You are on page 1of 752

Volume 82

ASPHALT PAVING
TECHNOLOGY
2013
JOURNAL OF THE ASSOCIATION
OF ASPHALT PAVING TECHNOLOGISTS

Denver, Colorado
April 710, 2013

DEStech Publications, Inc.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Asphalt Paving Technology 2013Volume 82

Produced by:
DEStech Publications, Inc.
439 North Duke Street
Lancaster, Pennsylvania 17602 U.S.A.

Copyright 2013 by the Association of Asphalt Paving Technologists


A l l r i g h t s r es er v e d

No part of this publication may be reproduced, stored in a


retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of the Association of
Asphalt Paving Technologists.

Printed in the United States of America


10 9 8 7 6 5 4 3 2 1

M ai n e n t r y u n d er t i t l e:
Asphalt Paving Technology 2013Volume 82

ISSN No. 0270-2932

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Journals of the Association were printed for the meetings listed below
and may be obtained from the Secretary-Treasurer, 6776 Lake Drive,
Suite 215, Lino Lakes MN 55014

1974 Williamsburg, Vol. 43 15.00 1999 Chicago, Vol. 68 75.00


1975 Phoenix, Vol. 44 20.00 1999 75th Ann, Volume 68A 25.00
1977 San Antonio, Vol. 46 25.00 2000 Reno, Vol. 69 75.00
1982 Kansas City, Vol. 51 40.00 2001 Clearwater, Vol. 70 80.00
1983 Atlanta, Vol. 52 45.00 2002 Colorado Springs, Vol. 71 80.00
1984 Scottsdale, Vol. 53 50.00 2003 Lexington, Vol. 72 80.00
1986 Clearwater, Vol. 55 50.00 2004 Baton Rouge, Vol. 73 90.00
1988 Williamsburg, Vol. 57 55.00 2005 Long Beach, Vol.74 90.00
1989 Nashville, Vol. 58 55.00 2006 Savannah, Vol.75 100.00
1990 Albuquerque, Vol. 59 55.00 2007 San Antonio, Vol. 76 100.00
1991 Seattle, Vol. 60 60.00 2008 Philadelphia, Vol. 77 125.00
1992 Charleston, Vol. 61 60.00 2009 Minneapolis, Vol. 78 150.00
1993 Austin, Vol. 62 60.00 2010 Sacramento, Vol. 79 150.00
1994 St. Louis, Vol. 63 60.00 2011 Tampa, Vol. 80 180.00
1995 Portland, Vol. 64 70.00 2012 Austin, Vol. 81 180.00
1996 Baltimore, Vol. 65 70.00 2013 Denver, Vol. 82 200.00
1997 Salt Lake City, Vol. 66 75.00
1998 Boston, Vol. 67 75.00

Prices to AAPT members for extra copies of the Journal are $5 less than
those listed.

"The Association of Asphalt Paving Technologists is not responsible for the


statements and opinions advanced in its publications."

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Copyright 2013
The Association of Asphalt Paving Technologists
ISSN 0270-2932
I.S.I. Certified

Reprints of Articles included in Volumes which are out-of-print may be


obtained from the AAPT Office.

Printed in the United States of America by DEStech, Lancaster, PA, US

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ASSOCIATION OF ASPHALT PAVING TECHNOLOGISTS
2013

W.J. Emmons Annual Award

The Board of Directors announces the recipients of the W.J. Emmons


Annual Award of the Association of Asphalt Paving Technologists for the
best paper presented at the 2012 Meeting in Austin, TX, held April 14,
2012, to be:

Winner:

Effects of Interface Condition Characteristics on Open-Graded


Friction Course Top-Down Cracking Performance

Yu Chen, Gabriele Tebaldi, Reynaldo Roque, George Lopp and


Yumin Su; University of Florida

Runner-up:
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Laboratory Evaluation of Asphalt Mixtures Containing Dry Added


Ground Tire Rubber and a Processing Aid

Gaylon L. Baumgardner, J. Michael Hemsley, Walter Jordan III,


Paragon Technical Services; Isaac L. Howard, Mississippi State
University

AAPT Scholarship Fund

The Board of Directors announces the 2013 winners of the AAPT


Scholarship to be:

Matthew Gersch, Auburn University

Mirella Villani, Delft University of Technology

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AAPT Board of Directors 2012-2013

From left:
1st Row: Maria Bonaquist, Jeyna Rowe, Marcia Prowell, Rebecca McDaniel,
Emma McDaniel Burley, Betty Skok, Amy Epps-Martin
2nd Row: Ramon Bonaquist, Geoffrey Rowe, Brian Prowell, Eileen Soler,
Mike Anderson, Rey Roque, David Newcomb, Gene Skok

2013-2014 Board of Directors

President: Ramon Bonaquist


First Vice-President: Brian Prowell
Second Vice-President: Geoffrey Rowe
Director-at-Large: Bill Buttlar
Director-at-Large: Randy West
Past President: Rebecca McDaniel
Past President: David Newcomb
--`,`,,,``,,```,,`,,,,,`,``

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Introduction of Mr. Ken Kandhal
as 2012 Honorary Member of AAPT
by E. Ray Brown

It is my pleasure to recognize Mr. Prithvi (Ken) Kandhal who has


been elected as honorary member in the Association of Asphalt
Paving Technologists (AAPT). Ken has been actively involved in
materials, mix design, and construction of asphalt pavements for over
40 years and is clearly deserving of this award. Ken has been very
active in all activities of AAPT since he joined the organization in
1972. As a result of his many contributions and leadership, he has
helped to advance AAPTs reputation to the preeminent organization
that it is today.

Kens interest in asphalt pavements began in India where he was first


introduced to asphalt as a highway agency employee. After working
there a few years, Ken moved to the US and began work as a graduate
student at Iowa State University where he received his Masters
Degree in Civil Engineering in 1969. After graduation he went to
work for the Pennsylvania Department of Transportation where he
served as the Chief Asphalt Engineer from 1970-1988. In 1988, he
began work at the National Center for Asphalt Technology where he
first served as Assistant Director and then Associate Director for many
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
years. When he retired in 2001, Ken was awarded the title of
Associate Director Emeritus of NCAT. Since his retirement Ken has
continued to work as a consultant on various asphalt projects
worldwide.

While with the Pennsylvania DOT and NCAT, Ken was very
successful in performing research and in publishing the results. Some
of his more significant accomplishments have involved work related
to: Superpave restricted zone, stripping of asphalt pavements,
performance testing of HMA mixtures, construction guidance for
longitudinal joints, improved tests for determining bulk specific
gravity of fine aggregate, testing and characterization of mineral
fillers, and aggregate tests related to HMA performance. He is also
one of the key authors of the textbook Hot Mix Asphalt Materials, Mix
Design, and Construction.

During his career, Ken has published more than 120 technical journal
articles as well as many other articles in magazines, conference
proceedings, and technical reports. Ken has published more than 20
AAPT papers during a 30-year period with his first paper being
published in 1973. During this time he published the AAPT papers in
four different decades: five in the 1970s, seven in the 1980s, six in
the 1990s and four in the 2000s. This shows a record of continued
active participation in AAPT activities.

Ken received the Emmons Award for best AAPT paper in 1989 for his
work on stripping of asphalt mixtures. He has been a member of
AAPT for over 40 years and served as President from 1999-2000.
During his year as President, Ken wrote personnel letters to all visitors
attending the annual AAPT conference personally thanking them for
their attendance and requesting that they consider joining AAPT. As a
result of his efforts there was a significant increase in AAPT
membership.

