IRC Highway Research Journal January-June 2018 PDF

HIGHWAY RESEARCH
JOURNAL
Highway Research Board

Indian Roads Congress
Volume 9 No. 1 January - June, 2018
https://www.irc.nic.in
Members of the Highway Research Board (2017-2020)
1 Chairman, HRB - Shri Manoj Kumar, Director General (Road Development) & Spl. Secy., MoRT&H, New Delhi
2 Secretary, HRB - Shri S.K. Nirmal, Secretary General, Indian Roads Congress, New Delhi
3 President, IRC - Shri K.S. Krishna Reddy, Managing Director, Karnataka Road Development Corporation Ltd., Bengaluru
Members
4 Prof. Satish Chandra, Director, CSIR-Central Road 29 Dr. P.K. Sarkar, Deptt. of Transport Planning, School
Research Institute, New Delhi of Planning & Architecture, New Delhi
5 Shri B.N. Singh, Additional Director General, 30 Dr. Dharmveer Singh, Assistant Professor, Deptt. of
MoRT&H, New Delhi Civil Engg., IIT Mumbai, Maharashtra
6 Shri I.K. Pandey, Additional Director General, 31 Dr. P. L. Patel, Head, Deptt. of Civil Engg., SVNIT,
MoRT&H, New Delhi Surat
7 Shri Y. Balakrishna, Chief Engineer (Roads and 32 Shri Bharat Chandra Pradhan, Chief Engineer, Rural
Bridges) SR&T, MoRT&H, New Delhi Works , Bhubaneswar
8 Lt. Gen. S.K. Shrivastava, Director General Border 33 Shri P.R. Patelia, Chief Engineer (Panchayat) & Addl.
Roads, New Delhi Secretary, R&B Deptt., Gandhinagar
9 Shri D.O. Tawade, Member (Technical), National 34 Shri C.V.S. Ramamurthy, Engineer-in-Chief,
Highways Authority of India, New Delhi Panchayati Raj Department, Hyderabad
10 Dr. I.K. Pateriya, Director (Technical), National Rural 35 Shri Ajit Upale, Chief Engineer (PMGSY), Rural
Roads Development Agency (NRRDA), New Delhi Development & Water Conservation Department,
11 Shri S.S. Nahar, Director, Indian Academy of Mumbai
Highway Engineers, Noida (U.P.) 36 Shri Shankar Prasad Singh, Engineer-in-Chief, Rural
12 Er. Bhuvana Kumar, Chief Engineer (NH), H.P. Works Deptt., Patna
P.W.D., Shimla (Himachal Pradesh) 37 Dr. M. Malakondaiah, Vice-Chairman & Managing
13 Shri H.K. Upreti, Engineer-in-Chief, PWD., Director, Andhra Pradesh State Road Transport
Dehradun Corporation, Vijayawada
14 Shri Govind Prasad Katare, Chief Engineer (NH), 38 Shri R.V. Chakrapani, Managing Director,
PWD, Bhopal, Madhya Pradesh M/s. Aarvee Associates Pvt. Ltd., Hyderabad
15 Shri M.S. Kidwai, Chief Engineer (NH), U.P. P.W.D., 39 Dr. Harshavardhan Subbarao, Chairman & Managing
Lucknow Director, Construma Consultancy Pvt. Ltd., Mumbai
16 Shri N. Lakshman Rao Peshve, Chief Engineer (NH), 40 Shri Sanjeev Kapoor, Director (Transportation),
PWD, Bangalore Ramboll India Pvt. Ltd., Gurgaon
17 Shri K.P. Prabhakaran, Chief Engineer (NH), Works 41 Maj. V.C. Verma, Chairman and Managing Director,
Deptt., Thiruvananthapuram Oriental Structural Engineers Pvt. Ltd., New Delhi
18 Shri Dipankar Ghosh, Chief Engineer (NH), PWRD, 42 S h r i Vi s h w a s J a i n , M a n a g i n g D i r e c t o r,
Kolkata Consulting Engineers Group, New Delhi
19 Smt. N. Shanthy, Chief Engineer (NH), PWD, 43 Shri I.P. Tantia, Chairman and Managing Director,
Chennai Tantia Constructions Ltd., Kolkata
20 Shri K. Lalsawmvela, Chief Engineer (Highway), 44 Shri D.P. Gupta, DG (RD) & AS (Retd.), MOST,
PWD Mizoram, Aizawl New Delhi
21 Shri Ch. Tiken Singh, Chief Engineer, P.W.D., 45 Shri A.V. Sinha, Former DG (RD) & SS, MoRT&H,
Manipur, Imphal New Delhi
22 Mrs. R. Geeta, Director, Highways Research Station, 46 Shri P.N. Jain, Former Secretary, R&B Deptt.,
Chennai Govt. of Gujarat, Ahmedabad
23 Prof. Shishir Kr. Sahu, Head, Deptt. of Civil Engg., 47 Shri Chaman Lal, Addl. Mission Director, NMCG,
NIT, Rourkela Panchkula
24 Shri K.A. Keshvani, Chief Engineer & Director 48 Dr. Sunil Bose, Former Head, FPC Divn., CRRI,
(Roads), Gujarat Engg. Research Institute (GERI), Faridabad
Vadodara 49 Shri Parimal Rai, President, IBC & Advisor to
25 Dr. Samson Mathew, Professor, Deptt. of Civil Engg., Administrator, Chandigarh UT, Chandigarh
NIT, Tiruchirappalli 50 Prof. K. Sudhakar Reddy, Deptt. of Civil Engineering,
26 Dr. M.S. Amarnath, Coordinator, Centre for Indian Institute of Technology Kharagpur, Kharagpur
Transportation Engg., Bangalore University, 51 Shri H.L. Mina, Former Secretary, PWD, Rajasthan,
Bangalore Jaipur
27 Dr. Praveen Kumar, Professor of Civil Engg., IIT 52 Shri V. L. Patankar, Former ADG, MoRT&H,
Roorkee, Roorkee New Delhi
28 Prof. A. Veeraragavan, Deptt. of Civil Engg., IIT 53 Shri U. Jaya Kodi, President, Technical & Arbitration ,
Madras, Chennai BSCPL Infrastructure Ltd., Hyderabad
Highway Research
Journal
Volume : 9 - 1 JANUARY - JUNE, 2018
Indian Roads Congress
Founded : On 10th December, 1934
Highway Research Board
Publisher & Editor: S.K. Nirmal, Secretary General, IRC

E-mail: secygen.irc@gov.in www.irc.nic.in
Headquarter: Satellite Office:

IRC Bhawan, Kama Koti Marg, Sector-6, Ground Floor, IDA Building, Jamnagar House,
R.K. Puram, New Delhi-110 022. Shahjahan Road, New Delhi-110 011.
Phone No.: +91-11-26171548 (Admn.), 23387140 & 23384543 (Membership), 23387759 (Sale),
26185273 (Tech. Papers, Indian Highways and Tech. Committees)
No part of this publication may be reproduced by any means without prior written permission from the Secretary General, IRC.
The responsibility of the contents and the opinions expressed in Highway Research Journal is exclusively of the author(s) concerned. IRC
and the Editor disclaim responsibility and liability for any statements or opinion, originality of contents and of any copyright violations
by the authors. The opinion expressed in the papers and contents published in the Highway Research Journal do not necessarily represent
the views of the Editor or IRC.
Printed at: M/s India Offset Press, New Delhi-110 064 `20
(All Rights Reserved. No part of this Publication shall be reproduced, translated or
transmitted in any from or by any means without the permission of the
Indian Roads Congress)
The opinions and conclusions in this Journal are those of the

Authors and not of the IRC Highway Research Board
Performance evaluation of Super-Pave and Marshall mix
design to suite Indian climatic and traffic conditions
Hemant Kumar1 Dr. Sangita2 Dr. (Mrs.) Vandana Tare3
ABSTRACT
In India, presently the bituminous mixes are designed based on empirical laboratory procedures
i.e. Marshall method of mix design without simulating the actual field conditions. This has led the
highway engineers to think of a performance based mix design method, which can predict the
fundamental properties of bituminous mixes such as rutting and fatigue. The primary objective of
the performance based mix design method to establish the appropriate amount of binder content
in the mix that will simultaneously satisfy the rut resistance and fatigue cracking requirements
for given traffic and environmental conditions. This lead to development of Superior performance
pavements (Super-pave) mix design by Strategic Highway Research Program (SHRP), USA. The
main objective of the study is to develop performance based mix design for Indian conditions
using Super-Pave Gyratory compactor (SGC) and compare the Super-pave asphalt mixture design
procedures with the Marshall Asphalt mixture design method. The comparison was based on several
issues including evaluation of materials prior to mixture design, the design asphalt content, and
the relationship between mixture design and pavement performance. After comparison between
mix design methods resilient modulus of different bituminous mixes prepared with conventional
as well as modified binder, are evaluated at different temperature. For performance deformation
of mixes, flow number test is performed through Asphalt Mixture Pavement Tester (AMPT) at
varying air voids and at different deviator stress level. It is found that the performance parameters
such as ITS, TSR, Resilient modulus and Flow number values are higher in case, sample prepared
by Super-pave mix design method as compare to those sample prepared by Marshall method of
mix design.
1. Introduction field compaction by road rollers and subsequently

In India, Marshall mix design is the sole method by the pneumatic wheels of vehicles during service
used for the design of bituminous concrete mixes. due to absence of kneading and shearing action
This is a standard laboratory method used by and during the compaction blows.
large in different parts of the world for determining Accordingly, there has been a growing concern
the strength and flow characteristics of bituminous among the highway engineers that the Marshall
paving mixes. It is also recognized that the impact method should be replaced with a better method,
compaction used in standard Marshall Method which will measure the fundamental properties
does not simulate the state of compaction and of bituminous mixtures. Such a method should
orientation of aggregate particles attained in the be able to predict pavement behavior in terms of
1. M.E Student in Transportation Engineering, S.G.S.I.T.S Indore, Email: mithktyagi2010@gmail.com

2. Senior Principal Scientist, Flexible Pavement Department CRRI, New Delhi, Email: sangita.crri@yahoo.com
3. Professor CE-AMD, S.G.S.I.T.S Indore, Email: vtare4@gmail.com
Highway Research Journal, January - June, 2018 1

Kumar, Dr. Sangita & Dr. Tare on
Performance Evaluation of Super-Pave and Marshall Mix Design to suite Indian Climatic and Traffic Conditions
fundamental properties such as fatigue, rutting  To compare Marshall Parameters obtained by

and low temperature cracking. This led to the using Super-Pave mix design and Marshall
development of Super-Pave (Superior Performing mix design method for polymer modified and
Asphalt Pavement) Asphalt Mixture Design neat bituminous concrete mixes.
Method in USA developed by Strategic Highway  To compare indirect tensile strength values
Research Program (SHRP) (Pandey 2002). This obtained at different temperatures for
method includes test equipment, test methods Marshall specimens prepared by using
and criteria. The key features in the Super-Pave Super-Pave mix design and Marshall mix
mix design method are laboratory compaction design method for polymer modified and neat
and performance testing. Super-Pave Gyratory bituminous concrete mixes.
Compactor (SGC) accomplishes laboratory  To compare the moisture susceptibility in
compaction. As it is possible to obtain similar terms of Tensile strength ratio and Retained
degree of reorientation of aggregate particles stability for specimens prepared using Super-
caused by the shearing action imparted to the Pave mix design and Marshall mix design
bituminous mixture in the field. (Sabita, 1993; method for polymer modified and neat
Rust et al 1992). bituminous concrete mixes.
Meor O. Hamzah et al. (2011) evaluated the effects  To compare permanent deformation of
of temperature on the resilient modulus of dense specimens prepared using Super-Pave mix
asphalt mixes incorporating steel slag either fully design and Marshall mix design method
or partially replacing natural granite aggregates or for polymer modified and neat bituminous
then subjected to STOA. The relationship between concrete mixes by using AMPT.
binder content and resilient modulus of aged 3. Materials and Experimental
SSDA and SSGDA specimens are presented, as Procedures
the binder content increases, the resilient modulus Two asphalt binder, VG-30(Viscosity Grade) and
of the specimens increase up to a maximum, then HiMA (Highly Modified Asphalt) were selected
decreases as the binder content continue to increase. in the study. The bituminous concrete (BC) mix
The Super-Pave system is applicable to virgin and was designed for 13 mm nominal size aggregate
recycled, dense-graded, Hot Mix Asphalt (HMA), gradation for Marshall Mix Design as per Indian
with or without modification. In addition, the Specifications (MoRTH 2013) and for Super-
Super-Pave performance tests are applicable to the Pave mix designed aggregate gradation was
characterization of a variety of specialized paving selected from Asphalt institute manual series No.
mixes such as Stone Matrix Asphalt (SMA). It can 02 (MS02).
be used when constructing new surface, binder, 4. Mix Design
and base layers, as well as overlays on existing
pavements. Through materials selection and 4.1 Super-Pave Mix Design for Asphalt
mix design, it directly addresses the reduction Mixtures
and control of permanent deformation, fatigue The following summarizes the current Super-Pave
cracking, and low-temperature cracking. It also mix design procedure:
explicitly considers the effects of aging and 4.1.1 Selection of Asphalt Binder
moisture sensitivity in promoting or arresting the
Selection of asphalt binder was based on seven-
development of these three distresses. (SHRP-
day average maximum temperature. Super-Pave
A-407).
mix design is done for Delhi where the Seven-day
2. Objectives average maximum air temperature was observed
 To compare the performance based mix to be 44°C and the one-day air temperature is
design with Traditional mix design method observed to be 4°C and the latitude of the Delhi
i.e. Marshall Mix design method for different is 28.62°.
bituminous mixes prepared with conventional From the above data, the Maximum pavement
as well as modified binder. temperature is calculated 66°C and the minimum
2 Highway Research Journal, January - June, 2018

pavement temperature is 4°C. In the present Table 1 Conformity of Available VG Grade

study traffic range is consider 30-100 ESALs Bitumen’s with Indian Tentative Specifications for
(millions) and fast transit. According to these Performance Grade Bituminous
data PG 70-10 grade bitumen should be used in S. No Grade Source Conformity with
mix design but for higher traffic load and higher Indian Tentative
traffic speed one higher grade of bitumen should Specifications
be used i.e. PG70-16. This performance grade 1 Gulf Petrochem- 1 PG70-15
2 Gulf Petrochem- 2 PG70-13
bitumen is equivalent to VG30 grade bitumen.
3 Ooms polymer PG70-15
For comparative purpose, highly modified
4 IOCL-Mathura PG76-16
bitumen is also taken in this study. Conformity Refinery- 1
of available VG grade bitumen’s with Indian 5 VG30 IOCL-Mathura PG70-16
Tentative Specifications for performance grade Refinery- 2
bituminous is given in Table 1. 6 IOCL-Mathura PG70-15
Refinery- 3
Mixing and compacting temperature of 7 ESSAR PG76-15
bituminous mixes is calculated by equi-viscous 8 Developed in lab PG70-15
method. Mixing temperature is determined 9 TIKI TAR PG 64-18
where the viscosity –temperatures line crosses 10 VG10 IOCL-Mathura PG 64-18
the mixing viscosity range of 0.17±.02 Pa- Refinery
s. Compaction temperature is determined 11 VG20 Developed in lab PG70-15
12 TIKI TAR PG82-13
where the viscosity –temperatures line crosses VG40
13 Developed in lab PG70-13
the mixing viscosity range of 0.28±.03 14 VG50 Developed in lab PG 76-12
Pa-s. The temperature corresponding to 0.28±.03
Pa-s i.e. compaction temperature was observed 4.1.2 Selection of Aggregate
to be 152°C and temperature corresponding The aggregate for super-pave should satisfy the
to 0.17±.02 Pa-s i.e. mixing temperature was following two aggregate properties:
observed to be 162°C for VG30 grade bitumen.  Consensus Properties
The temperature corresponding to 0.28±.03  Source Aggregate Properties
Pa-s i.e. compaction temperature was observed
to be 160°C and temperature corresponding 4.1.3 Selection of Design Aggregate Structure
to 0.17±.02 Pa-s i.e. mixing temperature was The selection of the design aggregate structure
observed to be 175°C for Highly Modified involved the following steps:
Asphalt (HiMA). Establishing Trial Blends: is given in Table 2.
Table 2 Aggregate Gradation for Super-Pave Mix Design Method
Combined Grading Adopted Control Points Restricted Zone

Sieve Size 0.45 Power
(mm)↓ Trial Trial Trial Values Min Max Min Max
Blend 1 Blend 2 Blend 3
19 100 100 100 100 100 100
13.2 90 89 94 84.90 90 100
9.5 80 76 87 73.20 90
4.75 62 57 71 53.60
2.36 47 44 55 39.10 28 58 36.1 36.1
1.18 38 35 44 28.60 25.6 31.6
0.6 26 24 29 21.10 19.6 23.6
0.3 18 17 20 15.50 15.5 15.5
0.15 13 12 15 11.30
0.075 4 4 4 8.30 2 10

Super-Pave uses 0.45 power gradation charts to

define a permissible gradation. This chart uses a (2)
unique graphing technique to judge the cumulative
Where, Vba = volume of absorbed binder, cm /cm of mix
3 3
particle size distribution of a blend of aggregate. Pb = percent of binder (assumed 0.05),
The ordinate of the chart is present passing. The Ps = percent of aggregate (assumed 0.95),
abscissa is an arithmetic scale of sieve size in Gb = specific gravity of binder (assumed 1.02),
millimeters, raised to the 0.45 power and shown Va = volume of air voids (assumed 0.04 cm3/cm3 of
mix)
in Table 3.
The volume of the effective binder (Vbe) can be
Table 3 Proportion of Aggregate for Different Trail
determined from this equation:
Blend
Vbe = 0.176-0.0675× [ln (Sn)] (3)
Proportions of Aggregate, %
Blends Stockpile Stockpile Stockpile Stockpile Where, Sn = the nominal maximum sieve size of
#1 #2 #3 #4
the aggregate blend (in mm)
Trial 15 20 26 39
blend #1 Finally, the initial trial asphalt binder (Pbi) content
Trail 18 22 24 36 is calculated from this equation:
Blend #2
Trail 10 12 34 44 (4)
Blend #3
4.1.4 Computation of trail binder content for each
Trial Blend Where, Pbi = percent (by weight of mix) of binder
 Effective Specific Gravity: The Gse of each
Ws = weight of aggregate, grams
blend is estimated using equation:
Gse = Gsb + 0.8 (Gsa-Gsb) . (1)
Where, Gse = Effective specific gravity of (5)
the aggregate blend.
Gsb = Bulk specific gravity of the
aggregate blend.
Gsa = Apparent specific gravity of 4.1.5 Specimen Compaction and Evaluation of
the aggregate blend. Trial Blends
 Bitumen Absorption: The percentage of The number of gyrations for initial compaction,
bitumen absorbed by the mineral aggregate design compaction, and maximum compaction are
expressed as percentage of bitumen absorbed taken from Table 5 and given below.
by the aggregate by the weight of the
aggregate. The binder absorbed into aggregate Nini = 9 gyrations; Ndes = 125 gyrations; Nmax = 205
(Vba) is estimated using the equation: gyrations
Table 4 Estimated Volumetric Properties @ 4% Air Voids for Different Mixes

Estimated volumetric properties @ 4% Air Voids for BC prepared with VG30
Blend Trial %AC Est. %AC %Air Voids %VMA %VFA D.P. %Gmm @ N = 9
1 4.4 5.34 4.0 15.12 73.55 0.73 86.48
2 4.4 5.03 4.0 14.72 72.83 0.74 86.22
3 4.4 4.92 4.0 14.42 72.26 0.88 85.78
Estimated volumetric properties @ 4% Air Voids for BC prepared with HiMA
Blend Trial %AC Est. %AC %Air Voids %VMA %VFA D.P. %Gmm @ N = 9
1 4.5 5.42 4.0 14.92 73.18 0.72 87.45
2 4.5 5.25 4.0 14.66 72.71 0.70 86.58
3 4.5 5.06 4.0 14.21 71.85 0.86 85.58

Estimated properties are compared against the example, Trial Blend 3 is selected as the design
mixture criteria. For the design traffic a nominal aggregate structure.
maximum particle size, the volumetric and 4.1.6 Select Design Asphalt Binder Content
densification criteria are:
A minimum of two specimens are compacted at
% Air Voids 4.0% each of the following asphalt contents:
% VMA 14.0% (12.5 mm nominal  Estimated binder content
mixture)  Estimated binder content ± 0.5%, and
12.0% (25.0 mm nominal  Estimated binder content + 1.0%.
mixture) For Trial Blend 3 of 12.5 mm NMAS with VG30,
% VFA 65% - 75% (≥30 × 106 ESALs) the binder contents for the mix design are 4.42%,
4.92%, 5.42%, and 5.92%. For Trial Blend 3 of 12.5
% Gmm @ Nini less than 89%
mm NMAS with HiMA the binder contents for the
Dust Proportion 0.6 - 1.2 mix design are 4.56%, 5.06%, 5.56%, and 6.06%.
After establishing all the estimated mixture A minimum of two specimens is also prepared for
properties, the designer can observe the values determination of maximum theoretical specific
for the trial blends and decide if one or more gravity at the estimated binder content. The average
are acceptable, or if further trial blends need to densification curves for each trial asphalt binder
be evaluated. All the Blends are acceptable. The content are graphed for comparative illustration in
VMA, VFA, D. P., and Nini criteria are met. For this Fig. 1.
Fig. 1 Average Denisification Curves for BC

Blend 3 with VG30 and HiMA, Varying Asphalt Binder Content
Table 5 Mixture Volumetric Properties @ Ndes for BC Mixes for Different Binder
Mixture Volumetric Properties @ Ndes for BC NMAS with VG30

%AC %Air Voids %VMA %VFA D.P. %Gmm @ %Gmm @
N=9 N = 125
4.4 5.98 16.14 62.98 0.98 84.50 94.02
4.9 4.60 15.70 70.70 0.88 85.21 95.40
5.4 3.98 15.26 78.19 0.80 85.91 95.98
5.9 3.50 15.36 85.28 0.73 87.08 96.50
Contd.....

