Construction and Building Materials 141 (2017) 598–607

Contents lists available at ScienceDirect

Construction and Building Materials

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

Investigation of waste ceramic tile additive in hot mix asphalt using

fuzzy logic approach
Çağdasß Kara ⇑, Murat Karacasu
Department of Civil Engineering, Faculty of Engineering and Architecture, Eskisehir Osmangazi University, Campus of Batı Meselik, Odunpazari, Eskisehir, Turkey

h i g h l i g h t s

Effects of WCA replacement on performance properties of HMA are evaluated.

Asphalt specimens with WCA could improve some performance characteristics of HMA.

Marshall test results of specimens were successfully estimated by fuzzy logic model.

Paper contributes to the encouragement of WCA re-use in HMA environmentally friendly.

Using WCA in HMA is a one of the effective practices of sustainable environment.

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

Article history: The aim of this paper is to explore the effects of waste ceramic tile additives on performance properties of
Received 9 December 2016 asphalt mixtures. Within this context varied rates of waste ceramic tile aggregates (WCA) were used as
Received in revised form 27 February 2017 aggregate for preparation and experiments for hot mix asphalt (HMA) specimens by using Marshall
Accepted 4 March 2017
Design Method. Static and Dynamic Creep Tests were applied to the new 30 specimens prepared accord-
Available online 17 March 2017
ing to optimum bitumen contents (OBC) and all experimental results were evaluated. Additionally, effects
of two important variables (WCA ratio and bitumen ratio) on Marshall Stability (MS), on air void content
Keywords:
(AVC) and on voids of filled with asphalt cement (VFA) were modelled. It was concluded that, since spec-
Waste
Asphalt
imens with HMA have good enough mechanical conditions according to Turkish Highway Construction
Fuzzy logic Specifications in HMA for wearing course, up to 30% of natural aggregates can be replaced by WCA. By
Ceramic tile using WCA, country resource will be used effectively within the context of sustainable environment.
Recycling Besides experimental results were evaluated and predicted with high accuracy by using fuzzy logic
Pollution (FL) approach. Thanks to the FL model, we could predict values which are wanted to have knowledge
about materials characterization with less specimens and less tests without spending much time and
workforce.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Recently, several studies are conducted for using of wastes in

The HMA, which is used in highway construction, is composed
of bitumen and aggregate. These materials are well-known for
their high production cost. Reduction of quality aggregate and
bitumen resources day by day, has led researchers to investigate
on different sources to conserve material. The issue of environmen-
tal pollution caused by industrial development has led reusing and
recycling of the waste materials to be highly considered. Therefore,
reuse in HMA would reduce the environmental impact, prevent the
pollution as well as it would be an alternative for the natural
resources.
⇑ Corresponding author.
E-mail addresses: ckara@ogu.edu.tr (Ç. Kara), muratk@ogu.edu.tr (M. Karacasu).
Portland cement concrete and HMA mixtures [1–4]. Van de Ven
et. al., examined the partial replacement of industrial sand as filler
and waste ceramic (CW) obtained from electrical insulators as
coarse aggregate in HMA. In the study, the modified specimens
showed similar results to control specimens in Indirect Tensile
Strength (ITS) and Marshall Stability tests, whereas there was an
increase in compressive strength values [5].
Torkittikul and Chaipanich studied the feasibility of using cera-
mic tile waste and fly ash as fine aggregate to produce mortar and
Portland cement concrete. From the result it was seen that the
workability is reduced by the usage of WCA. Also, the compressive
strength in the fly ash concrete increased with increasing waste
ceramic tile content [6]. Pacheco-Torgal and Jalali investigated
the feasibility of using WCA as aggregates (as filler, fine and coarse)

