Construction and Building Materials 149 (2017) 837–848

Contents lists available at ScienceDirect

Construction and Building Materials

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

Optimization of mechanical properties in concrete reinforced with fibers

from solid urban wastes (PET bottles) for the production of ecological
concrete
Erlon Lopes Pereira a,⇑, André Luis de Oliveira Junior b, Adonai Gomes Fineza b
a
Department of Chemistry, Chemistry Engineering Division, Federal University of Viçosa (Brazil), Peter Henry Rolfs Avenue, s/n, University Campus, Viçosa, Minas Gerais, Brazil
b
Department of Civil Engineering, Union of Higher Education of Viçosa (Univiçosa), Maria de Paula Santana Avenue, 3815, University Campus, Viçosa, Minas Gerais, Brazil

h i g h l i g h t s

The volume of fibers in the concrete affects its compression resistance while the fiber length does not affect this property.

The fiber volume factor and the fiber length can modify the concrete resistance to traction.

The ecological concrete load capacity is higher than the conventional concrete load capacity.

The ecological concrete produced with PET bottle fibers collected from urban solid waste is an alternative for waste management and it could reduce the
costs of constructions utilizing asbestos cement.

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

Article history: To solve structural problems such as cracks caused by plastic or hydraulic retraction and drying, the mod-
Received 17 January 2017 ification of concrete properties has been studied through the addition of polymer fibers. The objective of
Received in revised form 14 May 2017 this study was to evaluate the use of fibers from PET bottles on the process of manufacturing ecological
Accepted 17 May 2017
concrete, seeking to optimize the mechanical properties of compression and tension. Standardized mold-
Available online 1 June 2017
ing of specimens was performed according to the norms established by the Brazilian Association of
Technical Standards (Associação Brasileira de Normas Técnicas – ABNT) for conventional concrete. The
Keywords:
results obtained showed that the percent volume of fiber added to the concrete has a direct influence
Concrete compression
Concrete tension
on both the compressive and tensile strength of ecological concrete and that the fiber length has a direct
Concrete bending influence on the tensile strength of ecological concrete. The mathematical models generated in this work
Central composite design can be used to predict efficiency values as a function of the operating parameters: PET fiber length and
Recycling percent volume of PET fiber.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction
Aggregates, water and Portland cement are the main compo-
nents of concrete production and can be considered relatively
cheap and readily encountered anywhere in the world. The ease
of acquiring these raw materials stimulates the use of concrete in
construction sectors [19]. Although concrete presents diverse eco-
nomic, social and constructive advantages, Mehta and Monteiro
[25] mentioned that concrete mixtures with Portland cement pre-
sent structural problems such as cracking caused by plastic or
⇑ Corresponding author.
E-mail addresses: erlonlopes@gmail.com (E.L. Pereira), oliveiraandre66@yahoo.
com.br (A.L. de Oliveira Junior), engcivil@univicosa.com.br (A.G. Fineza).
hydraulic retraction and drying, which can make the its use unvi-
able for some purposes.
To resolve the structural problems mentioned the modification
of concrete properties has been studied through the addition of
suitable materials such as nylon fibers [24], polypropylene fibers
[29], PVC fibers [20], glass fibers [1], coconut fibers [23] and rubber
fibers [22]. In the present work, the use of a high-density
polypropylene (PVC) fiber has been proposed for this purpose. In
this context, fibers are highlighted as material for incorporation
into the concrete, indicated as a potential solution to solve the
problems of cracking, in addition to improvement of the structural
aggregation conditions between concrete components. According
to Mehta and Monteiro [25], the properties of fiber-reinforced con-
crete depend on the type of fiber, its quantity and its dimensions,
so there is still no ideal fiber concept in literature or standards