(Contd. from pre- page)

Mixture Volumetric Properties @ Ndes for BC NMAS with HiMA
%AC %Air Voids %VMA %VFA D.P. %Gmm@ %Gmm @
N=9 N = 125
4.56 5.38 15.73 65.82 0.95 83.53 94.62
5.06 4.17 15.44 73.01 0.86 85.11 95.83
5.56 3.25 15.32 78.79 0.78 86.01 96.75
6.06 3.25 15.49 85.40 0.72 87.02 96.75
Table 6 Design Mixture Properties at OBC for BC with VG30 and HiMA
Mix Property Result

Criteria
BC with VG30 BC with HiMA
Air voids 4 4 4%
%VMA 15.49 15.42 14%
%VFA 74.24 74.07 65-75
Dust proportion 0.84 0.87 0.6-1.2
% Gmm at Nini 85.54 85.27 <89%
Bulk density 2.39 2.40 -
4.1.7 Nmax Verification produce permanent deformation. Since Nmax
Super-Pave specifies a maximum density of 98% represents a compactive effort that would
at Nmax. Specifying a maximum density at Nmax be equivalent to traffic much greater than
prevents design of a mixture that will compact the design traffic, excessive compaction will not
excessively under traffic, become plastic, and occur.
Table 7 Percent Theoretical Specific Gravity at Maximum Number of Gyration
BC with VG30 BC with HiMA
Gyrations Specimen 1 Specimen 2 Avg. Specimen 1 Specimen 2 Avg.
Ht,mm %Gmm Ht,mm %Gmm Gmm Ht,mm %Gmm Ht,mm %Gmm Gmm
5 134.4 85.26 134.2 85.46 85.36 132.5 85.96 132.3 85.88 85.92
9 133.9 85.62 133.7 85.81 85.72 132.0 86.29 131.8 86.22 86.26
10 133.8 85.71 133.6 85.90 85.80 131.9 86.38 131.7 86.31 86.34
15 133.1 86.16 132.9 86.34 86.25 131.3 86.79 131.1 86.73 86.76
20 132.5 86.60 132.3 86.79 86.70 130.7 87.20 130.5 87.15 87.18
30 131.3 87.50 131.1 87.67 87.59 129.5 88.03 129.3 87.99 88.01
40 130.0 88.40 129.8 88.56 88.48 128.4 88.86 128.2 88.84 88.85
50 128.8 89.29 128.6 89.45 89.37 127.2 89.69 127.0 89.68 89.69
60 127.5 90.19 127.3 90.33 90.26 126.0 90.52 125.8 90.53 90.52
80 125.0 91.98 124.8 92.11 92.04 123.6 92.17 123.4 92.22 92.19
100 122.5 93.77 122.3 93.88 93.83 121.3 93.83 121.1 93.91 93.87
120 120.0 95.57 119.8 95.65 95.61 118.9 95.49 118.7 95.60 95.54
125 119.4 96.02 119.2 96.10 96.06 118.3 95.90 118.1 96.02 95.96
150 118.1 96.3 117.9 96.4 96.31 117.6 96.1 117.4 96.2 96.19
205 115.5 96.81 115.3 96.93 96.87 116.2 96.69 116.0 96.69 96.69
Gmb at 5 2.14 2.145 2.180 2.178
Gyration
Gmb at 125 2.41 2.412 2.432 2.435
Gyration
Gmb at 205 2.43 2.433 2.450 2.452
Gyration
Gmm 2.51 2.51 2.536 2.536

4.2 Marshall Method of Mix Design approximate design bitumen content is 5.32% by
Marshall Asphalt concrete mix design procedure weight of aggregate. Test should be planned on
(ASTM D1559) was used to find the optimum the basis of 0.5 % increment of asphalt content,
binder for both VG30 and HiMA binder for with at least two asphalt contents above and at
Bituminous concrete. In determining the design least two asphalt content below the approximate
binder content for a particular blend by Marshall design bitumen content. Prepare three specimens
Method, a series of test specimens is prepared for at each asphalt content by giving 75 blows on
a range of different asphalt contents so that the each face and each test specimen is subjected to
test data curve show well defined relationships. Specimen height determination, Bulk specific
The expected Design asphalt content, in percent gravity determination, Density and voids
by total weight of mix, could then be estimated analysis, Stability and flow test and the average
to be approximately equivalent to the percent of for the replicates at the same binder content were
aggregate in the final gradation passing the 0.075 also calculated and the test results are given in
mm sieve. For Bituminous concrete (Grade 2) Tables 8 and 9.
Table 8 Volumetric Properties of Bituminous Concrete Mix Prepared with VG30
% Bitumen by Average Gmm (gm/ % Va %VMA %VFB Marshall Marshall

Weight of Mix Gmb (gm/cc) cc) Stability (kN) Flow (mm)
4.45 2.39 2.55 6.27 16.93 62.95 8.65 2.96
4.95 2.402 2.53 5.06 16.95 70.16 11.25 3.03
5.45 2.411 2.52 4.33 17.08 74.68 12.19 3.25
5.95 2.42 2.5 3.20 17.21 81.41 11.48 3.48
6.45 2.409 2.475 2.55 17.92 85.80 10.16 4.93
Table 9 Volumetric Properties of Bituminous Concrete Mix Prepared with HiMA
% Bitumen Average Gmm % Va %VMA %VFB Marshall Marshall

by Weight of Gmb (gm/ (gm/cc) Stability (kN) Flow (mm)
Mix cc)
4.45 2.41 2.57 6.05 16.24 62.74 10.25 2.76
4.95 2.43 2.55 4.56 15.99 71.48 13.261 2.98
5.45 2.44 2.53 3.37 16.02 75.23 15.19 3.41
5.95 2.43 2.51 3.19 16.92 81.13 12.96 3.98
6.45 2.41 2.49 3.14 17.92 82.46 10.16 4.73
The optimum binder content for BC and DBM 5. VOLUMETRIC AND MECHANICAL
with VG30 & HiMA based on maximum density, PROPERTIES AT OPTIMUM BINDER CONTENT
maximum stability and 4% air voids is summarized The volumetric properties of compacted paving
in Table 10. mixture are important criteria by which the
Table 10 Optimum Binder for BC for Both the Mix quality of an asphalt mixture is evaluated.
Design Method and Binder Type The volumetric properties are determined
using mass and/or volume measurements of
Optimum Binder Content, %
mixture and its constituent components (binder,
Mixture Type Marshall Mix Super-pave Mix aggregate, air).
Design Design
6. Performance Test
BC with VG30 5.65 5.22
6.1 Moisture Susceptibility Tests
BC with HiMA 5.40 5.03
The moisture susceptibility of bituminous concrete

mixes for conventional as well as modified binder in the NCHRP 9-19 Project as a simple performance
was evaluated by measuring the Tensile Strength test to evaluate the rutting susceptibility of HMA
Ratio (TSR) as per ASTM: D 6931-12. The Indirect mixes. the FN test was conducted at a repeated
Tensile Strength of (ITS) of the mix is determined compressive Haver-sine loading (1 cycle with 0.1
before and after conditioning of Specimens and s loading time and 0.9 s resting time) to measure
tensile strength ratio is determined as the ratio vertical accumulated permanent strains as a function
of average tensile strength of dry subset to the of loading cycles. The NCHRP 9-19 Project
average tensile strength of Conditioned subset. The recommended conducting the FN test at an effective
air voids levels in all these tests were maintained pavement temperature in an unconfined mode with
at 7±1%. For conditioning the specimens, these a vertical stress between 69 and 207 kPa.
were immersed in water bath at 60˚C for 24h and
In this study, the FN test was performed by Asphalt
afterwards kept at 25˚C for 2h.
Mixture Performance Tester, in an environmental
Table 11 Volumetric Properties at Optimum Binder chamber at a temperature of 60°C (130°F) and
Content an applied low stress of 69 kPa (10 psi) and for
Marshall Mix Super-Pave Mix high stress of 207 kPa (30psi) All specimens were
Properties Design Design
conditioned at 60°C (140°F) for 4 hours prior to
VG30 HiMA VG30 HiMA
Marshall Stability, 11.85 13.09 13.64 15.69
testing. The testing was performed according to
kN @ 25 ˚C AASHTO TP 79, “Standard Method of Test for
Flow, mm 3.12 3.96 3.25 3.74 Determining the Dynamic Modulus and Flow
Air Voids, % 3.82 4.15 4.04 4.11 Number for Asphalt Mixtures Using the Asphalt
Specific Gravity 2.413 2.438 2.41 2.415
%VMA 14.71 14.26 15.49 15.42
Mixture Performance Tester (AMPT)” (AASHTO,
%VFB 74.04 70.89 74.24 74.07 2009) and the test was set to terminate at 10,000
loading cycles or accumulated 50,000 micro-
6.2 Resilient Modulus Test
strains, whichever came first.
Resilient modulus also known as Indirect Tensile
Stiffness Modulus (ITSM) is an important 7. Result and Discussion
parameter to evaluate performance of a mix and 7.1 Indirect Tensile Strength
to assess the pavement response to traffic loading. Table 12 shows the results of dry and wet ITS for
The repeated load indirect tension resilient bituminous concrete prepared by conventional
modulus test of bituminous mixtures is conducted as well as modified binder. The dry ITS value is
through repetitive application of compressive maximum compressive load born by sample before
loads in haver sine waveform. The compressive
failure while the wet ITS is load after soaking
load is applied along vertical diametric plane
the sample in water for 24h at 60°C. From these
of cylindrical specimen of mix and resulting
results it is clear that as the sample prepared by
deformation are measured. All the specimens for
Super-Pave mix design method have higher values
resilient modulus test were prepared at 5 % air
as compared those sample which are prepared by
voids. The specimen used for resilient modulus
using Marshall method irrespective of binder type.
test is a cylindrical specimen that is 100 mm
diameter by 63.5 mm thick. The bituminous Table 12 Indirect Tensile Strength of Different Mix
mixture specimen is normally loaded to a stress Design and Different Binder
level between 5 and 20% of indirect tensile Type of Mix design Avg. Dry Avg. Wet
strength using repeated pulse load with subsequent Binder Method ITS (kPa) ITS (kPa)
rest period usually 0.1 sec loading and 0.9 sec
Marshall Method 596 508
rest period. The specimens were tested at 25˚C to
VG30 Super-Pave 645 568
40˚C in accordance with ASTM: D7369-11 after
Method
their conditioning on environmental chamber at
selected test temperature for 6h. Marshall Method 798 710
6.3 Flow Number Test HiMA Super-Pave 824 763
Method
The FN test protocol was developed and introduced

7.2 Tensile Strength Ratio (TSR)

The moisture susceptibility is determined as ratio of
the average Wet ITS to Average Dry ITS. The TSR
for HiMA is higher than conventional binder i.e.
VG30 irrespective of mix design method. TSR
values are higher for Super-Pave mix design
method when compared to Marshall Mix design
method irrespective of Binder type. In other words,
Mix prepared by Super-Pave mix design method
shows higher resistance to moisture damage after
pavement had been laid and also during its service
life. Fig. 2 shows the variation of Tensile Strength
Ratio for mix design method and binder type.
Fig. 4 Resilient Modulus of HiMA for BC
improves the stiffness modulus of a mix substantially

when compared with mixes which are prepared
with conventional binder It is also found that as the
temperature increases the stiffness modulus decreases
irrespective of binder type and mix design method.
Resilient modulus values obtained by Marshall
Method of mix design are lower than by obtained
through Super-Pave mix design method irrespective
Fig. 2 TSR Variations with Binder Type and Mix of type of binder i.e. conventional or modified binder
Design Method at different temperature.
7.3 Resilient Modulus 7.4 Flow Number

The flow number are evaluated at different air
Figs. 3 and 4 shows the comparison of resilient
voids and at two deviator stress level for BC and
modulus for Conventional as well as modified
DBM mixtures prepared with conventional binder
binder sample prepared by Marshall and Super-Pave
as VG30 and Modified binder i.e. HiMA. The Flow
mix design method at 25˚C, 30˚C, 35˚C and 40˚C
number value for all these variables are evaluated
temperature. It is found that the modification in binder
for Marshall Mix Design and Super-Pave Mix
Design Method.
7.4.1 Effect of Air Voids on Flow Numbers
The relation between the air voids and flow number
are shown in Figs. 4-5. As indicated from the
experimental results, the flow number decreases
with the increase of air voids for BC sample
prepared with Conventional as well as Modified
binder irrespective of Mix design method.
7.4.2 Effect of Binder Type on Flow Number
The Figs. 6(a) and 6(b) shows that the Modification
of binder has a direct influence on the flow
numbers. Higher flow number values are obtained
with Modified binder mixes when compared with
conventional binder at different air voids and
Fig. 3 Resilient Modulus of VG30 for BC different deviator stresses.

7000 12000
207 kpa Marshall 207 kpa Marshall
6000 10000
69 kpa Marshall 69 kpa Marshall
207 kpa Superpave
5000 207 kpa Superpave
69 kpa Superpave
8000
Flow Number
Flow Number
4000 69 kpa Superpave
6000
3000
4000
2000
1000 2000
0 0
4 7 9 4 7 9
Air Voids% Air Voids%
Fig. 5 Flow Number Variation at Different Air Voids for VG30 and HiMA
Flow Number (cycles)
Binder Type Binder Type
Fig. 6 (a) Flow Number Variation with Binder Type.@ 4 % Air Voids and 7 % Air Voids
Flow Number (Cycles)
Binder Type
Fig. 6 (b) Flow Number Variation with Binder Type.@

9 % Air Voids .
Fig.7(b) Flow Number at 207 kPa
7.4.3 Effect of mix design method on Flow

Numbers: Figs. 7(a) and 7(b) shows a comparison
of the flow numbers for Marshall Mix design
method and Super-Pave method. The results
indicated that the Super-Pave mix method have
higher flow number values as compared Marshall
Fig. 7(a) Flow Number at 69 kPa mix design method for conventional binder as

well as modified binder at different air voids and  The VG30 and HiMA binders satisfied the
different deviator stresses. requirements of IS-73-2013 and IRC: SP 53-
7.4.4 Effect of Deviator Stress on Flow Number 2010 respectively. Softening point for HiMA
binder is 10˚C higher and complex modulus
When low deviator stress of 69 kPa (10 psi) is is found to be 5˚C higher than VG30.
applied, higher flow number values are obtained  Under the specified test conditions unaged
for mixes prepared by conventional as well as VG30 sustained up to 79.5˚C whereas unaged
modified binder. But when the stress level is HiMA binder sustained 84.5˚C i.e. 5˚C higher
increased to level of 207 kPa then the lower flow than VG30. similar trend is observed after
number values are observed for mixes. It shows aging these binders in rolling thin film oven.
that as the deviator stress are increased the Flow  Loss in mass after RTFOT is observed lower
number values decreased irrespective of binder for HiMA than VG30.
type and mix design method.
 Marshall method of mix design indicated
0.25% lower optimum binder content of
HiMA than VG30 for BC mixes.
 OBC is further found to be reduced 0.25 to
0.27 % in case of Super-Pave mix design
method for BC mixes. In general, OBC

for both the mixes with VG30 and HiMA
decreased with respect to Marshall method of
mix design.
 Retained stability and TSR are found to be
higher for Super-Pave mix design method
as compared to Marshall mix design method
irrespective of binder type.
 Resilient modulus value is higher for HiMA as
Fig. 8 (a) Flow Number for VG30 compared to VG30 for all mixes irrespective
of type of mix and method of mix design.
 Resilient modulus value obtained by Marshall
method are lower than by obtained through
Super-Pave mix design method irrespective
of type of binder used.
 The flow number decreases with the increases

of air voids irrespective of mix type, binder
type and method of mix design.
 The deviator stress (69kpa-207kpa) level
indicated that higher flow number is obtained
at lower stress level as compared to higher
stress level applied on BC mixes prepared
with VG30 and HiMA.
 The Flow Number for Marshall mix design
and Super-Pave mix design method indicated
Fig. 8(b) Flow Number for HiMA that the Super-Pave mix design method have
higher flow number values as compared to
8. Conclusion Marshall mix design method irrespective of
Based on the experimental study, following binder type and mix type at different air void
conclusions are drawn: level and deviator stress level.

References Vol. 11, No. 2, June 2013.

1. Neethu Roy, A. Veeraragavan & J. Murali 10 F. Jasim, “Comparison Between Marshall and
Krishnan, “Influence of Confinement Pressure and Super-Pave Mixtures Design”, Al-Qadisiya
Air Voids on the Repeated Creep and Recovery of Journal for Engineering Sciences, Vol. 5, no. 4,
Asphalt Concrete Mixtures”, International Journal pp. 394-406, 2012.
of Pavement Engineering Vol. 17, 2016 - Issue 2. 11. Meor O. Hamzah .et.al (2008) Anand J.
2. Mohammadreza Mirzahosseini; Yacoub M. Puppala, M.ASCE; Laureano R. Hoyos; and
Najjar; Amir H. Alavi; and Amir H. Gandomi, Ajay K. Potturi, “Resilient Moduli Response of
“Next-Generation Models for Evaluation of the Moderately Cement-Treated Reclaimed Asphalt
Flow Number of Asphalt Mixtures”, Internation Pavement Aggregates”, Journal of Materials in
Journal of Geomech., 2015, 15(6): -1—1 Civil Engineering © ASCE / July 2011
3. Veena Venudharan, Krishna Prapoorna Biligiri, 12. Amir Hossein Gandomi; Amir Hossein Alavi;
“Estimation of Phase Angles of Asphalt Mixtures Mohammad Reza Mirzahosseini; and Fereidoon
Using Resilient Modulus Test”, Construction and Moghadas Nejad, “Nonlinear Genetic-Based
Building Materials 82 (2015) 274–286. Models for Prediction of Flow Number of
4. Ekarizan Shaffie, Juraidah Ahmad, Ahmad Asphalt Mixtures”, Journal of Materials in Civil
Kamil Arshad, Dzraini Kamarunb, “Evaluation Engineering © ASCE / March 2011, 23(3):
of Volumetric Properties and Resilient Modulus 248-263
Performance of Nanopolyacrylate Polymer 13. Amir Hossein Alavi , Mahmoud Ameri ,
Modified Binder (NPMB) Asphalt Mixes” Journal Amir Hossein Gandomi , Mohammad Reza
Teknologi (Sciences & Engineering) 73:4 (2015) Mirzahosseini, “Formulation of Flow Number
21–25. of Asphalt Mixes using a Hybrid Computational
5. Khodaii, M. Khedmati, H.F. Haghshenas, and Method” Construction and Building Materials 25
M. Khedmati, “Statistical Evaluation of Hot (2011) 1338–1355.
Mix Asphalt Resilient Modulus Using A Central 14. M. Asi and Y. Khalayleh, “Adaptation of Super-
Composite Design”, International Journal of Pave Asphalt Concrete Mix Design Procedure
Pavement Research and Technology • December to Jordan Climatic and Traffic Conditions”,
2014. Technical Notes, International Journal of
6. Alex K. Apeagyei, “Flow Number Predictive Pavement Research and Technology, Vol. 4, no. 3,
Models from Volumetric and Binder Properties”, pp 154-161, 2010.
Construction and Building Materials Vol. 64, 14 15. Dr. Namir G. Ahmed Eng. Nour M. Ismai
August 2014, Pages 240–245 “Comparative Evaluation for Mix Design of
7. L F Walubita, J Zhang, A E Alvarez, X Hu, Marshall and Super-Pave Methods", Journal of
“Exploring the Flow Number (Fn) Index as a Engineering and Development, Vol. 13, No. 2,
Means to Characterize the HMA Permanent June 2009.
Deformation Response under FN Testing”, 16. S-J Lee, S. N. Amirkhanian, S-Z Kwon, “The
Journal of the South African Institution of Civil Effects of Compaction Temperature on CRM
Engineering Vol 55 No 3, October 2013, Pages Mixtures Made With The SGC and the Marshall
103–112, Paper 940. Compactor”, Construction and Building Materials
8. Jun Zhang, Allex E. Alvarez , Sang Ick Lee , Angela Journal (CBMJ), Vol. 22, 2008, pp. 1122–1128.
Torres , Lubinda F. Walubita, “Comparison of 17. Su Jian Ji and Dr. Mofreh Saleh, “Investigation
Flow Number, Dynamic Modulus, and Repeated of Factors Affecting Resilient Modulus for Hot
Load Tests for Evaluation of HMA Permanent Mix Asphalt”, Meor O. Hamzah, and Teoh C”,
Deformation”, Construction and Building Yi, Effects of Temperature on Resilient Modulus
Materials 44 (2013) 391–398. of Dense Asphalt Mixtures Incorporating Steel
9. H. Ziari, H. Divandari, “Presenting Asphalt Slag Subjected to Short Term Oven Ageing”,
Mixtures Flow Number Prediction Model Using International Journal of Civil, Environmental,
Gyratory Curves”, International Journal of Civil Structural, Construction and Architectural
Engineering, Transaction A: Civil Engineering, Engineering Vol: 2, No:10, 2008.

Effect of Polypropylene granules in Asphalt Mix
prepared with varying Packing characteristics
M. Satish Goud1 A. Ramesh2 V. Venkat Ramayya3 M. Kumar4
ABSTRACT
Bituminous concrete pavements are subjected to failures/damages during the design period. One
among the causes for damage is due to commercial vehicles carrying heavy loads with high frequencies.
Predominately failures in the pavement structure are in the form of rutting and fatigue and are termed
as structural failure. In order to improve the structural characteristics of hot mix asphalt layer,
additives are used as bitumen alone cannot cope up with structural characteristics of bituminous mix.
This study aimed to investigate the performance of Hot Mix Asphalt (HMA) mixtures designed using
Bailey method of gradation and compared with conventional method of gradation. Bailey method allows
aggregate interlocking which acts as backbone to the structure and develop strong aggregate skeleton
for durability and rut resistance.
Marshall Parameters are used to compare the packing characteristics of bituminous mix prepared
through Bailey method and Conventional method (MoRTH Specifications). Indirect Tensile Strength,
resilient modulus and rutting properties were used for comparing the packing characteristics of
bituminous concrete mix when added with Polypropylene. Polypropylene (PP) granules exhibited good
resistance towards rutting. The results infer that conventional mix prepared using mid-point gradation
exhibit less resistance towards rutting when compared with Bailey gradation and addition of PP.
Bituminous Concrete mix prepared with Bailey approach and when added with PP granules has good
stiffness value when compared with above mix combinations.
1. Introduction Vavrik et al (2002). The Bailey method of gradation

Hot Mix Asphalt (HMA) is the most commonly used selection takes into account the packing characteristics
material for paving applications in India. It primarily of aggregates through coarse aggregate ratio, coarser
consists of asphalt binder and mineral aggregates. portion of fine aggregate ratio and finer portion of
HMA is the composite material consisting of fine aggregate ratio. The parameters in this method
aggregate particle with different sizes, an asphalt are related directly to voids in the mineral aggregate
binder and air voids. When mineral aggregates are (VMA), air voids and compaction properties.
bound with asphalt binder, it acts as stone framework Pavement failure is defined in terms of decreasing
that provides strength and toughness to the mix serviceability caused by the development of cracks
Alshamsi (2006). The Bailey Method rely on how and ruts. Failures of flexible pavements are caused
the coarse and fine aggregates pack together to form due to many reasons or combination of reasons.
a strong aggregate skeleton. The Bailey Method Rutting and fatigue cracking distresses occurs during
was developed by Robert Bailey, a retired materials the service life of pavement. Rutting is defined as
engineer for the Illinois Department of Transportation permanent deformation along wheel path and mainly
1
PG Student Dept. of Civil Engineering, VNR Vignana Jyothi Institute of Engineering
2
Associate Professor, E-mail: aramesh28@gmail.com & Technology, Bachupally, Hyderabad.
3
Asst. Professor, Dept. of Civil Engineering, Mahatma Gandhi Institute of Technology, Gandipet, Hyderabad.
4
Professor, Dept. of Civil Engineering, University College of Engineering (Autonomous), Osmania University, Hyderabad.

