered
stress-strain plots (Figurefiber 3),
(Figure breakage and debonding
the failure
failure was sudden
sudden fromandthe matrix.
brittle. ThisAsrandom
shown
As shown in the stress-strain plots 3), the was and brittle. This random
in the stress-strain
brooming effect plots
effect confirms (Figure
confirms the 3), the
the stochastic failure
stochastic failure was
failure and sudden
and high and
high energy brittle.
energy release This
release of random
of the brooming
the fibers
fibers at effect
at multiple
multiple
brooming
confirms the
locations. The stochastic
The UD
UD steel failure
steel fiber and
fiber composites, high
composites, [S] energy release
[S]88,, exhibited of
exhibited failure the fibers
failure normal
normal toat multiple
to the locations.
the direction
direction of The
of loading,
loading, UD as
locations. as
steel fiber
shown in composites,
in Figure
Figure 7e. [S]
7e. The 8 , exhibited
The fracture failure
fracture cross-section normal
cross-section was to
was not the direction
not in-plane, of
in-plane, indicatingloading,
indicating there as shown
there was in
was fiber Figure
fiber pullout
pullout7e.
shown
The
and fracture
necking, cross-section
also observed was
by not in-plane,
Callens et al. indicating
[6]. At the there was
presence of fiber
a pullout
matrix crack, andthe necking,
fibers also
start to
and necking, also observed by Callens et al. [6]. At the presence of a matrix crack, the fibers start to
observed
yield and by Callens
and therefore et
therefore produce al. [6].
produce aa more At the presence
more localized of a
localized failure, matrix crack,
failure, visible
visible in the
in the fibers start
the microscope to yield
microscope image and
image of therefore
of Figure
Figure 88
yield
produce
(Carson adigital
more localized
handheld failure,
microscope visiblefrom
in the microscope
Carson Optical image
Inc. of Figure 8 (Carson
(Ronkonkoma, NY, digital
USA)). handheld
The steel
(Carson digital handheld microscope from Carson Optical Inc. (Ronkonkoma, NY, USA)). The steel
microscope
fibers from Carson
maintained the Optical
integrity of Inc.
the (Ronkonkoma,
composite after NY,matrix
the USA)). The started
cracks steel fibers maintained
forming. This is the
also
fibers maintained the integrity of the composite after the matrix cracks started forming. This is also
integrity
evident as of the
as composite
compositeafter
the composite the matrix
continues cracks
to yield
yield startedthe
beyond forming. This
failure strain
strain is of
also
theevident
epoxy.asThisthe composite
This composite
evident the continues to beyond the failure of the epoxy. composite
continues
showed to
high yield beyond
ductility, the
and failure
the strain
failure of
strainthe epoxy.
of 12.0% This composite
was more showed
than twice high
that ductility,
of pure and the
epoxy
showed high ductility, and the failure strain of 12.0% was more than twice that of pure epoxy
failure
(5.06%). strain of 12.0% was more than twice that of pure epoxy (5.06%).
(5.06%).

(a)
(a) (b)
(b) (c)
(c) (d)
(d) (e)
(e)
Figure7.
Figure 7.Images
7. Imagesof
Images offailure
of failurespecimens:
failure specimens:(a)
specimens: (a)[G]
(a) [G]55;;;(b)
[G] (b)[SGGGGS];
[SGGGGS];(c)
(c)[SGSGSGS];
[SGSGSGS];(d)
(d)[SSSGSSS];
[SSSGSSS];(e)
(e)[S]
[S]88...
Figure 5 (b) [SGGGGS]; (c) [SGSGSGS]; (d) [SSSGSSS]; (e) [S]8

Figure8.
Figure
Figure 8. UD
8. UD steel
UD steel composite
steel composite crack
composite crack bridging
crack bridging and
bridging andnecking
and neckingat
necking atmatrix
at matrixcrack.
matrix crack. Image
crack. Image taken
Image taken by
taken by Carson
by Carson
Carson
digital
digital handheld
handheld microscope.
microscope.
digital handheld microscope.

The three
The three different
different hybrid
hybrid composite
composite failure
failure patterns
patterns were
were observed
observed (Figure
(Figure 7b–d),
7b–d), and
and these
these
patterns lay
patterns lay within
within aaaspectrum
spectrumdepending
spectrum depending on
depending onthe
on therespective
the respectivefiber
respective fiberfractions.
fiber fractions. Hybrids
fractions. Hybrids withwith higher
higher
percentages
percentages of
percentages of glass
of glass exhibited
glass exhibited similar
exhibited similar failures
similar failures
failures toto [G]
to [G] , where
[G]55,, where
5 failure
where failure was
failure was distributed
was distributed along
distributed along the
along the length.
the length.
length.
Assumingthe
Assuming
Assuming theglass
the glassfibers
glass fibersfail
fibers failfirst
fail firstin
first inaaarandom
in randompattern,
random pattern,the
pattern, thestress
the stresswas
stress wastransferred
was transferredto
transferred tothe
to theneighboring
the neighboring
neighboring
steelfibers
steel fibersat
atmany
manydifferent
differentlocations
locationsalongalongthe
thegauge
gaugelength.
length.ItItisisclear
clearthat
thatthe
thefailure
failureisisdistributed
distributed
more axially. This resulted in larger failure strains than the [G]5 specimen (Figure 3), and aaspread
more axially. This resulted in larger failure strains than the [G] 5 specimen (Figure 3), and spreadofof
Polymers 2017, 9, 151 10 of 16