http://dx.doi.org/10.1016/j.conbuildmat.2017.05.148
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
838 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848
for use. Therefore, the properties of the cement matrix and the
fiber-matrix interface should be studied for each individual case.
As a result of the drastic increase in beverage consumption, the
production of polyethylene terephthalate (PET) bottles has
increased exponentially due to the favorable properties of this
plastic, including: low density, high resistance:weight ratio, high
durability, ease of conception/fabrication and low cost. The Brazil-
ian PET Industry Association (Associação Brasileira da Indústria do
PET – ABIPET) estimates a production of 840 thousand tons of this
plastic in 2016, 17% greater than production in 2015. In contrast,
Brazil currently recycles 60% of discarded PET bottles, being con-
sidered one of the countries with the highest rate of PET recycling
in the world [2].
Although PET recycling in Brazil is a reality, approximately 40%
of this product is destined for landfills and dumps after consump-
tion. In the context of using fiber as an aggregate in concrete, PET
bottle fibers present themselves as readily available material for
this use and of low cost, where this options provides benefits for
the elimination of these residues and reduction of environmental
damages.
According to Gu and Ozbakkaloglu [17], due to the fragile nat-
ure of concrete and facilitated availability of PET waste, several
researchers around the world have been producing ecological con-
crete using fibers from discarded PET bottles as concrete aggre-
gates to potentially neutralize or reduce their negative attributes.
However, the authors report the need for studies that correlate
characteristics and properties of fibers from PET residues (length,
volume, fraction or percentage of fiber inserted in the concrete)
with the mechanical properties of the concrete.
Considering the above, this work sought to evaluate the use of
fibers from PET bottle waste in the production of ecological con-
crete and to study the synergistic effect between length and per-
centage of these fibers on the tensile and compressive
mechanical properties of the ecological concrete in relation to tra-
ditional concrete (concrete mixture with Portland cement).
2. Material and methods
2.1. Experimental apparatus
The aggregates (gravel and sand) and binder (Portland cement) used in the con-
crete were tested in the laboratory in order to obtain the ideal parameters for use in
concrete.
As a binder Type II Portland Cement was used, composed of slag with resistance
class of 32 MPa at 28 days (CP II E 32), specific mass of 3 g cm3 and average unit
mass of 0.961 kg dm3. Specific mass and unit mass assays were performed accord-
ing to NBR NM 23 [7] and NBR NM 45 [8], respectively.
The sand used was natural quartz sand and according to the granulometric
composition test described in NBR NM 248 [11], the maximum characteristic size
of 4.8 mm and modulus of fineness of 2.66 were obtained. In the assay performed
according to the NBR NM 46 [9] which addresses the volume of fine material, a
result of 0.6% fine material was obtained. Using the NBR NM 45 [8], which addresses
the unit mass, a unit mass of 1.432 kg dm3 was obtained.
The bulk material used was Arkose gravel classified according to the NBR NM 52
[10] in terms of absorption and specific mass of saturated surface dry. The large
aggregate showed absorption of 0.4% and specific mass of saturated surface dry
of 2714 kg m3. To determine the unit mass and granulometric composition, the
gravel was submitted to the assays described in NBR NM 45 [8] and NBR NM 248
[11], resulting in 1.36 kg dm3 for the unit mass and the granulometric composition
showed a maximum characteristic size of 19 mm and fineness modulus of 6.91.
Polyethylene terephthalate (PET) fibers were obtained from PET bottles found in
trash prior to being sent to landfills and dumps after consumption of the product.
For determination of the mean absolute density (or specific mass) of the material,
the mass (m) and volume (V) of the material were quantified. To determine the
value of ‘‘m” an entire PET bottle with dimensions of 22.5 cm tall by 9.2 cm in diam-
eter was cut into smaller square pieces using stainless steel scissors. Each square of
the PET bottle had sides measuring 2 cm which was confirmed using a steel ruler
graduated in millimeters with precision of 0.01 mm. All the pieces were assembled
and weighed on a MARTE class II scale with precision of 0.01 g. To determine the
volume, the pieces were inserted into a glass beaker with working volume of 1 L,
settled and the remaining volume of the beaker was filled with distilled water.
The value of V was calculated by the difference between the total volume of the
beaker (1 L) and the volume of water required to complete it. The procedure was
performed 10 times (with 10 PET bottles) and the PET material density was calcu-
lated from the relationship between the m and V vales, obtaining an average abso-
lute density (or specific mass) of 1.3 g cm3.
For the assays described in Table 1, other PET bottles were transformed into
fibers with lengths of 7.9, 10, 15, 20 and 22 mm and with standard width of
3 mm. Transformation of the PET bottle into fibers was performed with the aid of
an electric PET bottle shredder with metering adjustment to allow for obtaining
the desired length.
The mechanical tests of the concrete described in this item were carried out at
the Civil Construction Materials Laboratory of the University Center for Higher Edu-
cation of Viçosa, located in the city of Viçosa (Latitude: 20° 450 1400 S, Longitude: 42°
520 5500 W, Elevation: 648 m), Minas Gerais state, Brazil. The research laboratory is
certified by the Ministry of Education in Brazil.
2.2. Experimental procedure
The concrete was dosed at a ratio of cement: sand: gravel of 1:2:2, and the Bau-
chemie Brazil curing retardant additive was used at the volumetric concentration of
0.6% in relation to the cement weight. Use of the additive was to increase the work-
ing time of the concrete, taking into account that 66 test pieces were molded (11
assays with 3 replicates for each test and 2 response variables, as described in item
2.2) and the curing time of conventional concrete is 3 h.
Preparation of the binder was conducted according to NBR 12655 [3], which
specifies the procedures that must be adopted with concrete in its fresh state. For
mixing the concrete a three-phase electric 600 L concrete mixer of the company
CSM was used whose drum rotated at 27 rpm. For molding the test pieces the slump
test set was used and Ø 10  20 cm cylindrical forms of the SOLOTEST were used. To
perform the compression and tension tests the Electro-hydraulic Press was used
with Graphic Digital Indicator, from SOLOTEST, with capacity of 100–200 ton force
and electro-hydraulic pump that allows for manual control of the test speed
increase.
After preparation of the concrete, six reference samples (Assay 11) were molded
for the compression and tension assays by diametral compression (Table 1). There-
after, specific amounts of concrete were separated for each fiber reinforced speci-
men as described in Table 2. Specific amounts of fibers were introduced into each
specimen, requiring that they were molded individually. Fiber incorporation was
done on a smooth and flat surface, and after homogenization the concrete was
molded in three layers, as established by the NBR 5738 [4]. In each layer the cement
was penetrated 12 times with a metal rod. The same procedure was repeated for the
specimens with regards to tension, not altering the molding procedure. Soon after
molding, the specimens were identified and left on a flat surface for hardening of
the concrete and this procedure took 48 h. For storage of the hardened concrete
after 48 h, the method of immersion in lime-saturated water was used. The speci-
mens remained immersed for 28 days to be ruptured as necessary (tension and
compression).
After the 28 days the 66 specimens were removed from the water and measure-
ments were obtained of length and diameter prior to placement in the Electrohy-
draulic Press to determine the values of compressive strength and tensile
strength of each specimen. With the average of each triplicate of the 11 assays,
the increases in compressive strength (R1) and the increases in tensile strength
by diametral compression of each specimen (R2) of the ecological concrete (assays
1–10) were calculated in relation to the reference concrete consisting only of sand,
gravel and Portland cement (Assay 11). The values of R1 and R2 were calculated
using Eqs. (1) and (2), respectively.
 
Cn  C11
R1 ¼  100 ð1Þ
C11
 
Tn  T11
R2 ¼  100 ð2Þ
T11
Table 1
Experimental matrix used in the CDD with values of the central, factorial and axial
points.
Assay F1 (mm) F2 (%)
1 10 0.1
2 10 0.3
3 20 0.1
4 20 0.3
5 7.9 0.2
6 22 0.2
7 15 0.05
8 15 0.34
9 15 0.2
10 15 0.2
11 0 0
 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848
where, Cn: Compressive strength of the concrete obtained for assays 1–10 reinforced
with fiber (MPa). C11: Compressive strength of the reference concrete represented in
assay 11 (MPa). Tn: Tensile strength of the concrete obtained for assays 1–10 rein-
forced with fibers (MPa). T11: Tensile strength of the reference concrete represented
in assay 11 (MPa). n: number of assays performed for compression and tension
(assays 1–10). R1: compression efficiency obtained for the ecological concrete in
relation to the reference concrete (%). R2: tensile efficiency obtained for the ecolog-
ical concrete in relation to the reference concrete (%).
The 11 assays presented in Table 1 were used to obtain an ordered pair, i.e., PET
fiber length and percent volume of PET fiber that gave the ecological concrete the
highest values of R1 and R2, where this condition was denominated the optimized
condition. In order to verify the optimized conditions, the experiment was repeated
using the optimal values of F1 and F2. In this experiment, in addition to analyzing
the response variables R1 and R2 the flexural strength was also evaluated, where in
this test 6 prismatic specimens were used (3 composed only of sand, gravel and
Portland cement and 3 with PET fibers and optimized F1 and F2 values). The value
of the increase in flexural strength (R3) was calculated using Eq. (3).
 
