See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261525291

Prediction of the Marshall Stability of Reinforced Asphalt Concrete with Steel

Fiber Using Fuzzy Logic, Journal of Intelligent and Fuzzy Systems

Article  in  Journal of Intelligent and Fuzzy Systems · June 2014

DOI: 10.3233/IFS-130873.

CITATIONS READS

0 314

1 author:

Serdal Terzi
T.C. Süleyman Demirel Üniversitesi
80 PUBLICATIONS   564 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Akıllı Servis Araçları konseptinin oluşturulması ve ulaşım sistemindeki türel değişim (modal shift) üzerine olan yararlı etkilerinin tahmini: pilot bir bölge üzerindeki
bilgisayar modellemesi View project

Investigating the engineering and visibility properties of hot mix asphalt with phosphorous modified bitumen View project

All content following this page was uploaded by Serdal Terzi on 08 October 2014.

The user has requested enhancement of the downloaded file.

Journal of Intelligent & Fuzzy Systems 26 (2014) 1943–1950 1943
DOI:10.3233/IFS-130873
IOS Press

Prediction of the marshall stability of

reinforced asphalt concrete with steel fiber
using fuzzy logic

PY
Sercan Serina,∗ , Nihat Morovab , Mehmet Saltanc , Serdal Terzic and Mustafa Karaşahind
a Civil Engineering Department, Faculty of Technology, Duzce University, Duzce, Turkey
b Manufacturing Engineering Department, Faculty of Technology, Suleyman Demirel University, Isparta, Turkey
c Civil Engineering Department, Faculty of Engineering, Suleyman Demirel University, Isparta, Turkey
CO
d Civil Engineering Department, Faculty of Engineering, İstanbul University, Avcilar İstanbul, Turkey

Abstract. In this study, Marshall Stability (MS) of steel fiber reinforced asphalt concrete has been predicted using steel fiber rate
OR

(0%, 0.25%, 0.50%, 0.75%, 1.0%, 1.5%, 2.0% and 2.5%), bitumen content (5%, 5.5% and 6.0%) and unit weights (2,465–2,515
(gr/cm3 )) by Fuzzy Logic (FL). Results have shown that developed FL model has a strong potential for predicting the MS of
asphalt concrete without performing any experimental studies.

Keywords: Asphalt concrete, steel fiber, fuzzy logic, marshall stability

TH

1. Introduction mance and lifetime of road pavements. It is aimed to

increase the performance and lifetime of roads by using
The pavements are under the effect of traffic loads different additive materials [2].
and natural circumstances. Traffic loads are composed The researchers that previously worked on this sub-
AU

of radial tensile and compressive stresses as well as ver-
tical compressive stresses occurring due to axle loads
during the movement of vehicles. The intensity and
magnitude of traffic loads are directly proportional to
the repetition of these stresses. The aim of the pavement
design is to determine the thicknesses of the layers that
are needed to bear safely the repetitive loads under envi-
ronmental circumstances without large deformations or
cracking, based on the characteristics of the materials
to be used in these layers [1].
Flexible pavements are designed so as to have at
least 20 years project life. The current research subjects
include the studies focusing on increasing the perfor-
∗ Correspondingauthor. Sercan Serin, Civil Engineering Depart-
ment, Faculty of Technology, Duzce University, Duzce, Turkey.
E-mail: sercanserin@duzce.edu.tr.
ject clearly indicated that the use of fiber in the
pavements and asphalt mixtures has a strengthening
effect. Fibers can be used especially in the mixtures
with continuous grading and in Stone Mastic Asphalt
(SMA) mixtures in order to overcome the deteriora-
tion of asphalt during carriage and construction of the
mixture, as well as in asphalt stabilization [3–5]. The
use of fibers alters the visco-elasticity characteristics
of the mixture [6]; enhances its dynamic modulus [7],
enhances sensibility against humidity [8], enhances
flow coherence, and provides resistance against the rut-
ting [9, 10]; as well as decreases the amount of reflective
cracks in asphalt mixtures and pavement [11–13].
In addition to these studies, a number of scholars
carried out researches regarding fatigue, performance,
rigidity, crack propagation etc. in fiber-reinforced
asphalt mixtures [14–21].