Polymers 2017, 9, 151 10 of 16

steel fibers at many different locations along the gauge length. It is clear that the failure is distributed
more axially. This resulted in larger failure strains than the [G]5 specimen (Figure 3), and a spread
plasticity along the length of the specimen. In contrast, hybrids with higher percentages of steel
of plasticity along the length of the specimen. In contrast, hybrids with higher percentages of steel
resembled the localized [S]8 failure more closely. Although the failure strain of the composites was
resembled the localized [S]8 failure more closely. Although the failure strain of the composites was not
not nearly as high as the [S]8 composite (Figure 3), these results suggest that the presence of steel
nearly as high as the [S]8 composite (Figure 3), these results suggest that the presence of steel yielding
yielding can afford warning to potential structural failure. An identical failure spectrum was
can afford warning to potential structural failure. An identical failure spectrum was observed in both
observed in both the open-hole tensile and open-hole cyclic specimens.
the open-hole tensile and open-hole cyclic specimens.
3.2. Open-Hole
3.2. Open-Hole Composites
Composites

3.2.1. Tensile
3.2.1. Properties
Tensile Properties
Figure 99presents
Figure presents a comparison
a comparison of theofmonolithic
the monolithic tensile stress-strain
tensile stress-strain relationships relationships
of composites of
composites with and without a hole. The ultimate nominal tensile strength (σ ),
with and without a hole. The ultimate nominal tensile strength (σult ), failure strain (εult ), and true
ult failure strain (ε ult ),
and true
tensile tensile(σstrength
strength (σtrue) are reported for each specimen in Table 4. The stiffness of the
true ) are reported for each specimen in Table 4. The stiffness of the open-hole (OHT)
open-hole (OHT) specimens
specimens remains unchanged remains unchanged
compared compared
to specimens to specimens
without without
a hole, but there isa a hole, but there
decrease in the is
a decrease
ultimate in the
tensile ultimate
strength tensile
of the OHT strength
specimensofdue thetoOHT specimens
the reduction duespecimen’s
in the to the reduction in the
cross-sectional
specimen’s cross-sectional area. The true stress of the OHT specimen was
area. The true stress of the OHT specimen was calculated using the reduced area of the specimen, and calculated using the
reduced area of the specimen, and the ratio of the true stress to the ultimate
the ratio of the true stress to the ultimate tensile stress of the specimen without a hole, σ tensile stress
σtrueof the
, is
𝜎true ult,no hole
specimen in
presented without
Table 4.a hole,
Though the OHT , isspecimens
presentedwere
in Table
fewer4. in
Though
number the OHT
and specimens
hence weretesting
no statistical fewer
𝜎ult,no hole
was possible
in number forhence
and these specimens, qualitatively
no statistical testing wasitpossible
can be noted that the
for these presencequalitatively
specimens, of steel decreased
it canthe
be
susceptibility of the composites to stress concentrations, as indicated by ratios closer
noted that the presence of steel decreased the susceptibility of the composites to stress to unity. Hybrid
(steel with glass)
concentrations, asepoxy composites
indicated by ratiosare thustobetter
closer unity.able to retain
Hybrid (steelmore of the epoxy
with glass) composite’s ultimate
composites are
strength even
thus better ablewith a stress
to retain concentration
more present ultimate
of the composite’s in contrast to the even
strength morewith
conventional glass/epoxy
a stress concentration
FRP composite.
present in contrast to the more conventional glass/epoxy FRP composite.

Figure 9.
Figure 9. Tensile
Tensilestress-strain
stress-strainrelationships
relationshipsofofspecimens
specimens without
without holes
holes and
and with
with holes
holes (denoted
(denoted by
by the
the label “OHT”).
label “OHT”).

Figure 10 presents open-hole specimen stress-strain relationships. The open-hole tensile curves
Figure 10 presents open-hole specimen stress-strain relationships. The open-hole tensile curves
(OHT) and the backbone of the open-hole half-cyclic curves (OHC) are shown for all compositions.
(OHT) and the backbone of the open-hole half-cyclic curves (OHC) are shown for all compositions.
The ultimate strength and failure strains are presented in Table 4. The OHT tests were
The ultimate strength and failure strains are presented in Table 4. The OHT tests were
displacement-controlled, and the OHC tests were load-controlled. No significant differences were
displacement-controlled, and the OHC tests were load-controlled. No significant differences were
observed between the stress-strain behavior during tensile and half-cyclic loading. This suggests
observed between the stress-strain behavior during tensile and half-cyclic loading. This suggests that
that the accumulated damage from loading/unloading below the ultimate stress does not have a
the accumulated damage from loading/unloading below the ultimate stress does not have a significant
significant effect on the composite’s mechanical properties.
effect on the composite’s mechanical properties.
Polymers 2017, 9, 151 11 of 16
Polymers 2017, 9, 151 11 of 16

Figure 10. Stress-strain relationships of specimens with open-holes during monolithic tensile
(denoted OHT) and half-cyclic loading (denoted OHC).