Fr  F11
R3 ¼  100 ð3Þ
F11
where, FR: Flexural strength of the concrete obtained for the experiment reinforced
with fibers (MPa). F11: flexural strength of the concrete obtained for the reference
concrete (MPa). R3: flexural efficiency obtained for the ecological concrete in rela-
tion to the reference concrete (%).
2.3. Experimental design
For process optimization the orthogonal and rotational central composite
design (CCD) statistical method was used, with two factors denominated PET fiber
length (F1) and fiber percentage in relation to the total concrete volume (F2). In
relation to the F1 factor, the PET fiber lengths of 10 mm, 15 mm and 20 mm were
adopted, respectively. In relation to factor F2 low, central and high percentages of
fibers were adopted in relation to the total concrete volume of 0.1%, 0.2% and
0.3%, respectively. The values adopted for factors F1 and F2 were based on the stud-
ies carried out by Pelisser et al. [26]. The CCD experimental matrix used for process
optimization is presented in Table 1.
The optimization experiment (Table 1) was performed to evaluate the synergis-
tic effects of factors F1 and F2 on two response variables (compression and tension),
so for each assay presented in Table 1 six samples were molded, three for each
response variable of the experiment. The triplicate is a procedure requested by
the ABNT NBR 7222/94 [6] and ABNT NBR 5739/94 [5], where the results of the
response variables are given by the arithmetic mean obtained in each assay. Deter-
mination of the reference values for the response variables was performed on the
28th day of molding, considering the date on which the concrete reaches its
designed resistance. At the end of the 28 days, the specimens were broken by com-
pression and tension via diametrical compression, obtaining resistance values for
each assay.
The specimen (SP) used had dimensions of 10 cm in diameter by 20 cm in
length, for a total volume of 1570.8 cm3. Thus, for the 0.1% content of PET fibers
in the total volume of a SP in terms of concrete, the PET volume in each specimen
with this respective content was 1.57 cm3. Therefore, using the premise that the
material density was 1.3 g cm3, the weight of PET for the specimen with 0.1% con-
tent was 2 grams. The same calculation was performed for the 0.2% and 0.3% vol-
umes and can be described by Table 2.
Descriptive statistics analyses were performed using the SISVAR program,
developed at the Department of Exact Sciences of the Federal University of Lavras
[15]. The analysis of variance (ANOVA) of the effects, mathematical models and sur-
face response graphs were conducted using the STATISTICA 8.0Ò program (StatSoft,
Table 2
Quantity of PET fibers per assay.
Assay F1 F2
1 10 0.1
2 10 0.3
3 20 0.1
4 20 0.3
5 7.9 0.2
6 22 0.2
7 15 0.05
8 15 0.34
9 15 0.2
10 15 0.2
11 0 0
Total
F1: Fiber length (mm). F2: Fiber percent volume (%). V: volume of fibers in each specimen (cm3). W: weight of fibers in each specimen (g). TW: total weight of fibers,
considering the construction of three specimens due to variable response (g).
839
Tulsa, OK, USA). The variable conversion procedure (factors) for application in the
models and process simulations were performed directly in the STATISTICA 10.0Ò
program, and for simulation the desirability tool was used.
3. Results and discussion
3.1. Results of the compression and tension assays
Table 3 presents the resistance values that characterize the
compressive and tensile strength of the ecological and reference
concretes used in assays 1–11.
By analyzing the results of Cn and Tn presented in Table 3, it can
be noticed that the insertion of fibers from PET bottles into the con-
crete increased the capacity of the concrete to withstand compres-
sion (Cn) and tensile (Tn) forces. According to Song et al. [29],
Khaloo et al. [19], Herki et al. [18], Silva et al. [28] and Lima
et al. [21] the insertion of fibers from various raw materials
denominated urban and industrial solid wastes positively or nega-
tively modify the mechanical properties of concrete. Table 4 shows
a summary of the studies performed by each author.
As shown in Table 4, the studies conducted by Khaloo et al. [19],
Herki et al. [18] and Silva et al. [28] indicate that the insertion of
fibers from tire rubber waste, expanded polystyrene/fly ash and
coconut fiber in concrete and mortars decreased their compressive
strength, attributing negative aspects to the quality of the ecolog-
ical mixtures produced, which makes use of the mentioned resi-
dues infeasible. In this context, the values found regarding the
use of fibers from PET bottles presented in Table 5 are promising
when compared to other solid wastes.
To discuss the results found in the experiment, the authors pre-
sented in Table 4 used values referring to the efficiency obtained
by concrete and mortar with addition of fibers. The values of com-
pressive (R1) and tensile (R2) efficiency were defined as the
marked degree of positive or negative difference between the
mechanical strength values of the ecological concrete assays
(Assays 1–10) and the reference concrete (Assay 11). In order to
compare the results obtained in the assays with addition of PET
fibers (1–10) to the results presented in Table 4 and to other refer-
ences in literature, Eqs. (1) and (2) were used to determine the val-
ues of R1 and R2 for the ecological concrete (Table 5).
From the values presented in Table 5 it is possible to observe
that the ecological concrete produced in this experiment was able
to withstand compressive (R1) and tensile forces (R2) higher than
those applied to the reference concrete, with maximum values of
14.3% and 16.6%, respectively. This proved that the loading capac-
ity of unreinforced concrete decreases faster than that of the fiber
reinforced concrete. Moreover, fiber-reinforced concretes exhibit
considerable post-cracking resistance and tenacity, since the fibers
‘‘sew” the cracks [25]. According to Song et al. [29], load transfer
V W TW
1.57 2 12
4.71 6.1 36.6
1.57 2 12
4.71 6.1 36.6
3.14 4.1 24.6
3.14 4.1 24.6
0.78 1 6
5.34 6.9 41.4
3.14 4.1 24.6
3.14 4.1 24.6
0 0 0
243 g
840 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848
occurs continuously between the fiber-matrix so the capacity to
withstand higher values of compression and tensile stresses is
improved, and therefore concretes reinforced with PET fibers are
more efficient than conventional concretes, justifying the men-
tioned values of R1 and R2.
