1 s2.0 S2090447921001623 Main
1 s2.0 S2090447921001623 Main
1 s2.0 S2090447921001623 Main
net/publication/351266278
CITATIONS READS
12 178
6 authors, including:
All content following this page was uploaded by Nur Izzi Md Yusoff on 02 May 2021.
Civil Engineering
a r t i c l e i n f o a b s t r a c t
Article history: This study seeks to determine the rheological properties of unaged and RTFO-aged bio-asphalt binders
Received 28 November 2019 using experimental and modelling methods. Crude palm oil (CPO) was used as a bio-oil at varying per-
Revised 8 May 2020 centages of 0, 5, 10 and 15% by total weight of asphalt binder. The dynamic shear rheometer (DSR)
Accepted 10 April 2021
was used to investigate the rheological properties of bio-asphalt binders. The multilayer feed-forward
Available online xxxx
neural network method was used to predict the complex modulus and phase angle of bio-asphalt binders
by virtue of its ability to learn and adapt. Result of the DSR analysis showed that the complex modulus of
Keywords:
bio-asphalt with 5% CPO is almost similar as that of the base asphalt binder, and that higher CPO content
Bio-asphalt
Crude palm oil
resulted in reduced complex modulus and higher phase angle. Result of the modelling shows that all
Rheological properties models have an R2 value greater than 0.99, thus indicating the good agreement between the predicted
Artificial neural network and the experimental results.
Ó 2021 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams Uni-
versity. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/
by-nc-nd/4.0/).
1. Introduction als have achieved a certain degree of success in using the biomate-
rials as an extender. A small number of researchers have
The global increase in the construction of new roads and main- investigated the use of crude palm oil (CPO) in bio-asphalt binders.
tenance and rehabilitation of existing roads require a huge amount Due to its wide availability, the use of CPO as an extender for base
of asphalt binder produced by crude oil refining [1]. The govern- asphalt binders is more cost-effective in contrast the use of other
ment, researchers and asphalt binder industry are always looking types of bio-oils. A total of 64.5 million metric tonnes of crude
for alternative materials that have the ability to improve the prop- palm oil was produced globally between 2016 and 2017 [12].
erties of asphalt road pavements whilst reducing their carbon foot- The production of palm oil in the international market is much
print [2]. Biotechnology is considered to be one of the key higher than any other vegetable oils [13]. Thus, there is an urgent
alternatives for a wide range of technologies [3,4]. Its utilisation necessity to develop the technologies for benefiting from this large
in the field of pavement engineering has been receiving serious quantity of palm oil in various biofuel applications, including its
attention due to its eco-friendliness, cost effectiveness, renewabil- use in bio-binders.
ity and energy efficiency in comparison to petroleum based asphalt A number of recent studies have investigated the rheological
binders [5–11]. Several experimental trials have been carried out properties of various bio-binder materials. Wang et al. (2018)
to determine the feasibility of using biomaterials as a partial or examined the chemical and rheological properties of bio-binders
total replacement for base asphalt binder. These experimental tri- by incorporating waste cooking oil (WCO) and styrene–butadienes
tyrene (SBS) as modifiers. The researchers observed that higher
⇑ Corresponding author. bio-oil content resulted in a marked improvement of the fatigue
E-mail address: abdulnaser_17005477@utp.edu.my (A.M Al-Sabaeei). performance of bio-asphalt at intermediate temperature despite
Peer review under responsibility of Ain Shams University. the reduced rutting resistance. It was concluded that WCO is a
promising bio-oil material for the production of a sustainable
asphalt binder [14]. Sun et al. (2016) reported that WCO has the
ability to significantly improve the fatigue and cracking resistance
Production and hosting by Elsevier properties of asphalt binders in spite of the reduced rutting
https://doi.org/10.1016/j.asej.2021.04.003
2090-4479/Ó 2021 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article as: A.M. Al-Sabaeei, M.B. Napiah, M.H. Sutanto et al., Determination of rheological properties of bio-asphalt binders through exper-
imental and multilayer feed-forward neural network methods, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2021.04.003
A.M. Al-Sabaeei, M.B. Napiah, M.H. Sutanto et al. Ain Shams Engineering Journal xxx (xxxx) xxx
resistance of bio-asphalts with higher WCO content [15]. The bio- nanosilica (NS) on the rheological performance of polymer modi-
binder derived from swine manure was used as a modifier in the fied binder (PMB), where ANN was used to predict the |G*| and d
investigation of rheological properties of bio-asphalt. Results of nanosilica-polymer modified binders (NS-PMBs) at different
showed that swine manure has the ability to enhance the low tem- temperatures, loading frequencies, and nanosilica concentrations.