http://dx.doi.org/10.1016/j.conbuildmat.2017.03.025
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
 Ç. Kara, M. Karacasu / Construction and Building Materials 141 (2017) 598–607
in traditional concrete. Results showed that replacement of WCA
increased the compressive strength value of concrete. And also
performance properties of concrete mixtures with WCA improved
against to control concrete mixtures concerning oxygen perme-
ability, capillary water absorption and chloride diffusion which
makes the concrete structures more durable [7].
Silvestre et al., analyzed the possibility of using WCA in HMA for
binder and wearing courses. It was observed that moisture suscep-
tibility, AVC, OBC and resistance of plastic deformation were
increased with increasing the waste ratio, although compressive
strength and MS were decreased. Up to 30%, by weight, addition
of RCA into HMA showed suitable results according to Spanish
Standards, the wearing course and binder course with RCA were
considered to be suitable for medium-low traffic volume roads
[8,9].
The production of ceramic tile in factories in Turkey leads to
approximately 8% of waste material. This rate is over 13% in the
factories engaged in first quality production. Ceramic tile produc-
tion in Bilecik-Eskisßehir-Kütahya regions make 43.2% of total cera-
mic tile production in Turkey [10]. Approximately 100,000 tons of
ceramic waste is produced in this region every year. The wastes
formed in the region can be converted to the aggregates at a low
cost. The aggregate obtained from the waste materials in the
region, cover a significant portion of the annual aggregate amount
used in road construction in Eskisßehir [11]. In some studies, the
effect of waste materials on performance properties of concrete
is modelled by FLA. [12–18]. Özgan modelled changing ductility
values of bitumen in various temperatures and waiting times by
fuzzy logic approach (FLA). By using this model, results showed
that ductility values depend on the temperature and waiting time
were able to be predicted with 96% accuracy [19]. Morova et al.,
investigated the effect of bitumen amount on strength of HMA
by FLA. Developed model was able to predict with 97.9% accuracy
[20]. Sayın and Tanyıldızı examined the stiffness of the HMA by
using fuzzy logic and Shell method. It was observed that, the
obtained results from fuzzy logic model have close results in com-
parison with calculated values by using Shell method [21]. WC is
produced at a tremendous rate and there exist lack of natural
resource for road construction all over the world. However, there
are only a few studies about WCA recycling in HMA. Besides paper
is the first study about that static and dynamic creep stiffness and
strain behavior of WCA in HMA is investigated and Marshall test
results of WCA in HMA are modelled by FLA. Within this context
varied rates of WCA, by weight, were used as aggregate for prepa-
ration of experiments for HMA specimens with Marshall Design
Method. Static and Dynamic Creep Tests were applied to the new
30 specimens prepared according to OBC and all experimental
results were evaluated. Finally, the effect of two important
variables (WCA ratio and bitumen ratio) on MS, AVC and VFA were
Fig. 1. Waste ceramic tile obtained from disposal area of Toprak Ceramic Factory.
599
100
Passing (%)
80
60
40
20
0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
Sieve size (mm)
lower limits upper limits Specimen
Fig. 2. Type II. aggregate grading curve for wearing course.
modelled by FLA. Predicted results obtained from fuzzy logic
model were compared with experimental results.
2. Materials
50/70 penetration bitumen was used in experiments. Bitumen
experiments are conducted and their compliance with specifica-
tions is explored. Natural aggregates and WCA were used in the
mixtures. WCA was obtained from plant waste disposal areas in
Eskisehir (Fig. 1). These wastes are ceramic tiles fractured ones
while glazing, packaging, transportation or these are caused by
manufacturing defects. According to chemical analysis of WCA
used in the experiments has small amount of glaze coat (approxi-
mately 5.5%, by weight).
The aggregate grading curve chosen as Type II wearing course
obtained from Turkish Highway Construction Specifications [22]
is shown in Fig. 2. Specimen grade values were obtained by taking
average values of the upper and lower limit values.
3. Methods and experimental study
Significant experiments were applied to natural and WCA and
bitumen which were used in the mixtures. Physical properties
and the required specifications related to bitumen and aggregates
are shown in Tables 1 and 2.
WCA were crumbled by a jaw crusher in laboratory and pre-
pared according to required Type II gradation as shown in Fig. 3.
Marshall Mix Design Method is used for HMA specimen’s prepa-
ration. Five different aggregate mixtures are prepared according to
the Type II gradation, including different ratios (0%, 10%, 20%, 30%
and 40%, by weight) of WCA. Seven different percentage of bitumen
(3.5%, 4%, 4.5%, 5%, 5.5%, 6%, and 6.5%, by weight) were added to the
mixtures. Three for each specimen’s combination were prepared;
in total 105 HMA specimens (ceramic percentage: 5 type, bitumen
percentage: 7 type and 3 for each specimen, respectively
5 * 7 * 3 = 105) were prepared. Marshall Stability & Flow Test was
600 Ç. Kara, M. Karacasu / Construction and Building Materials 141 (2017) 598–607

Table 1
Experimental results and specification limits related to aggregate.