1064-1246/14/$27.50 © 2014 – IOS Press and the authors. All rights reserved
1944 S. Serin et al. / Prediction of the marshall stability of reinforced asphalt concrete with steel fiber

Nowadays Artificial Intelligent methods have been 2. Experimental procedure

extensively used in civil engineering applications
[22, 23].
Fuzzy logic and neural networks are the widely
used artificial inference systems. Each approach has
its merits and drawbacks. To take advantage of both
approaches, integration of these systems has been pro-
posed by many researchers in recent years [24].
Designers utilize principles of science and math-
ematics to develop specific technologies. These
technologies are then used to create engineered tools
such as products, structures, machines, processes or
Crushed limestone aggregates were used in asphalt
mixtures. Aggregate material tests were carried out
based on American Codes, in order to obtain the physi-
cal and mechanical characteristics of the materials to be
used in the mixtures. Table 1 shows mixture ratios deter-
mined for hot mixtures. The physical and mechanical
characteristics of the aggregates used in the mixtures
are given in Table 2 [32].
AC 60/70 asphalt cement was used in asphalt mix-
tures as bitumen. The physical characteristics of this

entire systems. It has already been seen that different
tasks in engineering problem solving require different
analysis [25]. Recently, artificial intelligence and statis-
tical analysis have been extensively using in the fields
of civil engineering applications such as construction
management, building materials, hydraulic, geotechni-
cal and transportation engineering etc. [26–31].
In this study, it has researched that predictability of
MS of steel fiber reinforced asphalt concrete and had
varied mix properties with FL. With this aim, asphalt
concrete samples that added Steel fiber in accordance
OR
PY
selected bitumen are given in Table 3.
In this study, Dramix RC-80/60-BN fiber was used.
The characteristics of the mentioned steel fiber are given
Tables 4 and a photograph of these fibers is available in
Fig. 1.
CO
Asphalt mixtures were prepared in accordance with
the technical specifications required by Highway Gen-
eral Directorate of Turkey [33]. The upper and lower
limits required for the mixture grading of aggregate are
shown in Fig. 2.
A series of tests were carried out to determine the

with gradation determined in Highway Technical Spec- optimum bitumen content. Asphalt mixture samples
ification (HTS), had different percentage of steel fiber were produced using as 4.5%, 5.0%, 5.5%, and 6.0%
rate (0%, 0.25%, 0.50%, 0.75%, 1.0%, 1.5%, 2.0% and
2.5%), bitumen content (5%, 5.5% and 6.0%) and unit
weights (UW) (2.465–2.515 (gr/cm3 )) were prepared Table 3
and determined MS with Marshall test.
TH

Basic physical characteristics of the bitumen [31]

Characteristics of Bitumen
Test name Average values
Table 1 Penetration (25ºC) 60–70
Mixture ratios determined for hot mixtures Flash Point 180ºC
AU

Fire Point 230ºC

Sieve diameters
Softening Point 45.5◦ C
25–4.75 mm 4.75–0.075 mm Filler Ductility (5 cm/minute) >100 cm
Binder 40% 55% 5% Specific Gravity 1.034

Table 2
Physical and mechanical characteristics of aggregates
Sieve diameters Standard
0–4.75 mm 4.75–9.5 mm 9.5–25 mm Limit
Water Absorption % 3.54 1.63 0.81 Binder; %2.5 ASTM C 127
ASTM C 128
Abrasion loss (%) (Los Angeles) – – 23.804 Binder; %35 ASTM C 131
Fine Material % – 1.27 0.45 %0.5 ASTM C 117
Organic Material Clear Clear Clear Clear ASTM C 40
Freeze-Thaw % – – 6.69 Binder; %12 ASTM C 88
Peeling Strength. % – More than %50 More than %50 Binder; %50 *HTS Part 403App-A
Average Density (gr/cm3 ) 2.576 2.642 2.677 – ASTM C127, ASTM C128
Loose specific gr. (gr/cm3 ) 1.61 1.40 1.41 – ASTM C 29
Compact specific gr.(gr/cm3 ) 1.91 1.62 1.64 – ASTM C 29
∗ Highways Technical Specifications (HTS) in Turkey.
 S. Serin et al. / Prediction of the marshall stability of reinforced asphalt concrete with steel fiber 1945

Table 4
Properties of steel fiber
Length Max. Diameter Slenderness Specific Gravity Modulus of Tensile Strength The Number
Ratio Elasticity (MPa) (MPa) of Fiber (kg)
60 mm ± %5 0.75 mm ± %5 80 7480 200000 1035 4600

PY
CO
Fig. 3. Asphalt concrete specimens produced by using steel fibers.