perature properties of asphalt binder although the high tempera- The study employed a feed forward back propagation learning
ture grade decreased with higher bio-oil content [16]. It is technique. The LM and gradient descent (GDA) algorithms were
believed that bio-asphalt and base asphalt binder share some sim- used in the training stage. Varying number of neurons were used
ilarities with regard to their chemical and rheological properties, for each algorithm and the optimal number of neurons was identi-
and thus bio-oil is regarded as one of the appropriate modifiers fied based on maximum efficiency and minimum MSE during the
for improving the properties of asphalt binder used in the con- testing stage. It was concluded that ANN is one of the methods
struction of asphalt pavements [1,17,18]. Bio-binder is very effec- with the highest accuracy for modeling and predicting the |G*|
tive in reducing the viscosity of asphalt binders at high and d of NS-PMB.
temperatures, and this leads to better workability of asphalt mix- The testing of rheological properties of asphalt binder is expen-
tures and better energy efficiency [5,17,19]. Chen et al. (2014) sive, time consuming and requires skilled operators [23,31]. Fur-
studied the ability of waste edible oil to act as a rejuvenator of thermore, the use of bio-binder as a partial replacement for
the empirical and rheological characteristics of aged asphalt bin- petroleum-based binder can complicate engineering properties
ders modified with SBS. The findings of their study indicate that because of the different and complicated chemical component of
the improvement in the empirical and rheological properties of biomass sources [9]. As a consequence, a more in-depth knowledge
aged asphalt binder correspond with the dose of base binder reju- is required to evaluate the effect of replacing asphalt binder with
venator. The incorporation of SBS also resulted in enhanced rutting bio-binder through experimental investigation and modelling of
resistance of the rejuvenated asphalt binder. However, these rheological properties [33]. In this regard, ANN modeling could
improvements were observed for high-temperature properties be a valuable alternative tool for quantifying the rheological prop-
and thus there is a need to find ways to improve the low- erties of asphalts [24] and bio-asphalt binders. This study consists
temperature properties of the binders [20]. of two main parts. The first part of this study evaluates the effect of
Alamawi et al. (2019) investigated the use of palm kernel oil crude palm oil at varying concentrations of 0, 5, 10 and 15% by
polyol bio-based binder as a partial replacement for petroleum- weight of a 60/70 penetration grade binder before and after
based binder. The temperature sensitivity of bio-asphalt binders short-term aging. The conventional properties, namely the pene-
is reported to be very close to that of base asphalt binders even tration, ring and ball temperature, penetration index and high-
though there is a small loss of mass [21]. Rahman et al. (2017) temperature rheological properties, of the binders were investi-
investigated the empirical and rheological properties of asphalt gated. The second part uses the multilayer feed-forward neural
binders modified with WCO, crumb rubber and palm oil fuel ash network approach to develop new and efficient models for predict-
(POFA). The incorporation of 15% modifier as a replacement for ing the rheological properties of CPO modified asphalt binder in
base asphalt binder produced equal or better stability, flow and terms of |G*| and d.
rutting resistance performance [22]. Several models have been
developed for predicting the rheological properties of base asphalt
binders [23–25]. Even though very little research has been carried 2. Materials and methods
out on the modeling of rheological properties of bio-asphalt bin-
ders, extensive experimental studies have been carried out to 2.1. Materials and sample preparation
determine the rheological behavior of bio-asphalt binders and
the feasibility of using these binders in pavement applications. This study used a 60/70 penetration grade asphalt binder sup-
Artificial neural network (ANN) is a popular computational tool plied by Petronas Research and Scientific Services (PRSS), Malaysia.
for solving complex problems in preference to conventional tech- The conventional and rheological characteristics of the base
niques, and has been used to solve several civil engineering prob- asphalt binder were established based on the ASTM and AASHTO
lems that cannot be easily solved using conventional techniques standards shown in Table 1. The crude palm oil (CPO) was supplied
and engineering mechanics approaches. It is a very effective tech- by Felcra Processing and Engineering Sdn. Bhd, Perak, Malaysia.