3.2.2. Cyclic Properties

Figure 11 shows a sample of stress-strain behavior after half-cyclic tensile loading of an
open-hole specimen, [SSSGSSS]. In this testing, the loading protocol began and returned to a
benchmark of 445 N (100 lbs) as the applied load was increased by 0.1 σult for each cycle at a constant
load rate of 111 N/s (25 lb/s) until failure. For example, the maximum stress and strain of the seventh
cycle, where the sample was loaded to a stress equal to 0.7 σult, is indicated on the figure to
Figure 10.
demonstrate how Stress-strain
energy relationships
dissipation offound
specimensthe with ofopen-holes during monolithic
areatensile
Figure 10. Stress-strain relationships ofwas
specimens atwith end
open-holes eachduring
loading cycle.
monolithic Thetensile under the
(denoted
curve (denoted
is shaded OHT) to and half-cyclic
signify the loading
amount (denoted
of energy OHC).
absorbed in each loading cycle. Residual strain
OHT) and half-cyclic loading (denoted OHC).
caused from plastic deformation was assumed to be the strain at the end of each unloading cycle.
3.2.2.Figure
Cyclic 12aProperties
plots the amount of energy dissipated during open-hole half-cyclic loading of the
3.2.2. Cyclic Properties
composites.
Figure The x-axis represents
11 shows a sample the maximum strain
of stress-strain reached
behavior at each
after cycle prior
half-cyclic to unloading.
tensile loading ofThe an
y-axis
Figure
open-hole represents
11 shows
specimen, athe energy
sample dissipated
of stress-strain
[SSSGSSS]. during
In this testing, behavior each
the afterloading
loading cycle tensile
half-cyclic
protocol found by
andintegrating
began loading tothea
of an open-hole
returned
stress-strain
benchmark
specimen, ofcurve.
[SSSGSSS]. 445 N InThe
thispercentage
(100 lbs) as thethe
testing, adjacent
applied
loading toprotocol
load each curve
was increased
beganrepresents
by
and the
0.1returned
σult fortotal
eachtovolume
acycle atfraction ofof445 N
a constant
benchmark
steel
loadas within
rate the composite. For example, for a perfectly elastic material with no permanent
(100 lbs) theofapplied
111 N/s (25 loadlb/s)
was until failure. For
increased example,
by 0.1 theeach
σult for maximum
cycle at stress and strain
a constant loadof the
rateseventh
of 111 N/s
deformation,
cycle, where the
the energy
sample dissipated
was loaded willtobeanear zero
stress priortoto0.7
equal failure,
σ as indicated
ult, is the material on unloads
the along
figure to the
(25 lb/s) until failure. For example, the maximum stress and strain of the seventh cycle, where
the same linear path. This was observed in specimen [G] . In contrast,
demonstrate how energy dissipation was found at the end of each loading cycle. The area under the
5 plastic materials will dissipate
sample was loaded
energy
to a stress equal to 0.7 σ a, structural
loading.ofInult
is indicated on the figure to demonstrate how toenergy
curve isdue to yielding
shaded to signifyduring
the amount energy absorbed application, this could
in each loading potentially
cycle. Residualleadstrain
dissipation
warning was found
of damage at the end of
prior to failure, each loading cycle. The area under the curve is shaded to signify
caused from plastic deformation waswhichassumed is ideal
to be in
thestructural
strain at the design.
end of Higheachfailure strains
unloading of UD
cycle.
the amount of energy
steel Figure
composites12a plotsabsorbed
lead Callens
the ineteach
amount al.of[7]loading
to study
energy cycle.
energy
dissipated Residual
during strain
dissipation of caused
open-hole UD steel from plastic
composites
half-cyclic loading deformation
through
of the
was assumed
impact to be
testing. the strain at the end of each unloading cycle.
composites. The x-axis represents the maximum strain reached at each cycle prior to unloading. The
y-axis represents the energy dissipated during each loading cycle found by integrating the
stress-strain curve. The percentage adjacent to each curve represents the total volume fraction of
steel within the composite. For example, for a perfectly elastic material with no permanent
deformation, the energy dissipated will be near zero prior to failure, as the material unloads along
the same linear path. This was observed in specimen [G]5. In contrast, plastic materials will dissipate
energy due to yielding during loading. In a structural application, this could potentially lead to
warning of damage prior to failure, which is ideal in structural design. High failure strains of UD
steel composites lead Callens et al. [7] to study energy dissipation of UD steel composites through
impact testing.

Figure 11. Area under the curve of half-cyclic loading for [SSSGSSS]. The seventh loading cycle is
Figure 11. Area under the curve of half-cyclic loading for [SSSGSSS]. The seventh loading cycle is
demonstrated by an arrow.
demonstrated by an arrow.

Figure 12a plots the amount of energy dissipated during open-hole half-cyclic loading of the
composites. The x-axis represents the maximum strain reached at each cycle prior to unloading.
The y-axis represents the energy dissipated during each loading cycle found by integrating the
stress-strain curve. The percentage adjacent to each curve represents the total volume fraction of steel
within the composite. For example, for a perfectly elastic material with no permanent deformation,
the energyFigure 11. Areawill
dissipated under
bethe curve
near of half-cyclic
zero loading for
prior to failure, [SSSGSSS].
as the Theunloads
material seventh loading
along cycle is
the same linear
demonstrated by an arrow.
path. This was observed in specimen [G]5 . In contrast, plastic materials will dissipate energy due to
yielding during loading. In a structural application, this could potentially lead to warning of damage
prior to failure, which is ideal in structural design. High failure strains of UD steel composites lead
Callens et al. [7] to study energy dissipation of UD steel composites through impact testing.
A composite comprised of high strength elastic glass fibers and ductile steel fibers was expected
to have a large area under the stress-strain curve, due to a high amount of energy dissipated.
Polymers 2017, 9, 151 12 of 16