Goud, Ramesh, Ramayya & Kumar on
Effect of Polypropylene Granules in Asphalt Mix prepared with varying Packing Characteristics
occurs due to heavy wheel loads. Fatigue is caused distribution. The outcome of this study reveals that
by failure of the surface layer or base due to repeated aggregate size distribution played a significant
traffic loading. In order to improve the distress role in packing characteristics. Rutting is a form
additives are added in bituminous concrete mix. In of failure experienced on the bitumen surface
the year 2012, the global market for polypropylene road in most of the developing countries. Higher
was about 55 million tonnes; Polypropylene is the values of stiffness are taken to measure of better
world's second-most widely produced synthetic resistant to rutting. Manjunath and Poornachandra
plastic, after polyethylene. Polypropylene is an (2014) investigated the properties of HMA mixtures
economical material that offers combination of designed using Bailey method of gradation and
outstanding physical, chemical, mechanical, and compared with conventional method of gradation.
thermal properties when compared with any other The result indicates that there is an increase in VMA
thermoplastic materials. To improve the quality of all the mixtures in Bailey gradation and rutting
of bitumen, the rheological properties of bitumen characteristics were improved by Bailey method
are modified by blending with organic synthetic of gradation. Kockal and Kofteci (2016) worked
polymers like plastics and rubber. on durability aspects through different aggressive
1.1 Objectives environmental conditions in laboratory to minimize
deterioration due to the damage. Al-Hadidy and
The objectives set for the present research are
Yi-Qiu (2009) observed that the properties of PP
 To formulate aggregate gradation using Bailey modified asphalt mixtures and their performance
method and to achieve the required volumetric is better when compared to conventional mixtures.
properties over conventional method of The temperature susceptibility can be reduced
gradation. by the inclusion of PP in the asphalt mixture.
 To determine the optimum binder content Chaturabong and Bahia (2017) have used Hamburg
for Conventional mix and mix prepared with Wheel Tracking (HWT) is prominently used for
Bailey Approach, also to arrive on the optimum rutting resistance test. The results indicate that the
content of PP granules when added to above wet HWT currently specified in the evaluation of
mix combinations. AASHTO T 324 Tsai et al (2016) can be used as
 To investigate Tensile Strength and Resilient a substitute to the dry HWT test in the creep stage.
modulus of bituminous mixtures designed using The wet HWT test is used to detect the moisture
conventional method and Bailey approach. sensitivity by measuring cohesion or adhesion of
asphalt mastics. Vavrik et al (2002) provides Bailey
To determine rutting and stripping characteristics
method of gradation selection based on the packing

of above mix combination for development of

characteristics of aggregates. The parameters used
a better rut resistant bituminous mixture using
in Bailey method are coarse aggregate ratio, coarser
Bailey approach.
portion of fine aggregate ratio and finer portion
2. Literature Review of fine aggregate ratio are the three parameters
A brief literature on the research topic is presented considered. Aggregate interlock will provide a rut-
in this section. Graziani et al (2012) studied the resistant mixture. In order to ensure that the mixture
procedures and parameters of Bailey method contains adequate asphalt binder, VMA is changed
adopted in USA. Test results emphasize that usage by changing the packing of the coarse and fine
of Bailey parameters based on US and European aggregates. The method develops a strong aggregate
specifications and the study also examined different skeleton for rut resistance, durability, and adequate
types of sieves used in Bailey’s method of gradation. voids in the mineral aggregate. A strong aggregate
Aurilio et al (2005) described the methodology in structure is important as the aggregate supports
detail with some examples relating to real life to most of the compressive forces.
help designers and contractors to better understand
the mixes. Use of Bailey method will ensure that 2.1 Critical Appraisal
coarse aggregate interlock and control of aggregate Researchers have improved the performance of
packing, allowing the designer to specify desired the bituminous concrete mix using additives or
mixture properties. This will help to eliminate admixtures. In the present study, attempt was made
the trial and error method in gradation selection. to change the packing characteristics of Bituminous
Shen and Yu (2011) studied aggregate packing Concrete (BC) Grade-I using Bailey approach.
characteristics through bailey method in HMA PP granules were added at increment of 2% to the
mixture as it influences both particle size and shape weight of bitumen to the conventional mix and

bailey approach mixes. Different combinations are (i) Bailey Parameters will be evaluated for a given
made in Bituminous Concrete (BC) grade-I mix for blend as per the following formulae:
understanding the performance characteristics.
3. Methodology
The sequence of stages for achieving the study
objectives is presented below.
Stage 1 : Literature Study
Stage 2
: Aggregate gradation selection for
conventional and Bailey approaches. Where,
Stage 3 : Marshall Mix design for Conventional CA Ratio - Coarse Aggregate Ratio
mix, Bituminous mix prepared using PPhalf sieve - Percent Passing Half Sieve
Bailey approach and inclusion of PP PCS - Primary Control Sieve = 0.22 x NMPS
granules in above mix combinations.
SCS - Secondary Control Sieve = 0.22 x
Stage 4 : Performance of mix evaluated through PCS
Tensile strength, Resilient Modulus and TCS - Tertiary Control Sieve = 0.22 x SCS
Rutting test (Immersion type). PPPCS - Percent Passing Primary Control
Stage 5 : Interpretation of test results Sieve
PPSCS - Percent Passing Secondary Control
4. Experimental Program Sieve
In the current research, PP granules were added to
PPTCS -
Percent Passing Tertiary Control Sieve
bituminous concrete mix prepared with MoRTH
specifications and Bailey approach. Laboratory test The factor 0.22 was arrived after considering the
were performed on PP modified mixes of above analysis of 2-D and 3-D packing of different shaped
mix combinations for evaluation of performance particles. 2-D Analysis has shown the particle
characteristics. diameter ratio equal to 0.155 (all round) and 0.289
(all flat) with an average of 0.22 for angular and sub
4.1 Materials angular particles Vavrik et al. (2001).
Metamorphic type crushed stone aggregates and The blend was selected based on trial and error
stone dust is collected from local quarry. VG 30 method and was checked for Bailey parameters using
grade bitumen obtained from Indian Oil Corporation the formulae as indicated above. The selected JMF
Limited (IOCL), provided by GHMC (Greater was presented in gradation chart as shown in Fig. 1.
Hyderabad Municipal Corporation). Tests were
carried out in the laboratory to confirm the grades of 100
the bitumen. PP granules used in the present study
has a melting temperature of 1600C when tested in 80
the laboratory. PP granules were providing better

60
affinity for all mix combinations.
Percent Passing (%)
4.2 Gradation of the Mix 40

Bituminous Concrete, Grade-I was selected in the
present study as per MoRTH specifications. Bailey 20 Maximum Density Line
Bailey Gradation
method of gradation was implemented in the present Lower
Mid point Gradation
study apart from the conventional mix with mid-point 0
Upper
gradation adhering to MoRTH recommendations. 0 1 2 3 4 5
Bailey method considers the packing characteristics Sieve Size raised to 0.45
for achieving a strong aggregate skeleton for better
rutting resistance with adequate voids in mineral Fig.1 Gradation of the Mixes
aggregate for durability. The Bailey method of
mix design addresses issues related to mobility, 4.3 Preparation of Hot Mix Bitumen
compatibility and continuity of gradation through PP granules were added at 2% (w/w %) increments to
three parameters CA ratio, FAc ratio and FAf ratio the weight of bitumen up to 8% for Conventional mix
(AASTHO T-19). and Bailey approach (IRC:SP:53).

4.4 Preparation of Specimen motion until desired densities of mix are achieved as
4.4.1 Marshal Specimens explained in (BS EN 12697 33).
Marshall Specimens were prepared as per 4.4.2.1 Immersion type of wheel tracking device
(ASTM D 6926-04) to determine OBC (optimum An accelerated Immersion type wheel tracking test is
bitumen content) and OPC (optimum PP content). conducted in the laboratory for estimation of rutting
1200 gms of mix is required to prepare the specimen, resistance and is shown in Fig. 2. The wheel used in
aggregates were preheated before mixing with the study is steel wheel of 47 mm wide and the total
bitumen and PP granules. The mix is placed in the weight of the wheel assembly including surcharge
mould of 100 mm in diameter and compacted with is 710N and a LVDT is also provided to the side
4.54 kg rammer, falling from a height of 45.7 cm of wheel for monitoring rut depth to corresponding
with 75 blows on each face of the specimen. The number of wheel passes. Specimens were tested in
compacted specimen shall have a total thickness fully immersed condition in water at a temperature of
of 63.5 mm. Stability corrections were applied for 50oC (AASTHO T324). This test also explains on the
variation in thickness. The sample is allowed to cool stripping resistance of mix combinations used in the
for a few hours and then extracted using a sample study. The travel of the wheel path is 230 mm and the
extractor. The specimen is placed in the testing speed of the wheel for one pass is approximately 1.46
assembly after conditioning the sample in the water kmph (72 wheel passes per min and a failure criteria
bath kept at 60oC for 30 minutes duration. Load is of 20 mm rut depth is considered. B.C Grade I as per
applied at a rate of 50.8 mm/minute and tested at a MoRTH is adopted for the present study in preparation
temperature of 60oC. A load cell is used to apply the of specimens. 24102 passes will simulate traffic of
load on the specimen and an LVDT (Linear Variable 30 msa in the field Nahi et al (2011).
Displacement Transducer) was used for recording 4.4.3 Specimens for Resilient Modulus/ IDT
the flow values. The peak load was recorded by Marshall Specimens are used for conducting tensile
the DAS (Data Acquisition System) along with the test and determining the resilient modulus.
flow values. Three samples were prepared for each
percentage of bitumen varying between 4.5 to 6.5%. 4.4.3.1 Repeated load test set up
The peak load resisted by the specimen is termed Resilient modulus of all mix combinations are
as Marshall Stability value as outlined in (ASTM D determined on repeated load test set up providing
5581 – 07). The flow is noted down in terms of mm repetitive applications of compressive loads in a
or units (1 unit=0.25 mm) MoRTH. haver-sine waveform load is applied vertically for
0.1 seconds followed by a rest periods of 0.9 seconds.
4.4.2 Specimens for Rutting Test The frequencies, which control the length of the rest
The specimen was prepared in a slab compacter also period between load pulses, were selected based on
known as roller compacting device shown in Fig. the (ASTM D 4123) recommendations (1.0 Hz). The
2. The slab is provided with a volume of around magnitude of load applied was 10% of tensile strength
6000 cc (40x30x5 cm3) and with 4% air voids in obtained from IDT at 350C. The test is performed
slab. The mixes were prepared with densities arrived twice on one sample with 90 degree rotation. The
from Marshall Stability test. The specimens were horizontal and vertical deformations were measured
compacted using hydraulic pressure in oscillatory using the ±100 µm and ±5mm LVDT’s respectively.
Fig. 2 Slab Compactor and Immersion Type Wheel Rutting Equipment

Solenoid value Table 2 Physical Properties of Bitumen

cylinder
Consistency VG-30 Specifications Test
Characteristics of VG-30 standard
Penetration at 53 45 IS : 1203
Signal
250C,100 g,5 s,
Conditioner 0.1 mm, Min
Vertical Absolute Viscosity at 2507 2400-3600 IS : 1206
LVDT's 600C, Poises (Part- 2)
Load cell Kinematic Viscosity at 352.47 350 IS : 1206

1350C, cSt, Min (Part-3)
Horizontal
LVDT's Softening point 50 47 IS : 1205
(R&B), 0C, Min
Temperature
chamber Tests on residue from
rolling thin film oven
Fig.3 Repeated Load Test Set Up test:
a) Viscosity ratio at 2.5 4.0 IS : 1206
5. Results and Discussion 600C,Max (Part 2)
b) Ductility at 250C, 55 40 IS : 1208
5.1 Physical Properties of Aggregates cm, Min
Tests were conducted on aggregates and bitumen Table 3 Sample Codes for Different Mixes
for the preparation of the bituminous mixes.
Aggregates were tested for strength, toughness, Mix No. Description Code
hardness, shape and results are presented in 1 Mid-point gradation with virgin binder MV
Table 1 as per MoRTH Specification. 2 Bailey’s gradation with virgin binder BV
Table 1 Physical Properties of Aggregates 3 Mid-point gradation with PP modified binder MPp
4 Bailey’s gradation with PP modified binder BPp
Property Test Results MoRTH Method
(%) Specification of Test
(%) The Marshall’s mix design was carried out as detailed
in 4.4.1. The optimum binder content was first
Particle Combined Flakiness 17.57 Max 35% IS: 2386
shape and Elongation Part I
established for MV and BV mixes. The test results
Indices were presented through Table 4 & Fig. 4.
Strength Los Angeles abrasion 18.58 Max 30% IS:2386 Table 4 Results of Marshall’s Test
value or 17.41 Max 24% Part IV
Aggregate Impact Mix Binder Content (%)
Value Type
Max 4% Volume Max Bulk Average
Durability Soundness 8 Max 12% IS:2386 Stability of Voids Density Binder
5 cycles of Sodium Part V (kN) at (%) at (gm/cc) at Content (%)
Sulphate
MV 5% 5.4% 5.5% 5.3
Polishing Polished Stone Value 65 Min 55 Bs:812-
114 BV 5% 5.3% 5% 5.1
Water Water absorption 1.06 Max 2% IS:2386 Once the optimum binder was found for above mixes,
absorption Part III investigations were extended further to determine the
optimum PP content at the established OBC.
5.2 Bitumen Physical Properties The Marshall Quotient (MQ), the ratio of stability to
flow is used as a measure of the material’s resistance to
Grade of the bitumen was determined in the laboratory permanent deformation in service. The Marshall Quotient
as per IS : 73:2013 specifications and test results are is also known as “rigidity ratio”. Marshall Quotient for
shown in Table 2. four mix combinations is represented in Fig 5.
5.3 Marshall Stability Test Results Bituminous mixes were prepared at different
To achieve the objective in the current research and percentages of PP contents with an increment of 2.0%
to study rutting performance with reference to the to weight of bitumen. Bailey approach mix with 6%
effect of gradation and PP modification, suitable mix of PP granules was found to exhibit higher stability
sample codes have been assigned for bituminous which is 24.3 kN and the variation of Marshall
mixes and are presented in Table 3. parameters for PP modified mixes is shown in Fig 6.

8
6
BV 5
7 MV
Marshall Quotient
4 MV
6
3 BV
5 2 MPP
Air Void (%)
1 BPP
4
0
MV BV MPP BPP
3 Different Combinations
2 Fig.5 Marshall Quotient for all Combinations

1
4.5 5.0 5.5 6.0 6.5 24 MPP
BPP
Binder Content (%)
22
18
BV
MV
Stability (kN)
17 20
16
18
15
Stability (kN)
14 16
13 14
2 3 4 5 6 7 8
12 PP Content (%)
2.30
11
2.28 MPP
BPP
10 2.26
4.5 5.0 5.5 6.0 6.5
Binder Content (%) 2.24
2.40 2.22
Density (g/cc)
BV 2.20
MV
2.38
2.18
2.16
2.36
Density (g/cc)
2.14
2.12
2.34
2.10
2 3 4 5 6 7 8
2.32 PP Content (%)
7
MPP
BPP
2.30 6
4.5 5.0 5.5 6.0 6.5
Binder Content (%)
5
Air Void (%)
Fig. 4. Stability, Density and Air Voids for Different 4

MV and BV Mixes
3
5.4 Indirect Tensile Strength Characteristics
2
The indirect tensile strength test in accordance with
(AASHTO T 283) is used to determine the tensile 1
2 3 4 5 6 7 8
properties of the bituminous mixture which can PP Content (%)
further be related to the cracking properties of the
Fig. 6 Stability, Bulk Density and Air Voids for
pavement. Low temperature cracking, fatigue and
Different Percentages of MPP and BPP Mixes
rutting are the three major distress mechanisms.
A higher tensile strength corresponds to a stronger The formula for evaluating indirect tensile strength is
cracking resistance. At the same time, mixtures that presented below.
are able to tolerate higher strain prior to failure are St = 2P/π D t (4)
more likely to resist cracking than those unable to
tolerate high strains. Where,

St - IDT strength, MPa 5.6 Immersion Type Wheel Rutting Test

P - maximum load, N The rutting test is conducted in accordance with
AASHTO-T324 and used for evaluation of rutting
t - specimen height immediately before characteristics of the mix combinations. In this study
test, mm four slabs were prepared at mix densities, optimum
D - specimen diameter, mm binder content (OBC) and optimum PP content
(OPC) arrived from Marshall test. The specimens
5.4.1 Tensile Strength Ratio (TSR) were pre conditioned for about 2 hour’s duration in
Moisture damage in bituminous mixes refers to the order to achieve desired test temperature of 500C.
loss of serviceability due to the presence of moisture, Bituminous mix prepared with bailey approach and
the extent of moisture damage is called the moisture inclusion of PP granules exhibits better rutting and
susceptibility. The ITS test is a performance test which stripping properties when compared with other mix
is often used to evaluate the moisture susceptibility combinations as presented in Fig. 7
of a bituminous mixture. Tensile strength ratio (TSR) 0
MV
is a measure of water sensitivity and is the ratio of 2 MPP
BV
the tensile strength for water conditioned specimen, 4 BPP
to the tensile strength of unconditioned specimen 6
expressed as a percentage. A higher TSR value 8
Rut Depth (mm)

typically indicates that the mixture will perform 10 Stripping
well with a good resistance to moisture damage. The inflation point
12
laboratory results of mix combinations is tabulated in
14
Table 5.
16
TSR= S1/ S2(5) 18

20
5000 10000 15000
S1 - Tensile Strength of Conditioned sample (MPa), 0
No. of Passes
20000
S2 -Tensile Strength of Unconditioned sample Fig.7. Rutting Characteristics for Different Mix
(MPa) Combinations
Table 5 Tensile Strength Ratio for Different Mix
Combinations 6. Conclusion
Based on the experimental observations made in
Mixture Conditioned Un- TSR
Combination (S1)MPa Conditioned (S1/S2)
Marshall Test, Indirect Tensile strength Test, Resilient
(S2)MPa modulus and Immersion wheel test on four typical
combinations of mixes considered in this study, the
MV 0.796 0.95 0.84
following conclusions were made.
BV 0.857 0.934 0.92
MPP 1.231 1.449 0.85  The aggregate gradation obtained using Bailey
method of gradation was found almost in line
BPP 1.328 1.485 0.89
with the mid limits specified by the MoRTH
5.5 Resilient Modulus standards.
Repeated load test set up as per ASTM D 4123, test  The Marshall stability of Conventional and
is carried out for determining resilient modulus for Bailey method of gradation was found to be
each mix combinations in the study. The test results similar.
are presented in Table 6.  The Marshall Stability value is found to be
Table 6 Repeated Load Test Results 24.3kN for 6% Polypropylene granules at
5.1% bitumen content when designed using
Mix Binder Load Horizontal Tensile Resilient Initial Bailey method of gradation and is more than
Content (N) Deformation stress modulus tensile
% (µm) (MPa) (MPa) strain conventional bituminous mixes.
(µm)
 It is observed that air voids has decreased with
MV 5.3 950 0.006282 0.95 2416 0.000806
addition of Polypropylene, which is desirable
BV 5.1 934 0.005918 0.93 2533 0.000755 for better strength and good service life.
MPP 5.3 1449 0.007638 1.45 2672 0.001111
 The bulk density is found maximum with
BPP 5.1 1485 0.007435 1.49 2813 0.001082
2.379 gm/cc at 5% bitumen content for Bailey

Gradation mix and 5.5% bitumen content for 3. Graziani, A., Ferrotti, G., Pasquini, E., & Canestrari, F., "An
Conventional mix. Application to the European Practice of the Bailey Method
for HMA Aggregate Grading Design", Procedia-Social and
 The indirect tensile strength value of Behavioral Sciences, 53, (2012), 990–999.
unconditioned specimens has an increment of 4. Koçkal, N. U., & Köfteci, S., "Aggressive Environmental
with Polypropylene Bailey method of gradation. Effect on Polypropylene Fibre Reinforced Hot Mix Asphalt",
 Mixes has good resistance to moisture damage Procedia Engineering, 161, (2016), 963–969.
as the tensile strength ratio is meeting the 5. Manjunath, K., & NB, P. D. "Design of Hot Mix Asphalt
Using Bailey Method of Gradation", International Journal of
requirement.
Research in Engineering and Technology, 3(06), (2014),386–
 Resilient modulus for each mix combination 393.
was found using repeated load test. Test result 6. Nahi, M., Ismail, A., & Anffm, A., "Analysis of Asphalt
shows that Bailey approach using Polypropylene Pavement under Nonuniform Tire-Pavement Contact Stress
granules has higher stiffness compared to other Using Finite Element Method", Journal of Applied Sciences,
mix combinations. 11(14), (2011), 2562–2569.
7. Shen, S., & Yu, H., "Characterize Packing of Aggregate
 As the rutting temperature increases to 50˚C, Particles for Paving Materials: Particle Size Impact", Journal
the rut depth observed with BC Grade-1 after of Construction and Building Materials, 25(3), (2011), 1362–
20000 passes was 14.5mm for Conventional 1368.
method of gradation. The rut depth reduces 8. Tsai, B.-W., Coleri, E., Harvey, J. T., & Monismith, C. L.,
to 6.52 mm (44.96%) for Bailey gradation for "Evaluation of AASHTO T 324 Hamburg-Wheel Track
same number of passes. Device Test", Journal of Construction and Building Materials,
114, (2016),248–260.
 Polypropylene granules modified mix with
9. Vavrik, W. R., Pine, W. J., Huber, G., Carpenter, S. H., &
Bailey gradation was observed to be more
Bailey, R. The Bailey, "Method of Gradation Evaluation: the
resistant to stripping than other mixes evaluated Influence of Aggregate Gradation and Packing Characteristics
in this study. on Voids in the mineral Aggregate (with discussion)", Journal
 The fatigue life of PP modified mixes may be of the Association of Asphalt Paving Technologists, 70,
less than those without PP. Hence the fatigue (2001).
life studies are to be carried out to establish the 10. AASHTO T-19, "Standard Test Method for Bulk Density
efficacy of PP modification. (Unit Weight) and Voids in Aggregate".
11. AASHTO T 283, "Standard Method of Test for Resistance of
6.1 Usefulness of the Study for Highway Profession Compacted Asphalt Mixtures to Moisture-Induced Damage.
This article demonstrates the role of gradation in 12. Alshamsi, K., "Development of a Mix Design Methodology
bituminous concrete when provided as wearing for Asphalt Mixtures with Analytically Formulated Aggregate
course. The packing characteristics have put forth by Structures", Journal of Association of Asphalt Paving
Bailey method will influence the rutting behaviour Technology. (2006).
of bituminous concrete when used in construction of 13. ASTM D4123-82(1995), "Standard Test Method for Indirect
Tension Test for Resilient Modulus of Bituminous Mixtures".
flexible pavements.
14. ASTM D 5581-07, "Standard Test Method for Conduct of
This statement was demonstrated through the test results Resistance of Plastic Flow of Bitumen Mixes".
of immersion type wheel rutting test. 15. ASTM D 6926-04, "Standard Practice for Preparation of
Bituminous Specimens Using Marshall Apparatus".
The mix which fulfils Bailey approach was found to offer
16. Aurilio, V., Pine, W., & Lum, P. (2005), "The Bailey Method
better resistance to rutting and resilient characteristics
Achieving Volumetrics and HMA Compactability", (Vol. 50,
than the mix prepared with Midpoint conventional 1998. p. 159)
gradation (as per MoRTH for BC mix grade - 1). 17. BS EN 12697 33 (2007), "Standard Test Methods for Hot Mix
The study has opened a new area of research where in Asphalt Bituminous Mixtures- Specimen Prepared by Roller
the efforts can further be invested to demonstrate the Compactor".
efficacy of PP granules modification in bitumen mixes 18. IRC:SP-53, "Guidelines on Use of Modified Bitumen in Road
with a due consideration to fatigue life assessment. Construction".
19. IRC:SP:98-2013, "Guidelines for the Use of Waste Plastic in
References Bituminous Mixes (Dry Process) in Wearing Courses", Indian
1. Al-Hadidy, A., & Yi-Qiu, T., "Mechanistic Approach for Roads Congress, New Delhi.
Polypropylene-Modified Flexible Pavements", Journal of 20. "Specifications for Roads and Bridges", 4th Revision, Ministry
Materials & Design, 30(4), (2009), 1133–1140. of Road Transport and Highways, IRC New Delhi.
2. Chaturabong, P., & Bahia, H. U., "The Evaluation of Relative 21. Vavrik, W.R., Huber, G., Pine, W.J. & Carpenter, S.H. Bailey,
Effect of Moisture in Hamburg Wheel Tracking Test", Journal "Method for Gradation Selection in HMA Mixture Design",
of Construction and Building Materials, 153, (2017), 337– Transportation Research Circular No. E-C044. (2002),
345. Transportation Research Board.