nique for engineering applications as it has an acceptable degree of Table 2 presents the conventional and chemical characteristics of
accuracy, particularly for civil engineering problems [24]. In recent the CPO. Crude palm oil was added as an extender at 0, 5, 10,
years, many studies have used ANN to develop prediction models and 15% by weight of asphalt binder.
for asphalt binders and mixtures for application in pavement engi- The base asphalt binder was heated in a 160 °C oven for 60 min
neering [27–30]. Kok et al. [31] used ANN to develop a model for to make it sufficiently fluid for mixing. A homogeneous blend of
predicting the |G*| of conventional and SBS modified asphalt bin- bio-asphalt binders and CPO was obtained by mixing in a high
ders. The study used a 50/70 penetration grade asphalt binder, shear mixer at 1000 rpm mixing speed for 1 h at a temperature
and the data for the |G*| and d of conventional and SBS-modified
asphalt binder were obtained from experimental works which
used the DSR machine. The inputs for the ANN layers are testing Table1
Conventional and rheological properties of base asphalt binder.
temperature, load frequency and SBS contents, whereas the output
layer parameter is the complex modulus. The Levenberg Mar- Test Test Requirements Results
quardt (LM), scaled conjugate gradient and Pola Ribiere conjugate standard
gradient algorithms were used as variants of the algorithm. A tan- Penetration, at 25 °C, 0.1 mm ASTM D5 Min. 60 dmm 60
gent sigmoid function was also used for the output and hidden lay- Ring and Ball Temperature, °C ASTM D36 Min. 49 °C 49
ers. The statistical indicators used to compare the measured values Ductility at 25 °C, cm ASTM D113 Min.100 cm ˃100
Penetration Index, PI – – 1.04
with the predicted values to validate the model are root mean Mass loss, % ASTM D113 Max. 1% 0.12
square (RMS), coefficient of variation (COV), and correlation coeffi- G*/sind, kPa, at 64 °C and 10 rad/ ASTM Min. 1 kPa 1.03
cient (R2). The LM algorithm was found to be the most appropriate sec D7175-15
topology with a mean R value of 0.997, mean RMS value of 0.0039, RTFO - G*/sind, kPa, at 64 °C and ASTM Min. 2.2 kPa 3.2
10 rad/sec D7175-15
and COV mean value of 20.24. Dhawo [32] evaluated the effects of
2
A.M. Al-Sabaeei, M.B. Napiah, M.H. Sutanto et al. Ain Shams Engineering Journal xxx (xxxx) xxx
Table 2
Physicochemical properties of crude palm oil (CPO) and its components [34].
Characteristics Value
Appearance Deep orange-red color
Density at 40 °C 0.899–0.920
Softening point, °C 33–40
Refractive index at 50 °C 1.449–1.455
Acid value, mg KOH/ g oil 2–15
Typical solid fat content at 40 °C 1 – 6%
Carotene content 0.03 – 0.15%
Iodine value 50.0 – 55.0
Saponification value, mg KOH/ g oil 190–209
Nitrogen, % 0.41
Carbon, % 76.44
Hydrogen, % 13.14
Sulfur, % 0.019
Fig. 2. Phase angle of unaged bio-asphalt binders versus CPO content at various
Table 3 temperatures.
Conventional properties of base and bio-asphalt binders.
The penetration grade and ring and ball temperature tests are
the most frequently used conventional tests for determining the
Fig. 3. |G*| of RTFO-aged bio-asphalt binders versus CPO content at various
consistency of base and bio-asphalt binders [36]. The penetration temperatures.
index (PI) was calculated and analyzed using the data obtained
from the penetration and softening point tests [34]. The penetra-
tion grade test was conducted at 25 °C according to ASTM D5
[37] to determine the hardness and consistency of the asphalt bin-
der before and after modification with CPO. A 1 mm-diameter nee-
dle under a load of 100 g was dropped for 5 s into the asphalt
binder specimen. The average of three readings for each sample
was recorded to the nearest whole unit. The ring and ball temper-
ature test of the base and bio-asphalt binders was carried out
according to ASTM D36 [38] to establish the binders’ consistency
Fig. 4. Phase angle of RTFO-aged bio-asphalt binders versus CPO content at various
temperatures.