The [SGGGGS] hybrid, consisting of only 8.2% steel, was found to dissipate the most energy. Since
the strength of the glass is significantly higher than the steel fibers, this hybrid does not need a
Polymers 2017, 9, 151
large amount of steel to achieve high energy dissipation. To facilitate comparison among12composite
of 16

compositions with differing amounts of glass, the normalized energy value per 1% of steel present
in the composite is plotted versus maximum strain in Figure 12b. The results of Figure 12a,b are
consistent:
Polymershybrids containing less steel were able to dissipate more energy.
2017, 9, 151 12 of 16

(a) (b)
Figure 12. Energy dissipation of composites during open-hole half-cyclic loading: (a) calculated
energy values; (b) normalized energy values per 1% of steel. (Note) Percentage along curves
represents the steel fiber fraction in the composite. Lines connecting points used to display trend.
(a) (b)
A composite
Figure 12. Energycomprised
dissipationof high strength elastic
of composites during glass fibers and
open-hole ductileloading:
half-cyclic steel fibers was expected
(a) calculated
Figure
to have 12.aEnergy dissipation of composites during open-hole half-cyclic loading: (a) calculated energy
energy values; (b) normalized energy values per 1% of steel. (Note) Percentage along curves The
large area under the stress-strain curve, due to a high amount of energy dissipated.
values;
[SGGGGS](b) normalized
hybrid, energy
consisting values
of only per
8.2% 1% of
steel,steel.
was (Note)
found Percentage
to dissipate
represents the steel fiber fraction in the composite. Lines connecting points used to display trend. along
the curves
most represents
energy. Since the
the
steel fiber fraction
strength of the glass in theis composite.
significantlyLines higher connecting
than the points used to
steel fibers, thisdisplay
hybridtrend.does not need a large
amount of steel to
A composite achieve of
comprised high
highenergy
strength dissipation.
elastic glass Tofibers
facilitate comparison
and ductile among
steel fibers wascomposite
expected
compositions
to have a large with
areadiffering
under amounts
the of
stress-strainglass, the
curve, normalized
due to
The main goal of the nonmetal-metal hybrid composite system is to improve ductility and a energy
high value
amount ofper 1%
energy of steel present
dissipated. Theinenergy
the composite
[SGGGGS] is plotted
hybrid, consisting versus maximum
of only 8.2% steel,strainwas infound
Figureto12b. The results
dissipate the most of energy.
Figure Since
12a,b theare
absorption prior to failure. However, ductility causes permanent deformation and residual strain
consistent:
strength of hybrids
the glass containing less steel
is significantly werethan
higher abletheto dissipate
steel fibers, more energy.
this hybrid does not need a large
over the material lifecycle. A material that has re-centering capabilities after loading and unloading is
amountThe ofmainsteelgoal of the nonmetal-metal
to achieve high energy dissipation.hybrid composite
To facilitate system is to improve
comparison among ductility
compositeand
appealing
energyforabsorption
compositionsstructural applications
prior
with differing to failure.
amounts for ofboth
glass,strength
However, ductility
the and
causes
normalized serviceability
permanent
energy limit
valuedeformation
per 1% states.
of steel Figure
and 13inshows
residual
present
the residual
strain
the strain
over
composite thebehavior
plottedafter
ismaterial versus open-hole
lifecycle. A material
maximum half-cyclic
that in
strain loading.
hasFigure 12b.The
re-centering The x-axis
resultsrepresents
capabilities of after
Figure the maximum
loading
12a,b and
are
strain consistent:
reached
unloadingathybrids
iseach cycle
appealing prior to
lessunloading.
for structural
containing steelapplications
were able Thetoy-axis
for both represents
dissipate strength the serviceability
and
more energy. residual strain ratio,
limit which is
states.
Figure
calculated as 13
The the shows
main goalthe
residual residual
ofstrain
the strain
endbehavior
nonmetal-metal
at the of ahybrid after
loading open-hole
composite
cycle system
divided half-cyclic
is
bytothe loading.ductility
improve
cycle’s The x-axis
maximum and strain.
energy
The closer theabsorption
represents the is
ratio toprior
maximumzero,tothe failure.
strain However,
reached
better at each ductility
the re-centering causes
cycle capability
prior topermanent
unloading. deformation
The y-axis
of the material. and
Forrepresentsresidual
a perfectly theelastic
strain
residual over
straintheratio,
material
which lifecycle. A material
is calculated that hasstrain
as the residual re-centering
at the end capabilities
of a loading after
cycleloading
dividedand by
material with no plastic deformation, the residual strain ratio will be near zero. This behavior was
the cycle’s maximum
unloading is appealing strain. The closerapplications
for structural the ratio is tofor zero,
both thestrength
better the re-centering
and serviceability capability of the
limit states.
observed in the [G]5 specimen. In contrast, an inelastic material has a ratio closer to one, as observed
material.
Figure 13For a perfectly
shows elastic material
the residual with no plastic
strain behavior after deformation, the residual
open-hole half-cyclic strain ratio
loading. The will be
x-axis
in the represents
[S]8 zero.
near specimen.
This
the maximumOf the was
behavior three
strain hybridinat
observed
reached composites,
the
each [G]cycle the composite
5 specimen.
prior contrast,with
toInunloading. an
The the
inelastic
y-axislowest
materialpercentage
represents has
thea of
steel, [SGGGGS],
ratio closer
residual strain had
to one, the lowest
ratio,aswhich
observed residual
in the [S]
is calculated strain ratio,
specimen.
as8 the plateauing
residualOfstrainthe three around
at thehybrid 0.12.
end of composites, This
a loading cycle signifies
thedivided very
composite by good
re-centering
the capabilities.
withcycle’s
the lowest
maximum The
percentageother
strain. ofhybrids
The steel, alsoratio
closer[SGGGGS],
the hadismore
tohad favorable
zero,thethe lowest
betterre-centering
residual capabilities
strain
the re-centering ratio, with
plateauing
capability relatively
of the
around
low ratios that0.12.
material. For aThis
plateaued signifies
perfectly elastic
after very
reaching good
material re-centering
with no plastic
a threshold capabilities.
value deformation, Thethe
for the strain other
residual
ratio.hybrids strain
A materialalso had
ratiothat more
will be a low
has
favorable
near
and consistent re-centering
zero. residual
This behavior
straincapabilities
was observed
ratio along with
in relatively
the
loading [G]5cycles lowindicates
specimen. ratios that
In contrast, plateaued
greater an stability.
inelasticafter reaching
material
Structural has adesign
threshold
ratio
relies on closervalue
parameters, to one, forasthe
such asstrain
observed ratio.
in the
stability, A[S]
that material
8 specimen.
characterizethat Ofhasthe a material
the low
threeand consistent
hybridstrength residual
composites,
and thestrain
dictate theratio
compositeresponse
alongthe
with loading
lowestcycles indicates
percentage greater
of steel, stability. Structural
[SGGGGS], had the lowest designresidual
relies on parameters,
strain such as
ratio, plateauing
to loading.
stability,0.12.
around that This
characterize
signifiesthe material
very good strength
re-centering and dictate the response
capabilities. The other to loading.
hybrids also had more
favorable re-centering capabilities with relatively low ratios that plateaued after reaching a
threshold value for the strain ratio. A material that has a low and consistent residual strain ratio
along loading cycles indicates greater stability. Structural design relies on parameters, such as
stability, that characterize the material strength and dictate the response to loading.