The asphalt concrete samples, produced by using steel

fibers, are seen in Fig. 3.
Fig. 1. Steel Fibers. After experimental works, a fuzzy logic model devel-
oped by using MATLAB toolbox. While modeling
Bitumen (%), Steel fiber (%) and Unit weight (gr/cm3 )
OR

bitumen contents. Optimum bitumen content obtained were used as inputs and Marshall Stability of asphalt
as 5.5%. concrete were used as outputs.
Based on the obtained optimum bitumen con-
tent value, steel fibers were added in different ratio
in weights (0%–0.25%–0.50%–0.75%–1.0%–1.5%– 3. Theory of fuzzy logic
TH

2.0%– 2.5%) as three samples for each fiber rate.

Maximum Marshall Stability value was obtained for Fuzzy set theory was developed by Lotfi Zadeh
5.5% bitumen content and 0.75% fiber addition [32]. in 1965 to deal with the imprecision and uncertainty
AU

Fig. 2. Gradation curve of the aggregates used in mixture.

1946 S. Serin et al. / Prediction of the marshall stability of reinforced asphalt concrete with steel fiber

that is often present in real-world applications [34]. In

1974 [35], by applying Zadeh’s theories of linguistic
approach and fuzzy inference, successfully used the
‘IF-THEN’ rule on the automatic operating control of
steam generator. It needs only to set a simple controlling
method based on engineering experience. Therefore,
it is particularly useful in complicated structural sys-
tems. Fuzzy logic has been developing since 1965 and
become most successful in application [36].
In recent years, the number and variety of appli-
cations of fuzzy logic have increased significantly.
The applications range from consumer products such

PY
as cameras, camcorders, washing machines, and
microwave ovens to industrial process control, medical
instrumentation, decision-support systems, and portfo-
lio selection [37].
Fuzzy inference is the real process of mapping from
CO
a given set of input variables to an output relied upon a
set of fuzzy rules. The main process of a general fuzzy
inference system (FIS) includes four activities called as
fuzzification, fuzzy rule base, fuzzy inference engine
and defuzzification [38].
These parts are detailed below.
OR

1. Fuzzification: It converts each input data to

degrees of membership by a lookup in one or more
of several membership functions.
2. Fuzzy rule base: This contains rules that include
Fig. 4. Flow diagram. all possible fuzzy relationship between input and
TH

outputs using IF-THEN format.

AU

Fig. 5. General structure and membership functions of the model.

 S. Serin et al. / Prediction of the marshall stability of reinforced asphalt concrete with steel fiber 1947

Table 5
Some of examples rules
860. If (BITUMEN-% is B3) and (FIBER-% is F1) and (UNITWEIGHT is U2) then (MARSHALL-STABILITY is M14) (1)
861. If (BITUMEN-% is B3) and (FIBER-% is F1) and (UNITWEIGHT is U1) then (MARSHALL-STABILITY is M14) (1)
862. If (BITUMEN-% is B3) and (FIBER-% is F3) and (UNITWEIGHT is U1) then (MARSHALL-STABILITY is M14) (1)

3. Fuzzy Interference Engine: Collects all fuzzy In this study Mamdani type fuzzy logic model-
rules in the fuzzy rule base and learns how to ing studied for the prediction of Marshall Stability of
transform a set of inputs to related outputs. asphaltic concrete.
4. Defuzzification: This converts the resulting fuzzy
outputs from fuzzy interference engine to a num-
ber [39]. 4. Fuzzy modeling details