Ken has been very active in a number of other technical societies and
organizations. He is a registered engineer in the state of Pennsylvania.
He is a Fellow in ASTM and served as the Chairman of Committee
D04 on Road and Paving Materials (19981999). As a result of his

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
support to ASTM, he received the Award of Merit in 2003. He has
served in TRB where he was Chairman of Committee A2D02 from
1982-1988. He is now an emeritus member of A2D02. He is an
ASCE Fellow and served as Chairman of the Subcommittee on
Bituminous Materials from 1988-1992.

Ken is very worthy of this recognition and it is now my pleasure to


present to you, my good friend, Mr. Ken Kandhal.

Prithvi S. Kandhal Remarks

I am humbled by this honor bestowed upon me especially when it


comes from my peers in all continents of the world.

It is time to reflect back on circumstances and people who helped


increase my knowledge of asphalt paving technology during the last
40 years. It is not possible to name all the individuals; I will mention
only those with whom I worked closely. Over the years several people
assisted me in discharging my duty as President of the AAPT;
Chairman of the ASTM Main Committee D04 on Road and Paving
Materials; Chairman of the TRB Committee A2D02 on Asphalt
Mixtures; and Chairman of the ASCE Subcommittee on Bituminous
Materials.

As a young highway engineer in a desert region of India, which did


not have any stone quarry, I had the opportunity to construct about one
mile of road using dune sand and asphalt binder. Fortunately, I had a
copy of the Highway Engineering Handbook edited by K.B. Woods of
Purdue University, which I used for guidance. I used both Hubbard
Field and Marshall testing equipment at that time.

Having developed interest in asphalt, I came to Iowa State University


in Ames, Iowa, in 1968 for pursuing graduate study. Both Prof. Ladis
Csanyi, who invented the foamed asphalt process and Prof. Dah-yinn
Lee were engaged in asphalt research there. I even got the opportunity
to make a batch of foamed asphalt using steam. I completed my
graduate study under Prof. Lee researching asphalt absorption by

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
aggregates and development of an innovative method of determining
bulk specific gravity of aggregates using dyes.

Then in 1970 I joined the Pennsylvania Department of Transportation


(Penn DOT) as State Bituminous Engineer where both Jim Moulthrop
and Ron Cominsky were my colleagues. Later they played key roles in
administering SHRPs Asphalt Program. I was successful in
implementing up-to-date asphalt paving technology in Pennsylvania
with the able assistance of Carl Lubold of the Pennsylvania Asphalt
Pavement Association and Carlos Rosenberger of the Asphalt

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Institute. I am obliged to Bill Koehler who as Engineer of Tests gave
me the necessary environment and the freedom in my advancement as
an asphalt paving technologist.

Fortunately, in Penn DOT I was in charge of almost all asphalt-related


activities such as routine testing, research, development of
specifications and test methods, and troubleshooting in the field. This
helped me to get a well rounded experience in asphalt technology. One
thing really helped me in authoring a number of papers for AAPT,
TRB and ASTM while at Penn DOT. During the winter time, asphalt
technicians in my laboratory had little work to do. When visitors used
to tour the laboratory, it was a chronic complaint that technicians
(state employees!) are sitting idle. To address that problem, I started to
collect all types of aggregates from various quarries across the state
and also collect samples of AC-20 asphalt cement from various
refineries. Then I would draw up a research work plan. That would
keep all the technicians busy during the winter time and a lot of test
data were generated. Obviously, I had to sit down, analyze the test
data and write a research paper. We had just changed from penetration
grading to viscosity grading of asphalt cements in the US. There was a
lot to be researched to fully understand the viscosity graded asphalts.

There was another motivation to write research papers. Every time a


new governor came to office in Pennsylvania, the first thing which
was done was to restrict travel of state employees outside the state.
Only those who had a paper to present were allowed travel outside
Pennsylvania. By writing papers, I could attend AAPT and TRB
meetings regularly. During my tenure at Penn DOT of about 17 years,

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
I had the opportunity to author about 35 refereed papers. Very few
people in state DOTs publish technical papers. However, I was
inspired by Chuck Hughes and Gale Page, both working for state
DOTs, to do so. Gene Skok of AAPT, Bob Nady of Iowa, Bob
Dunning, and Byron Ruth of University of Florida also encouraged me
by appreciating my practical papers.

In 1987, I got a call from Dr. Freddy Roberts, the then Director of the
National Center for Asphalt Technology (NCAT) to consider joining
the newly established center at Auburn University, Alabama. I joined
NCAT in March 1988 because it was a challenge to establish a new
center with a skeleton staff of three engineers: Freddy Roberts as
Director and Ray Brown and myself as two Assistant Directors.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Fortunately, SHRP came at the right time. NCAT was awarded the
SHRP A-003 Project on Asphalt-Aggregate Interaction: Adhesion and
Absorption. I had the privilege of working with Bob Dunning, who
moved to Auburn to work on that project. Ray Brown and I were also
successful in obtaining several major NCHRP projects to sustain the
Center in the initial formative years. Both of us feel proud when we
see that NCAT has come a long way now to be recognized as a
premier asphalt research center in the world.

As Associate Director of NCAT and a member of the graduate faculty


of Civil Engineering at Auburn University, I had the opportunity of
advising and working with several bright Masters and Ph.D. students.
They were in alphabetical order: Shane Buchanan, Sanjoy
Chakraborty, Allen Cooley, Steve Cross, Kee Foo, John Haddock,
Mike Huner, Maqbool Khatri, Cynthia Lynn, Todd Lynn, Rajib
Mallick, S.S. Rao, Jay Winford, Yiping Wu, and Jingna Zhang. Rajib
Mallick and Allen Cooley, who are talented researchers, worked with
me on several research projects.

While at NCAT, I was also fortunate to work with Freddy Roberts,


Ray Brown, Doug Hanson. Frazier Parker, Mary Stroup-Gardiner,
Randy West, Don Watson, Brian Prowell, Buzz Powell, and Dave
Timm. Being a key instructor for the NCAT Professor Training
Course, I had the opportunity to meet many asphalt technologists in
the academia across the US.

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Although I retired from NCAT in 2001, it is difficult to get the asphalt
out of my system. Therefore, whenever my services are needed in any
country for troubleshooting prematurely distressed asphalt pavements
or training in asphalt technology I take up the challenge. All what I
have learned from colleagues like you has been helpful to me in
meeting such challenges.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

I am also thankful to my wife, Uma, who had to put up with my


obsession with asphalt all these years.

In closing, I again thank the AAPT Board of Directors and all


members for giving me this honor which I will cherish. Thanks.

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Key Note Address

Review and Implications of IARC 103 for the Asphalt Pavement


Industry

Anthony J. Kriech, Director of Research, Heritage Research


Group

Bitumen (asphalt in the U.S.) was reviewed by the International


Agency for Research on Cancer (IARC) in October of 2011. This
eight day comprehensive review of the literature resulted in IARC
classifying Occupational Exposure to straight-run bitumens and their
emissions during road paving as possibly carcinogenic to humans
(Group 2B). IARC determined this classification by reviewing the
relevant peer-reviewed journals related to cancer and bitumen. IARC
conducts these reviews through scientific working groups composed
of four areas of expertise. These include exposure, human, animal and
mechanistic subgroups. These subgroups evaluate the studies within
each area of expertise before taking it to the overall group for
discussion.
The classifications are assigned based on epidemiological evidence in
humans, experimental evidence in animals, mechanistic and other
relevant studies. IARC classifies agents as follows: Category 1 -
there is evidence of cancer in humans. Category 2A is probably
carcinogenic to humans, includes agents with limited evidence in
humans and sufficient evidence in experimental animals. Category 2B
- is used for agents with limited evidence of carcinogenicity in humans
and less than sufficient evidence in animals. Category 3 - is used for
agents where there is inadequate evidence in humans and experimental
animals. Category 4 - is for agents with evidence suggesting lack of
carcinogenicity in humans and experimental animals. In some cases,
such as bitumen, classification can be upgraded based on strong
evidence from mechanistic and other relevant data.
Human Data
By far the most extensive epidemiological study of bitumen workers, a
study of over 80,000 European bitumen workers in eight countries, did
not find any link to occupational activities and excess cancers (Olsson