Material’s properties Natural Aggregate Waste Ceramic Tile Standards Limits

Filler Fine Coarse Filler Fine Coarse
Specific gravity (gr/cm3) 2.731 2.723 2.71 2.58 2.50 2.31 TS3526 –
Water absorption (%) – 2.17 0.58 – 9.47 8.55 TS3526 <2
Los Angeles wearing test (%) 24.18 20.94 TS EN 1097-2 <30
Freezing and thawing test (%) 4.05 3.302 ASTM C 131 <10
TS 3694

Table 2 2500
Experimental results and specification limits related to bitumen. WCA 0%
2200
Material’s properties Results Standards Limits 1900 WCA 10%

MS, kg
Penetration (25 °C, 1/10 mm) 53.6 TS EN 1426 50–70 1600 WCA 20%
Softening point (°C) 50 TS EN 1427 46–54 1300 WCA 30%
Ductility (25 °C 5 cm/min) >100 TS 119 – 1000 WCA 40%
Flash point (°C) 322 TS EN ISO 2592 > 230 700
Loss on heating (%) 0.42 TS 121 <0.5
400
Specific gravity (gr/cm3) 1035 TS 1013 EN ISO 3675 – 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
Bitumen Rao, %
(a)

95 WCA 0%
85 WCA 10%
75
WCA 20%
VFA, %
65
55 WCA 30%
45 WCA 40%
35
25
3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
Bitumen Rao, %
(b)
2.50 WCA 0%
2.40 WCA 10%
PSG, gr/cm3

WCA 20%
2.30
WCA 30%
2.20 WCA 40%
2.10
Fig. 3. WCA crumbled and prepared according to required Type II gradation.
2.00
3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
Bitumen Rao, %
(c)
applied to the specimens. After that, Flow and Marshall Strength
values have been recorded. Practical Specific Gravity (PSG), MS, 20 WCA 0%
VFA, and AVC values were calculated for each CW ratio and graphs WCA 10%
15 WCA 20%
were drawn related to obtained values (Fig. 4a–d).
AVC, %

By taking average values of the following four, OBC values were 10 WCA 30%
determined for each waste ratio by the help of Fig. 4a–d (see in WCA 40%
Fig. 5). 5

0
1. BC corresponding to maximum MS, 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
2. BC corresponding to maximum PSG, Bitumen Rao, %
3. BC corresponding to the median of specification limits of VFA in (d)
the total mix (i.e. 4%),
4. BC corresponding to the median of specification limits of AVC in Fig. 4. Graphs related with Marshall Test results corresponding to bitumen and
the total mix (i.e. 70%). WCA rates; (a) MS, (b) VFA, (c) PSG, (d) AVC.

According to the Fig. 4a–d AVC and VMA, OBC values increase
with increasing waste ratio in HMA. VFA, MS and PSG values
decrease with increasing the waste ratio. Additionally, we can
say that, because ceramic aggregates are more porous than natural
aggregates, the specimens with ceramic ones are more porous. So,
especially PSG values of specimens with ceramic aggregates
decrease with increasing the waste ratio. According to determined
OBC values, the other Marshall parameters were recalculated for
each WCA ratio respectively. OBC values and obtained results are
shown together with the required limits of specifications from
General Directorate of Highways (GDH) in Table 3 [11,22]. By using
OBC values, 15 + 15 specimens were prepared for Dynamic and
Static Creep Stiffness Tests (3 specimens for each waste ratio)
(see in Fig. 6). The stiffness of asphalt-aggregate mixes has a para-
mount importance in determining how well a pavement performs
and it is fundamental to the analysis of pavement response to traf-
fic loading [23].
 Ç. Kara, M. Karacasu / Construction and Building Materials 141 (2017) 598–607 601

Opmum Bitumen Content (%)

4.91% 5.67% 6.48% 6.99% 8.59%

0% (CW) 10%
20%
(CW) 30%
(CW) 40%
(CW)
(CW)

Fig. 5. Optimum bitumen content.