3
A.M. Al-Sabaeei, M.B. Napiah, M.H. Sutanto et al. Ain Shams Engineering Journal xxx (xxxx) xxx
Fig. 5. |G*| and d of ANN architecture for unaged and RTFO-aged binders.
and d of the base and bio-asphalt binders were determined for the number of neurons was determined based on the maximum effi-
concentration of 0 to 15% CPO by weight of base asphalt binder. ciency and minimum MSE during the testing phase.
The performance of the model was established based on the val-
2.5. Multilayer feed-forward neural network ues of R2 and MSE. The R2 and MSE are expressed as follows:
3. Results and discussion asphalt binder modified with up to 5% of CPO, although the incor-
poration of higher percentages of CPO resulted in reduced |G*| and
3.1. Conventional characteristics larger d. This is most probably due to the higher aromatics content
in bio-oil compared to the content in the base binder [33].
Table 3 presents the conventional properties of bio-asphalt bin- The incorporation of between 5 and 10% CPO caused a higher
ders. The incorporation of higher amounts of CPO produced higher reduction in |G*| and higher increase in d, although there is a slight
penetration value and lower softening point and penetration index reduction in |G*| and a small increase in d with higher CPO content.
values. However, the incorporation of 5% CPO resulted in a very Generally, higher temperature resulted in lower |G*| and higher d.
small change in the conventional properties when compared to The binder with 5% CPO has the same |G*| value as the control
those of the base binder. Even though lower percentages of CPO asphalt binder although the value increased at intermediate tem-
can be used as a partial replacement for the base binder, the incor- perature. This indicates that CPO is a good bio-oil alternative for
poration of higher percentages of CPO without a polymer would the partial replacement for petroleum-based binder without
most probably have an adverse impact on the binders’ physical changing the binder’s rutting and fatigue resistance properties.
properties, especially in tropical countries where pavement tem- Higher test temperatures produced higher phase angle although
perature is high. Based on the specifications, the binder modified the phase angle for the 5% CPO asphalt binders decreased at inter-
with 5% CPO has a penetration value of 67 dmm, which is almost mediate temperature compared to that of the control binder. This
identical with the 60 dmm penetration value of the base binder; indicates that the addition of up to 5% crude palm oil improves
this indicates that the modified binder retains the 60/70 penetra- the G* and d at intermediate temperatures, which in turn enhance
tion grade of base asphalt binder. The penetration index of 5% the asphalt binder’s rutting properties. The reduced |G*| at high
CPO bio-asphalt binder is (1.59), which is within the required CPO content (more than 5%) enhanced the asphalt binder’s fatigue
(2 to 2) for temperature susceptibility. The softening point of properties. These findings are in agreement with those made by
5% CPO bio-asphalt binder is slightly lower compared to that of [35] which showed similar |G*| and d behavior of bio-asphalt bin-
the base binder. Result of the analysis shows that base asphalt bin- der incorporated with bio-oil extracted from palm oil empty fruit
der can be modified with 5% CPO without adversely affecting the bunch.
binders’ physical properties in high-temperature pavement
applications.
3.3. Rheological properties of RTFO-aged binders
3.2. Rheological properties of unaged binders Figs. 3 and 4 show the |G*| and d of RTFO-aged base binder and
bio-asphalt binders at various temperatures, respectively. It can be
Figs. 1 and 2 present the plot of |G*| and d of unaged base binder seen that higher CPO content resulted in reduced |G*| and higher d.
and bio-asphalt binders versus temperature, respectively. It can be This is most probably due to the higher aromatics content in bio-oil
seen that the |G*| and d remain almost unchanged for the base compared to the content in the base asphalt binder [33]. In con-
6
A.M. Al-Sabaeei, M.B. Napiah, M.H. Sutanto et al. Ain Shams Engineering Journal xxx (xxxx) xxx
Fig. 10. Experimental values versus predicted |G*| values of RTFO-aged binders.