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
et al., 2010). The authors investigated the contribution of exposure to

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
bitumen, other occupational agents, and tobacco smoking to lung
cancer risk. We found no consistent evidence of an association
between indicators of either inhalation or dermal exposure to bitumen
and lung cancer risk. A sizable proportion of the excess mortality
from lung cancer relative to the general population observed in the
earlier cohort phase is likely attributable to high tobacco
consumption, and possibly to coal tar exposure, while other
occupational agents do not appear to play an important role.
Animal
A significant 24-month inhalation study was conducted by Fuhst et al.
(2007) to evaluate possible carcinogenic effects of paving bitumen
emissions on rats. The fume condensate used for this study was
matched to paving industry field results. Data from this study showed
no increase in the number of tumor-bearing animals in any of the
bitumen-exposed groups as compared to the clean air control group at
concentrations up to 172.5 mg/m3. Typical asphalt paving exposure is
<1 mg/m3. Field-matched bitumen fume condensates collected from
the headspace above paving tanks (Kriech et al., 2007) were evaluated
in two-year dermal carcinogenicity assays in mice (Clark et al., 2011).
The authors concluded that the paving fume condensate was not
carcinogenic under the test conditions.
Mechanistic
The IARC subgroup on Mechanistics and Other Relevant Studies
reviewed a number of mechanistic studies involving paving asphalt.
The focus of many of these studies related to the presence of
Polynuclear Aromatic Hydrocarbons (PAHs) and Compounds (PACs).
There was evidence that some of these compounds were mutagenic
and carcinogens producing genotoxic activities. IARC determined
that bitumen fume caused cellular stress and disrupted cellular defense
systems.
Other mechanistic studies showed genotoxic effects in both animals
and humans. This was the first time that IARCs mechanistic subgroup
upgraded an agent which was a complex mixture (asphalt) rather than
a simple agent (chemical), especially when human and animal
evidence was inadequate (negative).

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Implications
IARC determines classification of hazards, not risk. Hazard and risk

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
distinctions are important. Hazard is the potential to cause harm; risk
is the likelihood of harm considering the degree of exposure. It is
possible to have a cancer hazard but not have a significant risk.
Normally risk is determined by regulatory agencies or established
scientific bodies.
The paving industry has been active for many years in reducing
exposure to workers. These include engineering controls on pavers as
well as warm mix technologies that reduce exposure through reducing
temperatures during paving operations. Recently published studies
also found that replacing diesel oil with biodiesel (B100) for cleaning
equipment and tools also greatly reduced workers exposure as well as
simply wearing gloves.
The industry should continue to look for ways to further reduce
exposure through reduction in temperatures. Improved methods of
measuring exposure focused on reduction in PAH exposures will help
reduce exposure in the workplace of paving operations.
References
Olsson, A., Kromhout, H., Agostini, M., Hansen, J., Lassen, C.F., Johansen, C.,
Kjaerheim, K., Langrd, S., Stckerm, I., Ahrens, W., Behrens, T., Lindbohl,
M.L., Heikkil, P., Heederik, D., Portengen, L., Shaham, J., Ferro, G., de Vocht,
F., Burstyn, I., Boffetta, P., A case-control study of lung cancer nested in a
cohort of European asphalt workers, Environ. Health Persp. 118 (2010) 1418-
1424.
Fuhst, R., Creutzenbeg, O., Ernst, H., Hansen, T., Pohlmann, G., Preiss, A. and
Rittinghausen, S., 24 Months Inhalation Carcinogenicity Study of Bitumen
Fumes in Wistar (WU) Rats, J. Occup. Environ. Hyg. 4(1) (2007) 20-43.
Kriech, A.J., Osborn, L.V., Wissel, H.L., Redman, A.P., Smith, LA. and Dobbs,
T.E., Generation of Bitumen Fumes Using Two Fume Generation Protocols
and Comparison to Worker Industrial Hygiene Exposures, J. Occup. Environ.
Hyg. 4(1) (2007) 6-19.
Clark, C.R., Burnett, D.M., Parker, C.M., Arp, E.W., Swanson, M.S., Minsavage,
G.D., Kriech, A.J., Osborn, L.V., Freeman, J.J., Barter, R.A., Newton, P.E., and
Stewart, C.W., Asphalt Fume Dermal Carcinogenicity Potential: I. Dermal
Carcinogenicity Evaluation of Asphalt (Bitumen) Fume Condensates, Regul.
Toxicol. Pharmicol. 61(1) (2011) 9-16.

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Table of Contents

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
TECHNICAL SESSIONS
Development of a Failure Criterion for Asphalt Mixtures under
Fatigue Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
JUN ZHANG, MOHAMMADREZA SABOURI, MURTHY N. GUDDATI
and Y. RICHARD KIM
Effect of Particle Mobility on Aggregate Structure Formation
in Asphalt Mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
NIMA ROOHI SEFIDMAZGI, POUYA TEYMOURPOUR
and HUSSAIN U. BAHIA
Hot Mix Asphalt Pavement Frictional Resistance as a Function
of Aggregate Physical Properties . . . . . . . . . . . . . . . . . . . . 59
JOHN E. HADDOCK and JOAN P. O'BRIEN
IlliTCLow Temperature Cracking Model for Asphalt
Pavements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
ESHAN V. DAVE, WILLIAM G. BUTTLAR, SOFIE E. LEON, BEHZAD BEHNIA
and GLAUCIO H. PAULINO
Effect of Short-Term Conditioning and Long-Term Aging on
Permanent Deformation Characteristics of Asphalt Mixtures . . . 127
HALEH AZARI and ALAEDDIN MOHSENI
Evaluating Photocatalytic Asphalt Pavement Effectiveness in
Real World Environments through Developing Models:
A Statistical and Kinetic Study . . . . . . . . . . . . . . . . . . . . . 151
HEATHER DYLLA, SOMAYEH A SADI, MARWA HA SSAN
and LOUAY N. MOHAMMAD

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Laboratory Conditioning Protocols for Warm-Mix Asphalt . . . . 177
FAN YIN, LORENA GARCIA CUCALON, AMY EPPS MARTIN,
EDITH ARAMBULA, ARIF CHOWDHURY and EUN SUG PARK
Merits of RAP Dominated Warm Mixed Flexible Pavement
Base Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
ISAAC L. HOWARD, JESSE D. DOYLE and BEN C. COX
Evaluation of High RAP-WMA Asphalt Rubber Mixtures . . . . . 253
WALAA S. MOGAWER, ALEXANDER J. AUSTERMAN, LOUAY MOHAMMAD
and M. EMIN KUTAY
Rutting and Moisture Damage Resistance of High RAP Warm
Mixed Asphalt: Loaded Wheel Tracking vs. Conventional
Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
JESSE D. DOYLE and ISAAC L. HOWARD
Effect of Long-Term Aging on RAP Mixtures: Laboratory
Evaluation of Plant Produced Mixtures . . . . . . . . . . . . . . . . 327
JO SIA S DANIEL, NELSON GIBSON, SEAN TARBOX, AUDREY COPELAND
and ADRIAN ANDRIESCU
Evaluation of Use of Recycled Asphalt Shingles in HMA . . . . . . 367
FUJIE ZHOU, HONGSHENG LI, SHENG HU, ROBERT LEE, TOM SCULLION,
GERMAN CLAROS, JON EPPS and JOE BUTTON
Evaluating the Effect of Rejuvenators on the Degree of Blending
and Performance of High RAP, RAS, RAP/RAS Mixtures . . . . . 403
WALAA S. MOGAWER, ABBA S BOOSHEHRIAN, SIAVA SH VAHIDI
and ALEXANDER AUSTERMAN
Effect of Rubber Characteristics on Asphalt Binder
Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
J. RICHARD WILLIS, PAMELA TURNER, CLAYTON PLEMMONS,
CAROLINA RODEZNO, TOM ROSENMAYER, CODRIN DARANGA
and DOUG CARLSON
Understanding Mechanisms Leading to Asphalt Binder Fatigue
in the Dynamic Shear Rheometer (DSR) . . . . . . . . . . . . . . . 465
C A S S I E H I N T Z a n d H US S A I N B A H I A
Fatigue Endurance Limit for HMA Based on Healing . . . . . . . . 503
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