PY
There are two types of FIS that can be implemented In this paper, a fuzzy logic model was developed
in the MATLAB’s FIS toolbox: Mamdani-type and to predict Marshall Stability of asphalt concrete using
Sugeno-type. Mamdani’s method is the most commonly experimental variables. The model has three inputs and
seen fuzzy methodology and it expects the output MFs an output. Inputs were bitumen (%), fiber (%) and unit
CO
to be fuzzy sets. After the aggregation process, there is weight (gr/cm3 ) and output was the Marshall stability
a fuzzy set for each output variable that needs defuzzi- of asphalt concrete. Flow diagram for study is given
fication [40]. below (Fig. 4)

(a) (b)
OR
MARSHALL STABILITY

1100
MARSHALL STABILITY

1200
1000
1000
900
800
TH

6.5 2.52
6 2.51
5.5 3
2.5
BI

UN

5
TU

0 2.49 2
IT

4.5
M

1
W

2.48
EN

-%
AU

4
EI

2 1
-% ER
FIBER
G
-%

3 FIB
HT

(c)
MARSHALL STABILITY

950
900
850
800
750
4
4.5
BIT 5 2.48
UM 5.5 2.49
EN 6 2.5
-% 2.51 GHT
6.5 WEI
2.52 UNIT

Fig. 6. Relationship between inputs and output of the model.

1948 S. Serin et al. / Prediction of the marshall stability of reinforced asphalt concrete with steel fiber

PY
CO

Fig. 7. Comparison of experimental and fuzzy logic results.

OR
TH
AU

Fig. 8. Matching figure of the values of experimental and fuzzy logic model.

While modeling, the membership functions for bitu-
men (%), fiber (%), unit weight, Marshall stability of
asphalt concrete were selected 4, 6, 6, 14 respectively.
The general structure of the model and membership
functions for input and output parameters used for fuzzy
logic modeling are given in Fig. 5 as a diagram.
In the rule base 870 rules formed using experimen-
tal results and experiences. Some of formed rules are
given Table 5. After determining membership functions
and forming rules (870), the FL model results obtained
using defuzzification monitor. The Marshall stability
from developed FL model as a function of inputs is seen
in Fig. 6. Centroid was used as defuzzification method.
5. Results and discussions
In this study, the predictability of the developed FL
model evaluated by using some statistical equations.
Equation (1) is the Root Mean Squared Error (RMSE)
 S. Serin et al. / Prediction of the marshall stability of reinforced asphalt concrete with steel fiber
Table 6
Statistics of Marshall Stability prediction using FL Statistics
COE Adjusted COE RMSE
0.8126 0.8059 53.09
and Equation (2) is the coefficient of simulation effi-
ciency (COE).