Although in this study fibers from PET bottles showed to be effi-
cient in increasing concrete compression and tensile strength,
some studies have also reported reductions in the compressive
strength of fiber-reinforced concrete. Pelisser et al. [26] reported
that with a percentage volume of PET fiber up to 0.30%, the com-
pressive strength is reduced by 10%. Fraternali et al. [16] reported
that the compressive strength of concrete containing 13.4 kg m3
of PET fiber was reduced by 8.2% compared to the strength of con-
ventional concrete.
Gu and Ozbakkaloglu [17] stated that the increase or decrease
of concrete strength is related to the modulus of elasticity of the
fiber (MEF). The use of plastic fibers with a low modulus of elastic-
ity results in a more significant reduction and plastic fibers with a
high modulus of elasticity result in a concrete with superior
strength. In addition to the MEF analysis, the use of PET for the pro-
duction of ecological concrete must be done under controlled and
optimized conditions, through protocols or mathematical models
for the determination of fibers in the concrete and prediction of
results. The results of these studies are presented in items 3.2
and 3.3.
3.2. Effect of fiber length (F1) and percent volume of fibers (F2) on the
compressive (R1) and tensile (R2) efficiencies of the ecological concrete
To investigate the synergistic influence of factors F1 and F2
(Table 1) on the increase in capacity of the ecological concrete to
withstand compressive (R1) and tensile (R2) forces, the efficiency
values obtained in assays 1–10 (Table 5) were submitted to analy-
sis of variance and the results are presented in Table 6.
The analysis of variance for the response variables R1 and R2 of
the ecological concrete (Table 6) shows which factor presented a
statistically significant difference at 5% (p < 0.05).
For the response variable R1, the analysis of variance shows that
the length of PET fibers (F1) does not have a statistically significant
influence at the level of 5% (p > 0.05) on this variable response. This
indicates that the F1 factor does not influence optimization and the
process of obtaining ecological concretes resistant to compression
(R1). The percent volume of PET fibers (F2) has a statistically signif-
icant influence at the level of 5% (p < 0.05) on the response variable
R1, for both linear and quadratic interaction. This shows that factor
F2 must be taken into account in optimization and in the process of
obtaining compression-resistant ecological concretes (R1). The lin-
ear interaction between the factors PET fiber length and percent
volume of PET fiber (F1  F2) was not statistically significant at
Table 3
Strength values encountered as response to the assays of compression (Cn) and tension (Tn).
Assay (n) Cn
1 46 ± 1
2 44 ± 2
3 44 ± 1
4 43.3 ± 0.2
5 43 ± 1
6 42 ± 1
7 45.5 ± 0.4
8 44 ± 1
9 42 ± 5
10 43 ± 2
11 40 ± 1
Cn: Compression strength of the concrete obtained for assays 1–11. Tn: tensile strength of the concrete obtained for assays 1–11. CV: coefficient of variation between the
repetitions of each assay (%).
5% (p > 0.05), since only the factor F2 interfered on the results with
respect to the response variable R1.
For the response variable R2, via the analysis of variance
(Table 6), it can be inferred that the linear factor of PET fiber length
(F1) and the linear factor of percent volume of PET fibers (F2) were
statistically significant at the level of 5% (p < 0.05). Therefore, the
increase or decrease in fiber length and percent volume of fiber
influenced the optimization and the process of obtaining tensile-
resistant ecological concretes (R2). Although both factors F1 and
F2 showed influence on R2 values when analyzed in linear condi-
tion, F1 (L) and F2 (L), the interaction between the factors fiber
length and percent volume of fiber (F1  F2) in this condition
was not statistically significant at 5% (p > 0.05).
To analyze the effect of the factor PET fiber length (F1) and per-
cent volume of PET fibers (F2) in quadratic and linear conditions on
increase in compressive efficiency of the ecological concrete (R1),
the results were expressed by means of a Pareto plot and illus-
trated by the surface response graph (Fig. 1).
Fig. 1a not only highlights the factors that showed to be influen-
tial on the compressive efficiency of the ecological concrete (R1),
but also indicates the tendency of each factor for its optimization.
As was described in Table 6, only the effect of fiber volume (F2)
crossed the Pareto frontier. The positive value for the quadratic
effect of F2 (Q) shows that as these values are increased, the com-
pressive efficiency of the ecological concrete is also increased on a
quadratic scale. The negative value for the linear effect of the factor
F2 (L) shows that the compressive efficiency of the ecological con-
crete increased linearly as the value of F2 is reduced. In addition,
the values of F2 (Q) and F2 (L) suggest that the quadratic model
has more influence on the prediction of data than the linear model,
indicating that the graph describing the effect of F1 and F2 on the
values of R1 is a Parabola with decreasing concavity (Fig. 1b). Ana-
lyzing Fig. 1b, because only the factor F2 was statistically signifi-
cant, it can be inferred that the ecological concrete achieves
efficiency >25% when the F2 values are less than 0.06%.
In order to analyze the effect of the factors PET fiber length (F1)
and percent volume of PET fiber (F2) in quadratic and linear condi-
tions on the increase of tensile efficiency of ecological concrete
(R2), the results were expressed on a Pareto plot and illustrated
by the surface response graph (Fig. 2).
Fig. 2a, in addition to highlighting the factors that proved to be
influential in the tensile efficiency of the ecological concrete (R2),
indicates the tendency of each factor for its optimization. As was
described in Table 6, only the linear effect values of F1 (L) and F2
(L) crossed the Pareto frontier. The positive value for F1 (L) demon-
strates that as the F2 values are increased, the tensile efficiency of
the ecological concrete also increased linearly. The negative value
for F2 (L) shows that as the F1 values are decreased, the tensile effi-
ciency of the ecological concrete is increased linearly. Furthermore,
CV Tn CV
2.95 3.8 ± 0.2 4.01
3.55 3.7 ± 0.4 9.77
1.4 4.1 ± 0.2 3.59
0.46 3.9 ± 0.2 4.46
2.24 3.8 ± 0.2 5.57
1.79 3.9 ± 0.1 3.55
0.94 4.1 ± 0.3 6.25
1.83 3.9 ± 0.1 1.5
10.9 3.9 ± 0.2 6.2
5.26 4.0 ± 0.3 7.43
3.17 3.5 ± 0.3 8.83