Performance and Fatigue behaviour of Alkali Activated
Slag Concrete with Steel Slag as Coarse Aggregate for
Concrete Pavement
Nitendra Palankar1 A.U. Ravi Shankar2 Mithun B. M.3
ABSTRACT
The present investigation is conducted to evaluate the possibility of replacing the natural coarse
aggregates with steel slag aggregates in Alkali Activated Slag Concrete (AASC) mixes. The
performance of AASC mixes incorporating steel slag as coarse aggregates is evaluated and is
compared with conventional cement concrete. Natural coarse aggregates are replaced with steel
slag aggregates at different replacement levels (25%, 50%, 75% and 100%) in AASC mixes.
The properties of fresh and hardened concrete mixes are studied in detail. The flexural fatigue
behaviour of hardened concrete mixes is also investigated in detail. Probabilistic distribution of
fatigue life data of concrete mixes are analyzed using Weibull distribution. The results indicate that
the AASC mixes with steel slag aggregates display slightly lower compressive and tensile properties
as compared to normal AASC. Modulus of elasticity values of AASC with steel slag were lower than
AASC with normal aggregates. AASCs with steel slag exhibit slightly higher water absorption and
volume of permeable voids. AASCs incorporating steel slag exhibit lower fatigue life than AASC
with normal aggregates. The results indicated that the performance of steel slag aggregates in
AASC is acceptable and it can be seen possible substitute to natural aggregates in pavement quality
concrete.
1. Introduction for OPC concrete and are further associated

The global concern associated with the emission with low CO2 emissions. AAS incorporate 100%
of green house gases and consumption of natural Ground Granulated Blast Furnace Slag (GGBFS)
resources related to augmented Portland cement as primary binder, offering partial solution for
manufacturing has led to the utilisation of industrial reduction of environment hazards occurring from
waste and by-products as primary ingredient as the disposal of slag.
binders in concrete. Cement-less binder such Alkali Activated Slag Concrete (AASC) has
as Alkali Activated Slag (AAS) system which been found to have some superior properties
include the use of an alkaline activator along with namely low heat of hydration, high early strength
silicate rich precursor material to form a final development and better durability in aggressive
product, having properties equivalent to Ordinary environments in comparison with OPC concrete
Portland Cement (OPC) concrete are gaining much (Palankar et.al 2016). It is possible to attain
attention amongst the researchers across the globe. compressive strength in the range 60 to 150 MPa
The energy required for the production of alkali for AASC activated with sodium silicate without
activated slag concrete is lesser than that required heat treatment or admixtures. The most important
1.
Associate Prof., Dept. of Civil Engg., Gogte Institute of Technology, Belgaum, E-mail: nitendrapalankar@gmail.com
2.
Professor, NIT Karnataka, Mangalore, E-mail: aurshankar@gmail.com
3.
Assistant Professor, NMAM Institute of Technology, Dakshina Kannada, E-mail: mithunbm@gmail.com

Nitendra, Ravi Shankar and Mithun on
Performance and Fatigue Behaviour of Alkali Activated Slag Concrete with Steel Slag as Coarse Aggregate for Concrete Pavement
reaction product formed in the activated slag is Researchers have investigated the use of steel slag
C–S–H gel, similar to the hydration product found in conventional concrete and reported superior
in OPC but having a lower Ca/Si ratio. The type, and encouraging properties than conventional
dosage and concentration of alkali activators aggregates (Alizadeh et al. 1996). Experimental
greatly influence the strength properties of results indicate OPC concrete incorporating steel
AASCs. A combination of liquid sodium silicate slag as coarse aggregates have higher compressive
and sodium hydroxide has proved to deliver strength, flexural strength and modulus of elasticity
the best strength performance in AAS. As the in comparison with concrete with limestone
dosage of alkalis in activating medium increases, aggregates (Patel 2008). Investigations carried
higher strength gain in AAS is attained (Rashad out by few researchers found the performance of
2013). Efforts to study the durability properties steel slag aggregates similar or slightly lower than
of alkali activated slag binders have been carried natural aggregates when replaced completely in
out by several researchers and have reported concrete (Ivanka et al. 2011).
better performance of AAS as compared to OPC Fatigue failure is one of the major modes of failure
(Rajamane 2013, Datta and Ghosh 2014) in structures like concrete pavements which are
Large scale concrete production has created subjected to repeated application of loads. The
depletion of natural aggregate resources. It is fatigue failure in concrete pavements occur under
necessary to recognize suitable alternatives to the influence of such repetitive loads or cyclic
natural aggregates in order to protect the natural load, whose peak values are considerably lower
aggregates for concrete in future. Steel slag may than the safe loads estimated through static tests.
be identified as a possible alternative to traditional The fatigue failure in concrete structures causes
aggregates in concrete. Moreover, the pollution progressive, localized and permanent damage
and disposal problems associated of steel slag due to dynamic or moving or cyclic loads (Lee &
could be minimised, along with conservation of Barr 2004). The fatigue life represented using S-N
natural aggregates. Steel slag also called as Linz curves, which provide the number cycles to failure
Donawitz slag is a by-product obtained during versus stress level; is an easy and effective method
the manufacture of steel. The steel slag may be to understand the fatigue behavior of concrete
available in two forms based on the manufacture specimen. The fatigue design process is associated
process namely Basic Oxygen Furnace (BOF) with lots of uncertainties from the numerous
slag and Electric Arc Furnace (EAF) slag. The assumptions made in analysis and material
chemical composition of BOF slag may differ variability and hence probabilistic concepts need
from that of EAF slag based on the manufacture to be implemented to model the fatigue data, in
process. The physical properties of steel slag order to evaluate fatigue resistance of concrete
aggregates also vary from source to source based structures (Naik et al. 1993).
on the production techniques and other parameters Despite of satisfactory performance of steel
(Fruehan 1985). slag as coarse aggregates in both normal and
The acceptance of steel slag in construction high performance concretes, there is no research
industry is associated with certain issues namely available relating to the incorporation of steel
volume deformation and high density. The volume slag as coarse aggregates to produce AAS based
deformation is mainly due to the reaction occurring concrete mixes. In this Paper, a study is conducted
between free lime or free magnesia present in steel to ascertain the mechanical, durability and
slag with water. This can be overcome by subjecting fatigue life properties of air cured AASC mixes
to the steel slag aggregates to weathering or aging prepared with steel slag as coarse aggregates for
for period of three to six months, allowing the free its use in Pavement Quality Concrete (PQC).
lime or magnesia content to bring within acceptable Initially OPC based concrete mix and AASC
limits (FHWA 2012). Steel slag aggregates when mixes were designed and then steel slag as coarse
subjected to natural weathering undergo slight aggregates were incorporated in the AASC mixes
alterations in their physical state with the formation at different levels of replacement (25%, 50%,
of thin film of calcite (CaCO3) that appears on the 75% and 100% by volume) to normal aggregates.
aggregates. Various properties such as slump, unit weight,

compressive strength, splitting tensile strength, polymer admixture (“Conplast SP 430”) supplied
flexural strength, modulus of elasticity, water by FOSROC, Chemicals (India) Pvt. Ltd.
absorption, Volume of Permeable Voids (VPV) Table 1 Chemical Composition of GGBS and Steel
were evaluated. The flexural fatigue behavior Slag (% by Weight)
of the concrete mixes with and without steel
slag was evaluated. Probabilistic distribution Constituents GGBFS Steel Slag
of fatigue life of concrete mixes was analyzed. CaO 34.22 45.52
The present experimental study is focussed on Al2O3 17.14 4.12
purpose of testing the properties of sustainable air
Fe2O3 1.22 21.54
cured AASC with steel slag as coarse aggregates
with the aim of conserving natural raw materials SiO2 32.52 11.04
and to recycle industrial by-products. Since, this MgO 9.65 8.17
kind of concrete requires air curing; it can be of Na2O 0.16 0.14
great benefit for places with severe shortage of K 2O 0.07 0.05
water resources. The utilization of GGBFS in
the construction industry will lead to reduced SO3 0.88 0.08
consumption of OPC which can be useful to Insoluble Residue 4.03 8.97
reduce green house gas emissions emerging from Loss of Ignition 0.04 0.25
the manufacture of Portland cement. Glass Content 92% -
2. Experimental Investigation Table 2 Physical Characteristics of Aggregates
2.1 Materials
Method
Sl. Crushed Steel Fine
In the present study, OPC 43 grade (specific No
Test
granite Slag aggregate
of Test,
gravity=3.14) meeting the requirements of reference
IS:8112–2013 was used. GGBFS in accordance to 1

Specific
2.69 3.35 2.64
IS:12089 – 1987 is procured from Iron and Steel Gravity
Plant, Bellary, India. The GGBFS has a fineness Bulk

IS:2386
Density
of about 370 m2/kg (Blaine) and specific gravity 1495 kg/ 1726 1475 kg/ (P-III)
a) Dry
of 2.9. The GGBFS can be classified as ‘acidic 2
loose
m3 kg/m3 m3 -1963
slag’ having basicity coefficient [Kb= (CaO + b) Dry 1653 kg/ 1935 1548 kg/
MgO) / (SiO2 + Al2O3)] of 0.88. The chemical compact m3 kg/m3 m3
composition of the GGBFS, are presented in Aggregate
Table 1. The alkaline activator for the activation 3 Crushing 24% 21% -
Value
of GGBFS was a solution prepared by combining IS:2386
Los
commercial grade NaOH flakes (97% purity, Angeles (PIV)-
4 20% 18% -
density = 2110 kg/m3) and sodium silicate solution Abrasion 1963
(14.7% Na2O + 32.8% SiO2 + 52.5% H2O by mass, value
and density = 1570 kg/m3) in water. Potable tap Aggregate

5 Impact 21% 16% -
water was used for preparation of alkali activator value
solution and for mixing as well as curing of
Water
OPCC specimen. Natural coarse aggregates used 6
Absorption
0.50% 2.0% 0.80%
were crushed granite aggregates with maximum
size of 20 mm (20 mm to 4.75 mm), while locally 2.1.1 Steel Slag Aggregates
available river sand conforming to requirements The steel slag aggregates (BOF slag) used in
of IS: 383–1970 was used as fine aggregates in this study is obtained from Iron and Steel Plant,
the present investigation. The aggregates are Bellary, India. The slag is in black colour, angular,
tested as per relevant Indian standard codes and stone like appearance having a porous texture
the results are presented in Table 2. The super surface. The slag was subjected to weathering
plasticizer used was commercially available treatment by exposing to outdoor conditions to
Sulfonated Naphthalene Formaldehyde (SNF) monsoon conditions for 180 days to bring the free

lime and free magnesia content within permissible forming calcium carbonate (CaCO3) which settles
limits. The aggregates after being subjected to down in the form of sediment appearing in the
weathering process, were found to undergone form of white powder on the aggregate surface
certain changes in the physical properties and (FHWA 2012). Fig. 1 presents the images of steel
appearance. The thin white film or coating of slag aggregates before and after undergoing the
calcite (CaCO3) appeared on the surface of the weathering process. Aggregates of maximum size
aggregate after being subjecting to weathering. A of 20 mm are considered in the present study. The
sample of weathered aggregates was washed in chemical composition of steel slag constitutes
order to check if whether the thin layer of calcite oxides of iron, calcium and silicon as the major
could be removed. However, the bond between composition along with oxides of magnesium,
the coating layer and aggregate was found to be aluminium and other metals in small quantities.
strong enough to be washed away. The presence Chemical composition and physical properties
of free lime more than 1% causes adverse effects of steel slag are presented in Tables 1 and 2
on the property of steel slag. The free lime reacts respectively. Table 3 presents the gradation of the
with water to form calcium hydroxide which when aggregates used for present study.
exposed to atmosphere reacts with carbon dioxide
(a) Steel Slag Aggregates before Weathering Process (b) Formation of thin Film of Calcite on Slag
Aggregates after Undergoing Weathering Process
(c) Enlarged Image of Steel Slag Aggregate Displaying the Presence of Coating of Calcite
Fig. 1. Image Showing Steel Slag Aggregates before and after Weathering Process

Table 3 Gradation of Aggregates
Fine Aggregates Coarse Aggregates

Passing (%)
Sieve No IS:383-1970 Passing Sieve No IS:383-1970
(mm) requirement (%) (mm) requirement Natural
Steel slag
aggregates
10 100 100 20 95-100 98 96
4.75 90–100 98.5 10 25-55 40 36
2.36 75–100 95.4 4.75 0-10 5 2
1.18 55–90 71.5
0.60 35–59 47
0.30 8–30 12
0.15 0–10 3.1
2.2 Mix Proportion and Specimen Details size (100x100x500) mm (for flexural strength
and fatigue testing), cylinders of size 100 mm
The mix design for OPC concrete is based on the
dia x 200 mm height (for split tensile strength),
procedure suggested by IS: 10262-2009. The total
cylinders of size 150 mm dia x 300 mm height (for
binder content is restricted to 425 kg/m3, with
modulus of elasticity) were cast. All specimens
water/binder ratio of 0.4 and coarse aggregate:
were demoulded after 24 hrs of casting. The
fine aggregate ratio of 0.64:0.36. The mixes are
OPCC specimens were cured in water tank at
designed to achieve a slump value of 25-50 mm.
A super plasticizer dosage of 0.4% (by weight of room temperature of (27±3°C) while the AASC
binder content) is added to the mix to arrive at the specimens were subjected to curing at relative
designated slump. humidity of 85±5% and room temperature of
(27±3°C). An average value of three specimens is
The AASC mixes are proportioned to contain recorded for all the tests.
same binder content (425 kg/m3) and water/binder
ratio (0.4) as that of OPCC mix. The total water 2.3 Flexural Fatigue Behaviour of Concrete
content in the activator solution for AASC mixes Mixes
constituted the sum of water readily available in The flexural fatigue tests on OPCC and AASC
liquid sodium silicate solution plus the extra water samples were carried out beam specimens of
added, to arrive at the required water/binder ratio. dimensions (100x100x500) mm on a MTS servo-
The alkali activator were proportioned to provide controlled hydraulic repeated load testing machine
a dosage of 4% Na2O (by weight of binder), with having a capacity of 5 tonnes (Fig. 2). The fatigue
an activator modulus (Ms) of 1.25 (weight ratio tests were conducted on OPCC and AASC samples
of SiO2/Na2O) and total water/binder ratio of 0.4. with 0%, 50% and 100% replacement with steel
No super-plasticizers are added for AASC mixes. slag aggregates. The static flexural strength of the
AASC mixes were prepared with steel slag as coarse mixes was recorded at 90 days before the fatigue
aggregate by replacing the natural aggregates by test was conducted and results are presented in
25%, 50%, 75% and 100% (by volume) keeping Fig. 3. The beam specimen was loaded at the same
the volume of total coarse aggregates constant. span (i.e. 400 mm) as it was loaded in case of
The aggregates used were in saturated surface static flexural tests. The specimen were subjected
dry condition. The details of mix proportions for to loading using constant amplitude half sinusoidal
OPCC and AASC are listed in Table 4. wave form at a frequency of 4 Hz without any rest
After proper mixing, the fresh concrete was period. The test setup was calibrated applying
poured in moulds for preparation of specimen for initial loading and the frequency of loading was
testing hardened properties. The cube specimens maintained constant throughout the test for all
of size (100x100x100) mm (for compressive specimens. The minimum load is maintained as
strength, water absorption and VPV), prisms of zero while the maximum load was adjusted based

on the required stress ratio (ratio of applied stress 85%, 80%, 75% and 70% to obtain a relationship
to the modulus of rupture of concrete). The fatigue between different stress ratios (S) and the number
tests were conducted at different stress levels i.e. of cycles to failure (N). The test was conducted at
Table 4 Details of Mix Proportions of Concrete Mixes (all quantities are in kg/m3)
Mix ID OPC GGBS Sand Coarse Steel Sodium NaOH Added Super-
aggregate slag silicate water plasticizer
OPCC 425 - 660 1195 - - - 170 1.7
AAS-0 - 425 647 1172 - 64.8 9.6 136 -
AAS-25 - 425 647 879 364 64.8 9.6 136 -
AAS-50 - 425 647 586 729 64.8 9.6 136 -
AAS-75 - 425 647 293 1094 64.8 9.6 136 -
AAS-100 - 425 647 - 1459 64.8 9.6 136 -
Note: OPCC - represents Portland cement based control mix; AAS - represents AASC mix with normal aggregates; AAS-X represents AASC
mixes with X (% by volume) of normal aggregates replaced with steel slag aggregates.
the end of 90 days of curing of concrete specimen Table 5. The OPCC and AASC mixes attained the
in order to eliminate the errors occurring due to the target slump values for which they were designed.
strength development of concrete mixes after 28 Incorporation of steel slag in the AASC mixes resulted
days of curing. The Fatigue life (N) i.e. the number slight decrease in slump value. This is due to the fact
of cycles up to failure of the sample was recorded. that steel slag aggregates are angular in shape, which
A total of 80 prism samples were tested with set cause reduction in the mobility of matrix thus reducing
of five samples for each mix. One set of fatigue the workability (Carlo et al. 2013). Concretes prepared
lives of the various concrete mixes are presented in with steel slag aggregates will require slightly higher
Table 7 while the complete data of fatigue lives are water/binder ratio to achieve a designated slump as
represented using S-N curves. The experimental compared to traditional aggregates. However, the
set up of fatigue testing machine is presented in slump loss was not significant at 25% replacement
Fig. 2. of steel slag in AASC. It can be observed from
Table 5, that the unit weight of the AASC mixes
increase with the incorporation of steel slag from
24.70 to 27.65 kN/m3. The higher density is attributed
to the higher specific gravity of the steel slag than
normal aggregates. The higher density of steel slag
AASC may not be much concern for application in
PQC. However, the unit weights of AASC and OPCC
are in the same range.
3.2 Compressive Strength
Compressive strength tests were conducted as per
IS:516-1959 at 3, 7, 28, and 90 days of curing
and the results are tabulated in Table 5. It can be
noticed that the OPCC and AASC specimens show
Fig. 2 Repeated Load Testing Set Up progressive strength development up to 90 days
3. Results and discussions of curing. The 28-days compressive strength of
AASC samples with and without steel slag are in
3.1 Workability and Unit Weight the range of 55±5 MPa. The AASC samples with
The slump test was performed according to the normal aggregates exhibit higher compressive
procedure suggested in IS:1199:1959. The slump strength as compared to OPCC samples. The AASC
values obtained for different mixes are presented in with 100% steel slag exhibit lower compressive

Table 5 Compressive Strengths Results of Various Concrete Mixes at Different Ages
Compressive strength (MPa) Slump Density

Mix ID
3 days 7 days 28 days 90 days (mm) (kN/m3)
OPCC 23.1 40.4 56.9 62.8 35 24.8
AAS -0 39.6 48.2 59.6 67.4 50 24.7
AAS-25 38.2 46.9 56.1 65.9 45 25.4
AAS-50 37.5 45.1 53.4 61.6 40 25.9
AAS-75 35.1 43.2 52.3 57.1 30 26.8
AAS-100 33.2 40.4 50.2 53.5 25 27.6
strengths as compared to natural aggregates. This necessary for application for PQC. It may be
may be due to the presence of weak aggregate noted that the AAS concrete mixes prepared with/
and paste bond occurring due to the presence of without steel slag aggregates attained compressive
thin film of calcite on the steel slag aggregates. strength greater than 50 MPa and hence satisfy the
Visual inspection of the failed samples after testing minimum strength requirement of M40 and for
exhibited incidents on debonding between the steel M30 for national/state highway and rural roads
slag aggregate and paste. It is a well established respectively as per IRC:58:2011. The high early
fact that the bond between the aggregate and the strength of AASC mixes is of great benefit for
paste significantly affect the mechanical properties PQC as it would allow the early opening of the
of concrete. If there exists a stronger bond constructed pavements to the vehicular traffic.
between the paste and coating layer than the bond
3.3 Static Flexural Strength and Split Tensile
between the coating and the aggregate surface, it
Strength
may lead to the development of weak coating–
aggregate interface thus leading to the reduction The static flexural strengths for all concrete
in mechanical and durability properties of the specimens were determined according to IS:
concrete (Forster 2006). The AASC with higher 516:1959. Table 6 presents 7, 28 and 90 days
proportions of steel slag displayed greater variation results of flexural strength of OPCC and AASC
of strength in the samples tested for the same mix. mixes. From Table 6, it can be observed that the
This may be attributed to the higher heterogeneity flexural strength decrease with the increase in steel
of slag particles as compared to natural aggregates. slag content in AASC. The flexural strength of
However the strength variation of AASC mixes AASC with 100% slag is lower than AASC with
with steel slag aggregates was within normal range normal aggregates at all ages. This may be due
of variability. There is no significant difference to decreased bonding between the paste and steel
in strength observed between AAS-0 and AAS- slag aggregate on account of presence of coating
25. The AASC mixes exhibit high early strength of calcite as in case of compressive strength.
as compared to OPCC i.e. the 3-days and 7-days The flexural strength of AASC show a similar
strength of AASCs higher than OPCC. The high decreasing trend with higher content of steel slag
early strength of AASC mixes may be attributed to aggregates as observed in compressive strength.
the physical and structural characteristics of binders AASC mixes with natural aggregates display
formed. The hydration process in case of AAS higher flexural strength than OPCC concrete at
depends upon the dissolution and precipitation all ages, which may due to the presence of highly
of GGBFS which in the presence of high alkaline dense interfacial transition zone between the paste
activator is quite faster than the hydration process and the aggregates and development of a distinct
occurring in Portland cement (Wang & Scrivener microstructure as compared to cement concrete
1995). However, rate of strength development (Susan et al. 2012). Table 6 presents the 28-day split
slowed down after 7-days and at the end of 90- tensile strength values for OPCC and AASC mixes.
days the OPCC and AASC show similar strengths. The spilt tensile strength test was conducted as per
AASC with normal aggregates achieved the IS : 5816:1999. The split tensile strength of AASC
highest strength at the end of 90 days. All mixes with normal aggregates is higher than OPCC. The
attained sufficient strength at 28 days of curing split tensile strength of AASC is found to decrease