MENA I. SOULIMAN, WALEED ZEIADA, MICHAEL MAMLOUK


and KAMIL KALOUSH
Investigation of Fracture Properties of California Asphalt
Mixtures Using Semi-circular Bending and Beam
Fatigue Tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
HAMED HAKIMELAHI, SHADI SAADEH and JOHN HARVEY

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development and Validation of a Rutting Model for Asphalt
Mixtures Based on the Flow Number Test . . . . . . . . . . . . . . 557
MARIA C. RODEZNO and KAMIL E. KALOUSH

Using Small Scale Specimens for AMPT Dynamic Modulus and


Fatigue Tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 579
XINJUN LI and NELSON GIBSON

A Mechanistic Permanent Deformation Model for Asphalt


Concrete in Compression . . . . . . . . . . . . . . . . . . . . . . . . 617
YEONG-TAE CHOI and Y. RICHARD KIM

SYMPOSIUMSUSTAINABLE ASPHALT TECHNOLOGIES


An Industry Perspective on Sustainable Asphalt
Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
HO W A R D M A R K S

Alternative Binders for the Production of Bituminous


Materials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
E L HA M F I N I

Recycling with RAP, RAS and Secondary Aggregates . . . . . . . . 661


J O S IA S D AN I EL

Sustainable Energy Use in Asphalt Production . . . . . . . . . . . . 683


AD A M HAN D

Binder Additives For Warm Mix Asphalt Technology . . . . . . . . 685


GAYLON L. BAUMGARDNER and GERALD R. REINKE

Warm Mix Asphalt TechnologyImplementation from a


State DOT Perspective . . . . . . . . . . . . . . . . . . . . . . . . . 711
B IL L S C H I E B E L

INTERNATIONAL FORUMSPECIFICATIONS AND USE OF


BITUMINOUS MATERIALS IN EUROPE
Specifications and Use of Bituminous Materials in Europe . . . . . 713 --`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

MIKE SOUTHERN, JEAN-LUC DELORME, JEAN-PA SCAL PLANCHE


and JAN VAN DER ZWAN

List of Award Winners . . . . . . . . . . . . . . . . . . . . . . . . . . 721


List of Officers and Life Members . . . . . . . . . . . . . . . . . . . . 730
Index of Contributors and Discussors . . . . . . . . . . . . . . . . . . 731

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
CD TABLE OF CONTENTS
In addition to the papers listed above, the following items are also
included in the CD
AAPT List of Members
Presentations of Papers Presented at 2013 Annual Meeting
2013 Annual Meeting Pictures

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion
for Asphalt Mixtures under Fatigue Loading
Jun Zhang, Mohammadreza Sabouri, Murthy N. Guddati,
and Y. Richard Kim*

Department of Civil, Construction, & Environmental Engineering, North Carolina State


University, Raleigh, NC 27695-7908

ABSTRACT: The failure criterion defines the applicable region associated with the continuum
damage model and is important in characterizing the service life of asphalt mixtures. A proper
failure criterion should consistently predict the failure of a material that reaches macro-fracture.
A previously developed criterion that uses the viscoelastic continuum damage (VECD) model
exhibits high variability and is considered to be inefficient because it requires calibration tests at
different temperatures. In this paper, a new concept that involves released pseudo strain energy is
introduced. This released pseudo strain energy concept focuses on the dissipated energy that is
related to stiffness reduction only and is fully compatible with and predictable using the VECD
model. A characteristic relationship is found between the stable rate of pseudo energy release
during testing and the final fatigue life of the same mixture, independent of strain amplitude and
temperature. Based on these observations, a new failure criterion is proposed. The proposed
failure criterion combines the advantages of the VECD model and this characteristic relationship,
which both originate from fundamental mixture properties, and is able to predict the fatigue life of
asphalt concrete mixtures across different temperatures and strain amplitudes.

KEYWORDS: Failure criterion, fatigue, asphalt, pseudo energy, viscoelastic, damage

1. Introduction
Fatigue cracking due to repeated traffic loading is one of the most significant distresses
found in asphalt concrete pavements. Depending on the dominant mechanism, the
development and propagation of fatigue cracks inside a pavement structure can be

The oral presentation was made by Professor Kim.

This is a reproduction of an article whose final and definitive form has been published in Road
Materials and Pavement Design 2013 Taylor & Francis. The article is available online at:
http://dx.doi.org/ 10.1080/14680629.2013.812843
1
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

categorized into two major phases: pre-localization and post-localization. In general, pre-
localization is manifested by the initiation and propagation of micro-cracks, and post-
failure represents the formation and propagation of macro-cracks.
The viscoelastic continuum damage (VECD) model is a continuum damage
mechanics-based model that has been applied successfully in simulating the behavior of
asphalt concrete mixtures during the pre-localization stages under both monotonic and
cyclic loading (Kim and Little, 1990; Lee and Kim, 1998; Daniel and Kim, 2002;
Underwood et al., 2006; Underwood et al., 2010). However, after reaching the fracture
region, the VECD model begins to lose the ability to characterize the material behavior
because of the formation of macro-cracks. The failure criterion defines the regions where
the VECD model is applicable and indicates the occurrence of fracture. In practice, the
failure criterion directly determines the fatigue life of the mixture, which is an essential
parameter in evaluating the mixtures ability to resist fatigue damage. Hence, it is crucial
to develop a failure criterion that can capture the failure of asphalt concrete accurately.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
The definition of fatigue failure of asphalt concrete in laboratory tests has always
been controversial, especially in the controlled strain cyclic loading mode when no
catastrophic failure or fracture is observed. Traditional fatigue analysis defines failure as
the point at which the materials modulus value reduces to 50% of its initial value, and
the corresponding number of cycles is denoted as Nf50. This failure criterion is
considered to be convenient for the implementation of continuum damage models but
may not provide a consistent prediction for the damage state due to its arbitrary
assumption (i.e., the 50% reduction in modulus value). Figure 1 (a) presents the point of
Nf50 on the normalized stiffness diagram of an asphalt mixture under fatigue testing, and
it is obvious that the fatigue failure defined by this approach for this mixture is still far
from the real damage capacity of the material.
As an alternative, Reese suggested a new approach by defining failure as the upper
limit of the phase angle (Reese, 1997). During cyclic loading, the measured phase angle
of asphalt concrete usually exhibits a stable increase followed by a sharp decrease. The
number of cycles at failure, Nf, is defined as the cycle when this sharp decrease occurs
(Figure 2). This approach is generally believed to have more theoretical support than the
traditional approach, because the definition of failure is tracked through the materials
viscoelastic behavior, and the huge reverse in phase angle must represent a
transformation in the dominant mechanism inside the material, which is most probably
macro-fracture. Reeses approach also proves to be consistent on the stiffness diagram.
As shown in Figure 1 (b), for all the fatigue tests the failure point as determined by the
phase angle generally is located at the transition point between the stable decrease and
accelerated decrease in the stiffness diagram. Hence, it is concluded that the drop in
phase angle is the better indicator of fracture (compared to the traditional approach), and

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

that the phase angle should be taken as the target failure point that the developed
criterion needs to predict.