Table 4
Experiments performed using recycled fiber to reinforce concretes and mortars.
Author Experimental procedure
Song The properties of reinforced
et al. concrete were investigated
[29] using coarse polypropylene
monofilaments and
polypropylene in staples. The
dosages used for the coarse
monofilaments were 3 kg m3,
6 kg m3 and 9 kg m3, and the
dosage for the polypropylene
staples was 0.6 kg m3
Khaloo The substitution of mineral
et al. aggregates for rubber fibers
[19] from tires was investigated,
replacing 12.5%; 25%; 37.5%
and 50% of the aggregate
mineral volume of the concrete
Herki The study sought to replace
et al. the small aggregate by
[18] expanded polystyrene at
volumes of 0, 60 and 100%,
together with the addition of
fly ash at 0, 20 and 40% by
weight in Portland cement.
Nine experiments were carried
out to combine these factors
Silva This study investigated the
et al. influence of coconut fiber
[28] length on the compressive
strength of mortars. The
lengths of 12.5, 25.0, 37.5, 50.0,
62.5 and 75.0 mm and volume
of 0.3% in relation to the total
volume of the mixture were
used
Lima The incorporation of sisal fiber
et al. residues for use in cement –
[21] based composites was
investigated. Compatibility
with the cement based matrix
was evaluated by a fiber
pullout test and flexural test in
reinforced composites with 2%
of residues
Table 5
Compressive (R1) and tensile efficiency (R2) values of the ecological concrete in relation to the reference concrete (Assay 11).
Assay R1 (%)
1 14.3
2 9.9
3 9.7
4 7.9
5 7.2
6 4.5
E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848
Results
A 14.6% increase was observed
in the compressive strength of
the concrete for coarse
monofilaments and a 17.31%
increase in the compressive
strength for polypropylene in
staples. Also observed were an
8.88% increase in the tensile
strength for coarse
monofilaments and a 13.35%
increase in strength for
polypropylene staples in
relation to the reference
concrete
A reduction of 78.81% in
compressive strength was
observed for the volume of
12.5%; a reduction in the
compressive strength of
95.15% for the volume of 25%;
reduction of 97.88% in the
compressive strength for the
volume of 37.5% and reduction
of 98.8% in the compressive
strength for the content of 50%
The reference concrete showed
a compressive strength of
16 MPa. The other experiments
that used the combination of
expanded polystyrene and fly
ash obtained mean reductions
of 50% in compressive
strength. The experiment that
used maximum concentrations
of polystyrene and fly ash
(100% substitution of the small
substrate by expanded
polystyrene and addition of
40% fly ash in the Portland
cement) showed the greatest
reduction in the compressive
strength of 84%
The addition of coconut fibers
to the mortar matrices resulted
in average reductions of 40% in
compressive strength when
compared to the reference
mortar
Addition of untreated sisal
fiber, scrap or field bush
increased the flexural strength
by about 24%, 65% and 16%,
compared to the reference
concrete, respectively
R2 (%)
8.9
6.2
16.5
9.8
7.4
11.7
841
the values of F1 (L) and F2 (L) suggest that the fiber length has a
greater influence than fiber volume on increasing the tensile effi-
ciency of ecological concrete. Analyzing Fig. 2b, it can be inferred
that the ecological concrete achieves efficiency >22% when using
percent volumes of fiber 0.06% and fiber lengths >24 mm. There-
fore, these are the values that provide ecological concrete with
greatest tensile efficiencies.
According to Pelisser et al. [26], PET fiber volumes greater than
0.30% of the total concrete volume present problems of homogene-
ity and workability. As expected, a greater loss of slump and effi-
ciency occurred when the fiber content increased, reducing its
tensile strength. Gu and Ozbakkaloglu [17] argued that both non-
fiber reinforced concrete and fiber reinforced concrete tend to
respond almost linearly up until the first crack point. After the first
crack, the PET fibers absorb the additional loads caused by the
crack, showing higher tensile strength than the reference concrete
due to the increased length of PET fibers, since the fibers ‘‘sew” the
cracks. For the effect of factor F2 (L), the negative value demon-
strates that as the fiber percent volume values are decreased, the
tensile efficiency of the ecological concrete (R2) increases linearly.
3.3. Mathematical model for the compression (R1) and tensile (R2)
efficiency data of the ecological concrete: generation, calibration and
analysis of the residues
3.3.1. Mathematical modeling and calibration of the obtained results
Results of the concrete compressive efficiency (R1) obtained for
each assay (Table 6) were correlated with the factors F1 and F2 to
determine the mathematical model that describes the behavior of
R1 as a function of each factor (Table 7).
Respecting the significance of each evaluated factor and using
the efficiency results for ecological concrete obtained for each
assay (Table 5) as the only response variable for R1, Eq. (4) was
generated.
R1 ð%Þ ¼ 35:4  1:3F1 þ 0:03F12  168:4F2 þ 334:4F22
þ 1:3F1F2 ð4Þ
Eq. (4) correlates the factors F1 and F2 with the compressive
efficiency of the ecological concrete. The model presented after cal-
ibration and validation can be used to predict efficiency values as a
function of operational parameters (factors F1 and F2). Calibration
of the model is shown in Fig. 3.
Fig. 3 shows that the results of R1 obtained experimentally are
similar to the results of R1 estimated by Eq. (4), with mean residual
squared value of 2.153 and R2 = 0.9113. Therefore, Eq. (4) is able to
predict the compressive efficiency results of ecological concrete
based on PET fiber length (F1) and percent volume of PET fiber
(F2) considering linear (L) and quadratic interactions (Q).
The tensile efficiency (R2) results obtained for each assay
(Table 5) were correlated with the factors simultaneously for
determination of the mathematical model that describe the effi-
ciency behavior of the concrete as a function of each factor
(Table 8).
Assay R1 (%) R2 (%)
7 13.4 16.6
8 9.2 12.1
9 5 10.4
10 6.1 12.6
11 – –
842 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848
Table 6
Analysis of variance (ANOVA) performed with the results of Table 5.
Factors SS
Analysis of variance performed with the results for the response variable R1
F1 (L) 13.56782
F1 (Q) 1.75018
F2 (L) 18.42153
F2 (Q) 51.11161
F1 (L)  F2 (L) 1.69000
Error 8.61190
STS 97.11600
Analysis of variance performed with the results for the response variable R2
F1 (L) 37.1786
F1 (Q) 8.8510
F2 (L) 31.6755
F2 (Q) 4.6740
F1(L)  F2(L) 4.1330
Error 11.6619
STS 107.9707
F1: factor referring to the length of the fibers (mm). F2: factor referring to the percentage volume of fiber (cm3). (L): condition of linear interaction for each factor. (Q):
condition of quadratic interaction for each factor. SS: sum of squares; DOF: degrees of freedom; MS: mean square; F: calculated f; STS: sum of total squares. * Statistically
significant at 5% (p < 0.05).
Respecting the significance of each factor evaluated and using
the tensile efficiency results of the ecological concrete obtained
for each assay (Table 5) as the only response variable for R2, Eq.
(5) was generated.
R2 ð%Þ ¼ 5:6 þ 2:5F1  0:06F12  29:9F2 þ 101:1F22
 2F1F2
Eq. (5) correlates the factors analyzed with the tensile efficiency
of the ecological concrete (R2). The model presented (Eq. (5)) after
calibration and validation can be used to predict efficiency values
as a function of the operating parameters: PET fiber length (F1)
and percent volume of PET fiber (F2). Calibration of Eq. (5) can
be verified in Fig. 4.
Fig. 4 shows that the experimentally obtained R2 results are
similar to the R2 results estimated by Eq. (5), with the mean resid-
ual squared value of 2.915 and R2 = 0.892. Therefore, Eq. (5) is able
to predict the tensile efficiency results of the ecological concrete
based on the factors PET fiber length (F1) and percent volume of
PET fiber (F2), considering the linear (L) and quadratic interactions
(Q).
3.3.2. Residual analysis of the models
Calado and Montgomery [14] stated that in addition to the pre-
dictive capacity provided by the model calibration (Figs. 3 and 4),
the adequacy of the regression model (Eqs. (4) and (5)) should be
synergistically investigated using the residual analysis of the mod-
els generated for the compressive and tensile efficiencies (Fig. 5).
Calado and Montgomery [14] noted that in analysis of the resi-
dues it is necessary to initially evaluate the homoscedasticity of the
residues in addition to the outliers with studentized values.
The diagnostic results of the homoscedasticity analysis for the
residual values are shown in Fig. 5a and e. Homoscedasticity is
the hypothesis that the errors between the efficiency values of
the ecological concrete observed experimentally (Table 5) and the-
oretical values predicted for R1 and R2 by Eqs. (4) and (5), respec-
tively, have zero mean and constant variance when constructing a
statistical model. If this requirement is met, calibration of the
model which predicts the values of R1 and R2 is considered statis-
tically valid (Figs. 3 and 4). If the homoscedasticity assumption is
not valid, the models described by Eqs. (4) and (5) fit into the
heteroscedasticity condition. For the diagnosis of heteroscedastic-
ity, it is attempted to find some trend in the homoscedasticity diag-
nosis graphs of the residuals for R1 and R2 (Fig. 5a and e,
respectively). Therefore, if the points are randomly distributed
DOF MS F p
1 13.56782 6.30190 0.066032
1 1.75018 0.81291 0.418242
1 18.42153 8.55632 0.043023*
1 51.11161 23.74000 0.008206*
1 1.69000 0.78496 0.425680
4 2.15297
9
1 37.17861 12.75217 0.023354*
1 8.85097 3.03586 0.156404
1 31.67554 10.86463 0.030042*
1 4.67403 1.60318 0.274170
1 4.13302 1.41761 0.299624
4 2.91547
9
around 0, without any behavior or tendency, there is an indication
that the variance of the residuals is homoscedastic. However, the
presence of a ‘‘funnel” is an indication of the presence of
heteroskedasticity [14,12].
Analyzing Fig. 5a and e, it is observed that the points are ran-
domly distributed around 0, without any tendency or linear rela-
ð5Þ tionship, indicating that the variance of the residuals is
homoscedastic. According to Calado and Montgomery [14], in this
case the standard errors of estimates obtained by the least squares
method are correct and therefore the statistical inference is valid. It
is also possible to affirm that the coefficients of the models (Tables
7 and 8) are the best estimators of minimum variance because they
are not biased.
After the homoscedasticity test the values of studentized resi-
dues were used to evaluate if there were outliers (Fig. 5b and f).
Studentized residuals have constant variances, which makes it very
practical to search for outliers. According to Calado and Mont-
gomery [14], outliers of the residues are atypical points, i.e., pre-
senting behavior different from the others. The authors
mentioned that this analysis is of great importance because if an
outlier is influential, it interferes on the adjusted regression func-
tion (the inclusion or not of the point substantially modifies the
adjusted values). But a single observation, even if considered an
outlier, does not mean that consequently it is an influential point.
Therefore, a point may be outlier with respect to Y or X and may or
may not be an influential point.
According to Box et al. [13], any observation outside the range
of 3 to 3 can be considered an outlier and the impact of this dis-
turbance should be evaluated on estimates of the response variable
by the model. If there are no influential points, one can rely more
on the proposed model. Based on Fig. 5b and f, it can be observed
that none of the 10 analyzed data points were outside the men-
tioned range, so there were no outliers for R1 and R2.
In addition to the diagnosis of homoscedasticity and residual
outliers, Rodrigues and Iemma [27] stated that the models
described by Eqs. (4) and (5) must meet the statistical conditions
in which the adjustment errors are normally distributed (Fig. 5c
and g) and independent (Fig. 5d and h).
The third analysis conducted was the diagnosis of normality of
the residuals (Fig. 5c and g). According to Calado and Montgomery
[14] and Barros Neto et al. [12], the models described in Eqs. (4)
and (5) can only be used if the data related to the residuals meets
the normality condition, so if the test indicates that this premise is
invalid the regression results (Tables 7 and 8) and the models
described by Eqs. (4) and (5) for estimates of the R1 and R2 values,
 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848 843