Table 6 Tensile Properties and Modulus of Elasticity of Concrete Mixes

Split Tensile Strength Modulus of Elasticity
Flexural Strength (MPa)
Mix ID (MPa) (GPa)
7 days 28 days 90 days 28 days 28 days
OPCC 4.62 6.01 6.39 3.90 34.9
AAS-0 6.08 6.93 7.46 4.42 34.2
AAS-25 5.96 6.59 7.33 4.26 33.7
AAS-50 5.69 6.24 6.79 4.09 32.8
AAS-75 5.41 6.06 6.29 3.96 32.2
AAS-100 5.29 5.88 6.05 3.81 31.1
with replacement by steel slag aggregates. At Voids (VPV) were evaluated according to ASTM
lower levels of replacement up to 25% steel slag; C 642-06. A set of three cubes of (100x100x100)
no appreciable change in split tensile strength is mm size were tested for each mix. The water
noticed, however split tensile strength of AASC at absorption and VPV of different mixes at 28
100% replacement is found to be marginally lower days are presented in Figs. 3 and 4 respectively.
than AASC with normal aggregates. AASC mixes with normal aggregates having
The test for modulus of elasticity was conducted same binder content and water-binder ratio as
as per IS : 516:1959. Table 6 presents the static OPCC show reduced water absorption and VPV
modulus of elasticity of OPCC and AASC mixes at values at both 28 days of curing. This may be
28 days of curing. It can be observed that the static probably due to the presence of very refined
modulus of OPCC mixes is higher than that of AASC closed pore structure in the AASC samples
with traditional aggregates. The relationship of 28- which restrict the water from penetrating into
day compressive strength with other mechanical the structure (Mithun & Narasimhan 2015).
such as elastic modulus, tensile properties will However, the water absorption and subsequent
differ as a function of binder chemistry. AASC VPV increase with the replacement of natural
vary in their strength development rates and micro aggregates with steel slag. This may be attributed
structural differences which may lead differences to the higher water absorption of steel slag
in flexural and elastic properties (Provis 2013). The aggregates as compared to normal aggregates.
modulus of elasticity of AASC mixes decrease with The presence of pores in steel slag may have
the replacement of steel slag. However, at lower resulted in the increase in VPV in the mixes. At
levels (upto 25%) of replacement, the variation lower level of replacement of steel slag i.e. 25%
in modulus of elasticity is marginal. The type and (AAS-25), the water absorption is marginally
properties of coarse aggregates affect the modulus higher than normal AASC but beyond that the
of elasticity of high performance concretes. The water absorption increased considerably.
nature of the constituent material and nature of
interfacial transition zone between the paste and 110
Weight of water absorbed (gms)
the aggregates affect the elastic modulus of the 105

concrete. The influence of aggregates on modulus 100
of elasticity becomes smaller with decreasing 95
strength of concrete. All the mixes with/without
90
steel slag aggregates achieved flexural strength
85
and modulus of elasticity greater than 4.5 MPa and
30 GPa (minimum requirement as per IRC : 80
58:2011) and hence qualify its utility for national 75

OPCC AAS-0 AAS-25 AAS-50 AAS-75 AAS-100
and state highways.
Mix ID
3.4 Water Absorption and Volume of
Permeable Voids Fig. 3 Water Absorption of Various Concrete Mixes
The Water Absorption and Volume of Permeable at 28-days

general represent the progressive deterioration

11
of the structure and breaking of the bonds. The
10.5
S-N curves obtained by plotting stress level v/s
Volume of permeable voids (%)
10
number of cycles up to failure, for OPCC and
9.5
AAS with and without steel slag are presented
9
in Fig. 5. The equations obtained from the S-N
8.5
curves can be utilized for estimation of fatigue
8 cycle at any stress level. The equations generated
7.5 from S-N curves for different concrete mixes are
7 tabulated in Table 8.
OPCC AAS-0 AAS-25 AAS-50 AAS-75 AAS-100
Mix ID From Table 7, it can be noticed that the AASC mix
with normal aggregates display higher resistance
Fig. 4 Volume of Permeable Voids of Concrete Mixes to fatigue failure as compared to OPCC and other
at 28-days AASC mixes irrespective of the applied stress level.
This may be due to the highly dense interfacial
3.5 Fatigue Life of Concrete Mixes transition zone between the paste and the aggregate
The fatigue life i.e. the number of cycles up to occurring in alkali activated binders as compared to
failure for OPCC and AASC mixes with/without that occurring in conventional cement (Susan et al.
steel slag are presented in Table 7. The use of S-N 2012). The fatigue life of AASC mixes decreased
cures or Wohler curve is the most basic method with the inclusion of steel slag aggregates in AASC
of representing the fatigue behaviour of concrete mixes. This may due to the presence of weak bond
specimen. S-N curve is an important parameter in between the aggregate and the paste which may
the analysis of fatigue data in which ‘S’ denotes lead to higher propagation of the crack leading to
the stress amplitude and ‘N’ denotes the number earlier failure. The mix AAS-100 display lower
of cycles to complete failure. The S-N curves in failure cycles compared to all other mixes. It was
(a) S-N curve for OPCC (b) S-N curve for AAS-0
(c) S-N curve for AAS-50 (d) S-N curve for AAS-100
Fig. 5 S-N Curves for Various Concrete Mixes

observed that specimens exhibit low fatigue life verification and relative ease in use (Oh 1991)
when subjected to higher stress levels. However, and hence the same is adopted for this study.
at low stress levels exhibited higher fatigue lives. The Weibull distribution is characterised by two
The fatigue lives of all the concrete mixes satisfied important parameters firstly ‘α’ which describes
the minimum fatigue cycles as recommended by the shape of the distribution and second ‘µ’ which
IRC:58:2011. The failures of the specimens were defines the characteristic life. The use of graphical
visually examined and were found to have failed method is done to estimate the distribution
within the middle one third spans. The statistical parameters and to indicate that the distribution of
correlation coefficient values from the Table 8 the fatigue life of OPCC and AASC mixes at given
for different concrete mixes and different stress stress levels (S) follow Weibull distribution.
levels were found to be in the range 0.88 to 0.91 3.7 Estimation of Weibull Distribution
indicating statistical significance. From Fig. 5 it Parameters using Graphical Method
can be noticed that all the concrete mixes display
The survival function of Weibull distribution can
similar behaviour.
be expressed as follows:
Table 7 Fatigue Life of OPCC and AASC Specimens (1)
Mix IRC: OPCC AAS-0 AAS-50 AAS -

n represents specific value of random variable,
ID 58-2011 100 α represents shape parameter or Weibull slope at
Stress stress level ‘S’ and µ characteristic life at stress
Level Number of cycles to failure (N) level ‘S’.
0.85 30 317 364 350 291 Taking twice the logarithm of both sides of Eq. 1
0.80 119 1986 2124 1698 1478 (2)
0.75 477 57835 61214 49175 54014 Eq. 2 may be written in the following form:
Y = α X – B (3)
0.70 9700 97587 107587 92475 85412
Where,
Table 8 Relationship between Fatigue Cycle (N) and
Stress Level (SR)
Mix ID Equations R2 X = ln(n) and B = α ln(μ).
OPCC ln(N)=0.953-SR/0.02 0.903 It can be noticed from the Eq. 3 fatigue data follows
Weibull distribution if the relationship between X
AAS-0 ln(N)=0.956-SR/0.02 0.902 and Y is linear and the distribution parameters can
AAS-50 ln(N)=0.954-SR/0.02 0.879 be obtained from the straight line. Therefore linear
regression analysis is performed for fatigue data
AAS-100 ln(N)=0.946-SR/0.02 0.881
to arrive at the relation for each stress level ‘S’ as
3.6 Probabilistic Analysis of Fatigue Test Data in Eq. 3.
The fatigue life of concrete mixes display The empirical survivorship function LN (n) at
significant scatter and variability in their nature a given stress level for each fatigue life data is
due to the heterogeneity of the material. In such calculated using the following Eq.4 (Mohammadi
conditions individual fatigue life of structures are & Kaushik 2005).
difficult to predict and hence it is desirable to adopt (4)
probabilistic approach in the design, to acquire
information for satisfactory fatigue resistance Where ‘i’ represent failure order number and ‘k’
for pavements. The two parameter Weibull represents sample size under consideration at a
distribution is the most widely adopted method particular stress level.
for statistical description of fatigue data due to The two parameters ‘α’ and ‘μ’ of Weibull
its better developed statistics, sound experimental distribution can be directly obtained from plot

with/without steel aggregates can be described

between and ln (n). The slope of the line
using Weibull distribution. However it would be
provides shape factor ‘α’ while the characteristic
convincing to carry out a goodness-of-fit test in
life ‘μ’ can be obtained from the equation,
B=α ln(μ). The graph between and order to confirm that it is a valid distribution model
ln (n) is plotted for all the stress levels and for for statistical description of fatigue life of OPCC
all concrete mixes and the Weibull distribution and AASC. Hence the Kolmogorov-Smirnov test
parameters are calculated. The Weibull distribution was carried out for this purpose (Mohammadi &
parameters for OPCC and AASC mixes at different Kaushik 2005). The Kolmogorov–Smirnov test
stress levels are presented in Table 9. A sample can be performed by using Eq. 5.
plot for OPCC at stress level of 0.85 is presented (5)
in Fig.6. From Table 9, it can be noticed that the
statistical coefficient of correlation ranges 0.87 to Where, F+(Xi)=i/k=observed cumulative
0.99 at different stress levels which indicates that histogram.
the fatigue life data for OPCC and AASC mixes
i = order number of the data point.
follows two parameter Weibull distribution.
k = total number of data points in the
sample under consideration at a
given stress level.
F (Xi) = hypothesized cumulative
distribution function given by
Eq. 6
(6)
where, n is the specific value of a random variable
N; α=shape parameter or Weibull slope at stress
level S; µ=scale parameter or characteristic life at
Fig. 6 Graphical Analysis of Fatigue for OPCC at given stress level S; and n0=location parameter or
Stress Level of 0.85
minimum life at stress level S.
3.8 Goodness-of-Fit Test for Fatigue Data Table 10 shows the results calculated by
From the analysis, it is established that the fatigue Kolmogorov–Smirnov test of fatigue life for AAS-
life at various stress levels of OPCC and AASC 0 at stress level S=0.85.
Table 9 Values of Weibull Parameters for Concrete Mixes at Different Stress Levels
Mix ID OPCC AAS-0

Stress Level α µ R 2
α µ R2
0.85 1.585 208 0.979 1.408 247 0.929
0.80 1.568 1365 0.963 1.584 1515 0.937
0.75 1.185 42183 0.955 1.339 46482 0.936
0.70 1.752 71790 0.988 1.761 79488 0.997
Mix ID AAS-50 AAS-100
Stress Level α µ R 2
α µ R2
0.85 1.468 217 0.971 1.268 206 0.959
0.80 1.888 1208 0.910 1.141 1210 0.879
0.75 1.143 38222 0.930 0.951 42721 0.869
0.70 0.992 68810 0.923 1.774 65010 0.986

Table 10 Kolmogorov–Smirnov Test of Fatigue Life

for AAS-0 at S=0.85
Stress F + (Xi) | F +(Xi) - (7)

i Xi F (Xi)
Level = i/k F (Xi)| Where, Pf is probability of failure
1 91 0.2 0.2174 0.0174
Since the fatigue life data of concrete mixes
2 111 0.4 0.2769 0.1231 follow Weibull distribution, it can be used for
AAS-0 3 163 0.6 0.4271 0.1729 the calculation of fatigue lives corresponding
4 302 0.8 0.7348 0.0652 to different survival probabilities. The Eq.7
5 364 1 0.8221 0.1779 gives the fatigue life (n) for different failure
probabilities. Using the values of Weibull
From Table 10, it can be observed that the maximum
distribution parameters (‘α’ and ‘μ’) obtained
difference is 0.1779 (for i=5) for this case. The
earlier for the fatigue life data of concrete mixes,
critical value Dc for n=5 and 5% significance level
the Eq.7 can be used to calculate the fatigue lives
is found to be 0.563 from Kolmogorov–Smirnov
corresponding to different failure probabilities
Table. Since Dt < Dc (0.0.1779< 0.563), the present
(Pf). The Fig.10 presents the predicted flexural
two parameter Weibull distribution model for
fatigue life of OPCC and AASC mixes calculated
fatigue life at stress level S=0.85 is acceptable at
at different survival probabilities (0.05, 0.5 and
the 5% significance level. The goodness-of-fit test
0.95) using Eq.5. From Fig. 7, it can be noticed
is performed for the fatigue life data for OPCC and
that for a particular stress level, the number of
AASC at different stress levels and the model was
expected cycles which a concrete mix can sustain
found to be acceptable at 5% level of significance.
decreases with the increasing probability of
3.9 Survival Probability and S–N Relationship failure. The expected number of cycles is greater
The fatigue life for different probability f failure at lower probability of failure i.e. at 0.95 (failure
can be expressed as follows: probability=5%). The graphs presented in Fig. 7
(a) Predicted Fatigue Life for OPCC (b) Predicted Fatigue Life for AAS-0
(c) Predicted Fatigue Life for AAS-50 (d) Predicted Fatigue Life for AAS-100
Fig. 7. Predicted Fatigue Lives Corresponding to Different Survival Probabilities for Various Concrete Mixes

can be used to predict the number of cycles which costlier as compared to AASC-0. On the other
the OPCC and AASC mixes can sustain at any hand, cost of AASC-100 is further reduced with
desired stress level at a particular probability of incorporation of steel slag (AASC-100) since steel
failure. slag is available at free of cost. However, based
4. Cost analysis on source of procurement and mode of transport,
the cost of steel slag may vary from location to
The cost analysis of concrete mixes for one cubic location. AASC with/without steel slag display
metre of concrete is carried out and tabulated in cost benefits along with satisfactory strength
Table 11. The costs of ingredients considered performance for rigid pavements as compared to
include the local purchase cost of material including OPCC. Such a type of eco-friendly concrete also
the handling and shipping charges. The steel slag leads to reduction in cost of disposal of industrial
is available free of cost. From Table 11, it may be waste materials.
noticed that cost of AAS-0 for one cubic metre of
concrete is Rs 5388, as compared to OPCC which 5. Recommendations
is Rs 5784. The OPCC is further associated with Special case and quality-control procedures need
cost of water curing upto 28 days (not included to be taken during the production and processing
in cost analysis) which will make OPCC more of slag to ensure the separation of unwanted
Table 11 Cost Analysis of Concrete Mixes (one cubic metre of concrete)
Mix ID OPC GGBS Sand Granite Steel Sodium NaOH SP
Slag Silicate
Quantity (kg/m3)
OPCC 425 - 660 1195 - - - 1.7
AAS-0 - 425 647 1172 - 64.8 9.6 -
AAS-100 - 425 647 - 1459 64.8 9.6 -
Cost Analysis
Cost Rs Rs Rs Rs free 16/kg 40/kg 40/kg Total
390/50 190/50 1100/ 1400/ Rs
kg kg ton ton
OPCC 3315 - 726 1673 - - - 70 5784
AAS-0 - 1615 712 1640 - 1037 384 - 5388
AAS-100 - 1615 712 - 0 1037 384 - 3748
matter such as furnace brick, incompletely fused optimum activator modulus and sodium oxide
fragments, metal fragments etc., from the steel slag dosage before finalising the mix for AAS with steel
aggregates before its use in concrete. Steel slag slag aggregates. Extensive laboratory tests have to
aggregates must be crushed and screened to produce be carried to determine the physical and chemical
a desirable aggregate gradation. Before using the properties and compare their performance to control
steel slag aggregates extensive testing needs to sections built with natural aggregates. Additional
be conducted in the laboratory to determine the safety precaution needs to be taken while handling
expansive properties and free lime content to ensure the liquid sodium silicate and sodium hydroxide,
volumetric stability. The steel slag aggregates need since they are corrosive in nature and their contact
to undergo for a minimum period of 180 days to with skin leads to formation of rashes and burning.
ensure the hydration of free lime content to avoid
volumetric expansion. Conventional pavement 6. Conclusion
design and field test procedures can be followed for The following are the major conclusions drawn
PQC containing steel slag aggregates. Traditional from the present experimental investigation:
methods and equipment may be used to produce (i) The replacement of traditional aggregates
concrete with steel slag aggregates as that used for with steel slag in AASCs results in slight
conventional aggregates. It is necessary to identify reduction of the workability whereas the
the optimal content of steel slag aggregates, the density of AASCs increases.

(ii) The AASC mixes exhibit higher early and 6. Ivanka, N., Dubravka, B. and Goran, V. (2011),
ultimate strengths as compared to OPCC. “Utilisation of Steel Slag as an Aggregate in Concrete”,
AASC mixes with steel slag aggregates Materials and Structures, Vol. 44, 1 pp. 565–1575.
exhibit lower compressive and tensile 7. Mithun, B.M. and Narasimhan, M.C. (2015),
strengths as compared to AASC with normal “Performance of Alkali activated Slag Concrete Mixes
aggregates due to the presence of a weak Incorporating Copper Slag as Fine Aggregates”,
aggregate-paste interface. Journal of Cleaner Production, pp. 1–8. http://dx.doi.
org/10.1016/j.jclepro.2015.06.026.
(iii) The modulus of elasticity of AASC with steel
8. Lee, M. K. and Barr, B.I.G. (2004), “An Overview of
slag is lower than AASC with conventional the Fatigue Behaviour of Plain and Fiber Reinforced
aggregates. Concrete”, Cement and Concrete Composites,Vol. 26,
(iv) AASC mixes with steel slag exhibit higher pp. 299-305.
VPV and water absorption as compared to 9. Mohammadi, Y. and Kaushik, S.K. (2005), “Flexural
traditional aggregates due to higher water Fatigue Life Distributions of Pain and Fibrous Concrete
absorption and presence of pores in steel slag at Various Stress Levels”, Journal of Materials in Civil
aggregates. Engineering, ASCE, Vol.17, pp. 650-658.
(v) The AASC with steel slag aggregates display 10. Naik, T.R., Singh, S. S., and Congli, Y. (1993),“Fatigue
lower fatigue life as compared to AASC Behavior of Plain Concrete Made with or Without Fly
with normal aggregates. The two parameter Ash”, Center for By-Products Utilization University of
Wisconsin-Milwaukee Department of Civil Engineering
Weibull distribution can be approximately
& Mechanics 3200 North Cramer Street Milwaukee,
used for modeling the fatigue life data of WI 53211
OPCC and AASC mixes with a statistical
correlation coefficient greater than 0.9. 11. Oh, B.H. (1991), “Fatigue Life Distributions of
Concrete for Various Stress Levels”, ACI Material
(vi) The AASC mixes with steel slag show Journal, Vol. 88, pp. 122-128.
satisfactory results for its use in PQC and 12. Palankar, N., Shankar, A. R., and Mithun, B. M.
displays economical benefits over OPCC. (2016), “Durability Studies on Eco-Friendly Concrete
Production of air cured AAS concrete will Mixes Incorporating Steel Slag as Coarse Aggregates”,
result in reduction of OPC and in turn reduces Journal of Cleaner Production, Vol.129, pp.437-448.
carbon dioxide emissions. Utilization of steel 13. Patel, J. P. (2008), “Broader use of Steel Slag Aggregates
slag in concrete creates opportunity for the in Concrete” Doctoral dissertation Thesis, Cleveland
conservation of natural aggregates and solves State University,
the disposal problem of slag in steel industry. 14. Provis, J.L. (2013), “Alkali-activated Binders and
References Concretes: the Path to Standardization Geopolymer
Binder Systems”, STP 156620120078, ASTM, pp. 185-
1. Alizadeh, R., Chini, M., Ghods, P., Hoseini, M., 195.
Montazer, Sh. and Shekarchi, M. (1996), “Utilization
15. Rajamane, N.P. (2013), “Studies on Development of
of Electric Arc Furnace Slag as Aggregates in concrete
Ambient Temperature Cured Fly Ash and GGBS Based
– Environmental Issue”, Tehran: CMI Report.
Geopolymer Concretes.” PhD thesis, VTU, Belgaum,
2. Datta, D. and Ghosh, S. (2014), “Durability Study of India.
Geopolymer Paste Blended with Blast Furnace Slag”,
16. Rashad, M.A. (2013), “A Comprehensive Overview
IOSR Journal of Mechanical and Civil Engineering,
about the Influence of Different Additives on the
Vol.11, pp.73-79.
Properties of Alkali-Activated Slag–A Guide for Civil
3. FHWA, (2012), User Guidelines for Byproducts and Engineer”, Construction and Building Material, Vol.
Secondary use Mterials in Pavement Construction, 47, pp.29–55.
Publication Number: FHWA-RD-97-148.
17. Susan, A.B., Ruby, M. de. G., John, L.P. (2012),
4. Forster, S.W. (2006), “Significance of Tests and “Engineering and Durability Properties of concretes
Properties of Concrete and Concrete Making Materials”, Based on Alkali-Activated Granulated Blast Furnace
Chapter 31- Soundness, Deleterious Substances and Slag/Metakaolin Blends”, Construction and Building
Coatings. STP 169D.ASTM. Materials, Vol. 33, pp.99–108.
5. Fruehan, R.J. (1985), “The Making Shaping and 18. Wang, S.D. and Scrivener, K.L. (1995), “Hydration
Treating of Steel”, Association of Iron and Steel Products of Alkali-Activated Slag Cement”, Cement
Engineers”, ISBN: 0–930767–02–0. and Concrete Research, Vol.25, pp.561–571.