Repeated (Dynamic) loading creep test, under square shape load

pulse, was conducted for specimens in accordance with the British
Standard DD 226. UTM-5P-14P was used to carry out dynamic load
at a certain periodic rate on to the asphalt samples and deforma-
tion values were measured by using a Linear Variable Displace- Fig. 6. Dynamic and static creep stiffness test machine.
ment Transducer (LVDTs) [24]. The test conditions are preloading
for 60 s at 100 kPa as a conditioning stress to obtain a proper bed-
ding of the specimen due to a shape that might not really be flat, Dynamic Creep Sffness

Creep Sffness(MPa)
and the test was conducted at 40 °C for one-hour loading time 320
and 500 kPa applied stress. The results were given as permanent
deformation after 1800 cycles and 1 h expressed in strain. 80
Periodically repeated loading consists of 1000 ms pulse width
and a 2000 ms pulse period. Dynamic load was applied to 15 spec- 20
imens for a period of 1 h and the deformation values of specimens
were measured by LVDTs. Accumulated permanent strain of spec- 5
imens were calculated by using Eq. (1): 0 500 1000 1500 2000 2500 3000 3500 4000
Pulse count
h
e¼ ð1Þ 0% 10 % 20 % 30 % 40 % (waste raos)
H0
Fig. 7. Calculated dynamic creep stiffness values related to pulse counts and waste
where, e is accumulated permanent strain, h is axial deformation ratios.
and H0 is initial height of specimen. Test was performed at temper-
ature of 40⁰ and 500 kPa stress [25]. Finally, graphs related to calcu-
Accumulated Strain
Accumulated Strain (με)

lated deformation and stiffness values were drawn averaging 3 test

80000
results of specimens which have same waste ratio as seen in Figs. 7
and 8 (3 specimens prepared for each waste ratio, in total 15 spec- 60000
imens were prepared).
40000
Considering the relationship related with creep stiffness and
accumulated strain between specimens with and without WCA, 20000
specimens with 10% WCA and 5.67% OBC have best creep stiffness
0
and strain results. After the point of direction change of curve, test
0 500 1000 1500 2000 2500 3000 3500 4000
was finished cause the stress led to specimen’s permanent defor- Pulse count
mation. Considering accumulated strain graphs at the beginning, 0% 10 % 20 % 30 % 40 % (waste raos)
approximately up to pulse of 1650, specimens with 20%, 30% and
40% had worse experimental results than specimens with 10% Fig. 8. Calculated accumulated strain values related to pulse counts and waste
ratios, under dynamic loading.
and 0%. However, after pulse counts of 1650, specimens with
WCA of 20% and 30%, showed better results than specimens with
0% WCA and had better resistance and creep stiffness values in along 4200 s to test specimens constantly in static test. In this
the long term as shown in Figs. 7 and 8. investigation static load was applied to specimens (3 specimens
After preloading of 100 kPa along 10 min, the only differences prepared for each waste ratio, in total 15 specimens) and deforma-
between dynamic and static stiffness tests is loading of 500 kPa tion values of specimens were measured by LVDTs. Creep stiffness

Table 3
Experimental results and HMA design criteria for wearing course (GDH, General Directorate of Highways, [22]).

WCA content (%) OBC (%) PSG (gr/cm3) MS AVC VFA (%) Flow VMA
(kg) (%) (mm) (%)
0 4.91 2.45 1761.6 3.1 77.8 2.49 14.19
10 5.67 2.38 1685.6 4 75.7 2.42 16.56
20 6.48 2.31 1484.1 4.77 73.71 2.96 18.43
30 6.99 2.20 1545.5 5.98 68.9 2.980 20.6
40 8.59 2.12 1387.9 7.4 70.2 4.32 24.02
Standard Limits
Min – – 900 3 65 2 14
Max – – – 5 75 4 –
602 Ç. Kara, M. Karacasu / Construction and Building Materials 141 (2017) 598–607

Stac Creep Sffness Table 4

100 The used expressions in the FL model.
Creep Sffness

90 INPUTS OUTPUTS
Bitumen rate, % Waste rate, % MS, kg VFA, % AVC, %
80
B1 W1 MS1 VFA1 AVC1
70 B2 W2 MS2 VFA2 AVC2
B3 W3 MS3 VFA3 AVC3
60 B4 W4 MS4 VFA4 AVC4
0 500 1000 1500 2000 2500 3000 3500 4000 4500 B5 W5 MS5 VFA5 AVC5
Pulse count B6 MS6 VFA6 AVC6
0% 10 % 20 % 30 % % 40 (waste raos) B7 MS7 VFA7 AVC7
MS8 VFA8 AVC8
Fig. 9. Calculated static creep stiffness values related to pulse counts and waste
ratios.

with 0%, 20% and 30% WCA were close to each other. The studies
were conducted before, it was realized that the test results
Accumulated Strain achieved by dynamic and static creep tests don’t have to have sim-
8000
ilar correlation with each other. In general, it is believed that the
Accumulated Strain