Fig. 1. Analysis of the effect of PET fiber length (F1) and percent volume of PET fiber (F2) on the increase in compressive efficiency (R1). a – Pareto plot constructed with the
results of compressive efficiency (R1) for the factors F1 and F2 presented in Table 1. b – Response surface graph describing the effect of the volume of PET fibers (F2) of the
fiber-reinforced concrete on the compressiven efficiency (R1). Experiment performed with 2 factors, 1 block, 10 runs, residual MS = 2.152975.

respectively, cannot be used. In this case, the experiment to gener-
ate the model must be modified introducing other explanatory
variables, or changing the type of model or planning) so that the
errors behave as a normal variable. Calado and Montgomery [14]
and Rodrigues and Iemma [27] showed that when points follow
the behavior of the line (they are not distant from it) it is an indi-
cation that errors are normally distributed. Based on this, when
observing the behavior shown in Fig. 5c and g it is possible to infer
that data referring to the residuals of the models (Eqs. (4) and (5))
follow the normality behavior required for their use to predict and
simulate the results.
The fourth analysis performed was the diagnosis of residue
independence (Fig. 5d and h). According to Rodrigues and Iemma
[27], to verify the residual independence hypothesis the relation-
ship of residuals versus the order of data collection for the 10
experiments or number of cases should be used. When evaluating
Fig. 5d and h it was not possible to perceive a tendency of the 10
points, i.e., there is no indication of dependence of the residuals,
allowing use of the models presented in Eqs. (4) and (5).
Based on the results presented in Fig. 5 it can be inferred that
the residuals of the models reported in Eq. (4) for prediction of
the R1 values and in Eq. (5) to predict the values of R2 meet the
conditions of normality, homoscedasticity and independence. Fur-
thermore, the outliers of the residuals with studentized values did
not influence the predictive capacity or indicate perturbation in the
estimates of the parameters, indicating reliability of the proposed
model.
3.4. Validation of the proposed models, mathematical simulation and
optimum point evaluation
To understand the robustness of the models described in Eqs.
(4) and (5), the mathematical validation of these models was per-
formed using four different tests as described in Table 1. The levels
844 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848

Fig. 2. Analysis of the effect of fiber length (F1) and percent volume of fiber (F2) on the increase in tensile efficiency (R2). a – Pareto plot elaborated with the results of tensile
efficiency (R2) for the factors F1 and F2 presented in Table 1. b – Surface response graph that describes the effect of length (F1) and volume (F2) of concrete reinforced with
fibers on the tensile efficiency (R2). Experiment performed with 2 factors, 1 block, 10 runs, residual MS = 2.915474.

Table 7
Regression coefficient values for the model that describes the increase in compressive strength (R1) as a function of factors F1 and F2.

Factor Regression coefficient Standard error t4 p Lower limit of the CI95% Upper limit of the CI95%
Mean/Interc. 35.336 8.37696 4.21818 0.013497 12.077 58.5937
F1 (L) 1.263 0.88038 1.43456 0.224728 3.707 1.1814
F1 (Q) 0.025 0.02745 0.90162 0.418242 0.051 0.1010
F2 (L) 168.425 35.56505 4.73568 0.009066 267.169 69.6802
F2 (Q) 334.375 68.62676 4.87237 0.008206 143.837 524.9134
F1  F2 1.300 1.46730 0.88598 0.425680 2.774 5.3739

F1: factor referring to the fiber length (mm). F2: factor referring to the percent volume of fibers (cm3). (L): condition of linear interaction for each factor. (Q): condition of
quadratic interaction for each factor.

of the factors used in the validation tests and the results obtained the theoretical values predicted by the models (Eqs. (4) and (5))
are presented in Table 9. when given the same input variables used in the assay (factor
As shown in Table 9, the results of R1 and R2 obtained experi- levels). The values of R1 and R2 obtained experimentally are within
mentally in the validation assays A, B, C and D were compared with the confidence intervals (CI95%) predicted by the models, confirm-
 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848 845

16

15

14

13

12
Predicted Values
11

10

4
3 4 5 6 7 8 9 10 11 12 13 14 15 16
Observed Values

Fig. 3. Results of R1 observed experimentally vs results of R1 estimated by the model of Eq. (4).

Table 8
Regression coefficient values for the model that describes the increase in tensile strength (R2) as a function of factors F1 and F2.

Factor Regression coefficient Standard error t4 p Lower limit of the CI95% Upper limit of the CI95%
Mean/Interc. 5.5495 9.74814 0.56929 0.599603 32.615 21.5156
F1 (L) 2.5075 1.02449 2.44757 0.070630 0.337 5.3519
F1 (Q) 0.0557 0.03194 1.74237 0.156404 0.144 0.0330
F2 (L) 29.8500 41.38648 0.72125 0.510660 144.757 85.0573
F2 (Q) 101.1160 79.85987 1.26617 0.274170 120.611 322.8426
F1  F2 2.0330 1.70748 1.19064 0.299624 6.774 2.7077

F1: factor referring to the fiber length (mm). F2: factor referring to the percent volume of fibers (cm3). (L): condition of linear interaction for each factor. (Q): condition of
quadratic interaction for each factor.