modified overlay mixture, Transport Res Rec 1545 (2001),
 n
1 
RMSE =  (Yi(m) − Yi(p) )2 (1)
N properties of Asphalt mixtures, J Wuhan Univ Technol – Mater
i=1
n
[(xm )i − (xp )i ]2
COE = 1 − 1=1
n (2)
1=1 [(xm )i − (X̄)]
2
In equations 1 and 2 subscripts m, p, N and X̄indicate
measured, predicted, pattern and average measured data
CO
respectively.
Also experimental studies and FL model results com-
parison graphics are illustrated in Figs. 7 and 8.
As shown in Figs. 7 and 8, the results obtained
FL model is very close to the experimental results.
This situation is demonstrated clearly by the statistical
OR
parameters shown in Table 6 (COE and RMSE).
6. Conclusions
In this study, a model was developed by fuzzy logic
TH
method and it’s usability for predicting the Marshall
Stability of steel fiber reinforced asphalt concrete was
investigated. While modeling 30 experimental results
were used to develop the model. Coefficient of deter-
mination (R2) and Total Root Mean Square Error
AU
(RMSE) criteria were used for comparison of exper-
imental results with the results predicted by the model
of FL. When the results were compared, RMSE and R2
values were found as 53,09 and 0,8126 respectively. As
a result, Marshall Stability values of steel fiber rein-
forced asphalt concrete could be predicted practically
with very low error rates in a very short time without
performing any experiments.
References
[1] E. Ağar, G. Öztaş and I. Sütaş, Concrete roads, Istanbul Tech-
nical University Press, Istanbul, 1998.
[2] F. Umar and E. Ağar, Pavement Structure, İstanbul Technical
University Civil Engineering Faculty Press, İstanbul, 1991.
[3] H.F. Hassan, S.A. Oraimi and A. Taha, Evaluation of open-
graded friction course mixtures containing cellulose fibers and
1949
styrene butadiene rubber polymer, J Mater Civil Eng 17 (2005),
416–422.
[4] J.P. Serfass and J. Samanos, Fiber-modified asphalt concrete
characteristics, applications and behavior, J Assoc Asphalt
Paving Technol 64 (1996), 193–230.
[5] P. Peltonen, Wear and deformation of characteristics of
fiber reinforced Asphalt pavements, Construct Build Mater 5
(1991), 18–22.
[6] H. Huang and T.D. White, Dynamic properties of fiber-
98–104.
[7] S. Wu, Q. Ye, N. Li and H. Yue, Effects of fibers on the dynamic
Sci Ed 22 (2007), 733–736.
[8] B.J. Putman and S.N. Amirkhanian, Utilization of waste fibers
PY
in stone matrix Asphalt mixtures, resources, Conserv Recycl
42 (2004), 265–274.
[9] R.S. McDaniel, Asphalt additives to control rutting and crack-
ing, PHD thesis, Purdue University, Indiana, 2001.
[10] H. Chen, N. Li, C. Hu and Z. Zhang, Mechanical performance
of fibers-reinforced asphalt mixture, J Chang’an Univ (Nat Sci
Ed) 24(2) (2004), 1–5.
[11] S. Tapkin, The effect of polypropylene fibers on asphalt per-
formance, Building Environment 43(6) (2008), 1065–1071.
[12] D.A. Maurer and G.J. Malasheskie, Field performance of
fabrics and fibers to retard reflective cracking, Geotextiles
Geomembranes 8 (1989), 239–267.
[13] C. Huaxin, X. Qinwu, C. Shuanfa and Z. Zhengqi, Evaluation
and design of fiber-reinforced asphalt mixtures, Materials and
Design 30 (2009), 2595–2603.
[14] G. Yang, A fatigue test on fiber reinforced asphalt mixtures,
Critical issues in transportation systems planning, develop-
ment, and management ICCTP, 2009.
[15] C. Hua-Xin, Z. Zheng-Qi and H. Chang-Shun, Low-
temperature anticracking performance of fiber-reinforced
asphalt mixture, J South China Univ Technol, 2004–04.
[16] M. Xiang, N. Fu-Jian, G. Xing-Yu and S. Heng, Study and
analysis on pavement performance of the polyester fiber-
reinforced asphalt mixture, J High Transport Res Develop,
2006–01.
[17] G. Nai-Sheng, Z. Ying-Hua and Z. Hong-Tao, Equivalent stiff-
ness moduli of fiber reinforced asphalt concrete, J High Transp
Res Develop, 2006–09.
[18] G. Nai-Sheng, Z. Ying-Hua and S. Lue-Lun, Dynamic param-
eters of fiber reinforced asphalt concrete, J High Transport Res
Develop, 2007–10.
[19] S. Bueno Benedito de, R. da Silva Wander and C. de
Lima Dario, ASCE M, Minete Enivaldo, Engineering prop-
erties of fiber reinforced cold asphalt mixes, J Environ