7000 higher bitumen contents provide the higher plastics flow suscepti-
bility. The highly plastic susceptibility may lead to the high perma-
6000 nent deformation, due to too much bitumen in the HMA that
causes the loss of internal friction between aggregate particles
5000 [25,26]. Yet the specimens with 10%, 20% and 30% have better
0 500 1000 1500 2000 2500 3000 3500 4000 4500 results under dynamic load because of higher bitumen absorption
Pulse count capacity and OBC of WCA. We can say that under dynamic loading,
0% 10 % 20 % 30 % 40 % (waste raos)
specimens with that waste ratios have better results at their own
Fig. 10. Calculated accumulated strain values related to pulse counts and waste OBC (5.67%, 6.48%, 6.99% respectively). Likewise, under static load-
ratios, under static loading. ing, higher waste and higher bitumen content together provide
mixtures high deformation resistance.
and accumulated strain values of specimens were calculated and
by taking average of 3 test results, creep stiffness and accumulated 3.1. Fuzzy logic
strain graphs were drawn as shown in Figs. 9 and 10.
According to graphs related to static creep test, we can say that Fuzzy sources are incomplete and uncertain information
specimens with 10% WCA have worst stiffness and accumulated resources appearing in different ways. FLA aims to carry out
strain values. However, specimens with 40% WCA have the best numerical predictions for complex situations based on some
creep stiffness results, so have even better results than specimens assumptions [20]. FLA was firstly suggested by Lotfi A. Zadeh in
with 0% WCA. In addition to this test result values of specimens 1967. This method helps to achieve and identify uncertain sets or

Fuzzy
base
rule

Input Fuzzy Output

Fuzzificaon inference Defuzzificaon
Data data
engine

Fig. 11. A Typical architecture of fuzzy logic process.

Fig. 12. General structure of evaluated model.

 Ç. Kara, M. Karacasu / Construction and Building Materials 141 (2017) 598–607 603

Fig. 13. Membership functions related to; (a) bitumenrate, (b) wasterate, (c) MS, (d) VFA, (e) AVC.
604 Ç. Kara, M. Karacasu / Construction and Building Materials 141 (2017) 598–607

Fig. 14. Formulated rules

Fig. 15. Surface maps related to; (a) MS, (b) VFA, (c) AVC depend on bitumen rate and waste rate
 Ç. Kara, M. Karacasu / Construction and Building Materials 141 (2017) 598–607 605

suspicious concepts. Due to these aspects, fuzzy set theory has
been developed quickly from that times to the present days and
its practical applications have been carried out successfully. There-
fore the main feature of this fuzzy set theory is intuitive and natu-
ral [27].
While setting the membership functions in application of FL
approach, different types of functions can be used at the same time
(such as triangular, trapezoidal, Gaussian and Cauchy etc.). The
membership function of a fuzzy set is a generalization of the indi-
cator function in set elements. This function takes any number
between 0 and 1 [19]. A FL model is composed of fuzzifier, infer-
ence engine fuzzy base rule and defuzzifier as shown in Fig. 11.
Moreover, input and output data could be added to the system.
The fuzzifier decides how the required input will be converted into
a fuzzy logic input to be used by the inference engine.
Fuzzy rule base contains rules that include all possible fuzzy
relation between inputs and outputs. These rules are expressed
in the If-Then format. In the fuzzy approach, there are no mathe-
matical equations and model parameters. All the uncertainties,
model complications and nonlinear relationships are included in
the fuzzy inference procedure in the form of If-Then conditions
[28,29]. The inference engine decides how to process the inputs
from the fuzzifier and applies reasoning to compute fuzzy output.
Also, it displays the results of input data and if-then condition.
Defuzzifier converts the outputs into precise values.
4. Evaluation of Marshall experimental results with FL
MATLAB software was used to evaluate the experimental
results and the Mamdani-type Fuzzy Logic Inference System (FIS)
was used. In this study, the effect of WCA ratio and bitumen ratio
on MS, AVC and VFA were modelled by FL approach. For each waste
and bitumen ratio 3 specimens each, in total 105 specimens were
prepared. After taking the average of each 3 same conditioned
specimens, 175 data points were used for the FL model. Fig. 12
shows the general structure of the evaluated model.
The Mamdani method was used in this paper [30,31]. Based on
the past experience, the triangular functional form was chosen.
Shortcuts were used in the FL model as shown in Table 4.
The WCA ratio and bitumen ratio were used as input and MS,
AVC and VFA values were used as output data. The membership
functions are as shown in Fig. 13a–e.
The precision of homogeneity in membership functions were
investigated and obtained acceptable base value results and ranges
are used by trial and error method. During the process, the base
values are divided into equal intervals and the if-then rules were
formulated.
In total 35 if-then rules were used for the developed model. An
example of a rule is shown below;
If (bitumenrate is B1) and (wasterate is W1) then (MS is M7)
(VFA is VFA4) (AVC is AVC4) (1).
In the expression mentioned above, bitumenrate and wasterate
were defined by numerical values and were grouped in verbal
expression in the form of B1, B2. . .B7 and W1, W2. . .W5. In this
model, bitumen rate and waste rate rules are defined as input in
‘‘if” part and the other parameters (MS, VFA and AVC) located in
‘‘Then” part are defined as output. Some formulated rules of this
model and surface maps are shown in Figs. 14 and 15a–c.
Finally, experimental results and predicted results obtained
from FL models are shown in Table 5.