1.0 1.0
VTe30LC VTe30LC

0.8 Nf50 Nf
0.8

0.6 0.6
E*/E50

E*/E50
0.4 0.4

0.2 0.2
(a) (b )
0.0
0.0
0.E+00 5.E+04 1.E+05 2.E+05 2.E+05
0.E+00 5.E+04 1.E+05 2.E+05 2.E+05
Number of Cycles Number of Cycles
Figure 1. Determined failure point by different approaches:(a) traditional fatigue
analysis, and (b) phase angle approach.

55
Phase angle(degree)

50

45

40

35

Nf
30
0 200 400 600 800 1000 1200

Number of Cycles

Figure 2. Definition of failure by Reese's approach.


However, the change in time dependency is not included in most continuum damage
models, including the VECD model, because: first, the change in phase angle for asphalt
mixtures usually is observed not to be that significant compared to asphalt binders;
second, it is the degradation of stiffness that is of more interest for engineering purposes;
and finally, the exact mechanism that causes the variation in phase angle, whether it is
nonlinear viscoelasticity or damage due to fatigue cracking or plasticity, remains
unclear. In this way, the VECD model is not able to sense the variation in phase angle or
to recognize its drop. That is, the model itself cannot predict failure automatically, and
so the development of another failure criterion is necessary.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

The development of the failure criterion starts with the simple assumption that the
failure of the material occurs at a critical damage state or at an equivalent critical
stiffness value. Hou tried to develop unified failure envelopes through experimental
observations of twelve different mixtures (Hou et al., 2010). The pseudo stiffness at
failure, which is the pseudo stiffness at the point of the phase angle drop, is expressed as
a function of reduced frequency, and the coefficients of the function are related to the
nominal maximum aggregate size (NMAS) of the mixture and whether or not reclaimed
asphalt pavement (RAP) is included. Even though the basic idea behind this approach
seems to be logically reasonable, the high variability in the experimental data makes it
very difficult to develop a convincing envelope among the different mixtures. In reality,
due to the complexity and heterogeneity of asphalt concrete, it is quite possible that a
slight alteration in the composition of the material, such as the asphalt content or
aggregate properties, may induce a substantial change in the distribution of the failure
stiffness. Hence, investigating failure stiffness may not be the best option for developing
a failure criterion.
Another type of research called the dissipated energy approach also is used in the
study of failure criteria for asphalt concrete. The dissipated energy approach tries to
relate the energy associated with damage to the final fatigue life and is said to be able to
reduce the variability commonly observed in fatigue testing (Castelo Branco et al.,
2008). Currently, most of the dissipated energy approaches focus on the quantification of
the dissipated energy via the hysteresis loop in the domain of stress-strain/stress-pseudo
strain. Ghuzlan and Carpenter (2000) developed the concept of ratio in dissipated energy

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
change (RDEC) and state that it is the change in dissipated energy between consecutive
cycles that represents damage. A unique relationship between the plateau value (PV) in
the RDEC and the final fatigue life also is proposed (Shen and Carpenter, 2005). The so-
called dissipated pseudo strain energy (DPSE) is introduced by uncoupling the effect of
viscoelasticity first via the correspondence principle (Schapery, 1984). Kim et al. (2003)
employed the concept of DPSE in the characterization of micro-crack growth. Masad et
al. (2008) applied the rate of change in the DPSE to the J-integral in fracture mechanics
and assessed the fatigue cracking potential of different asphalt mixtures.
Although these approaches have all shown positive results to some extent, one
significant obstacle remains that prevents any of them from being incorporated into the
VECD model as the failure criterion. Up to now, these dissipated energy approaches
have been investigated through experiments only, in which the histories of stress, strain
and phase angle are all known in advance. However, as mentioned earlier, the variation
in phase angle cannot be captured or predicted using the VECD model. Hence, the
dissipated energy inside the hysteresis loop cannot actually be calculated from the
model, and thus, any criterion based on it cannot be implemented.

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

Given the above considerations, the objective of this paper is to develop a failure
criterion that can be applied using the VECD model and can predict the fatigue failure
that is equivalent to the drop in phase angle in experimental observations with both
consistency and accuracy.

2. Materials and Test Methods

In this study, experiments were performed on four different types of mixtures to


investigate possible failure criteria. The first mixture is a Superpave surface 9.5 mm mix
(S9.5C) that is commonly used in North Carolina, and the other three are 9.5 mm
Superpave mixes from Vermont that are part of the New England RAP project. The
Vermont mixes are designated as VTeXXLC, with XX representing the percentage of
RAP. Table 1 presents a summary of the properties of these four mixtures.
All the specimens were compacted using the Superpave gyratory compactor (SGC)
with a diameter of 150 mm and to a height of 178 mm. To obtain specimens of uniform
air void distribution, these samples were cored and cut to a height of 150 mm and
diameter of 100 mm for testing. Prior to testing, the air void content was measured using
the CoreLok method for each specimen. All the test specimens used in this study have an
air void ratio within the range of 5.5 0.5% for the S9.5C mixture and 6.0 0.5% for
the VTeXXLC mixtures. To minimize the aging effect, specimens were sealed in plastic
bags and stored carefully in a cabinet if they were not tested immediately after
fabrication. No specimens were tested later than two weeks after they were cored and
cut.
Table 1. Summary of study mixture properties.
Target
NM AS
Mix Type Binder AC (%) RAP (%) Air Void
(mm)
(%)
S9.5C PG70-22 5 .2 0 9 .5 0 5 .5
VTe00LC PG64-28 6 .5 0 9 .5 0 6 .0
VTe30LC PG64-28 6 .6 1 9 .5 30 6 .0
VTe40LC PG64-28 6 .5 5 9 .5 40 6 .0
Two main tests were carried out in this study: (1) the dynamic modulus test was
performed to determine the linear viscoelastic characteristics using both AMPT and
MTS 810 closed loop servo hydraulic machines, and (2) the CX cyclic direct tension test
was implemented to describe the viscoelastic damage characteristics using both the
AMPT and MTS 810 machines.

5
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

Dynamic modulus testing on the S9.5C mix was performed using the MTS 810
machine in load-controlled mode in axial tension-compression, following the protocol
given in AASHTO T342-11. Tests were completed for all mixtures at -10C, 5C, 20C,
40C, and 54C and at frequencies of 25, 10, 5, 1, 0.5, and 0.1 Hz. The VTeXXLC mix
dynamic modulus testing was performed using the AMPT in load-controlled mode in
axial compression following the protocol given in AASHTO TP 79-11. Tests were
completed for all mixtures at 5C, 20C, 40C, and 54C and at the same frequencies as
for the MTS. For tests using both machines, the testing order is from low to high
temperatures and from high to low frequencies in order to minimize damage to the
specimens. Load levels were determined by a trial and error process so that the resulting
strain amplitudes were between 50 and 70 microstrain to prevent damage to the
specimens.
All direct tension test specimens were glued to metal plates at both ends using epoxy
before they were placed in the machine for testing. Data acquisition programs were
prepared using LabView software for data collection and analysis. Vertical deformations
were measured in the middle of each specimen using four linear variable differential
transducers (LVDTs) at intervals of 90 degrees with a gauge length of 70 mm for the
AMPT specimens and 100 mm for the MTS specimens. DEVCON steel putty was
used to glue the steel end plates and targets for the LVDTs that were used for testing the
specimens.
The viscoelastic damage characteristics were determined by conducting CX cyclic
direct tension tests at three different temperatures and multiple strain levels for each
temperature with the frequency of 10 Hz. Fingerprint dynamic modulus tests were
conducted to check the variability of the test specimens before running the CX cyclic
direct tension tests. Checking the specimen variability is important so that the linear
viscoelastic properties obtained from the dynamic modulus tests can be used properly in
the VECD analysis. The CX cyclic direct tension test allows the specimen to run to
complete failure while still effectively limiting the viscoplasticity. Hence, viscoelastic
damage is the dominant mechanism. Details of the experimental procedure can be found
in Underwood et al. (2010) and Hou et al. (2010).