A Review of Ant Algorithm and its Variants in Reference
to Vehicle Routing Problems in Transportation
Dr. Praveen Kumar1 Ms. Shambhavi Mishra2
Abstract
A dependable and reliable information source for transportation route planning and ready response
to traveler demand, have become a hot topic of research these days. Transportation route planning,
analysis and processing of traveler demands are among most important factors affecting response to
traveler demand. Because of highly dynamic networks and frequent discontinuity, it is desirable to
establish routes for fast delivery of people and goods, having a low probability of disconnection. Thus
public transport network design problem is becoming one of the most important problems faced by
transport management officials in the world. This problem belongs to the class of difficult combinatorial
optimization problem, whose optimal solution is difficult to discover. When designing the network, the
interests of both the operator and the traveler must be taken into account. Opposing nature of these
interests, make transport network design a multi criteria decision-making problem. While designing the
transport network, the number of satisfied travelers is to be maximized, the total number of transfers,
and the total travel time of travelers is to be minimized. The concept of agent-based modeling like ant
algorithm can be applied to these complicated problems. In this Paper we will discuss basics of ant
algorithm and its variants. Application of basic and modified versions of these algorithms to routing
problems is also presented. Ant Algorithm is a hybrid algorithm, mimicking the foraging behavior of
social ants that can be used for route planning. Ants basically use pheromone as a chemical messenger
in order to direct path to future ants foraging for food and the pheromone concentration can be treated
as the indicator of quality solutions to a problem of interest. The visibility is also a parameter used that
makes the probability of choosing nearer points higher than farther points. Ant-based algorithms are
particularly suitable for discrete optimization problems. Application of basic ant system algorithm to a
22 node network from Indore city for shortest route calculations is also presented, in order to show its
efficiency in route calculations.
1. INTRODUCTION of traffic innovative technologies are required. The
Road transport is vital to economic development, traffic congestion on highway could be reduced by
trade and social integration. Due to a huge reduction in use of private vehicles i.e. modal shift
development in every field, transport demand is also from private to public vehicles could be one possible
growing expeditiously and private vehicle ownership solution. But for modal shift to public transport,
has recorded significant growth over the recent past. the public transportation system should excel in all
This growth of vehicular traffic on roads has been far respects, use of public vehicles should be better than
greater than the growth of the highway infrastructure private vehicles use so that it can attract commuters.
which results in saturation of major highways. To Thus public vehicles should be good in comfort,
solve this severe problem of traffic, construction of frequency and above all network should be well
a new highway could be a possible solution but that connected covering all possible points of interest
is not feasible due to economic constraints and lack within normal travel times. This could be done by
of space. Thus to solve this long existing problem proper network design of public transit, expansion
1.
Professor & Coordinator, Transportation Engg Group, E-mail : pkaerfce@iitr.ac.in
2.
Research Scholar, E-mail : shambhavimishra89@gmail.com Civil Engineering Department, IIT Roorkee

Dr Kumar & Ms Mishra on
A Review of Ant Algorithm and its Variants in Reference to Vehicle Routing Problems in Transportation
of existing lines, increasing the availability of direct 2. ANT ALGORITHM

service among various origin-destination pairs, Ant Colony Algorithm was been first proposed by
frequency increase, development of the dedicated Dorigo, first invented for the classical Travelling
bus lanes separated from the rest of the traffic, extra Salesman Problem later used for routing in networks.
transit services on peak days, improving passenger’s Ant algorithm, mimic the foraging behavior of
comfort and schedule reliability, better information social ants[7]. This technique originally developed
systems for passengers, etc. All the above mentioned to solve NP (Non deterministic Polynomial time)
points could only be achieved with use of advanced hard combinatorial optimization problems. These
technologies in network designing, which can provide are the problems in which indifferent variables
the information about the planning of shortest route governing the objective function can take only
between the desired origin and destination points, discrete values e.g. Vehicle routing, Travelling
utilize the latest web based technologies and reliable Salesman Problem, Scheduling etc. Ants basically
path routing algorithms. Many of the factors that use pheromone as a synthetic messenger, and
should be taken into account when designing bus this pheromone deposits can be considered as the
network are mutually contradictory. That is why this guide of quality solutions to a problem of interest.
problem of network design belongs to the class of The search algorithms produce routes and paths
difficult combinatorial optimization problem, whose marked by the higher pheromone concentrations
optimal solution is troublesome to discover. While because that is the more favorable path in link to
designing a traffic network, we try to maximize the pheromone quantity. Thus algorithms based on
number of satisfied passengers, to minimize the total behavior of ants are particularly suitable for discrete
number of transfers, and to minimize the total travel optimization problems. The movement of an ant
time of all served passengers. Due to the contradicting totally depends on the pheromone concentration
nature of these interests, we treat the bus network on different paths, which will evaporate over time.
design problem as a multi-criteria decision-making Thus ant algorithms should be made considering
problem. this time-dependent evaporation of pheromone
otherwise it will lead to premature convergence
Ant algorithms is one of the agent based method
of the algorithm to the local (often wrong)
used for network design, many researchers have
solutions. With proper pheromone evaporation,
been working on it to devise new techniques for
these algorithms usually behave very well. The
network designing and routing problems faced
self-organization of the ants is based on relatively
by travelers. Ant algorithm is a part of swarm
simple rules of individual insect’s behavior[5].The
intelligence, the branch of artificial intelligence
ants successful at finding food leaves a pheromone
that studies algorithms inspired by the observation
trail behind those other ants follow in order to
of the behavior of natural insects. In swarm reach the food. As new ants follow the pheromone
intelligences special kinds of artificial agents are trail that path becomes stronger. In this way ants
created with correlation to social insects (bees, can communicate among themselves, and they are
wasps, ants, and termites) which have lived on capable to solve complicated problems. It has been
Earth for many years. These insect colonies shown by the experiments that ants are capable to
teach us that simple organisms can form systems find out the shortest path between any two points
capable of performing highly complicated tasks by in the space.
dynamically interacting with each other. Swarm
intelligence is study of behavior of individuals There are two important issues in ant-algorithm
in various decentralized systems, it can be used are: the probability of choosing a route and the
to model complex traffic and transportation evaporation rate of pheromone. There are not many
processes[2]. An ant algorithm takes inspiration ways of solving these problems; it is still an area
from the foraging behavior of some ant species. of research. For a network routing problem, the
probability of ants at a particular node i to choose
These ants leave pheromone on the ground in order
the route from node i to node j is generally given
to mark some positive path that should be followed
by-
by other members of the colony. Rest of the Paper
explains Ant algorithm and its six variants. Use of
these variants to different transportation problems
is also mentioned along with. (1)

Where, α>0 and β >0 are the influence parameters,

and their typical values are α ≈ β≈ 2. Here, τij is (3)
the pheromone concentration of the route between Where,
i and j, and ηij the desirability of the same route,  m - number of ants
also known as visibility function. Sometimes  ρ is pheromone evaporation rate having value
a prior knowledge about the route such as the less than 1, and (1-ρ) represents the fraction
distance dij is often used so that ηij is proportional of pheromone remained on the link
to 1/dij, which implies that shorter routes will be
 Δτijk=Q/Lk if the ant k traveled the path cijk
selected due to their shorter traveling time, and
between cities i and j; Q is some constant, and
thus the pheromone concentrations on these routes
are higher. This is because the traveling time is Lk is the length of the kth ant's travel
 Δτijk=0 otherwise.
smaller and because of this, lesser amount of
pheromone has been evaporated during this cycle. 2.2 Elistist Ant System
The representation of problem is important, it is The solution construction and evaporation are as
to be in form of graph containing nodes and links defined in the Ant System Algorithm.
connecting nodes in order to be solved by Ant
System. The pheromone update is a little bit different the
best to date ant deposits an additional quantity of
There are total six variants of Ant System used by pheromone on paths it traveled
various researchers and it is still a developing field.
The definitions and methodology in these variants (4)
are taken from different sources mentioned in
Here, τij =Q/C if the path ij is from τ , τ is
bs bs bs bs
references. These variants are as follows-
the best to date round trip, e is an algorithm's
2.1 Ant System parameter. It was reported, that the best value for e
The Ant System Optimization algorithm is the is between four and six[7].
first algorithm proposed and analyzed by Marco 2.3 Ranked Ant System
Dorigo[8]. Initially every link between nodes has
some initial amount of pheromone. Each ant starts Again, the difference is in the pheromone update.
from a randomly assigned node and goes from a For each of iterations, the best to date ant and
node to the next node until all nodes are visited additionally the w-1 best ants for this iteration are
exactly once. selected. The best to date and each of the selected
ranked ants deposit pheromone on paths they
The ant at node i selects the next node to visit by travel[7].
calculating probabilities:
(5)
(2)
Here,
Q is a constant, Lr is the length of the rth ant trip, e
 s - partial solution
p
is the additional multiplier. Again, the best results
 N - set of all paths from the node i to all are at e≥1. The member w*Δτijbs is similar to the
adjacent nodes still not visited by the ant Elitist Ant Algorithm parameter for the best to date
 c - path from the city i to the city j
ij ant.
 p - probability
2.4 Best-Worst Ant System
 τij - amount of pheromone on the path cij
 ηij - some heuristic factor, usually ηij=Q/dij, The solution construction and evaporation rule is
where dij is a distance between cities i and j, as defined in Ant System.
Q is some constant But only the best to date ant updates pheromones:
 α and β - algorithm parameters.
(6)
The ant choses path with highest probability.
After each ant made its route to destination Here, if the path ij ϵ τ , τ is the best to
bs bs
node, pheromones on paths connecting the date round trip, Cbs -length of the trip.
nodes is updated according to the formula: In addition, the paths of the round trip of the worst

ant for the current iteration that are not in the best to  Tik is a set of cities adjacent to the current city
date solution are subject to additional evaporation: i at step k,
 τil is the amount of pheromone on the path cil,
(7)
 Q is a constant (in this app it is a
Here, ρw is an additional evaporation factor for
city's extent), dil is a distance between the
all crsϵTw and crs∉Tw∩Tbs, Tw is the worst solution
cities i and l,
for the given iteration, and Tbs is the best to
 β - algorithm parameter
date solution. Usually this algorithm includes a
mutation of the pheromone on selected paths of the After the ant moves from the city i to city j,
trail, and the algorithm we resets all pheromones the pheromone on the path cij is evaporated
to the initial value and restart search for the best and updated according to
solution when the process seems stuck.[7] (11)
 ξ -the local evaporation factor,
2.5 Min-Max Ant System
 τ0 - additional pheromone (an algorithm
The solution construction is according to the
parameter).
equation Ant System. There are variants in the
selection of the ants allowed to update pheromones: After the first ant has made the kth step and has
the best to date ant, or the best for current iteration, modified a pheromone, it does not begin the next
or the best after latest reset ant, or the best to date step it waits until all ants complete the kth step and
ant for even iterations, and the best for iteration for pheromones are modified on their travelled path.
odd iterations.[7] After all ants complete iteration global evaporation
is applied to paths.
There are min and max pheromone limits to the
quantity of pheromone on the paths between cities, Ant colony optimization (ACO) algorithms are the
τmin and τmax. It is believed that prevents local loops. currently state of art for solving the many complex
So, evaporation is: optimization problems in regard to vehicle routing,
route planning and scheduling.
(8) 3. APPLICATION OF ANT ALGORITHM
The update is: AND ITS VARIANTS IN VEHICLE
ROUTING AND NETWORK DESIGN
(9) PROBLEMS:
Here τij =Q/C if the path ijϵ T , T is a selected
bs sel sel sel
 Shao-han et.al. (2005) investigates a
ant's round trip, Csel is the length of the trip. shortest-path network problem using an
2.6 Ant Colony annealed ant system algorithm, in which an
annealing strategy is embedded to find out the
In all previous algorithms pheromones were
probabilities to decide which path the ants will
updated only after all ants completed travel for a
select next. The shortest-path problem is to
given iteration. Ant Colony System is different.
determine the shortest route between a source
After an ant moves from one city to the next,
and a destination in a transportation-network
the pheromone on the traveled path is reduced
topology. In this approach, two globally
by some value. The solution construction is also
optimizing annealed ant algorithms that are
different[8].
Concentrated Model and Distributed Model are
The algorithm has a parameter q0ϵ[0,1]. The constructed. The Concentrated Model (CM)
parameter sets the boundary between two rules to means all ants are initially concentrated in the
select the next city. On each step, the algorithm source node while all ants randomly select a
generates a random number 0 ≤ q ≤1. If q0<q, the node except the destination as their starting
next city to move is selected according to Ant point initially and at least one must appear
System; else index j of the next city to move is: in the source node for the Distributed Model
j = argmax kT k {τil*ηilβ} (10) (DM)[15].
i
 Yan and Xiyu (2007) studied another variant
Where,
of ant system is the ant-clustering algorithm,
 i is the index of the current city, which is used in macroscopic planning of
 j is the index of the city to move next, transportation hub. They reported an unbalance

in construction of highway and transportation industrial application with 57, 80, 96 and 132
hub. So highway main hub cities are clustered cities, respectively. For better comparability
using cluster analysis. K-means algorithm is simulated annealing algorithm, classical meta-
the most widely used algorithm in clustering heuristic and genetic algorithm also applied to
analysis. A new clustering algorithm is five test problems. For each six algorithm 30
developed with fusion of K-mean algorithm runs were performed on each of the five test
and ant clustering algorithm. The experimental problems. For the 30 city problem the results of
results show this algorithm can more all methods were very good, with the classical
effectively solve clustering problem than K meta-heuristics being slightly better than the
means algorithm and ant clustering algorithm ant system algorithms. Fairly the same was true
used alone[19]. for the 57-city-problem, all methods were able
 Ant algorithms can also be used to find out to solve the problem optimally, except basic
wide path on a large map. The conventional AS, which had the highest average deviation
ant model is modified by Krist of and Balazs from the optimum. With increasing problem
(2004). The paths found by the algorithm size, the results achieved by the ants became
consist of multiple alternative routes near better compared to the other procedures. For
each other without relevant difference in the problem with 80 cities, the optimal solution
length (or cost). (The result looks like a wide could still be found by simulated annealing and
path on a topographical map.) This leads to the improved ant strategies, while basic AS
a loosely defined path, which is very useful, and the genetic algorithm failed. The classical
because obstacle avoidance maneuvers can be methods generated tours that exceeded the
completed without leaving the path itself. This optimum on average by more than 2.5%
means that the one doesn’t have to worry about whereas for elitist ant system and rank-based
getting back on the route, which would need ant system the corresponding value was less
additional searches. Using the method of the than 1%. The larger problems (n = 96 and n
ants this way leads to a very different search = 132) could not be solved optimally by any
type (in search space handling and in result of the methods within the allowed runtime, but
as well), than the conventional approaches. the best solutions found were still very close to
A further advantage of the algorithm is that it the optimum. Thus the ant system could for all
provides further information about the search problem instances compete with the classical
space in connection with traffic load[12]. meta-heuristics regarding speed and quality,
 Majid, et.al. (2012) used a modified version of and that the ranking improved the performance
elitist ant system in order to avoid local optimum of the ant system algorithm in every respect[4].
points in Travelling Salesman Problem. In  Zhang et.al. (2011) studied Best-Worst ant
contrast to the classical Elitist Ant System system in application to Travelling Salesman
algorithms, the proposed algorithm uses only a Problem. The Best-Worst Ant System (BWAS)
global updating, which will increase pheromone algorithm achieves a strong exploitation of
on the edges of the best (i.e. the shortest) route the search history by allowing both the best
and will at the same time decrease the amount solutions and the worst solutions to change
of pheromone on the edges of the worst (i.e. the pheromone during the pheromone trail update.
longest) route. The algorithm like its natural It not only makes effective use of the positive
version makes use of pheromone evaporation feedback of iteration (global)-best ant, but
in order to prevent rapid convergence of ants to also makes full use of the negative feedback
a sub-optimal path. In other words, pheromone of iteration (global)-worst ant. It improves
density is reduced in each iteration by the efficiency significantly. Results obtained
0≤ρ≤ 1[13]. with an application to the TSP have shown
 Bullnheimer et.al. in 1997 compared three that BWAS is very effective in finding optimal
version of ant system. These versions of the solutions and achieves a strongly improved
ant system algorithm, i.e. Ant System, Elitist performance compared to ACO[18].
Ant System and rank based ant system, were  Stutzle and Holger (2000) applied max-min
applied to five different TSP instances: a 30 city ant system to Travelling Salesman Problem
example and four real-life problems from an and Quadratic Assignment problem in order

to know its benefits over classical ant system customer locations. A linear programming (LP)
algorithm. The Max –Min Ant System base algorithm is also applied to MDVRP in
(MMAS) can easily be extended by adding order to compare the optimality of solution.
local search algorithms. MMAS shows better Based on simulation studies and statistical
performance for symmetric TSP instances than evaluations, it was found that the ant colony
ACS while ACS and MMAS reach the same optimization technique leads to more optimal
level of performance on Asymmetric TSPs. results as compared to the existing LP based
The computational results with the MMAS, method. Furthermore, the ant colony based
Elitist and Rank based ant systems suggest method used provides much consistent solution
that the best computational results are obtained over different runs showing convergence of the
when in addition to the pheromone trail limits solution to a small neighborhood [14].
effective diversification mechanisms based on  Gao et.al. (2016) used the Ant Colony
pheromone re-initialization are used. In general, Optimisation in Dynamic Vehicle Routing
we found that an effective search diversification Problem. Ant colony algorithm can resolve
is necessary to achieve best performance when dynamic optimization problems due to its
applying Ellitist Ant System or Rank-based robustness and adaptation. In this study, a
Ant System with additional local search[16]. clustering ant colony optimization (KACO)
 Narasimha et.al. (2013) applied ant colony with three immigrant’s schemes is used to
optimization technique to Multi-Depot Vehicle address the DLRP. The DLRP is constituted
routing problem (MDVRP). The Multi-Depot by a location allocation problem (LAP) and a
Vehicle Routing Problem (MDVRP) involves vehicles routing problem (VRP) in dynamic
minimizing the total distance traveled by environments. The K-means algorithm was
vehicles originating from multiple depots so used to optimize the location of problem so as to
that the vehicles together visit the specified determine the depots and nearby cities divided
customer locations (or cities) exactly once. This into the same class. Then ant search algorithm
problem belongs to a class of Nondeterministic with three immigrants, i.e., random immigrants,
Polynomial Hard (NP Hard) problems. In this elitism-based immigrants and memory-based
study a variant of MDVRP is used, called immigrants aimed at resolving the vehicle
min–max MDVRP, where the objective is routing problem in the dynamic environments
to minimize the tour-length of the vehicle including random and cyclic environments.
traveling the longest distance in MDVRP. Min– Experimental results suggest that KACO with
max MDVRP is of specific significance for three immigrant schemes can find promising
time-critical applications such as emergency solutions for the DLRP within reasonable
response, where one wants to minimize the computational time. To verify the effect of
time taken to attend any customer. An extension the K-means on KACO, the performance of
of an existing ant-colony technique for solving KACO is also compared with the ACO without
the Single Depot Vehicle Routing Problem K-means algorithm which is called WKACO.
(SDVRP) is applied to solve the multiple depots Compared with the solution found by WKACO
and min–max variants of the problem. The the solution obtained by KACO can generate
approach is based on decomposing the overall shorter routes by starting from multiple depots.
problem into several min–max SDVRPs via The comparative results demonstrate that the
equitable partitioning of the region consisting of K-mean algorithm improves the solution found
depots and cities. The min–max SDVRP is then by the ant search and reinforces the effect of
solved using the ant colony method that finds immigrants’ schemes for three different scales
out the minimum value of distance constraint of LRP instances. Results also demonstrate
that yields a solution to the traditional SDVRP. that the clustering algorithm can significantly
This optimal distance constraint, when used in improve the performance of KACO in terms
a traditional SDVRP, minimizes the maximal of the qualities and robustness of solutions,
distance traveled by a vehicle. The proposed suggesting that the proposed algorithm
method was implemented in MATLAB may lead to a new technique for tracking
for obtaining the solution for the min–max the environmental changes by utilizing its
MDVRP with any number of vehicles and clustering and evolutionary characteristics [10].

 Suman as part of his PhD thesis (2011) length of the route is minimized then it may
developed Ant System algorithm based be possible that none of the shortest route will
decision support system for public transport be satisfied by the designed route. Therefore
management officials for bus route design. modification in pheromone update rule of basic
An example problem is taken to validate the Ant System algorithm function done where Lk
methodology and later this design can be route length of kth ant travel should be between
applied to any city bus network. The network a specified maximum and minimum route
with Origin-Destination data is assumed (same length value. Thus, by specifying the maximum
could be generated by ticketing data for any and minimum permissible route length values,
network). O and D data represents the travel the route can be designed with the maximum
behavior of the travelers and a traveler always Weightage and least route length between the
prefers the shortest route between the origin specified values[17].
and the destination. Thus the maximum trip 4. APPLICATION OF ANT SYSTEM
in the O and D table should be served by the ALGORITHM TO REAL NETWORK
shortest route. A computer program based upon
the C# programming logics was developed The basic ant system algorithm is applied to a 22
for the ant algorithm to find the shortest route node network from Indore city of Madhya Pradesh
between the nodes. The trips made by travelers India. The selected network consists of three main
are assigned as a Weightage factor to the links arteries of the city and connecting roads. One is
that are used in their shortest route. For bus the busiest road of the city AB road, second inner
route design the basic Ant system algorithm road with high traffic and third outer ring road
used in finding the shortest route is modified. with comparatively less traffic. Travel time data
The probability function is modified to contain is also collected on these roads for both weekdays
the weightage factors assigned to links. With and weekend and peak and off peak hours. The
the above modified probability function the network selected has effect of office and leisure
resulted optimal route, will be a route having time activities both.
maximum Weightage value with least route The basic ant system algorithm is the used to find
length. If the Weightage value is maximized out shortest route between selected origin and
then the trips satisfied by their shortest route destination using both distance and travel time as
may also be maximized. But, if the total impedance. The program is designed using C-sharp
Fig. 1 Selected 22 Node Network of City Indore

programming language in Visual Studio.