FigureFigure 13. Residual strain ratio of composites during open-hole half-cyclic loading. Note: Percentage
13. Residual strain ratio of composites during open-hole half-cyclic loading. Note: Percentage
along curves represents the steel fiber fraction in the composite. The lines connecting points are used
along curves represents the steel fiber fraction in the composite. The lines connecting points are used
to display trends.
to display trends.
Figure 13. Residual strain ratio of composites during open-hole half-cyclic loading. Note: Percentage
along curves represents the steel fiber fraction in the composite. The lines connecting points are used
to display trends.
unloading of each cycle. The x-value of point 3 represents the maximum strain attained at that
respective cycle. The slopes of line 3–4 and line 5–6 are equal to that of line 1–2 and the vertical
distance is equivalent to 2 σy. Point 7 represents a targeted design strain. The 7–8–9–10 parallelogram
signifies the outermost boundary of the ideal bilinear hysteresis behavior.
This
Polymers ideal
2017, hysteresis model was applied to the hybrid composites, [SGGGGS], [SGSGSGS],
9, 151 13and
of 16
[SSSGSSS], along with the steel composite, [S]8 (Figure 15). In this work, only half-cyclic testing was
performed due to experimental limitations that did not permit the acquisition of compression data.
3.2.3. Theoretical Hysteresis Model
However, the stress-strain curve does appear to follow kinematic hardening behavior. Using this
assumption, the model
A theoretical completewas usedhysteresis
to furtherbehavior,
analyze theincluding
hybrid compositestress-strain relationships
cyclic behavior. in the
The combination
compressive
of an elastic andstress region,
plastic was predicted
material and shown
results in unique in Figure
hysteresis 15. The
behavior. redhalf-cyclic
Under portion oftesting,
the bilinear
it was
model followed each experimental tensile loading and unloading cycle.
observed that the hybrid composites resemble the behavior of lead-rubber bearings (LRBs), commonly The black portion of the
curve
used as serves
base as the predicted
isolators extension
in structural seismic forapplications
full-cycle behavior
[33]. Theby extending
novelty in thisthe curvelies
system intoin the
the
compression
re-centering region.
force. ThisThishybrid
bilinearcomposite
model predicted
behaviorthe mayhysteresis behavior
be predicted usingof all
thehybrid
bilinear composites
hysteresis
by finding
model the elastic
shown in Figure and14.hardening
Point 1 slopes,
begins at the(0,
yield stress,
0) and theand the residual
x-value of Pointstrain. Composites
2 is located at the with
yield
astrain
higheroffraction
the steeloffibers,
steel were found
εy . The slope to of
follow the prediction
line 1–2, more
or the elastic closely.
slope, The calculated
represents the moduluselasticofand
the
hardening slopes align
hybrid composite in thewith theregion.
elastic paths The
of the experimental
slope of line 2–3,hysteresis behavior.
or the hardening Hybrid
slope, composite
is equivalent to
cycles that lieofwithin
the modulus the glasstheandideal parallelogram
epoxy constituents, suggest
as wellmaterial stability. Once
as the post-yielding theofstress
slope reaches
the steel. The
greater than 2slope
post-yielding σy, the
wassteel fibers may
determined yield in compression.
by bilinearization of the steelThis is not desirable
stress-strain relationship forusing
structural
equal
materials
areas. Thisbecause
hardening subsequent
slope is whatloading cycles
dictates the will also
return pathyield,
at theand the material
unloading of eachwill become
cycle. more
The x-value
unstable
of point 3over time. the
represents Themaximum
area under each
strain idealized
attained loading
at that respectivecyclecycle.
was The
found, and
slopes the 3–4
of line energy
and
dissipation
line 5–6 arefollowed
equal to thatthe same
of linetrend as the
1–2 and theexperimental values.
vertical distance As this model
is equivalent to 2 is
σyan idealized
. Point shape,
7 represents
ita targeted
can be seendesign how it canThe
strain. under-
7–8–9–10or over-predict
parallelogram thesignifies
actual energy dissipation
the outermost of the of
boundary composite
the ideal
depending on the curve
bilinear hysteresis fit.
behavior.