3. Development of Failure Criterion

The development of the failure criterion in this study starts with mix S9.5C. Because the
investigation focuses on viscoelastic damage, the signals of stress and strain are focused
mainly on magnitude.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

3.1 Preliminaries

For asphalt concrete under controlled CX cyclic loading, the stress and strain
histories can be described using the sinusoidal functions given in Equations 1 and 2.
i sin (t )
= [1 ]

= i sin (t + i ) [2 ]

where i , i , and i are the stress amplitude, strain amplitude, and phase angle measured
at cycle i, respectively.
If the stress and strain values are plotted against each other, the formed hysteresis
loop represents the dissipated energy, which is the total energy consumed during that
cycle. However, this consumed energy does not correspond to damage exclusively; it
also includes the energy that is associated with viscoelasticity. According to the
correspondence principle, the effect of viscoelasticity can be eliminated by replacing the
physical strain with the equivalent pseudo strain. For the physical strain given in
Equation 1, the corresponding pseudo strain R is written as:

= R i E * i si n ( t + ) [3 ]

where Ri is the pseudo strain amplitude at cycle i ; E is the undamaged complex


modulus for the given reduced frequency; and is the phase angle that is related to
viscoelasticity only.
The newly formed hysteresis loop in the stress-pseudo strain space is the DPSE
(WiR . Its value is equal to the area of the enclosed ellipse and can be calculated using
Equation 4.

=Wi R i iR sin (i ) [4 ]
During the controlled CX cyclic tests, the stiffness and phase angle both change with
the cycles, and so the DPSE will also change shape simultaneously. As the cycles
continue, the incline of the loop will gradually lower due to the reduction in stiffness,
and the enclosed area of the loop also will increase in size because of the increase in the
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

phase angle. Figure 3 shows the typical trend of the pseudo hysteresis loop during
controlled CX cyclic testing.

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

CX cyclic test

Increase in
phase angle

Stiffness
reduction

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Figure 3. Pseudo hysteresis loops for controlled CX cyclic tests.
3.2 Rate of Pseudo Strain Energy Release
For Equation 4, the calculation of the DPSE requires information about both the stiffness
and phase angle. In reality, the reduction in stiffness is known to be induced by damage.
However, with respect to the increase in phase angle, details regarding the underlying
mechanism remain unclear.
There are theories that attribute the increase of phase angle to nonlinear
viscoelasticity, damage or a combination of the two (Masad et al., 2008; Si et al., 2002).
Therefore, instead of calculating the DPSE, which involves a parameter that is both
uncertain and unpredictable, the dissipated pseudo strain energy is evaluated only as it
relates to the reduction in stiffness.
During cyclic loading, the maximum stored pseudo strain energy at each cycle
appears at the point of peak stress, which is also the point that corresponds to the
maximum pseudo strain and maximum experienced damage at that cycle. Hence, the
maximum pseudo strain energy reflects the materials current damage state and also its
ability to store energy. Based on the definition of pseudo stiffness for the VECD model
(Figure 4), the maximum stored pseudo strain energy at cycle can be calculated as:
1
(WmRax =
)i ( 0,ta )i ( R 0,ta )i [5 ]
2

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

where 0, and 0, are the tension amplitudes of the stress and pseudo strain. The
relationship between these two amplitudes is linked through the magnitude-based pseudo
stiffness, which is:

0,= F * R 0,ta [6 ]
ta

Hence, the maximum stored pseudo strain energy at cycle can be rewritten as:
1 [7 ]
(WmRax =
)i ( F )i ( R 0,ta )i 2
2
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Figure 4. Schematic view of pseudo stiffness in the VECD model.

The maximum stored pseudo strain energy at each cycle reflects the ability of the
material to store energy at that particular time. As the damage accumulates, the material
loses the stored energy for the same magnitude of applied pseudo strain due to the
reduction in pseudo stiffness. The difference between the current stored maximu m
pseudo strain energy and the corresponding undamaged state is referred to as the total
released pseudo strain energy, which represents the cumulative loss of pseudo energy
due to the damage process, and is denoted as WCR .

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

Because the phase angle is excluded from the analysis, the elliptical hysteresis loop
in the stress-pseudo strain space can be simplified into a straight line that represents the
pseudo stiffness only. The concept of total released pseudo strain energy can be
interpreted in a two-step process. During the first step, it is assumed that all the damage
is temporarily frozen and the material is loaded to the applied maximum pseudo strain.
However, because of damage, the material cannot actually hold that amount of energy.
So, during the second step, the frozen damage is released and some of the stored energy
is released. In the stress-pseudo strain plot, the material will return to the point that
corresponds to the current damage state. The released energy during this second step is
the total released pseudo strain energy, which is presented by the shaded triangular area

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
in Figure 5. Its formulation is also given in Equation 8.
1 R
( 0,ta ) (1 Fi )
2
(WCR )i
= [8 ]
2 i

No damage = 1

With damage

Figure 5. Schematic representation of total released pseudo strain energy,WCR .

Equation 8 indicates that the total released pseudo strain energy is affected by two
factors: one is the applied pseudo strain amplitude, R0,ta , and the other is the reduction in
the material pseudo stiffness. In this way, the same amount of total released pseudo
strain energy can be obtained at either a high applied pseudo strain amplitude or with a
relatively large decrease in stiffness. From this point of view, the total released pseudo
strain energy, WCR , is a comprehensive energy measure that quantifies the dissipated
energy using both the external loading and the material itself.

10

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

Figure 6 shows the history of the total released pseudo strain energy, WCR ,
calculated from experiments using Equation 8. Because WCR measures the dissipated
energy in a cumulative sense, the rate of change in WCR with respect to the number of
cycles also is evaluated in the same plot. A local window of a fixed size, e.g., 50 cycles,
is chosen and moves along with the loading history. The rate of WCR is calculated at each
window by considering the history of WCR inside as an approximate linear function and
the value of the rate equal to the fitted slope of the linear function.

Figure 6. History of and its corresponding rate for controlled CX cyclic tests.

It is interesting to observe that for all the tests, the evolution of is generally
categorized into three regions. During the initial cycles, the rate of diminishes in a
relatively quick fashion, which usually relates to the transition state of on-specimen
stress/strain for CX cyclic test. When the specimen is loaded close to failure, the rate of
also jumps and it suggests a start of unstable damage propagation. If those two
severe regions are neglected, there is a significant stable region for the rate of ,
during which the rate of release of the pseudo strain energy, or released pseudo strain
energy per cycle, is almost constant. To be more precise, the stable region is defined as
the region where the relative variation in rate of per cycle stabilizes within certain
limits, e.g. 10%, between adjacent windows. Given that this stable region occupies the
majority of the loading history, its corresponding stable rate of pseudo strain energy
release characterizes the overall rate of energy loss during this fatigue test and is referred
to as ( . Because the damage propagates very slowly during cyclic loading, each state
of the material is taken as quasi-static, and the rate of energy release should be equal to
the rate of damage accumulation. Given that the total released pseudo strain energy is
calculated based on stiffness reduction, which is related definitively to damage, the

11
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

stable rate of the pseudo strain energy release ( can be regarded as the steady rate of
damage accumulation as well.