 The algorithm is initialized with the
following parameters: (12)
 Ant Count m: Numbers of ants that will  The total number of ants placed at origin
be exploring the population, node then transferred to next valid nodes
 Iterations I: Number of generations for and process is iterated till all ants reach
which ants will be searching for solutions, the destination node.
 Initial Pheromone τ0: The constant  After that pheromone intensities on links
pheromone value in starting applied are updated using following rule:
equally to all links,
 Pheromone Decay Constant ρ: Strength
with which pheromone decays in a path (13)
0< ρ<1 If minimum route length of best ant or ants

 Alpha α: Pheromone strength while

is less than the value of global minimum
finding the next node route length, this value is set as L0. The
 Beta β: Heuristic strength while finding process is then iterated to specified number
the next node of iterations or when all ants started
 Global Minimum Route length L0: travelling through one single route.
checked after each iteration with The code developed gives Optimum results within
minimum route length and updated, permissible calculation time. When distance is
initially set to a very large value, used as impedance the results are verified by
 Network Details: number of nodes and network analyst extension of ArcGIS software and
link lengths found to be correct when number of iterations is
 Origin and Destination node 20 or more. However at less number of iteration
results deviate from optimum value. Thus when
 Ant Route Generation
travel time is used as impedance number iteration
 All ants placed at origin node

selected is 20 or more than that. To verify results
 Probability of transfer to next node the algorithm is also applied to a 10 and 15 point
calculated by rule: example network, where optimum results are
Fig. 2 Code Window and Application Interface for Finding Shortest Route

obtained at lesser number of iterations. Thus with Vehicle Routing Problem. In: Evolutionary Computation in
increase in number of nodes number of iterations Combinatorial Optimization". Proceedings of the Lecture
Notes in Computer Science, vol. 3004, pp. 72–83.
provided should be large or end condition should
7. Dorigo, M. and Stützle, T.,2004, “Ant Colony Optimization”
be when all ants follow the same route to reach
Bradford Book, ISBN 0262042193.
destination.
8. Dorigo, M., 1992, “Optimization, Learning and Natural
5. SUMMARY AND CONCLUSION Algorithms [in Italian]". PhD thesis, Dipartimento Die
Elettronica, Politecnico di Milano, Milan.
Complex traffic and transportation problems call 9. Dorigo, M., Maniezzo, V. and Colorni, A., 1996, “The Ant
for development of modern systems that merge System: Optimization by a Colony of Cooperating Agents”
knowledge, techniques, and methodologies from IEEE Transactions on Systems, Man, and Cybernetics-Part
various scientific areas. B, 26:29–41.
The application of the ant based algorithm to 10. Gao, S., Wang, Y., Cheng, J., Inazumi, Y., and Tang, Z.,
2016, “Ant colony Optimization with Clustering for
traffic and transportation systems will prove
Solving the Dynamic Location Routing Problem”, Applied
to be beneficial, for future research in terms of Mathematics and Computation, vol-285, pp 149-173
computation time and quality of the solutions 11. Koushik Venkata Narasimha, Elad Kivelevitch, Balaji
discovered. It can be successfully applied to Sharma, Manish Kumar, 2013, “An Ant Colony
routing as well as network design problems, Optimization Technique for Solving Min–Max Multi-
and can be successfully applied to other traffic Depot Vehicle Routing Problem”, Swarm and Evolutionary
problems too. Solving multi-objective vehicle Computation 13, elseviers.com, pp. 63–73
routing, scheduling and dispatching problems by 12. Kristof Csorba and Balazs Todor, 2004, “Ant Search
ant algorithm approach should be also real research on Large Maps to Find Wide Paths”, Second IEEE
International Conference on Intelligent Systems.
challenge in years to come.
13. Majid, Y., Farza, D. and Farhad, R.,2012, “Modification
Solutions can be developed for new and existing of the Elite Ant System in Order to Avoid Local Optimum
network design problems based on hybrids of Points in the Traveling Salesman Problem”, Cornell
different problems. The successful application University, Ithaca, New York.
of ant algorithms to difficult combinatorial 14. Narasimha, K. V., Kivelevitch, E., Sharma, B., and Kumar,
M., 2013, “An Ant Colony Optimization Technique
optimization problems in various parts of world
for Solving Min–Max Multi-Depot Vehicle Routing
is also very encouraging. One could expect Problem”, Swarm and Evolutionary Computation,
development of more ant based models for vol-13, pp.63-73
solving real-time vehicle routing and public transit 15. Shao-Han Liu, Jzau-Sheng Lin, and Zi-Sheng Lin, “A
problems. Shortest-Path Network Problem using an Annealed Ant
System Algorithm”, Department of Computer Science and
REFERENCES Information Engineering, National Chin-Yi Institute of
1. Bell, J.E., McMullen, P.R., 2004. "Ant Colony Optimization Technology,
Techniques for the Vehicle Routing Problem". Advanced 16. Stutzle, T, and Holger H. H., 2000, “MAX-MIN Ant
Engineering Informatics 18, 41–48. System”, Future Generation Computer Systems, elseviers.
2. Beni, G., Wang, J., 1989. "Swarm Intelligence". In: com, pp. 889-914.
Proceedings of the Seventh Annual Meeting of the Robotics 17. Suman, S., 2011, “Web GIS Based Advanced Public
Society of Japan. RSJ Press, Tokyo, pp. 425–428. Transport System Using Mapserver And Ant Algorithm”,
3. Bernd Bullnheimer, Richard F., Hartl Christine Straut, Ph.D. Thesis, Indian Institute of Technology Roorkee,
1997, “A New Rank Based Version of the Ant System - A India.
Computational Study”, Department of Management Science, 18. Van Zhang, Hao Wang, Yonghua Zhang, Yun Chen,
University of Vienna, Austria 2011, “BEST-WORST Ant System”, 3rd International
4. Bullnheimer, B., Kotsis, G., Strauss, C., 1998, Conference on Advanced Computer Control, IEEE.
"Parallelization Strategies for the Ant System". In: De 19. Yan Meng and Xiyu Liu, “Application of K-means
Leone, R., Murli, A., Pardalos, P., Toraldo, G. (Eds.), Algorithm Based on Ant Clustering Algorithm in
"High Performance Algorithms and Software in Nonlinear Macroscopic Planning of Highway Transportation Hub”,
Optimization", Applied Optimization. Kluwer Academic Department of Management and Economy, Shandong
Publishers, Boston, pp. 87–100. Normal University, Jinan, China.
5. Deneubourg, J.L., Aron, S., Goss, S., Pasteels, J.M., 1990. 20. Zhishuo, L., Yueting, C., 2005. Sweep Based Multiple
"The Self-Organizing Exploratory Pattern of the Argentine Ant Colonies Algorithm For Capacitated Vehicle Routing
Ant". Journal of Insect Behavior 3, 159–168 Problem. In: Proceedings of the 2005 IEEE International
6. Doerner, K.F., Hartl, R.F., Kiechle, G., Lucka, M., Reimann, Conference on e-Business Engineering (ICEBE’05). IEEE
M., 2004. "Parallel Ant Systems for the Capacitated Computer Society, Beijing, pp. 387–394.

Logarithmic Matching for Road Culvert Performance
Curve
Subir Kumar Podder1
Abstract
Applications of the logarithmic matching have been successfully tried in several cases in open-
channel flows, coastal hydrodynamics and sediment transport. The asymptotes for Headwater
Depth at inlet for flows through a culvert under Outlet Control Type and Inlet Control Type can
be expressed with logarithmic functions. This Paper does not deal with hydraulics of culverts.
But it presents an application of the asymptotic matching method, the logarithmic matching, for
matching the two asymptotic solutions as a single composite one. In doing so it uses output data
from hydraulic analysis of a specific pipe culvert configuration. It is intended to assist a practicing
engineer to develop a single expression, for other cases, to expediently use in his spreadsheet
computations.
1. Introduction However, the immediately following section is

Applications of the logarithmic matching have intended at clarifying how performance curves are
been successfully tried in several cases in open- used by practicing engineers.
channel flows, coastal hydrodynamics and sediment 2. Use of Culvert Performance
transport (Junke Guo, 2002)[3]. This Paper applies Curves
it to performance curve of culverts.
Basic Requirements of Culvert Hydraulic
A performance curve is a plot of headwater Design – In hydraulic design of a culvert, the
depth or elevation versus flow rate. The resulting following are the required inputs:
graphical depiction of culvert operation is useful (a) Design Discharge, Q
in evaluating the hydraulic capacity of a culvert
(b) Design pipe diameter, D
for various headwaters. In developing a culvert
performance curve, both inlet and outlet control (c) Inlet configuration (with or without wing-
curves must be plotted. This is necessary because walls, whether or not projected beyond the
the dominant control at a given headwater is hard embankment toe, depressed or not etc.)
to predict. Also, control may shift from the inlet to (d) Cross section of the channel at the pipe outlet
the outlet, or vice-versa, over a range of flow rates to determine the tail-water depth
(HDS -5, FHWA)[1]. (e) Height of embankment (HEmb) at the culvert
location
This Paper however does not deal with the
hydraulics of culverts, but it only delves onto How are the aforesaid inputs used in design –
a mathematical application which will allow The application is essentially carried out by a
engineers to use the ‘performance curves’ more practicing engineer in the following steps:
expediently in their spreadsheet computations. (i) The first four mentioned above are those that
1. Associate Director, LEA Associates South Asia Pvt. Ltd., New Delhi, E-mail: subir@lasaindia.com

Podder on
Logarithmic Matching for Road Culvert Performance Curve
decide the hydraulics, which essentially get the above. HDS-5, FHWA[1] is a readily available
used for computing the Head Water Depth at resource that describes the equations involved.
Inlet (HWi). What this Paper intends – Actually the ‘Culvert
(ii) This HWi is then compared with HEmb to Performance Curve’ comprises of two parts, one
determine whether the computed HWi is that corresponds to ‘Inlet Control’ and the other
acceptable or not. ‘Outlet Control’. Between them is the transition
(iii) Further several codes limits the value of the zone. A practicing engineer is therefore required
ratio HWi/D for new roads /upgraded roads to use two regression equations, and also an
being designed. Often a value of HWi = 1.2D approximation if the transition zone applies to his/
is used for new road designs, and often this her case.
even gets stipulated in manuals/contracts. This Paper delves on a method to combine all these
(iv) But for repair/rehabilitation road projects situations under a single curve. The advantage is
the decision of a culverts’ acceptability is that then the engineer is required to use only a single
practically decided by the extent of HWi equation in his spreadsheet. The Paper resorts to
vis-à-vis HEmb (such that the water level in ‘logarithmic matching’ which is a practice adopted
the worst scenario does not adversely affect by engineers working for many other hydraulic
the pavement). That a stringent requirement applications (say weirs)[3].
would either require a much higher
embankment or deny reuse of an existing The example given in the Paper is for the specific
culvert, in repair projects only the ‘chances case of a pipe culvert with only one barrel of
of affecting the pavement’ get considered, i.e. 1.2 m in diameter, and of length 16.0 m. For other
HWi Vs HEmb comes into play. cases, hydraulic computations will first have to
be undertaken. Thereafter, a similar approach, as
Basics of Culvert Hydraulics – A same culvert this Paper outlines, can be used to derive a single
can operate as “Outlet Control” or “Inlet Control”, representative expression for engineers to readily
and that depends on a number of factors. Further, use them in their spreadsheet computations.
for either of the controls, inlets can be submerged
or un-submerged, and that too has a bearing on the 3. Fundamentals of Culvert
computations involved. Controls
In practice an engineer has to determine HWi for Inlet and outlet control are the two basic types of
both the conditions (for different Q), and then the flow control. The basis for the classification system
maximum HWi is adopted. is the location of the control section. However it
is reiterated that the following are for reference
FHWA[1] gives elaborate details of equations
only, and that this Paper only uses data that gets
involved, and the same get reproduced in many
derived from computations carried out based on
other codes/manuals.
these principles.
What is Culvert Performance Curve – Major Inlet control occurs when the culvert barrel is
manuals and codes stipulate the following: Plot capable of conveying more flow than the inlet will
HWi/D Vs Q/AD0.5, where ‘A’ is the cross sectional accept. The control section of a culvert operating
area of the culvert barrel. This plot is called under inlet control is located just inside the entrance.
Culvert Performance Curve. A practicing engineer Critical depth occurs at or near this location, and
is required to use this for deciding the HWi. The the flow regime immediately downstream is
engineer however can resort to regression equation supercritical. Hydraulic characteristics downstream
(of the curve) to facilitate his/her spreadsheet of the inlet control section do not affect the culvert
computation. capacity. The upstream water surface elevation
Are these detailed in this Paper – This Paper and the inlet geometry represent the major flow
does not detail these, because these are standard controls. The inlet geometry includes the barrel
computations which find reproduction in codes shape, cross-sectional area, and the inlet edge.
and manual practiced in many countries. It is The two basic conditions of inlet control depend
considered that a practicing engineer is exposed to upon whether the inlet end of the culvert is or is

Podder on
not submerged by the upstream headwater. If the of flow through the pipe, D = diameter for pipe
inlet is not submerged, the inlet performs as a culvert, or barrel depth for box culvert]. Details,
weir. If the inlet is submerged, the inlet performs with examples along with nomographs, are given
as an orifice. Equations are available for each of in HDS-5, FHWA[1].
the above conditions in Appendix-A, HDS-5,
4. Logarithmic Matching – The
FHWA[1].
Concept
Outlet control flow occurs when the culvert
Suppose two asymptotic solutions for a
barrel is not capable of conveying as much flow
nonlinear problem, using an analytical method
as the inlet opening will accept. The control or experimental method, exist. Then these two
section for outlet control flow in a culvert is asymptotes can be expressed by or transformed
located at the barrel exit or further downstream. into the following (Junke Guo, 2001)[3].
Either subcritical or pressure flow exists in the
culvert barrel under these conditions. All of the у = K1 lnx + C1 (1)
geometric and hydraulic characteristics of the for x << x0, and
culvert play a role in determining its capacity. y = K2 lnx + C2 (2)
These characteristics include all of the factors for x >> x0. In the two equations above, x is an
governing inlet control, the water surface independent variable, y is a dependent variable, K1
elevation at the outlet, and the slope, length, and and K2 are two slopes based on a logarithmic scale,
hydraulic roughness of the culvert barrel. Tail- shown in Fig. 1, C1 and C2 are two intercepts, and
water is defined as the depth of water downstream x0 is a reference of x.
of the culvert measured from the outlet invert.
To merge the two asymptotes Eq. 1 and Eq. 2 into a
It is an important factor in determining culvert
single composite equation, two logarithmic models
capacity under outlet control conditions. Outlet
can be developed.
control flow conditions can be calculated based
on energy balance. The total energy required Model I is
to pass the flow through the culvert barrel is y = K1 lnx + α ln [1 + (x/x0 )β ] + C1(3)
made up of the entrance loss, the friction losses and
through the barrel, and the exit loss. Other losses, Model II is
including bend losses, losses at junctions, and
y = K2 lnx – α ln {1 – exp [- (x/x0 )β ]} + C2 (4)
loses at grates should be included as appropriate.
In order to avoid tedious backwater calculations, In the above two equations α and x0 are determined
approximate methods have been developed with K1, K2, C1 and C2, and β is a transitional shape
for outlet control analysis. Based on numerous parameter that is determined by a collocation
backwater calculations performed by FHWA, it method or a least-squares method (Griffiths and
was found that a downstream extension of the Smith 1991).
full flow hydraulic grade line pierces the plane For Model I, it is easy to see that for x << x0,
of the culvert outlet at a point one-half way ln [1 + (x/x0 )β >>> 0 (5)
between critical depth and the top of the barrel. then Eq. 3 reduces to Eq. 1
Therefore, it is possible to begin the hydraulic For x >> x0, one has
grade line at a depth of (dc+D)/2 above the
ln [1 + (x/x0 )β >>> β lnx – β lnx0 (6)
outlet invert and extend the straight, full flow
hydraulic grade line upstream to the inlet of the Substituting the equation above into Eq. 3 gives
culvert at a slope of Sn (the barrel slope). If the that
tailwater exceeds (dc+D)/2, the tailwater is used y = (K1 + αβ) lnx + (C1 – β lnx0) (7)
to set the downstream end of the extended full Comparing Eq. 7 with Eq. 2 yields
flow hydraulic grade line. The inlet losses and
K1 + αβ= K2 (8)
the velocity head are added to the elevation of
the hydraulic grade line at the inlet to obtain the and
headwater elevation. [Here, dc= critical depth C1-αβ lnx0 = C2(9)

Podder on
The two equations above lead to many possible connections. In practice, one may
α = (K2 - K1)/β (10) compare the two expressions by an error analysis
and pick up the better one.
and
x0 = exp [ (C1 – C2) / (K2 – K1) (11) 5. Framework for Model
Development
Now one can see that x0 is the cross point of
the two asymptotes, shown in Fig. 1. Therefore, The preliminary step in the development of the
Model I can be further written as model is the identification of two appropriate
y=K1lnx + [(K2–K1)/β] ln{1+(x/x0 )β }+C1 (12) asymptotic solutions, which the model then
matches as a single composite one using logarithmic
in which x0 is determined by Eq. 11, and β is a matching technique.
transitional shape parameter that is the only
undetermined parameter. Note that the logarithmic In this Paper the case of a pipe culvert with a single
matching fails when K1 = K2. 1.2 m dia barrel and length 16m is considered. Inlet
Headwater depths (Hwi) were first determined for
Similarly, one can show that for Model II, Eq. 4
a range of design discharges (0.2 – 3.6 m3/s), for
can be written as
both Inlet Control and Outlet Control conditions.
y = K2 lnx + [(K1–K2) / β] ln{[1 - exp Depending on which condition yields a larger
[-(x/x0)β]}+C2 (13) Hwi (for each discharge) determined the dominant
in which x0 and β are similar to those in Eq. 12. control (as described in the section Introduction at
Eq. 12 or 13 is the solid line in Fig. 1. It is to be noted the onset).
that Eq. 12 and Eq. 13 are only two simple ways in It is to be noted that for the Inlet Condition up to a
y +C2 certain discharge, the inlet end of the culvert is not
Inx submerged. Beyond a certain discharge the inlet
K 2
y=
end of the culvert is submerged by the upstream
headwater. Therefore there remains a transition
range even for the Inlet Control condition itself (as
described in the section Fundamentals of Controls
1
C
at the onset).
x+
In
Table 1 gives the details of the calculations that

1
K
y=
form the basic asymptotes. It is noted that the

In x0 In x
Paper does not detail the computations of the data
presented in Table 1, since the focus is on the
Fig. 1 application of the data.
Table 1 Input Parameters for the Asymptotes

Inlet Control Outlet Control Design Control Type
y x (corresponds to
Q Q/AD0.5 HWi/D HWi/D HWi/D max. HWi/D)
0.2 0.331 0.482 0.836 0.836 Outlet
0.4 0.663 0.573 0.975 0.975 Outlet
0.7 1.160 0.753 1.150 1.150 Outlet
1.0 1.657 0.988 1.315 1.315 Outlet
1.3 2.154 1.277 1.488 1.488 Outlet
1.4 2.320 1.385 1.549 1.549 Outlet
2.5 4.142 2.971 2.508 2.971 Inlet (submerged)
2.8 4.639 3.529 2.881 3.529 Inlet (submerged)
3.3 5.468 4.580 3.608 4.580 Inlet (submerged)
3.6 5.965 5.282 4.105 5.282 Inlet (submerged)

Podder on
then computed (see Fig. 3). It is appreciated that

these errors result from the assumptions that went
into the expressions for HWi for Inlet and Outlet
Controls (as described in the section Fundamentals
of Controls at the onset), apart from the transitions
that exist between these conditions as well as
those for Inlet Controls condition itself (between
Fig. 2 Inlet Control Curve (Q/AD0.5 Vs HWi /D), the inlet end of the culvert submerged and not
HWi /D as the x-axis submerged). The ‘error’ (ɛmodel) plotted against x
showed reasonable correlation (Fig. 3)
ɛmodel = -0.75 ln (x) + 0.96 (15)
Continuing from above, the single asymptotic
relationship for the design discharge was derived
using
ydesign = ymodel – ɛmodel
Fig. 3 Outlet Control Curve (Q/AD0.5 Vs HWi /D) Using the data set, the relationship that gets derived
HWi /D as the x-axis follows (Fig. 4).
Figs. 2 and 3 give the basic asymptotes, for Inlet ydesign = 3.0465 ln(x) + 0.8996 (16)
and Outlet control respectively. Fig. 4 is also intended at how the model (Eq. 16)
6. Application of the Framework represents a single composite relationship by
and Development of the Model addressing, both the Inlet Control and the Outlet
Control conditions, as well as the transition regions
From Figs. 2 and 3, the following can be worked
between them.
out:
7.000
K1= 2.3230, C1 = 1.7992, K2 = 3.6514, C2 = 0.7622.
Then, x0 = 2.18 follows, derived from Eq. 11.
6.000
Thereafter β the transitional shape parameter was
focussed upon. Applying Eq. 12, and using trial and
error, β was developed for each set of x=HWi/D 5.000
and y=Q/AD0.5.
A polynomial was fitted to derive the relationship
4.000
between β and x. Using this, the following
relationship between x and y was then established Outlet Control
(Fig. 4): 3.000
ymodel = 2.2965 ln (x) + 1.7996 (14) Inlet Control
2.000
Model, Eqn 16
1.000
Eqn 14
0.000
Fig. 4 ɛmodel Vs HWi/D
0.000 2.000 4.000 6.000
(HWi/D in the x-axis)
Fig. 5 The Modelled Single Asymptotic Relationship
Comparing the derived y values using Eq. 14
with design y values, errors in the estimates were This relationship is applicable for a pipe culvert

Podder on
with a single 1.2 m dia barrel and length 16m. Height for box culvert);
Similar expressions can be modelled for different A = Total area of the culvert barrel;
barrel sizes (both box and pipes) and barrels
lengths. The advantage is that these then allow HWi = headwater depth at culvert inlet;
avoiding rigorous computations and facilitate a dc = flow critical depth;
ready numerical application.
x = independent variable of asymptotic relations;
7. Application of the developed
y = dependent variable of asymptotic relations;
model
For a pipe culvert with a single 1.2 m dia barrel K1 and K2 = are two slopes based on a logarithmic
and length 16 m, the following demonstrates scale for the asymptotic relationships;
the application of the model [Eq. 16] developed C1 and C2 = are intercepts based on a logarithmic
using logarithmic matching) for a specific design scale for the asymptotic relationships;
discharge.
x0 = a reference of x used in the single asymptotic
Q = 2.8 m3/s relation developed;
D = 0.900 m β = transitional shape parameter.
A = 0.636 m2 References
Y = Q/AD0.5 = 4.639 1. FHWA, "Hydraulic Design Series Number
5, Hydraulic Design of Highway Culverts",
HWi = 3.072 m Publication No. FHWA-NHI-01-020, September
X = HWi /D = 3.413 2001, (Revised May 2005)
Ydesign = 4.639 2. Griffiths, D.V. AND Smith, I.M (1991),

"Numerical Methods for Engineers", CRC Press,
(Using Model, Eq. 16) 264-266.
Nomenclature 3. JUNKE GUO, "Logarithmic Matching and
its Applications in Computational Hydraulics
Q = Design flow rate; and Sediment Transport", Journal of Hydraulic
D = Culvert dimension (Diameter for pipe, or Research, Vol. 40, 2002, No. 5
GENERAL REPORT ON ROAD RESEARCH WORK DONE IN INDIA - CALL FOR

SUBMISSION OF REPORTS ON ROAD RESEARCH CARRIED OUT
DURING THE YEAR 2017-18
One of the objectives of the Indian Roads Congress is to disseminate and propagate technical knowledge
and make Civil Engineers aware about National/International research studies. To propagate importance
of research and make available all research related information under single publication, IRC annually
compiles research reports on Road & Bridge Research works being done in India, from various organisations
like, MORTH, NHAI, CPWD, BRO, NRRDA, IITs, NITs, Engineering Colleges, Contractors, Consultants
and Researchers. With the help of Central Road Research Institute, the compiled data is published by IRC
as “General Report on Road Research Work Done in India” every year.
Organisations concerned with research and development, construction, monitoring and maintenance of
Road & Bridge works, Traffic, Transportation and Geotechnical Engineering, etc are requested to report
the findings of Research & Development Projects carried out during the year 2017-18 in the relevant
Proformae, which will prove beneficial to the members of the highway profession. The Proformae are
available on the Website of IRC (www.irc.nic.in) and can be downloaded from there. The Reports may
please be sent to the Secretary General, Indian Roads Congress, Kama Koti Marg, Sector 6, R.K. Puram,
New Delhi – 110 022 by post or e-mail at: tc.irc@gov.in latest by 25th June, 2018.