18

16

14
Predicted Values

12

10

4
4 6 8 10 12 14 16 18
Observed Values

Fig. 4. Tensile efficiency of the ecological concrete (R2) observed experimentally vs efficiency results of R2 estimated by the model of Eq. (5).
846 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848

Predicted vs. Residual Values Predicted vs. Residual Values

2 factors, 1 Blocks, 10 Runs; MS Residual=2,152975 2 factors, 1 Blocks, 10 Runs; MS Residual=2,915474
DV: R2
DV: R1
2,5
2,0

2,0
1,5
1,5

Raw Residuals
Raw Residuals 1,0
1,0

0,5
0,5

0,0 0,0

-0,5
-0,5

-1,0
-1,0
-1,5
-1,5
-2,0

-2,0
-2,5
4 5 6 7 8 9 10 11 12 13 14 15 16
4 6 8 10 12 14 16 18
Predicted Values Predicted Values

(a) (e)
Predicted vs. Residual Values Predicted vs. Residual Values
2 factors, 1 Blocks, 10 Runs; MS Residual=2,152975 2 factors, 1 Blocks, 10 Runs; MS Residual=2,915474
DV: R2
DV: R1
3
2,5

2,0
2

Studentized Del. Residuals

Studentized Del. Residuals

1,5

1,0 1

0,5
0
0,0

-0,5
-1
-1,0

-1,5 -2

-2,0
-3
-2,5

-3,0
-4
4 5 6 7 8 9 10 11 12 13 14 15 16
4 6 8 10 12 14 16 18
Predicted Values Predicted Values
(b) (f)

Normal Prob. Plot; Studentized Del. Residuals Normal Prob. Plot; Studentized Del. Residuals
2 factors, 1 Blocks, 10 Runs; MS Residual=2,152975 2 factors, 1 Blocks, 10 Runs; MS Residual=2,915474
DV: R2
DV: R1
3,0
3,0

2,5 2,5
,99 ,99
2,0 2,0
Expected Normal Value
Expected Normal Value

,95 ,95
1,5 1,5

1,0 1,0
,75 ,75
0,5 0,5
,55 ,55
0,0 0,0

-0,5 ,35 ,35

-0,5

-1,0 ,15 -1,0 ,15

-1,5 -1,5
,05 ,05
-2,0 -2,0
,01 ,01
-2,5 -2,5
-3,0
-3,0
-3,0 -2,5 -2,0 -1,5 -1,0 -0,5 0,0 0,5 1,0 1,5 2,0 2,5
-4 -3 -2 -1 0 1 2 3
Residual Residual

(c) (g)
Residuals vs. Case Numbers Residuals vs. Case Numbers
2 factors, 1 Blocks, 10 Runs; MS Residual=2,152975 2 factors, 1 Blocks, 10 Runs; MS Residual=2,915474
DV: R1 DV: R2
2,0 2,5

2,0
1,5
1,5
1,0
Raw Residuals

Raw Residuals

1,0
0,5
0,5

0,0 0,0

-0,5
-0,5
-1,0
-1,0
-1,5
-1,5
-2,0

-2,0 -2,5
0 1 2 3 4 5 6 7 8 9 10 11 0 1 2 3 4 5 6 7 8 9 10 11
Case Number Case Number

(d) (h)
Fig. 5. Residual analysis of the model presented in Eqs. (4) and (5). Homoscedasticity diagnosis of residuals for R1 and R2. a;e – Homoscedasticity diagnosis of the residuals. b;
f – Diagnosis of residual outliers. c;g – Normality of data referring to the residual of the statistical model. d;h – Independence diagnosis of the residuals.
 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848 847

Table 9
Data for the comparison between compressive (R1) and tensile efficiency values (R2) obtained for each validation assay and values predicted by the model.

Validation assay Factor levels Value obtained Value predicted by the model (CI95%)
experimentally
F1 F2 R1 R2 R1 R2
A 10 0.2 6.7 7.8 7.4 (3–12.2) 8 (2.5–13.5)
B 20 0.2 4.7 8.6 4.9 (0.1–9.6) 12.3 (6.8–17.8)
C 15 0.1 7.5 13.5 10.4 (5.7–15.1) 14.5 (9.0–20)
D 12 0.15 13.0 10.7 8.4 (4–13.1) 10.7 (5.2–16.2)

F1 F2 Desirability
20,0

1,

14,3
,5

R1
9,4
5,6 0,

4,5
0,0
22,0

6,2 11,4 16,6

1,

11,5

R2
,5

0,
0,0

Desirability
0,23

7,93 15,0 22,01 0,06 0,2 0,34

Fig. 6. Mathematical simulation of the behavior of the R1 and R2 results in function of the F1 and F2 values, using the desirability tool in the software StatisticÒ 8.0.

Table 10
Theoretical values for R1 and R2 and experimental values for R1, R2 and R3 obtained using assays with ecological concrete at the optimized condition with F1 of 22 mm and F2 of
0.06%.

Specification Obtaining resistance (MPa) Experimental values for R1 and Value predicted by the model (%)
R2 (%)
Cn Tn Fn R1 R2 R3 R1 R2
Ecological concrete 47 ± 1 4.3 ± 0.2 4.4 ± 0.3 17.5 22.86 4.76 13.4 (5.4–21.3) 19.2 (10–28.4)
Conventional concrete 40 ± 1 3.5 ± 0.3 4.2 ± 0.1