trast, the binder modified with 5% CPO has a lower d value than 3.4. ANN model analysis
that of the base binder, which may be due to the compatibility of
CPO with base asphalt binder at lower CPO content. The addition The accuracy of an ANN model is influenced by the network
of 5 and 10% CPO resulted in even smaller |G*| values and larger architecture. There is no standard for the number of neurons in
d values, although higher CPO content caused a slight decrease in the hidden layers. In this study, an ANN with 3 and 11 neurons
|G*| and larger d. This indicates that the addition of CPO does not for the first and second hidden layers, respectively, have been
have a significant effect on aging, especially at higher CPO content. shown to be the optimal topology for |G*| and d of unaged and
Higher temperature produced lower |G*| value and higher d value, RTFO-aged bioasphalt binders, as can be seen in Fig. 5. Figs. 6
indicating the consistent effect of CPO on the rheological properties and 7 present the results of ANN training performance for unaged
of base asphalt binder before and after aging. This indicates that and RTFO-aged binders, respectively. The best validation perfor-
bio-asphalt binders modified with CPO have a consistent rutting mance for unaged binders based on mean square error (MSE) is
resistance. In addition, the smaller complex modulus at high CPO 7.9487 105 at epoch 1000 for |G*| and 0.0015984 at epoch
content (more than 5%) enhanced the asphalt binder’s fatigue 108 for d. The best validation performance for RTFO-aged binders
properties. This is in agreement with the findings made by [20] are 0.0033017 at the epoch 1000 and 0.00065665 at epoch 615
that showed an almost similar trend for the |G*| and d of a bioas- for G* and d, respectively.
phalt binder modified with waste vegetable oil. It can be concluded Based on the difference between the predicted and actual val-
that the reduced |G*| and larger d of bio-asphalt binders could ues of |G*| and d in unaged and RTFO-aged conditions, the pro-
enhance the workability of the binder, which in turn improve mix- posed neural network is able to learn the relationship between
ture durability and lower the mixing and compaction temperatures CPO content and test temperature as input parameters and |G*|
of asphalt mixes. and d as output parameters. The values of |G*| and d determined
7
A.M. Al-Sabaeei, M.B. Napiah, M.H. Sutanto et al. Ain Shams Engineering Journal xxx (xxxx) xxx
by the ANN model are close to the experimental values, which is an tion. These results are consistent with those obtained by [24]
indication of high prediction accuracy. which showed that the behaviour of |G*| and d of the modified
Figs. 8 and 9 show the relationship between the predicted val- asphalt binder are similar with those of nano-silica.
ues generated by the developed ANN models with the actual values Figs. 10 and 11 show the relationship of the values predicted by
from experimental work of unaged binders. The predicted values the developed ANN models and the actual values from the experi-
are very close to the actual values, indicating a strong correlation mental work of RTFO-aged binders. Similar with the unaged bin-
between CPO content and test temperature as input and |G*| and ders, the predicted values are almost identical with the actual
d as outputs. The R2 and MSE values were used to determine the values, thus indicating a strong correlation between CPO content
significance of the developed ANN models. Fig. 8 shows that the and test temperature in short term aging condition. The R2 values
R2 values for the training, testing, and validation of the complex for training, testing, and validation of the complex modulus for the
modulus for the ANN model are 0.99994, 1 and 1, respectively. ANN model of RTFO-aged binders are 0.99999, 1 and 0.99999,
Fig. 9 shows that the R2 values for training, testing, and validation respectively. Fig. 11 shows that the R2 values for training, testing,
of the phase angle for the ANN model are 0.99976, 0.99995 and and validation of the phase angle for the ANN model are
0.99995, respectively. The R2of the ANN modeling is greater than 0.99997, 0.99999 and 0.9999, respectively. The R2 of the RTFO-
0.99% for all data sets for |G*| and d. This indicates the ability of aged binders determined using ANN modeling is also greater than
the developed ANN model to represent at least 99% of the experi- 0.99% for all data set of |G*| and d. This indicates the ability of the
mental results, and thus can be used to make an accurate predic- developed ANN model to represent the experimental results for an
8
A.M. Al-Sabaeei, M.B. Napiah, M.H. Sutanto et al. Ain Shams Engineering Journal xxx (xxxx) xxx
accurate prediction of |G*| and d bio-asphalt binders under short [12] Hamada HM et al. The present state of the use of palm oil fuel ash (POFA) in
concrete. Constr Build Mater 2018;175:26–40.
term aging condition.
[13] Yasin MHM et al. Potentials of palm oil as new feedstock oil for a global
alternative fuel: A review. Renew Sustain Energy Rev 2017;79:1034–49.
4. Conclusion [14] Wang C et al. Laboratory investigation on chemical and rheological properties
of bio-asphalt binders incorporating waste cooking oil. Constr Build Mater
2018;167:348–58.