Eng 129 (2003), 952. http://dx.doi.org/10.1061/(ASCE)0733-
9372(2003)129:10(952
[20] S.M. Abtahi, M. Sheikhzadeh and S.M. Hejazi, Fiberrein-
forced asphalt-concrete – a review, Construct Build Mater
24(6) (2010), 871–877.
[21] S.M. Hejazi, S.M. Abtahi and M. Sheikhzadeh, Introducing
two simple models for predicting fiber-reinforced asphalt con-
crete behavior during longitudinal loads, Journal of Applied
Polymer Science 109 (2008), 2872–2881.
[22] M.E. Emiroglu, O. Kisi and O. Bilhan, Predicting discharge
capacity of triangular labyrinth side weir located on a straight
channel by using an adaptive neuro-fuzzy technique, Advances
in Engineering Software 41(2) (2010), 154–160.
[23] M. Mashrei, N. Abdulrazzaq, T.Y. Abdalla and M.S. Rahman,
Neural networks model and adaptive neuro-fuzzy inference
 1950 S. Serin et al. / Prediction of the marshall stability of reinforced asphalt concrete with steel fiber
system for predicting the moment capacity of ferrocement
members, Engineering Str 32-6 (2010), 1723–1734.
[24] K. Kucuk, C.O. Aksoy, H. Basarir, T. Onargan, M. Genis and
V. Ozacar, Tunnelling and Underground Space Technology
incorporating Trenchless Technology Research 26(1) (2011),
38–45.
[25] M. Emiroğlu, A. Beycioğlu and S. Yildiz, Anfis and Statis-
tical Based Approach to Prediction the Peak Pressure Load
of Concrete Pipes Including Glass Fiber, Expert Systems with
Applications 39(3) (2012), 2877–2883.
[26] M. Yilmaz, B.V. Kok, B. Sengoz, A. Sengur and E. Avci,
Investigation of complex modulus of base and EVA modified
bitumen with Adaptive-Network-Based Fuzzy Inference Sys-
tem, Expert Systems with Applications, In Press, Available
online, 2010.
[27] D. Moazami, H. Behbahani and R. Muniandy, Pavement reha-
bilitation and maintenance prioritization of urban roads using
fuzzy logic, Expert Systems with Applications 38 (2011),
12869–12879.
[28] S. Tapkin, A. Çevik and Ü. Uşar, Prediction of Marshall test
results for polypropylene modified dense bituminous mixtures
using neural networks, Expert Systems with Applications 37(6)
(2010), 4660–4670.
[29] T.E. Karakasidis, D.N. Georgiou and J.J. Nieto, Fuzzy regres-
sion analysis: An application on tensile strength of materials
and hardness scales, Journal of Intelligent and Fuzzy Systems
23(5) (2012), 177–186
[30] S.A. Kalogirou, Artificial intelligence for the modeling and
control of combustion processes: A review, Progress in Energy
OR
and Combustion Science 29 (2003), 515–566.
[31] P. Samaras, A. Kungolos, T. Karakasidis, D. Georgiou and
K. Perakis, Statistical evaluation of PCDD/F emission data
TH
AU
View publication stats
during solid waste combustion by fuzzy clustering tech-
niques, Journal of Environmental Science and Health - Part A
Toxic/Hazardous Substances and Environmental Engineering
36 (2001), 153–161.
[32] S. Serin, N. Morova, M. Saltan and S. Terzi, Investigation of
usability of steel fibers in asphalt concrete mixtures, Construc-
tion and Building Materials 36 (2012), 238–244.
[33] General Directorate of Highways. State highways technical
specifications (HTSs), Ankara, 2006.
[34] L.A. Zadeh, Fuzzy set, Information and Control 8 (1965),
338–353.
[35] E.H. Mamdani, Application of fuzzy algorithms for control
of simple dynamic plants, Proc IEEE 121 (1976), 1585–
1588.
[36] H. Tanyildizi, Fuzzy logic model for prediction of mechanical
PY
properties of lightweight concrete exposed to high Tempera-
ture, Materials and Design 30 (2009), 2205–2210.
[37] H. Işik and S. Arslan, The design of ultrasonic therapy device
via fuzzy logic, Expert Systems with App 38 (2011), 7342–
7348.
[38] H. Kuşan, O. Aytekin and İ. Özdemir, The use of fuzzy logic
CO
in predicting house selling price, Expert Systems with Appli-
cations 37 (2010), 1808–1813.
[39] M. Omid, Design of an expert system for sorting pistachio nuts
through decision tree and fuzzy logic classifier, Expert Systems
with Applications 38 (2011), 4339–4347.
[40] I. Akkurt, C. Başyigit, S. Kilincarslan, and A. Beycioglu, Pre-
diction of photon attenuation coefficients of heavy concrete by
fuzzy logic, Journal of the Franklin Institute 347(9) (2010),
1589–1597.