Table 5
Experimental results and predicted results obtained from FL models.

Bitumen Ceramic Ratio (%) Experimental Predicted

Ratio (%)
MS (kg) VFA (%) AVC (%) MS (kg) VFA (%) AVC (%)
3.5 0 1980.77 50.98 7.49 2004.50 55.10 8.53
3.5 10 1380.20 41.67 10.47 1327.25 45.30 10.98
3.5 20 1027.77 35.29 13.20 1106.75 34.80 13.43
3.5 30 807.36 30.74 15.64 870.50 34.80 16.05
3.5 40 810.91 28.51 16.99 870.50 27.80 16.05
4 0 2121.94 62.44 5.41 2162.00 64.90 3.45
4 10 1455.78 48.12 9.24 1547.75 45.30 8.53
4 20 1121.51 40.38 12.15 1106.75 45.30 10.98
4 30 977.74 35.50 14.44 870.50 34.80 13.43
4 40 958.07 33.02 15.75 870.50 34.80 16.05
4.5 0 1950.62 71.09 4.12 2004.50 74.70 3.45
4.5 10 1655.90 55.32 7.82 1547.75 55.10 8.53
4.5 20 1275.84 46.45 10.72 1327.25 45.30 10.98
4.5 30 1173.19 40.48 13.19 1106.75 45.30 13.43
4.5 40 1054.66 37.15 14.78 1106.75 34.80 16.05
5 0 1761.09 78.74 3.04 1768.25 74.70 3.45
5 10 1690.44 64.06 6.08 1768.25 64.90 6.08
5 20 1446.52 53.69 8.98 1547.75 55.10 8.53
5 30 1234.54 45.05 12.16 1327.25 45.30 10.98
5 40 1250.03 41.24 13.83 1327.25 45.30 13.43
5.5 0 1380.60 85.71 2.06 1327.25 85.20 1.70
5.5 10 1716.42 74.58 4.10 1768.25 74.70 3.45
5.5 20 1500.91 59.64 7.76 1547.75 55.10 8.53
5.5 30 1387.25 50.72 10.71 1327.25 55.10 10.98
5.5 40 1335.44 45.72 12.70 1327.25 45.30 13.43
6 0 1158.10 89.06 1.64 1106.75 85.20 1.70
6 10 1708.17 83.02 2.69 1768.25 85.20 3.45
6 20 1549.74 67.61 6.04 1547.75 64.90 6.08
6 30 1496.46 56.00 9.47 1547.75 55.10 8.53
6 40 1376.58 50.26 11.57 1327.25 55.10 10.98
6.5 0 982.30 90.89 1.44 870.50 92.20 1.70
6.5 10 1431.20 87.29 2.06 1327.25 85.20 1.70
6.5 20 1482.11 73.89 4.83 1547.75 74.70 6.08
6.5 30 1562.45 62.80 7.80 1547.75 64.90 8.53
6.5 40 1429.39 53.65 10.90 1327.25 55.10 10.98
606 Ç. Kara, M. Karacasu / Construction and Building Materials 141 (2017) 598–607

5. Results & discussion

Based on the experimental results, the following could be

specified:

- Having looked at experimental results related to aggregate, in

comparison to conventional aggregate, recycled WCA present
15.5% more wearing resistance and 22.7% freezing and thawing
resistance. Also WCA, natural aggregate and bitumen tested
could be used in the mixtures as shown in Tables 1 and 2.
- As a result of the porous structure of the ceramic tile, adding
WCA present an increase of AVC and VMA, OBC values, a
decrease of VFA, MS and PSG values.
- Also according to the results, specimen which contains 30%
WCA, by weight, was lighter (11%) than the conventional spec-
imen in equal volume.
- Although there is no specification about dynamic and static
creep test results, it is possible to make comparison between
test results and specimens with 0% WCA. Specimens with 20%,
(a) 30% and %40 have worse results for dynamic creep stiffness test
than specimens with 0% and 10% in the beginning of the test,
yet with the passing of time, specimens with 20% and 30% show
better results than specimens with 0% WCA, that is to say have
better resistance and creep stiffness values in the long term.
According to static creep test graphs, specimens with 10%
WCA have worst results. Yet, specimens with 40% WCA have
the best creep stiffness and strain results. In addition to this test
result values of specimens with 0%, 20% and 30% WCA were
close to each other. Finally, we can say that under dynamic
loading, specimens with waste ratios demonstrate better
results at their own OBC (5.67%, 6.48%, 6.99% respectively).
Likewise, under static loading, higher waste and higher bitumen
content provide high deformation resistance to mixtures.

According to values of Table 5, comparison graphs related to

relationships between the predicted values and experimental
results were drawn by MATLAB as shown in Fig. 16a–c.
Slopes of the line (optimum = 45⁰) and R2 (optimum = 1) values
could give us information about relationship between the pre-
(b)
dicted values and experimental results. Drawn graphs with linear
equation have highly regression (R2: AVC = 97.26%, VFA = 97.77%,
MS = 95.99%) with predicted and experimental values. From slopes
of the linear equations, we can say that AVC, VFA and MS values are
predicted with 98,19%, 97.6%, 92,38% accuracies, respectively.