3.3 Characteristic Relationship

Because the stable rate of pseudo strain energy release (GR characterizes the overall rate
of damage accumulation during fatigue testing, it is reasonable to hypothesize that a
correlation must exist between GR and the final fatigue life, Nf (number of cycles to
failure), because the faster the damage accumulates, the quicker the material should fail.
Figure 7 presents the relationship between GR and the fatigue life Nf for mix S9.5C. It is
found that except for the case of quick failure with fatigue life less than 1000 cycles (one
at 5C and one at 27C), the GR and the fatigue life Nf are highly correlated, and
surprisingly, this correlation is not sensitive to temperature, which seems to be a
characteristic function of the given mixture.

1.0E+05
S9.5C-5C
S9.5C-19C
1.0E+04
S9.5C-27C

1.0E+03
GR

1.0E+02

1.0E+01

1.0E+00

1.0E-01

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
500 5000 50000 500000
Nf

Figure 7. Relationship between GR and Nf for mix S9.5C.

The concept behind the characteristic relationship is actually quite similar to that of
the damage characteristic proposed in the VECD model. Instead of stating that a unique
relationship exists between damage and stiffness for all viscoelastic damage dominating
cases, the theory states that a unique relationship exists between the rate of damage
accumulation and fatigue life, which means that no matter what the loading condition is,
once the rate of damage accumulation is determined, the corresponding fatigue life for
the given mixture is determined as well.

12

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

If the characteristic relationship between GR and Nf is proven to be present


consistently, then it can be used as the criterion for the VECD model to predict the
fatigue life for asphalt mixtures. Because the calculation of WCR relies only on the
variation in pseudo stiffness, its value can be evaluated from the VECD model directly.
Moreover, the corresponding stable energy release rate, GR, can be obtained as well. For
consistency, the same methodology is applied for the determination of GR, which
evaluates the rate of WCR in a moving local window first. The value of GR is picked up
when the fluctuations between adjacent windows are lower than a predefined threshold.
Once the value of GR is obtained, the fatigue life of the asphalt concrete can be predicted
immediately through the characteristic relationship. Figure 8 compares the results
between the predicted GR obtained from the VECD model and the measured GR obtained
from experiments. It is found that the two values are generally quite close to each other.
In reality, because GR is a rate-based value, it is sometimes predicted with even less error
compared to the point-to-point prediction of stiffness. The consistency between the
experimental measurements and the model prediction for GR ensures the linearity of the
proposed characteristic relationship in the model application and also confirms the
applicability of this characteristic relationship for the development of a failure criterion.

4000 100000
S9.5C-5C S9.5C-5C
Predicted Stable rate GR
Predicted Stable rate GR

S9.5C-19C S9.5C-19C
3000 S9.5C-27C S9.5C-27C
LOE LOE

2000 100

1000

(a) (b )
0 0.1
0 1000 2000 3000 4000 0.1 100 100000
Measured Stable rate GR Measured Stable rate GR

Figure 8. Comparison of GR between measured and predicted results: (a) arithmetic


scale, and (b) log scale.

4. Application of the Proposed Failure Criterion

4.1 Parameter Determination for the Characteristic Relationship

Because the characteristic relationship between GR and Nf relates to the fundamental


--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

properties of the material in terms of damage tolerance, and is independent of

13

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

temperature, the development of this relationship can be simplified by conducting tests


at only one appropriate temperature. This temperature should be low enough to minimize
the effect of viscoplastic strain during the cyclic testing. The following rule of thumb is
used to determine the test temperature based on the performance grade (PG) of the base
binder:
T ( C ) 0.5 (High temperature binder PG grade - Low temperature binder PG grade) - 2 [9 ]

For a given mixture, the values of GR can be calculated using the VECD model as
follows:

(1) For a given pseudo strain history, predict the evolution of pseudo stiffness F
using the VECD model;
(2) Incorporate the value of F and pseudo strain tension amplitude R0,ta at each
cycle into Equation 8, and calculate the history of total released pseudo strain
energy WCR ;
(3) Identify the stable linear region during WCR and evaluate the stable rate of
pseudo strain energy release GR.

The fatigue life is obtained through experimental observations of the drop in phase
angle (based on Equation 9 where 13C is chosen as the test temperature for the
VTeXXLC mixtures with PG 64-22 binder, and 19C is chosen for the S9.5C mixture
with PG 70-22 binder). Figure 9 through 11 present the characteristic relationship
developed for mixtures VTe00LC, VTe30LC and VTe40LC, respectively. According to
these graphs, a significant relationship is found between GR and Nf for all three mixtures.
From the fits shown in Figure 9 through Figure 11, a power-law relationship is derived
between GR and Nf, which is later used to predict the fatigue life, as presented in Section
4 .2 .

14
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

10000
VTe00LC-13C

1000

100
R
G

10

R2 = 0.95
1
1.E+03 1.E+04 1.E+05 1.E+06
Nf (Cycle)

Figure 9. Relationship between GR and Nf for VTe00LC mix at 13C.

10000
VTe30LC-13C

1000

100
GR

10

R2 = 1.00
1
1.E+03 1.E+04 1.E+05 1.E+06
Nf (Cycle)
Figure 10. Relationship between GR and Nf for VTe30LC mix at 13C.

15
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

10000
VTe40LC-13C

1000
R

100
G
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

10

R2 = 0.99
1
1.E+03 1.E+04 1.E+05 1.E+06

Nf (Cycle)

Figure 11. Relationship between GR and Nf for VTe40LC mix at 13C.

By plotting the relationships for all three VTeXXLC mixtures and the S9.5C mixtures
together, as shown in Figure 12, it is interesting to observe that the characteristic
relationships for the three VTe mixtures do not seem to have any significant difference.
This finding leads to the hypothesis that the characteristic relationship is not sensitive to
the percentage of RAP content. However, a difference does exist between the S9.5C
mixture and the other three mixtures, which may indicate that the relationship is affected
by other factors, such as binder characteristics, asphalt content or aggregate properties. It
should be noted that the fact that the characteristic relationship is not affected by the
amount of RAP does not necessarily mean that during a similar loading history the
fatigue life measurements of the mixtures with different percentages of RAP are the
same. For this phenomenon, the material behavior should be simulated in a pavement
system that may result in different stable rates of the pseudo strain energy release (GR)
for different mixtures with different percentages of RAP and, therefore, result in
different fatigue life measurements by using the same characteristic relationship.

16

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

10000
VTeXXLC-13C
1000 S9.5C-19C
R2 = 0.98

100
R
G

10 R2 = 0.99

0
1.E+03 1.E+04 1.E+05 1.E+06

Nf (Cycle)
R
Figure 12. Relationship between G and Nf for VTeXXLC mixes at 13C and S9.5C at
19C.