Suman & Sinha on
Pavement Surface Condition Prediction by Markov Chains
Pavement Surface Condition Prediction

by Markov Chains
S.K. Suman1 S. Sinha2
ABSTRACT
A model for prediction of pavement surface condition plays a vital role in Pavement Management System
(PMS). In developing countries like India, PMS is the needed approach for the optimum utilisation of the
available scarce resources. Pavement management System is concerned with optimal use of materials in
time and space, leading to cost optimization.
This Paper focuses on methodology involved in the prediction of pavement condition using probabilistic
techniques. As pavement condition prediction parameter like traffic loading, pavement materials,
construction methods and environmental condition are not deterministic therefore probabilistic
techniques is used i.e. the present study. Markov chains have the property that probabilities involving the
process will evolve in the future depend only on the present state of the process and so are independent
of the events in the past. The state of the transition matrix will be defined based on the Pavement
Condition Index (PCI) and element of the transition matrix will be determined by the probability
method. Fifteen cycles was carried out because the expected design life of the bituminous roads is
fifteen years. The probabilistic model requires only a minimal amount of data such as pavement class,
pavement condition of two consecutive years and the pavement length. PCI shall be utilized as index
and the present technique in pavement management systems will create good systems which may lead
to more savings of the roads maintenance funds and enhance the ability of the roads network to provide
better level of service.
1. INTRODUCTION pavement performance in developing country like
Prediction of pavement surface condition models India in general and the eastern province of Bihar
reflects the deterioration process of pavement in particular where such prediction models are not
conditions over the age and provides forecasting of existent. The aim of this study is to provide the
pavement condition over time. It plays a vital role in overall process for prediction of pavement condition
pavement management system. The pavement surface using Markov Chains, to compute the Pavement
condition is affected by parameters like traffic axle Condition Index (PCI) while incorporating the every
loads, environmental conditions, moisture content types of distress, their density and severity level for
etc. are themselves uncertain in nature hence the rate each sections of the considered pavement, to develop
of pavement deterioration is uncertain. Modelling a model based on temporal data and to validate the
uncertainty requires the use of probabilistic operation developed model by degree of fit and information
research techniques. In this context, the use of available with Road Construction Department, in the
Markov chain in prediction of model captures the Eastern state of Bihar in India.
uncertainties behaviour of pavement deterioration.
The most important advantage of this technique is 2. MARKOV CHAINS
that it requires only two years or any two threshold A Markov Chain is a special type of discrete-time
justified temporal data. The present study can be stochastic process, when the state of a system Xt+1 at
taken as the first step regarding assessment of time t+1 depends on the state of the system Xt at some
1. Assistant Professor, E-mail : sksuman@nitp.ac.in
Department of Civil Engineering, NIT Patna
2. Professor

Suman & Sinha on
previous time t but does not depend on how the state The entry of 1 in the last row of the transition matrix
of the system Xt was obtained. This can be expressed corresponding to state 100(PCI of 0 to 100) indicate
as P(Xt+1 = j|Xt = i), where P is the probability of the a holding or trapping state. The pavement condition
state at time t+1 being j given that the state at time cannot transit from this state unless repair action is
t was i, assuming that the probability is independent performed. The state vector for any duty cycle t is
of time. This assumption is formally known as the obtained by multiplying the initial state vector X (0)
stationary assumption. This stationary assumption is by the transition matrix P raised to the power of t.
used because of the limited time period of the data Thus,
collected. If data were collected over a large period, X (1) = X (0).P
the quantity and type of materials used might change
over time and influence how a typical pavement X (2) = X (1).P = X (0).P2
section would determine. A Markov Chain can be ………………………..
summarized through a probability transition matrix
and the initial state probabilities. To model pavement ………………………..
deterioration with time, it is necessary to establish a X (t) = X (t-1).P = X (0).Pt
Transition Probability Matrix (TPM), denoted by P.
With this procedure, if the transition matrix
The general form of P is given below: probabilities can be estimated, the future state of the
road at any duty cycle t can be predicted. After the
several cycle operations, rows of the matrix come
to identical entries, the reason is that probabilities
in any row are the steady state probabilities for the
Markov chain i.e. the probabilities of the state after
enough time has elapsed that the initial state is no
The matrix contains all of the information necessary longer relevant.
to model the movement of the process among the
condition states. The transition probabilities pij 3. MARKOV TRANSITION PROBABILITY
indicate the probability of the portion of the network MATRICES
in condition i moving to condition j in one duty cycle. These are useful in representing the change in
Two more conditions apply to the process when it is condition of the system from one state to another
used to simulate pavement deterioration. First, pij = 0 over time. The key elements of any Markov transition
for i> j signifying the belief that roads cannot improve matrices are states and transitions. A method is
in condition without first receiving treatment. Second, presented herewith for computing the transition
pnn=1, signifying a holding state whereby roads that probability matrix from temporal data of pavement
have reached their worst condition cannot deteriorate
surface condition deteriorates over time. The element
further. Consequently, in pavement deterioration, the
of the transition probability matrix (pij) may be
general form of the transition matrix P is denoted by
estimated by using Eq. 1.
P 1.
(1)
Where, Nij = number of road sections in the network
that moved from condition i to condition j during one
duty cycle and Ni = total number of road sections
that started the year in condition i. The proportions
are likely to vary from year to year thereby requiring
A further restriction allowing the condition to an average to be determined for each pij to ensure
deteriorate by no more than one state in one duty accuracy in the model.
cycle is commonly used in pavement deterioration
modelling. The transition probability matrix is then 4. PAVEMENT CONDITION INDEX
denoted by P2. Pavement Condition Index (PCI) is a numerical
indicator of present pavement surface condition
that reflects the riding quality of pavement
surface. The PAVER method (Shahin M.Y. 1994)
for computing the PCI is a function of the types,
severity and the density of the distresses. It is
a numerical index that uses a scale from 0 to 100.

Suman & Sinha on
Fig. 1 shows schematic diagram on which PCI existing practice prevalent is to take up the maintenance
depends and PCI rating with rating score scale. work of any deteriorated pavement by simple observation
by the engineers of the concern department. Lack of any
scientific basis for denudation has led towards irrational
prioritizing the maintenance works leading to severe
wastage of inadequate resources. In the present study
adopted methodology was implemented on black topped
road in Patna, Bihar. As a study area a stretch of NH30
was selected for measurement of surface condition
because it caters maximum traffic with their axle loads
incoming and outgoing from the Patna city.
Visual surface condition survey was carried out on
NH30 for ten km of this road length and total considered
road length is divided into small section of 100 m
length towards approaching to rational data. PCI for
Fig. 1 Schematic Diagram for Rating Pavement
every selected road section was determined based on
Condition the measured value of density and severity level of the
distresses, according to the PAVER method guidelines.
5. DATA COLLECTION
For the development of transition probability matrix,
In spite of non-existence of pavement condition PCI was evaluated at the beginning of the year 2011
prediction model in developing country like India, and the end of the year 2011 as details in this contest is
there has not been much attention given to the area. The tabulated in Table 1.
Table 1 PCI Data Along the Chainage in Jan, 2011 & Dec., 2011
PCI PCI PCI PCI

Chainage (X102)
Chainage (X102)
Chainage (X102)
Chainage (X102)
Dec, 2011
Dec, 2011
Dec, 2011
Dec, 2011
Jan, 2011
Jan, 2011
Jan, 2011
Jan, 2011
0-1 52 47 25-26 54 48 50-51 90 81 75-76 84.5 76
1-2 78 72 26-27 95.5 86 51-52 46 41 76-77 70 63
2-3 86 77 27-28 97.5 88 52-53 97 87 77-78 100 90
3-4 94 84 28-29 62 56 53-54 95.5 86 78-79 60 54
4-5 91 82 29-30 94 83 54-55 100 90 79-80 95 85.5
5-6 97 87 30-31 91 88 55-56 95 86 80-81 97.5 87
6-7 90.5 80 31-32 92.5 83 56-57 90.5 81 81-82 96.5 86
7-8 100 90 32-33 91 82 57-58 95 85.5 82-83 70 63
8-9 91 82 33-34 96.5 87 58-59 90 81 83-84 91.5 82
9-10 74 67 34-35 56 50 59-60 71.5 64 84-85 94 84
10-11 100 91 35-36 97.5 87.5 60-61 100 90 85-86 100 90
11-12 96 86 36-37 95 85 61-62 82 73 86-87 92 82
12-13 90 81 37-38 100 97 62-63 95 85.5 87-88 91.5 82
13-14 96 86.5 38-39 96.5 86 63-64 55 49 88-89 95.5 86
14-15 100 90 39-40 81.5 73 64-65 91.5 82 89-90 86 77
15-16 84 75.5 40-41 96.5 87 65-66 74 66 90-91 94 84
16-17 85 77 41-42 97.4 86 66-67 92 82 91-92 74 66
17-18 96 86 42-43 92.5 82 67-68 97.5 87.5 92-93 56 50
18-19 95 85.5 43-44 97.5 88 68-69 87 78 93-94 74 66
19-20 90.5 81 44-45 60 54 69-70 90.5 81 94-95 96 86
20-21 88 79 45-46 91.5 82 70-71 77 69 95-96 93 83
21-22 92 82 46-47 86 77.5 71-72 90 81 96-97 96 86
22-23 100 90 47-48 74 66 72-73 90 80 97-98 96.5 87
23-24 90.5 80 48-49 100 90 73-74 84 75.5 98-99 91 82
24-25 98 88 49-50 96 87.5 74-75 78 70.5 99-100 78 71

Suman & Sinha on
6. DEVELOPMENT OF MODEL Pavement transition matrix is generated based on

The pavement surface condition vector of 1x7 was initial pavement condition index and final pavement
obtained based on the measured and calculated PCI condition index. During this process some data
value as visual surface condition survey was conducted are remain in initial state after transition and some
in the year Jan, 2011. Here X (0) is represented by are moves to lower state of conditions. The details
condition vector whose element is the PCI mentioned process can be seen from Table 2 to Table 4.
against the rating score as shown below.
100-85 85-70 70-55 55-40 40-25 25-10 10-0

X(0) =
73 16 7 4 0 0 0
Table 2 Pavement Condition Rating Distribution

Sl. No. Condition state Transition Corresponding Condition Rating Number of pavement
sections
1 Excellent→Excellent 100-85→100-85 40
2 Excellent→Very Good 100-85→85-70 33
3 Very Good→Very Good 85-70→85-70 9
4 Very Good→Good 85-70→70-55 7
5 Good→Good 70-55→70-55 3
6 Good→Fair 70-55→55-40 4
7 Fair→Fair 55-40→55-40 4
8 Fair→ Poor 55-40→40-25 0
9 Poor →Poor 40-25→ 40-25 0
10 Poor →Very Poor 40-25→ 25-10 0
11 Very Poor→Very Poor 25-10→ 25-10 0
12 Very Poor →Failed 25-10→10-0 0
13 Failed →Failed 10-0→ 10-0 0
14 Total No. of sections(100 m) 100
Table 3 Pre Transition Probability Matrix for Pavement Condition

Future pavement condition in year(t+1) Total No.
Excellent Very Good Good Fair Poor Very Poor Failed of Sections
condition in year(t)
Excellent 40 33 0 0 0 0 0 73
Present pavement
Very Good 0 9 7 0 0 0 0 16
Good 0 0 3 4 0 0 0 7
Fair 0 0 0 4 0 0 0 4
Poor 0 0 0 0 0 0 0 0
Very Poor 0 0 0 0 0 0 0 0
Failed 0 0 0 0 0 0 0 0
Total No. of Sections 100
Table 4 Transition Probability Matrix for Pavement Condition

Future pavement condition in year(t+1)
Present pavement condition
Excellent Very Good Good Fair Poor Very Poor Failed

Excellent 0.548 0.452 0 0 0 0 0
in year(t)
Very Good 0 0.563 0.438 0 0 0 0

Good 0 0 0.429 0.571 0 0 0
Fair 0 0 0 1.000 0 0 0
Poor 0 0 0 0 0 0 0
Very Poor 0 0 0 0 0 0 0
Failed 0 0 0 0 0 0 1

Suman & Sinha on
Table 5 Predicted Pavement Condition Vector from Year 2011 to 2025

2011 = 40 42 10 8 0 0 0
2012 = 21.91781 41.71 22.66 13.71429 0 0 0.00
2013 = 12.00976 33.37 27.96 26.66327 0 0 0.00
2014 = 6.580689 24.20 26.58 42.63963 0 0 0.00
2015 = 3.605857 16.59 21.98 57.82872 0 0 0.00
2016 = 1.975812 10.96 16.68 70.38803 0 0 0.00
2017 = 1.082637 7.06 11.94 79.91725 0 0 0.00
2018 = 0.593226 4.46 8.21 86.7412 0 0 0.00
2019 = 0.325055 2.78 5.47 91.4303 0 0 0.00
2020 = 0.178112 1.71 3.56 94.55483 0 0 0.00
2021 = 0.097596 1.04 2.27 96.58809 0 0 0.00
2022 = 0.053477 0.63 1.43 97.8867 0 0 0.00
2023 = 0.029303 0.38 0.89 98.70369 0 0 0.00
2024 = 0.016056 0.23 0.55 99.21135 0 0 0.00
2025 = 0.008798 0.13 0.33 99.52357 0 0 0.00
The pavement condition vector after completion of 8. CONCLUSION

one year i.e. at the end of the year 2011 is obtained Present study demonstrates the overall process
by multiplying the pavement condition row vector for developing the pavement performance model
(1x7) to transition probability matrix (7x7). Finally based on Markov chain. The case study gives the
a 1x7 matrix is obtained. For every one year cycle idea how Markov’s chain can be used for prediction
pavement condition predicted and represented by the of pavement condition. Base year pavement
row vector as shown in Table 5.
conditions have been identified using pavement
The developed deterioration of pavement model for condition index computed by using PAVER
the bituminous topped road under consideration over method. Markov chains process is applied for
the year is indicated in Fig. 2. predicting the horizon year pavement conditions.
The developed model of six degree parabola with
coefficient of regression equal to 0.9995 and error
of 1.25% in predicted and observed value, shows
the satisfactory result with a reasonable degree
of accuracy. Therefore, Developed model can be
used for prediction of pavement surface condition
in future. Further this model can be used by the
road construction department in planning of
Fig. 2 Model for Prediction of Pavement Condition for maintenance and repair of roads.
NH30 (10km) Over Year 2011-25 REFERENCES
7. VALIDATION OF MODEL 1. Butt A.A., Shahin M.Y., Feighan K.J. & Carpenter S.H.
(1987), "Pavement Performance Prediction Model using the
The developed model is validated by first Coefficient Markov Process", Transportation Research Record, TRB,
of regression (R2) which value is 0.9995, indicates Washington D.C.,1123,12-19.
approaching to one and secondly this is done by 2. Hillear F.S. & Lieberman G.J.(2009), "Introduction to
comparing the observed condition rating for the Operations Research", Eighth Edition, Tata McGraw Hill
Education Private Ltd., New Delhi
month of June of the year 2012 with the condition
rating predicted by the deterioration model developed 3. Joseph N.T. & Chou Y.J. (2001), "Pavement Performance
Analysis Applying Probabilistic Deterioration Methods",
for the same year. The reason for comparing the Transportation Research Record, TRB, Washington D.C.,
condition rating in year 2012 is that this is the 1769, 01-2962, 20-27.
only year for which the data is available with Road 4. Ortiz-Garcia J.J., Costello S.B. & Snaith M.S. (2006),
Construction Department (RCD), Bihar. In year 2012, "Derivation of Transition Probability Matrices for Pavement
the pavement condition rating obtained by using the Deterioration Modelling", Journal of Transportation
Engineering, Vol. 132, No. 2, ASCE, 141-161
model is 67.25%, whereas the value obtained from
RCD for the same year is 66%.The difference in the 5. Shiyab A.M.S.H. (2007), "Optimum Use of the Flexible
Pavement Condition Indicators in Pavement Management
value is 1.25% .This result shows that the developed System", Ph.D. Thesis, Curtin University of Technology,
model is able to predict future condition rating with Online at www.adt.curtin.edu.au/thesis/available/adt,
reasonable degree of accuracy. Accessed on January 18, 2012.

Copyright ©
Regd. No. DELENG/2008/26432 with the Registrar of Newspapers
CONTENTS
PAVEMENT ENGINEERING
Ø “Performance Evaluation of Super-Pave and Marshall Mix Design to 1

Suite Indian Climatic and Traffic Conditions”
By Hemant Kumar, Dr. Sangita & Dr. Vandana Tare
Ø “Effect of Polypropylene Granules in Asphalt Mix Prepared 13

with varying Packing Characteristics”
By M. Satish Goud, A. Ramesh, V. Venkat Ramayya & M. Kumar
Ø “Performance and Fatigue Behaviour of Alkali Activated Slag Concrete 21

with Steel Slag as Coarse Aggregate for Concrete Pavement”
By Nitendra Palankar, A.U.Ravi Shankar & Mithun B.M.
TRAFFIC ENGINEERING
Ø “A Review of Ant Algorithm and its Variants in Reference to

Vehicle Routing Problems in Transportation” 35
By Dr. Praveen Kumar & Ms. Shambhavi Mishra
Ø “Logarithmic Matching for Road Culvert Performance Curve” 44

By Subir Kumar Podder
ASSET MANAGEMENT
Ø “Pavement Surface Condition Prediction by Markov Chains” 50

By S.K. Suman & S. Sinha

IRC Highway Research Journal January-June 2018 PDF

Uploaded by

Copyright:

Available Formats

IRC Highway Research Journal January-June 2018 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

IRC Highway Research Journal January-June 2018 PDF

Uploaded by

Copyright:

Available Formats

HIGHWAY RESEARCH

Highway Research Board

Volume 9 No. 1 January - June, 2018

Highway Research Board

Publisher & Editor: S.K. Nirmal, Secretary General, IRC

Headquarter: Satellite Office:

The opinions and conclusions in this Journal are those of the

Hemant Kumar1 Dr. Sangita2 Dr. (Mrs.) Vandana Tare3

1. Introduction field compaction by road rollers and subsequently

1. M.E Student in Transportation Engineering, S.G.S.I.T.S Indore, Email: mithktyagi2010@gmail.com

Highway Research Journal, January - June, 2018 1

fundamental properties such as fatigue, rutting  To compare Marshall Parameters obtained by

2 Highway Research Journal, January - June, 2018

pavement temperature is 4°C. In the present Table 1 Conformity of Available VG Grade

Table 2 Aggregate Gradation for Super-Pave Mix Design Method

Combined Grading Adopted Control Points Restricted Zone

Highway Research Journal, January - June, 2018 3

Super-Pave uses 0.45 power gradation charts to

Table 4 Estimated Volumetric Properties @ 4% Air Voids for Different Mixes

4 Highway Research Journal, January - June, 2018

Fig. 1 Average Denisification Curves for BC

Mixture Volumetric Properties @ Ndes for BC NMAS with VG30

Highway Research Journal, January - June, 2018 5

(Contd. from pre- page)

Mix Property Result

6 Highway Research Journal, January - June, 2018

% Bitumen by Average Gmm (gm/ % Va %VMA %VFB Marshall Marshall

% Bitumen Average Gmm % Va %VMA %VFB Marshall Marshall

Highway Research Journal, January - June, 2018 7

8 Highway Research Journal, January - June, 2018

7.2 Tensile Strength Ratio (TSR)

improves the stiffness modulus of a mix substantially

7.3 Resilient Modulus 7.4 Flow Number

Highway Research Journal, January - June, 2018 9

Flow Number (cycles)

Binder Type Binder Type

Flow Number (Cycles)

Fig. 6 (b) Flow Number Variation with Binder Type.@

Fig.7(b) Flow Number at 207 kPa

7.4.3 Effect of mix design method on Flow

10  Highway Research Journal, January - June, 2018

method for BC mixes. In general, OBC

 The flow number decreases with the increases

Highway Research Journal, January - June, 2018 11

References Vol. 11, No. 2, June 2013.

12  Highway Research Journal, January - June, 2018

M. Satish Goud1 A. Ramesh2 V. Venkat Ramayya3 M. Kumar4

1. Introduction Vavrik et al (2002). The Bailey method of gradation

Highway Research Journal, January - June, 2018 13

of above mix combination for development of

14  Highway Research Journal, January - June, 2018

the laboratory. PP granules were providing better

4.2 Gradation of the Mix 40

Highway Research Journal, January - June, 2018 15

Fig. 2 Slab Compactor and Immersion Type Wheel Rutting Equipment

16  Highway Research Journal, January - June, 2018

Solenoid value Table 2 Physical Properties of Bitumen

Load cell Kinematic Viscosity at 352.47 350 IS : 1206

Highway Research Journal, January - June, 2018 17

Highway Research Journal, January - June, 2018 1

2 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 3

4 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 5

6 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 7

8 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 9

10 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 11

12 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 13

14 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 15

16 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 17

18 Highway Research Journal, January - June, 2018

TSR= S1/ S2(5) 18

Highway Research Journal, January - June, 2018 19

20 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 21

22 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 23

24 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 25

26 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 27

28 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 29

30 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 31

32 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 33

34 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 35

36 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 37

38 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 39

40 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 41

42 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 43

44 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 45

46 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 47

48 Highway Research Journal, January - June, 2018

Highway Research Journal, January - June, 2018 49