6. Conclusions

It is concluded that WCA, natural aggregate and bitumen tested

could be used in the HMA according to specification limits from
GDH.
In the long term, specimens with 20% and 30% show better
deformation resistance and dynamic creep stiffness values than
specimens with 0% WCA and specimens with 40% WCA have the
best static creep stiffness and strain results.
Finally, according to the all test results in HMA, up to 30% of
natural aggregates can be replaced by WCA for wearing course
because of presenting good enough mechanical characteristics in
compliance with Turkish Highway Construction Specifications.
(c) AVC value of HMA with 30% WCA exceed max specification limit.
However, it is only value that exceed specification limit. According
Fig. 16. Relationship between the predicted values and experimental results to mixtures test like MS tests, Dynamic and Static Creep stiffness
depend on; (a) AVC, (b) VFA, (c) MS.
tests applied to the asphalt specimens with 30%, it was observed
 Ç. Kara, M. Karacasu / Construction and Building Materials 141 (2017) 598–607
that this condition hasn’t had bad influence on performance prop-
erties of HMA. In the light of these observations it was thought that
exceeding of it could be ignored. So it is suggestable that speci-
mens which have up to 30% WCA could be used in the mixtures
used on expressway because of having good enough performance
properties. Using higher ratio from 30% WCA could make perfor-
mance properties of HMA worse, because more OBC and particle
count of glazed WCA decrease friction of surface resistance. And
also higher OBC increases cost of HMA. Recommended usable max-
imum WCA ratio in HMA is 30% and specimens with this ratio have
small amount (approximately 1.6%) glaze coat by weight. Addition-
ally, it was observed that when the ceramic tile was crumbled,
glaze coat has lost slickness feature substantially.
This paper contributes to the encouragement of WCA re-use in
pavement constructions. Using WCA, country resource will be used
effectively within the context of sustainable environment. This
study is a comprehensive experimental-based study that about
using WCA in HMA. Additionally, experimental results are sup-
ported by part of fuzzy logic modeling. Also intermediate and
extreme values of experimental results could be predicted high
accuracy by the FL approach without the need to perform many
tests. So thanks to the evaluated FL model, we can predict and
obtain results which are wanted to have knowledge about materi-
als characterization with less specimens and less tests, without
spending much time and workforce. In future work, the results
could be evaluated and compared by using methods like Artificial
Neural Network, Adaptive Neuro-Fuzzy Inference Systems,
Regression.
Acknowledgment
This research did not receive any specific grant from funding
agencies in the public, commercial, or not-for-profit sectors.
References
[1] M. Behera, S.K. Bhattacharyya, A.K. Minocha, R. Deoliya, S. Maiti, Construction
and Building Materials 68 (2014) 501–516.
[2] R.V. Silva, J. de Brito, R.K. Dhir, Construction and Building Materials 65 (2014)
201–217.
[3] M. Arabani, S.M. Mirabdolazimi, Materials Science and Engineering: A 528
(2011) 3866–3870.
607
[4] A. Modarres, M. Rahmanzadeh, Construction and Building Materials 66 (2014)
476–483.
[5] M. Van de Ven, A. Molenaar, M. Poot, Asphalt mixtures with waste materials:
possibilities and constraints, in: Conference on Asphalt Pavements for
Southern Africa [CAPSA11], 10th, 2011, KwaZulu-Natal, South Africa, 2011.
[6] P. Torkittikul, A. Chaipanich, Cement and Concrete Composites 32 (2010) 440–
449.
[7] F. Pacheco-Torgal, S. Jalali, Construction and Building Materials 24 (2010) 832–
838.
[8] R. Silvestre, E. Medel, A. García, J. Navas, Materials & Design 50 (2013) 471–
478.
[9] R. Silvestre, E. Medel, A. García, J. Navas, Construction and Building Materials
45 (2013) 115–122.
[10] G. Saatcioglu, Turkish Ceramic Industry Report, in: T.C. Federation (Ed.)
_
Istanbul, 2010.
[11] Ç. Kara, Fayans Atıklarının Bitümlü Sıcak Karısßımların Performans
Özelliklerine Etkisi, Civil Engineering, vol Master, Eskisßehir Osmangazi
Üniversitesi, 2012.
[12] F. Özcan, C.D. Atisß, O. Karahan, E. Uncuoğlu, H. Tanyildizi, Advances in
Engineering Software 40 (2009) 856–863.
[13] M. Sarıdemir, Advances in Engineering Software 40 (2009) 350–355.
[14] M. Sarıdemir, Advances in Engineering Software 40 (2009) 920–927.
_
[15] M. Sarıdemir, I.B. Topçu, F. Özcan, M.H. Severcan, Construction and Building
Materials 23 (2009) 1279–1286.
_ Topçu, C. Karakurt, M. Sarıdemir, Materials & Design 29 (2008) 1986–1991.
[16] I.B.
_ Topçu, M. Sarıdemir, Computational Materials Science 42 (2008) 74–82.
[17] I.B.
_ Topçu, M. Sarıdemir, Construction and Building Materials 22 (2008) 532–
[18] I.B.
540.
[19] E. Özgan, Journal of the Faculty of Engineering & Architecture of Gazi
University 24 (2009).
[20] N. Morova, S. Serin, S. Terzi, Bitüm Miktarının Asfalt Betonu Dayanımına
Etkisinin Bulanık Mantık Yaklasßımıyla Değerlendirilmesi, 6th International
Advanced Technologies Symposium (IATS’11), 2011.
[21] E. Sayin, H. Tanyildizi, Gazi Üniversitesi Mühendislik-Mimarlık Fakültesi
Dergisi 21 (2006).
[22] Anonymous, General Directorate of Highways, Ankara, 2013.
[23] A.A. Tayebali, B.-W. Tsai, C.L. Monismith, Stiffness of Asphalt-Aggregate Mixes,
Strategic Highway Research Program, National Research Council Washington
DC, 1994.
[24] S. Mirzapour Mounes, M.R. Karim, A. Khodaii, M.H. Almasi, Geotex. Geomembr.
44 (2016) 109–116.
[25] T. Baghaee Moghaddam, M. Soltani, M.R. Karim, Construction and Building
Materials 65 (2014) 487–494.
[26] S. Tayfur, H. Ozen, A. Aksoy, Construction and Building Materials 21 (2007)
328–337.
[27] S. Subasßı, A. Beycioğlu, M. Emiroğlu, Bilimde Modern Yöntemler Sempozyumu-
Bmys, 2008.
[28] S. Akkurt, G. Tayfur, S. Can, Cement and Concrete Research 34 (2004) 1429–
1433.
[29] F. Demir, Cement and Concrete Research 35 (2005) 1531–1538.
[30] E.H. Mamdani, S. Assilian, International Journal of Man-Machine Studies 7
(1975) 1–13.
[31] L.A. Zadeh, Systems, IEEE Transactions on, Man and Cybernetics, 1973, pp.
28–44.