Figure14.
Figure Lead-rubberbearing
14.Lead-rubber bearingideal
idealbilinear
bilinearhysteresis
hysteresisbehavior.
behavior.

This ideal hysteresis model was applied to the hybrid composites, [SGGGGS], [SGSGSGS], and
[SSSGSSS], along with the steel composite, [S]8 (Figure 15). In this work, only half-cyclic testing was
performed due to experimental limitations that did not permit the acquisition of compression data.
However, the stress-strain curve does appear to follow kinematic hardening behavior. Using this
assumption, the complete hysteresis behavior, including stress-strain relationships in the compressive
stress region, was predicted and shown in Figure 15. The red portion of the bilinear model followed
each experimental tensile loading and unloading cycle. The black portion of the curve serves as
the predicted extension for full-cycle behavior by extending the curve into the compression region.
This bilinear model predicted the hysteresis behavior of all hybrid composites by finding the elastic and
hardening slopes, the yield stress, and the residual strain. Composites with a higher fraction of steel
were found to follow the prediction more closely. The calculated elastic and hardening slopes align
with the paths of the experimental hysteresis behavior. Hybrid composite cycles that lie within the ideal
parallelogram suggest material stability. Once the stress reaches greater than 2 σy , the steel fibers may
yield in compression. This is not desirable for structural materials because subsequent loading cycles
will also yield, and the material will become more unstable over time. The area under each idealized
loading cycle was found, and the energy dissipation followed the same trend as the experimental
values. As this model is an idealized shape, it can be seen how it can under- or over-predict the actual
energy dissipation of the composite depending on the curve fit.
Polymers 2017, 9, 151 14 of 16
Polymers 2017, 9, 151 14 of 16

(a) (b)

(c) (d)
Figure 15. Ideal hysteresis behavior: (a) [SGGGGS]; (b) [SGSGSGS]; (c) [SSSGSSS]; (d) [S]8.
Figure 15. Ideal hysteresis behavior: (a) [SGGGGS]; (b) [SGSGSGS]; (c) [SSSGSSS]; (d) [S]8 .

4. Conclusions
4. Conclusions
The conclusions presented herein are based on the results of monolithic tensile and half-cyclic
testing conclusions
The of glass/steelpresented hereinepoxy
fiber reinforced are based on thecoupons.
composite results of monolithic
Three differenttensile and half-cyclic
fiber ratios of steel
testing of glass/steel fiber reinforced epoxy composite coupons. Three different
reinforcement were studied: (1) [SGGGGS] (8.2% steel); (2) [SGSGSGS] (15.7% steel); fiber ratios
andof(3)
steel
reinforcement were studied: (1) [SGGGGS] (8.2% steel); (2) [SGSGSGS] (15.7% steel);
[SSSGSSS] (22.8% steel). The mechanical properties were characterized and gave rise to the and (3) [SSSGSSS]
(22.8% steel).findings:
following The mechanical properties were characterized and gave rise to the following findings:

•  TheThe tensile
tensile strength
strength of of
thethe hybrid
hybrid composites
composites was
was directly
directly proportionaltoto
proportional the
the respectiveglass
respective glassand
and steel fiber percentages. The strengths from highest to lowest
steel fiber percentages. The strengths from highest to lowest were as follows: [G]5 , [SGGGGS],were as follows: [G]5,
[SGGGGS], [SGSGSGS], [SSSGSSS], and [S] 8. This order held true for tensile, open-hole tensile,
[SGSGSGS], [SSSGSSS], and [S]8 . This order held true for tensile, open-hole tensile, and open-hole
and open-hole half-cyclic loading.
half-cyclic loading.
 The rule of mixtures proved valid in the elastic region and predicted the stiffness values
• The rule of mixtures proved valid in the elastic region and predicted the stiffness values accurately.
accurately. However, in the post-yield region, the ROM consistently over-predicted the
However, in the post-yield region, the ROM consistently over-predicted the stress-strain
stress-strain relationship. More research is needed on theoretical models of nonmetal-metal
relationship. More research
hybrid composites is needed
in the inelastic on theoretical models of nonmetal-metal hybrid composites
region.
 in Composites
the inelasticwithregion.
a higher percentage of steel had localized failure. In contrast, composites with
• Composites
higher percentageshigher
with a of glasspercentage
had a more ofdistributed
steel had localized failure.making
failure pattern, In contrast, composites
it difficult with
to predict
higher percentages
failure location. The of glass
additionhadofa steel
morehelped
distributed failure
maintain the pattern,
integritymaking it difficultafter
of the composite to predict
the
failure
failurelocation. The addition
of the glass fibers. The of hybrid
steel helped maintain
composites the integrity
experienced of the
a ductile composite
failure, whichafter
maythe
provide
failure warning
of the to structural
glass fibers. failure.
The hybrid This wasexperienced
composites due to the aspread
ductileoffailure,
plasticity over
which maya larger
provide
length of steel fibers.
warning to structural failure. This was due to the spread of plasticity over a larger length of
 steel
Thefibers.
addition of steel fibers to glass/epoxy composites decreased the vulnerability to stress
• concentrations.
The addition of steel Accumulated
fibers todamage from cyclic
glass/epoxy loading decreased
composites does not havethe avulnerability
significant effect on
to stress
the composite stress-strain relationship. This behavior is potentially beneficial
concentrations. Accumulated damage from cyclic loading does not have a significant effect on the in structural
elements stress-strain
composite that are subject to repeatedThis
relationship. dynamic loading.
behavior is potentially beneficial in structural elements
 [SGGGGS] outperformed [SGSGSGS] and [SSSGSSS] and offers balanced mechanical
that are subject to repeated dynamic loading.
properties. This composite had the highest strength, dissipated the most energy during loading,
• [SGGGGS] outperformed [SGSGSGS] and [SSSGSSS] and offers balanced mechanical properties.
This composite had the highest strength, dissipated the most energy during loading, and
Polymers 2017, 9, 151 15 of 16

showed the most consistent re-centering capabilities. This may signify that the amount of steel
reinforcement may be optimized to achieve target structural performances.
• The hybrid composite half-cyclic behavior may successfully be predicted using the bilinear
hysteresis model of lead-rubber bearings. This model suggested that the hybrid composites had
greater stability. This model also justified the experimental energy dissipation and residual strain
ratio results.
• Overall, glass/steel fiber hybrid composites show promise in structural applications because of
their high strength, energy absorption during loading, and re-centering capabilities. More research is
needed to optimize the composite design to achieve higher failure strains.

Acknowledgments: The work at the University of Connecticut leading to this publication has received funding
from PFI:AIR-TT National Science Foundation Grant #1500293. The authors thank Hexion, Owens Corning and
Bekaert for providing materials and technical support for this research.
Author Contributions: Amanda K. McBride was a master’s student at the University of Connecticut for the
duration of the research. Arash E. Zaghi was Amanda K. McBride’s major advisor and the Principal Investigator
for this research. Amanda K. McBride developed the research program with the guidance of Arash E. Zaghi.
Amanda K. McBride performed the experiments and processed the data with Samuel L. Turek. Kelly A. Burke
served as a technical advisor, as well as a Co-Principal Investigator for this research. Amanda K. McBride drafted
the manuscript and all authors contributed to the final paper.
Conflicts of Interest: The authors declare no conflict of interest. The founding sponsors had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the
decision to publish the results.

References
1. Bray, D.J.; Dittanet, P.; Guild, F.J.; Kinloch, A.J.; Masania, K.; Pearson, R.A.; Taylor, A.C. The modelling of
the toughening of epoxy polymers via silica nanoparticles: The effects of volume fraction and particle size.
Polymer 2013, 54, 7022–7032. [CrossRef]
2. Swolfs, Y.; Meerten, Y.; Hine, P.; Ward, I.; Verpoest, I.; Gorbatikh, L. Introducing ductility in hybrid carbon
fibre/self-reinforced composites through control of the damage mechanisms. Compos. Struct. 2015, 131,
259–265. [CrossRef]
3. Asgarinia, S.; Viriyasuthee, C.; Phillips, S.; Dubé, M.; Baets, J.; Van Vuure, A.; Verpoest, I.; Lessard, L.
Tension-tension fatigue behavior of woven flax/epoxy composites. J. Reinf. Plast. Compos. 2015, 34, 857–867.
[CrossRef]
4. Tian, Z.; Song, H.; Wan, Z.; Du, X. Fatigue properties of steel cord-rubber composite. J. Elastom. Plast. 2001,
33, 283–296. [CrossRef]
5. Ahmed, S.F.U.; Maalej, M.; Paramasivam, P. Analytical model for tensile strain hardening and multiple
cracking behavior of hybrid fiber-engineered cementitous composites. J. Mater. Civ. Eng. 2007, 19, 527–539.
[CrossRef]
6. Callens, M.G.; Gorbatikh, L.; Verpoest, I. Ductile steel fibre composites with brittle and ductile matrices.
Compos. Part A Appl. Sci. Manuf. 2014, 61, 235–244. [CrossRef]
7. Callens, M.G.; De Cuyper, P.; Gorbatikh, L.; Verpoest, I. Effect of fibre architecture on the tensile and impact
behaviour of ductile stainless steel fibre polypropylene composites. J. Compos. Struct. 2015, 119, 528–533.
[CrossRef]
8. Callens, M.G.; Gorbatikh, L.; Bertels, B.; Smet, M.; Verpoest, I. Tensile behaviour of stainless steel fibre/epoxy
composites with modified adhesion. Compos. Part A Appl. Sci. Manuf. 2015, 69, 208–218. [CrossRef]
9. Allaer, K.; De Baere, I.; Lava, P.; Van Paepegem, W.; Degrieck, J. On the in-plane mechanical properties of
stainless steel fibre reinforced ductile composites. J. Compos. Sci. Technol. 2014, 100, 34–43. [CrossRef]
10. Mosleh, Y.; Clemens, D.; Gorbatikh, L.; Verpoest, I.; Van Vuure, A.W. Penetration impact resistance of novel
tough steel fibre-reinforced polymer composites. J. Reinf. Plast. Compos. 2015, 34, 624–635. [CrossRef]
11. Faes, J.C.; Rezaei, A.; Van Paepegem, W.; Degrieck, J. Influence of matrix toughness and interfacial strength
on the toughness of epoxy composites with ductile steel fabric reinforcement. In Proceedings of the 16th
European Conference on Composite Materials (ECCM16), Seville, Spain, 22–26 June 2014; Volume 22,
pp. 779–793.
Polymers 2017, 9, 151 16 of 16