This research has investigated the base and rheological proper- [15] Sun Z et al. Properties of asphalt binder modified by bio-oil derived from waste
ties of unaged and RTFO-aged bio-asphalt binders incorporated cooking oil. Constr Build Mater 2016;102:496–504.
[16] Fini EH et al. Partial replacement of asphalt binder with bio-binder:
with CPO. A multilayer feed-forward neural network method was characterisation and modification. Int J Pavement Eng 2012;13(6):515–22.
used to develop the models for establishing the rheological perfor- [17] Samieadel A, Schimmel K, Fini EH. Comparative life cycle assessment (LCA) of
mance of bio-asphalt binders in terms of |G*| and d at varying CPO bio-modified binder and conventional asphalt binder. Clean Technol Environ
Policy 2018;20(1):191–200.
contents and test temperatures. The base and rheological proper-
[18] Yang S-H, Suciptan T. Rheological behavior of Japanese cedar-based biobinder
ties of bio-asphalt binders with low CPO content are almost similar as partial replacement for bituminous binder. Constr Build Mater
as those of the base binder; however, the binders incorporated 2016;114:127–33.
[19] Hill B et al. Low-temperature performance characterization of biomodified
with more than 5% CPO have a smaller complex modulus value.
asphalt mixtures that contain reclaimed asphalt pavement. Transp Res Rec
The optimum CPO content was determined to be 5%. The experi- 2013;2371(1):49–57.
mental values of |G*| and d were compared with the predicted val- [20] Chen M et al. Physical, chemical and rheological properties of waste edible
ues generated by the ANN models by using the R2 correlation vegetable oil rejuvenated asphalt binders. Constr Build Mater
2014;66:286–98.
coefficient and mean square error (MSE). The unaged binders have [21] Alamawi MY et al. Investigation on physical, thermal and chemical properties
R2 and MSE, complex modulus and phase angle values of 0.9999, of palm kernel oil polyol bio-based binder as a replacement for bituminous
0.9996, 7.9487 105 and 0.0015948, respectively. The R2, MSE, binder. Constr Build Mater 2019;204:122–31.
[22] Rahman MT, Hainin MR, Bakar WAWA. Use of waste cooking oil, tire rubber
|G*| and d of the RTFO-aged bio-binders are 0.99998, 0.99992, powder and palm oil fuel ash in partial replacement of bitumen. Constr Build
0.0033017 and 0.00065665, respectively. It can be concluded that Mater 2017;150:95–104.
the proposed multilayer feed-forward neural network technique [23] Khamis N et al. Predicting the rheological properties of bitumen-filler mastic
using artificial neural network methods. Jurnal Teknologi. 2018;80(1):71–8.
is satisfactory for a fast and accurate modeling of the |G*| and d [24] Yusoff NIM et al. Engineering characteristics of nanosilica/polymer-modified
of bio-asphalt binders. The experimental values are almost similar bitumen and predicting their rheological properties using multilayer
with the values predicted by the ANN models. Therefore, it can be perceptron neural network model. Constr Build Mater 2019;204:781–99.
[25] Yusoff NIM, Shaw MT, Airey GD. Modelling the linear viscoelastic rheological
concluded that the developed ANN models are important tools for
properties of bituminous binders. Constr Build Mater 2011;25(5):2171–89.
determining whether CPO should be used as an extender in asphalt [26] Sebaaly H, Varma S, Maina JW. Optimizing asphalt mix design process using
binders. artificial neural network and genetic algorithm. Constr Build Mater
2018;168:660–70.
[27] Shafabakhsh G, Ani OJ, Talebsafa M. Artificial neural network modeling (ANN)
Declaration of Competing Interest for predicting rutting performance of nano-modified hot-mix asphalt mixtures
containing steel slag aggregates. Constr Build Mater 2015;85:136–43.
The authors declare that they have no known competing finan- [28] Esra’a IA, Abo-Qudais SA. Modeling of creep compliance behavior in asphalt
mixes using multiple regression and artificial neural networks. Constr Build
cial interests or personal relationships that could have appeared Mater 2018;159:635–41.
to influence the work reported in this paper. [29] Firouzinia M, Shafabakhsh G. Investigation of the effect of nano-silica on
thermal sensitivity of HMA using artificial neural network. Constr Build Mater
2018;170:527–36.