ing that there is no significant statistical difference at the level of
5% (p > 0.05) between data predicted by the model and that
obtained experimentally. Therefore the models described by Eqs.
(4) and (5) present robustness, predictability and can be used for
simulation in compressive and tensile tests when using concrete
containing fibers from PET bottles. For optimization of the R1
and R2 values, Eqs. (4) and (5) were used to simulate different val-
ues of F1 and F2, until obtaining maximum values of R1 and R2 as
shown in Fig. 6.
Mathematical simulation for the prediction of R1 and R2 was
conducted as a function of the F1 and F2 values present in Table 1.
The curve profiles of R1 and R2 presented in Fig. 6 indicate that the
values of F1 and F2 leading to the value of 1 on the desirability axis
(to the right in the figure) are the optimal points. As can be
observed in Fig. 6, using the models presented in this work (Eqs.
(4) and (5)) the maximum values of the predicted R1 and R2 are
14.3 and 16.6, respectively.
For the response variable R1, when analyzing the factor F1 it can
be observed that all the values tested resulted in responses within
the same range of significance (blue band), indicating that the fiber
size does not influence the increase in compressive efficiency of the
ecological concrete, as previously explained in Table 6. However,
848 E.L. Pereira et al. / Construction and Building Materials 149 (2017) 837–848
when analyzing factor F2 it can be observed that the value of R1 is
maximized when the value of factor F2 is 0.06% and decreases as
the value of F2 increases; this describes the linear effect shown
in Fig. 1a, where the smaller the value of F2 (L) the greater the
value of R1. After reaching the inflection point with a value of 0.2
for F2, it is noted that the increase in factor F2 promotes an
increase in the value of R1 which describes the quadratic interac-
tion shown in Fig. 1a, where the larger the value of F2 (Q) the
greater the value of R1. Fig. 6 also shows that fiber lengths close
to 22 mm can provide values of R1 near the optimal point. Because
PET bottle grinding is a parameter that influences the cost of the
project, where the larger the fiber size the lower the cost of milling,
the use of fibers with sizes close to 22 mm should be investigated.
For the response variable R2, both the F1 factor and the F2 fac-
tor were shown to be influential. As the value of F1 increases it is
possible to notice an increase in the value of R2 (Fig. 6), character-
izing the linear interaction F1 (L) shown in Fig. 2a. For F1 values
between 7.93 and 11.5 the R2 values found were below 6.2% and
above 11.5; all other F1 values tested resulted in responses (R2)
within the same significance range (blue range), showing that in
order to achieve an increase in the tensile efficiency of ecological
concrete, PET fibers with a minimum size of 11.5 mm should be
used. As the value of F2 increases it is possible to notice a reduction
in the value of R2 (Fig. 6), characterizing the linear interaction F2
(L) shown in Fig. 2a. For F2 values of 0.06%, the R2 values were pre-
dicted to be close to 16.4% and all other F1 values tested resulted in
responses (R2) within the same range of significance (blue range),
showing that in order to achieve an increase in tensile efficiency of
the ecological concrete, a percent volume of PET fibers close to
0.06% should be used. After simulation of the optimal values, the
desirability tool present in the software indicated that the values
of 0.06% and 22 mm for the factors F1 and F2, respectively, are
the values that allow for obtaining the highest values of R1 and
R2 possible to be found in ecological concrete using PET fibers. In
this context another experiment was carried out, molding test
specimens with fibers using the optimized F1 and F2 values. In
addition to the increase in compressive (R1) and tensile efficiency
(R2) of this ecological concrete, the increase in flexural strength
(R3) calculated using Eq. (3) was also verified.
As can be observed in Table 10, the mean values obtained
experimentally are within the interval predicted by the models
for the optimum point. Furthermore, it is noticed that the average
values obtained experimentally exceed the maximum values
expected in the simulation (Fig. 6); this indicates that not only
are they robust, but the models described by Eqs. (4) and (5) make
conservative predictions, since the value obtained experimentally
is always greater (greater efficiency) than the value predicted by
the model.
When analyzing the value of R3 an increase in the ability of the
concrete to withstand bending can be observed, which in relation
to reinforced concrete can contribute to the use of rebar.
4. Conclusion
The load capacity of an ecological concrete produced with PET
fibers is greater than the load capacity of conventional concrete.
This is because the insertion of fibers from PET bottles into con-
crete improved the mechanical properties of the ecological con-
crete in terms of compression, tension and bending.
The synergistic effect between the factors fiber volume and
fiber length were shown to influence the concrete tensile effi-
ciency. However, only the fiber volume factor influenced the com-
pressive efficiency of the ecological concrete.
The mathematical models generated in this work can be used to
predict values of compressive and tensile efficiency in function of
the operational parameters of length and volume of PET fibers,
which can stimulate the use of ecological concrete by construction
companies and contractors, since the results can be estimated and
the economic viability determined before the work.
References
[1] P. Asokan et al., Improvement of the mechanical properties of glass fibre
reinforced plastic waste powder filled concrete, Constr. Build. Mater. 24 (2010)
448–460.
[2] Brazilian Association of the PET Industry, Census of PET Recycling in Brazil,
2016. <>, 2016 (accessed on November 3, 2016).
[3] Associação Brasileira de Normas Técnicas, NBR 12655: Portland cement
concrete – Preparation, control and receipt – Procedure, Rio de Janeiro, 2016.
[4] Associação Brasileira de Normas Técnicas, NBR 5738: molding and curing of
concrete cylindrical specimens, Rio de Janeiro, 1994.
[5] Associação Brasileira de Normas Técnicas, NBR 5739/94: Concrete.
Compression test of specimens, Rio de Janeiro, 1994.
[6] Associação Brasileira de Normas Técnicas, NBR 7222/94: concreto.
Determination of tensile strength by diametrical compression, Rio de Janeiro,
1994.
[7] Associação Brasileira de Normas Técnicas, NBR NM 23/01: Portland cement
and other powdered materials – determination of the specific mass, Rio de
Janeiro, 2001.
[8] Associação Brasileira de Normas Técnicas, NBR NM 45/06: aggregates –
determination of unit mass and volume of voids, Rio de Janeiro, 2006.
[9] Associação Brasileira de Normas Técnicas, NBR NM 46/03: aggregates –
determination of fine material passing through the 75 lm sieve per wash, Rio
de Janeiro, 2003.
[10] Associação Brasileira de Normas Técnicas, NBR NM 52/02: determination of
specific mass and apparent specific mass, Rio de Janeiro, 2002.
[11] Associação Brasileira de Normas Técnicas, NBR NM 248/03: aggregates –
determination of the granulometric composition, Rio de Janeiro, 2003.
[12] B. Barros Neto, I.S. Scarmínio, R.E. Bruns, How to Do Experiments, Editora
Bookman, Porto Alegre, 2010, p. 413.
[13] E.G.P. Box, W.G.E. Hunter, J.S. Hunter, Statistics for Experimenters: An
Introduction to Design, Data Analysis, and Model Building, John Wiley, New
York, 1978.
[14] V. Calado, D.C. Montgomery, Experiment Planning, Editora e-papers, Rio de
Janeiro, 2003, p. 260.
[15] D.F. Ferreira, Sisvar: a computer statistical analysis system, Ciência e
Agrotecnologia 35 (6) (2011) 1039–1042.
[16] F. Fraternali, S. Padea, V.P. Berardi, Effects of recycled PET fibres on the
mechanical properties and seawater curing of Portland cement-based
concretes, Constr. Build. Mater. 61 (2014) 293–302.
[17] L. Gu, T. Ozbakkaloglu, Use of recycled plastics in concrete: a critical review,
Waste Manage. 51 (2016) 19–42.
[18] B.A. Herki, J.M. Khatib, E.M. Negim, Lightweight concrete made from waste
polystyrene and fly ash, World Appl. Sci. J. 21 (2013) 1356–1360.
[19] A.R. Khaloo, M. Dehestani, P. Rahmatabadi, Mechanical properties of concrete
containing a high volume of tire–rubber particles, Waste Manage. 28 (2008)
2472–2482.
[20] S. Kou et al., Properties of lightweight aggregate concrete prepared with PVC
granules derived from scraped PVC pipes, Waste Manage. 29 (2009) 621–628.
[21] P.R.L. Lima et al., Characterization and treatment of sisal fiber residues for
cement-based composite application, Rev. Bras. Eng. Agric. 34 (2014).
[22] X. Liu et al., Feasibility analysis on application of modified concrete contains
rubber powder of straddle type monorail train waste tire, Proc. Environ. Sci. 31
(2016) 804–811.
[23] M.L. Marques et al., Compatibility of vegetable fibers with Portland cement
and its relationship with the physical properties, Rev. Bras. Eng. Agric. 20
(2016) 466–472.
[24] G. Martínez-Barrera et al., Concrete reinforced with irradiated nylon fibers, J.
Mater. Res. 21 (2006) 484–491.
[25] P.K. Mehta, J.M. Monteiro, Concrete: Structure, Properties and Materials,
second ed., Editora PINI, São Paulo, 2008.
[26] F. Pelisser et al., Mechanical properties of recycled PET fibers in concrete,
Mater. Res. 15 (2012) 279–286.
[27] M.I. Rodrigues, A.F. Iemma, Planejamento de experimentos e otimização de
processos – uma estratégia sequencial de planejamentos, Casa do Pão Editora,
Campinas, 2005. 325 p..
[28] E.J. Silva et al., Compressive strength of mortars in function of the addition of
coconut fiber, Revista Brasileira de Engenharia Agrícola e Ambiental 18 (12)
(2014) 1268–1273.
[29] S. Song, S. Hwang, B.C. Sheu, Strength properties of nylon- and polypropylene-
fiber-reinforced concretes, Cem. Concr. Res. 35 (2003) 1546–1550.