12. Sinmazçelik, T.; Avcu, E.; Bora, M.O.; çoban, O. A review: Fibre metal laminates, background, bonding types
and applied test methods. J. Mater. Design 2011, 32, 3671–3685. [CrossRef]
13. Botelho, E.C.; Silva, R.A.; Pardini, L.C.; Rezende, M.C. Review on the development and properties of
continuous fiber/epoxy/aluminum hybrid composites for aircraft structures. Mater. Res. 2006, 9, 247–256.
[CrossRef]
14. Moussavi-Torshizi, S.E.; Dariushi, S.; Sadighi, M.; Safarpour, P. A study on tensile properties of a novel
fiber/metal laminate. Mater. Sci. Eng. 2010, 527, 4920–4925. [CrossRef]
15. Higgins, A. Along the Bond Line: Groundbreaking Aircraft Structures; Technical Bulletin for Fokker Technologies:
Papendrecht, The Netherlands, 2016.
16. Rubio-González, C.; Velasco, F.; Martínez, J. Analysis of notched woven composites and fiber metal laminates
with previous fatigue damage. J. Compos. Mater. 2015, 50. [CrossRef]
17. Kretsis, G. A review of the tensile, compressive, flexural and shear properties of hybrid fibre-reinforced
plastics. Composites 1987, 18, 13–23. [CrossRef]
18. Swolfs, Y.; Gorbatikh, L.; Verpoest, I. Fibre hybridisation in polymer composites: A review. Compos. Part A
Appl. Sci. Manuf. 2014, 67, 181–200. [CrossRef]
19. Hayashi, T. On the improvement of mechanical properties of composites by hybrid composition.
In Proceedings of the 8th International Reinforced Plastics Conference, Brighton, UK, 1972; British Plastics
Federation: Brighton, UK; pp. 149–152.
20. Marom, G.; Fischer, S.; Tuler, F.R.; Wagner, H.D. Hybrid effects in composites: Conditions for positive or
negative effects versus rule-of-mixtures behaviour. J. Mater. Sci. 1978, 13, 1419–1426. [CrossRef]
21. Hull, D.; Clyne, T.W. An Introduction to Composite Materials, 2nd ed.; Cambridge University Press: New Delhi,
India, 1996; p. 310.
22. Breuer, U.P.; Schmeer, S.; Eberth, U. Carbon and metal fiber reinforced airframe structures—A new approach
to composite multifunctionality. In Proceedings the of Deutscher Luft-und Raumfahrtkongress, Stuttgart,
Germany, 10–12 September 2013.
23. Satish, K.G.; Siddeswarappa, K.; Kaleemulla, K.M. Characterization of in-plane mechanical properties of
laminated hybrid composites. J. Min. Mater. Char. Eng. 2010, 9, 105–114. [CrossRef]
24. Ahmed, T.J. Hybrid Composite Structures: Multifunctionality Through Metal Fibers. Ph.D. Thesis, Technical
University of Delft, Delft, The Netherlands, 2009.
25. Thysen, S. Mechanical Behavior of Hybrid Steel and Glass Fibre Composites, Master Thesis, KU Leuven,
Leuven, Belgium, 2013.
26. Stainless Steel Filler Materials for Plastics; Technical Bulletin for NV Bekaert SA; Bekaert: Zwevegem,
Belgium, 2013.
27. De Bondt, S.; Decrop, J. Bundle Drawn Stainless Steel Fibers. U.S. Patent 7166174 B2, 2007.
28. Epon Resin 828 Technical Data Sheet; Technical Bulletin for Hexion, Inc.: Batesville, AR, Canada, 2005.
29. Astm Standard Test Method for Tensile Properties of Plastics; ASTM International: West Conschohocken, PA,
USA, 2002.
30. Astm Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials; ASTM International:
West Conshohocken, PA, USA, 2007; Volume ASTM D3039.
31. Astm Standard Test Method for Open Hole Tensile Strength of Polymer Matrix Composite Laminates; ASTM
International: West Conshohocken, PA, USA, 2011.
32. Harik, V.M.; Klinger, J.R.; Bogetti, T.A. Low-cycle fatigue of unidirectional composites: Bi-linear s-n curves.
International J. Fatigue 2002, 24, 455–462. [CrossRef]
33. Hu, J.W. Response of seismically isolated steel frame buildings with sustainable lead-rubber bearing (LRB)
isolator devices subjected to near-fault (NF) ground motions. Sustainability 2015, 7, 111–137. [CrossRef]

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).