Acknowledgments [30] Ziari H et al. Predicting rutting performance of carbon nano tube (CNT) asphalt
binders using regression models and neural networks. Constr Build Mater
The authors would like to express their gratitude to Universiti 2018;160:415–26.
[31] Kok BV et al. Investigation of complex modulus of base and SBS modified
Teknologi PETRONAS (UTP) for providing all necessary facilities bitumen with artificial neural networks. Expert Syst Appl 2010;37
for conducting this research and Universiti Kebangsaan Malaysia (12):7775–80.
(UKM) for the financial support of this work (GUP-2018-094). [32] Alhamali DI. Development of the rheological properties modelling system for
nanosilica-polymer modified bitumen based on artificial neural network, in
Department of Civil and Structural Engineering. Doctor of
References Philosophy: Universiti Kebangsaan Malaysia; 2017.
[33] Su N et al. Productions and applications of bio-asphalts–A review. Constr Build
[1] Li J et al. Preparation and properties of soybean bio-asphalt/SBS modified Mater 2018;183:578–91.
petroleum asphalt. Constr Build Mater 2019;201:268–77. [34] Tan C-P, Nehdi IA. The physicochemical properties of palm oil and its
[2] Nejad FM et al. Rutting performance prediction of warm mix asphalt components, in Palm Oil. Elsevier. 2012:377–91.
containing reclaimed asphalt pavements. Road Mater Pavement Des 2014;15 [35] Poh CC et al. Effect of fast pyrolysis bio-oil from palm oil empty fruit bunch on
(1):207–19. bitumen properties. IOP Conference Series: Materials Science and
[3] Fini EH et al. Fractional viscoelastic study of low-temperature characteristics of Engineering. IOP Publishing; 2018.
biomodified asphalt binders. J Mater Civ Eng 2016;28(9):04016078. [36] Khairuddin FH et al. Physicochemical and thermal analyses of polyurethane
[4] McCready N, Williams R. Utilization of biofuel coproducts as performance modified bitumen incorporated with Cecabase and Rediset: optimization
enhancers in asphalt binder. Transp Res Rec: J Transp Res Board using response surface methodology. Fuel 2019;254:115662.
2008;2051:8–14. [37] ASTMD5, Standard test method for penetration of bituminous materials, USA,
[5] Azahar WNAW et al. Chemical modification of waste cooking oil to improve ASTM International, 2013.
the physical and rheological properties of asphalt binder. Constr Build Mater [38] ASTM. D36, Standard Test Method for Softening Point of Bitumen (Ring-and-
2016;126:218–26. Ball Apparatus). Annual Book of Standards 2006.
[6] Mamat R et al. Areview of performance asphalt mixtures using bio-binder as [39] ASTM. D2872, Standard test method for effect of heat and air on a moving film
alternative binder. Jurnal Teknologi. 2015;77(23):17–20. of asphalt (rolling thin-film oven test). USA: Annual Book of ASTM Standards
[7] Pouget S, Loup F. Thermo-mechanical behaviour of mixtures containing bio- 2012.
binders. Road Mater Pavement Des 2013;14(sup1):212–26. [40] Das AK, Panda M. Investigation on rheological performance of sulphur
[8] Sun D et al. Evaluation of optimized bio-asphalt containing high content waste modified bitumen (SMB) binders. Constr Build Mater 2017;149:724–32.
cooking oil residues. Fuel 2017;202:529–40. [41] Al-Sabaeei AM et al. Influence of nanosilica particles on the high-temperature
[9] Gong M et al. Physical–chemical properties of aged asphalt rejuvenated by bio- performance of waste denim fibre-modified bitumen. Int J Pavement Eng
oil derived from biodiesel residue. Constr Build Mater 2016;105:35–45. 2020:1–14.
[10] Hill B et al. Evaluation of low temperature viscoelastic properties and fracture [42] AASHTO T 315, Standard method of test for determining the rheological
behavior of bio-asphalt mixtures. Int J Pavement Eng 2018;19(4):362–9. properties of asphalt binder using a dynamic shear rheometer (DSR), American
[11] Lei Z, Yi-qiu T, Bahia H. Relationship between glass transition temperature and Association of State and Highway Transportation Officials, 2012.
low temperature properties of oil modified binders. Constr Build Mater [43] Hornik K, Stinchcombe M, White H. Multilayer feedforward networks are
2016;104:92–8. universal approximators. Neural Networks 1989;2(5):359–66.