4.2 Fatigue Life Prediction

Once the characteristic relationships are obtained, they can be applied to predict the
fatigue life of asphalt mixtures for any other loading conditions. As mentioned earlier,
the characteristic relationship between the stable rate of pseudo energy release GR and
fatigue life Nf is not limited to only one temperature, as long as viscoelastic damage is
the dominant mechanism. Once the VECD parameters and the applied pseudo strain
history are known, the corresponding fatigue life can be readily determined using the
characteristic relationship between GR and Nf, as shown in Figure 9 to Figure 11.
Once the characteristic relationships are obtained, they can be applied to predict the
fatigue life of asphalt mixtures for any other loading conditions. As mentioned earlier,
the characteristic relationship between the stable rate of pseudo energy release GR and
fatigue life Nf is not limited to only one temperature, as long as viscoelastic damage is
the dominant mechanism. Once the VECD parameters and the applied pseudo strain
history are known, the corresponding fatigue life can be readily determined using the
characteristic relationship between GR and Nf, as shown in Figure 9 to Figure 11.
The advantage of using the characteristic relationship is that it can significantly
reduce the number of tests required for the calibration of the failure criterion. In order to

17
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

obtain the parameters of the previously developed criterion that is based on failure
stiffness, fatigue tests had to be conducted at multiple temperatures. With the new
characteristic relationship proposed here, calibration tests are required at only one
temperature, because the relationship among the different temperatures has been proven
to be statistically unique.
In order to verify the applicability and accuracy of the proposed failure criterion,
fatigue test predictions were performed for the three VTeXXLC mixtures by using the
derived characteristic relationship at 13C to predict the fatigue life for the other
conditions.
Table 2 presents the overall prediction results. According to this table, the average
prediction error is only around 20%, which is well within the variability of fatigue life,
whereas the percentage errors in only two cases exceed 50 percent.

400000 1000000
S9.5C S9.5C
VTe00LC VTe00LC
VTe30LC VTe30LC
300000
VTe40LC VTe40LC
Predicted Nf

Predicted Nf

LOE 100000
LOE

200000

10000
100000

(a) (b )
0 1000
0 100000 200000 300000 400000 1000 10000 100000 1000000
Measured Nf Measured Nf

Figure 13. Comparison of measured and predicted fatigue life:(a) arithmetic scale, and
(b) log scale.

Figure 13 presents the final predicted results for all four mixtures. The figure shows
that the proposed failure criterion is able to provide a reasonable prediction across all
temperatures. Hence, the proposed failure criterion is considered to be both efficient and
accurate.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

18

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

Table 2. Fatigue Life Prediction for VTeXXLC Mixtures at Multiple Conditions.

Specimen
Mix Type Temperature Initial Straina () Nf, Experiment GR Nf, Predicted Error %
Name
1054 343 4097 485 5668 38
7C 1041 260 26343 46 30195 15
1050 224 99482 12 76952 -23
1037 218 174026 3 210218 21
1043 379 14122 253 8981 -36
13C
1051 277 59610 21 52593 -12
VTe00LC 1041 409 5701 291 8147 43
1046 282 127821 4 182418 43
1044 301 107400 6 125981 17
1015 517 13114 167 12085 -8
20C
1045 465 15992 144 13399 -16
1013 678 6065 744 4187 -31
1009 674 6035 842 3833 -36
1026 222 65609 17 70634 8
1068 312 12518 122 16012 28
7C
1052 348 3697 520 5385 46
1070 266 21534 47 32891 53
1102 366 5507 555 5130 -7
1029 436 1898 1835 2087 10
13C
1044 244 33124 48 32478 -2
VTe30LC 1059 202 110263 8 119182 8
1027 244 43159 33 42562 -1
1043 876 1098 4767 1018 -7
1130 1032 897 7872 698 -22
20C 1046 700 3697 1310 2689 -27
1090 428 15713 144 14164 -10
1077 221 164122 4 210872 28
1028 213 230093 2 360499 57
1019 325 10304 179 12491 21
7C
1011 244 83362 10 88401 6
1025 218 130961 6 133333 2
1060 323 14921 180 12443 -17
13C
1033 330 16929 101 18394 9
1003 499 3498 1089 3652 4
VTe40LC
1038 310 83268 8 106295 28
1032 300 105617 7 113921 8
1007 541 12112 225 10687 -12
20C
1031 541 12921 237 10313 -20
1009 716 4298 1248 3331 -22
1010 726 6894 641 5242 -24
a
On-specimen strain at the 50th loading cycle

19
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

5. Summary and Conclusions

In this study, a new energy-based failure criterion is developed for the prediction of
fatigue life using the VECD model. The (power-law) characteristic relationship that
relates the stable rate of pseudo strain energy release GR and fatigue life Nf is found to be
a fundamental property of the mixture. In contrast to the existing failure criteria that
often require characterization tests at multiple temperatures, the proposed failure
criterion requires characterization tests at only a single temperature, thus significantly
reducing the costs associated with testing. Using this failure criterion and the S-VECD
model, the fatigue life of asphalt concrete at different temperatures and strain amplitudes
can be predicted from dynamic modulus tests and CX cyclic direct tension tests at four
strain amplitudes. The effectiveness of the proposed failure criterion is illustrated by
comparing predicted and actual fatigue life data for the four different mixes.
The proposed model is focused on predicting fatigue life under constant repeated
loading conditions at different temperatures. However, it may be possible to build on the
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

proposed ideas to develop a complete VECD-based model for predicting fatigue life
under complicated stress histories that are experienced in the field. Development of such
a model is the focus of on-going research.

6. Acknowledgements

This research was funded by Texas A&M University as part of the Asphalt Research
Consortium project and by the University of New Hampshire as part of the New England
RAP project. The authors gratefully acknowledge their support.

7. References

Castelo Branco, V. T. F., E. Masad, A. Bhasin, and D. N. Little. (2008). Fatigue


Analysis of Asphalt Mixtures Independent of Mode of Loading. Proceedings, 87th
Annual Transportation Research Board Meeting, Transportation Research Board of
the National Academics, Washington, D.C.
Daniel, J. S. and Y. R. Kim. (2002). Development of a Simplified Fatigue Test and
Analysis Procedure Using a Viscoelastic Continuum Damage Model. Journal of
the Association of Asphalt Paving Technologists, Vol. 71, pp. 619650.

20

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Development of a Failure Criterion for Asphalt Mixtures under Fatigue Loading

Ghuzlan, K. A. and S. H. Carpenter. (2000). Energy-Derived, Damage-Based Failure


Criterion for Fatigue Testing. Transportation Research Record, TRB, No. 1723,
pp. 141149.
Hou, T., B. S. Underwood, and Y. R. Kim. (2010). Fatigue Performance Prediction of
North Carolina Mixtures Using Simplified Viscoelastic Continuum Damage Model.
Journal of the Association of Asphalt Paving Technologists, Vol. 79, pp. 3580.
Kim, Y. R. and D. N. Little. (1990). One-Dimensional Constitutive Modeling of Asphalt
Concrete. ASCE Journal of Engineering Mechanics, Vol. 116, No. 4, pp. 751772.
Kim, Y. R., D. N. Little, and R. L. Lytton. (2003). Fatigue and Healing Characterization
of Asphalt Mixtures. Journal of Materials in Civil Engineering, Vol. 15, No. 1, pp.
7583.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Lee, H. J. and Y. R. Kim. (1998a). A Uniaxial Viscoelastic Constitutive Model for


Asphalt Concrete under Cyclic Loading. ASCE Journal of Engineering Mechanics,
Vol. 124, No. 1, pp. 3240.
Lee, H. J. and Y. R. Kim. (1998b). A Viscoelastic Continuum Damage Model of
Asphalt Concrete with Healing. ASCE Journal of Engineering Mechanics, Vol. 124,
No. 11, pp. 12241232.
Masad, E., V. T. F. Castelo Branco, D. N. Little, and R. L. Lytton. (2008). A Unified
Method for Analysis of Controlled-Strain and Controlled-Stress Fatigue Testing.
International Journal of Pavement Engineering, Vol. 9, No. 4, pp. 233246.
Reese, R. (1997). Properties of Aged Asphalt Binder Related to Asphalt Concrete
Fatigue Life. Journal of the Association of Asphalt Paving Technologists, Vol. 66,
pp. 604632.
Schapery, R. A. (1984). Correspondence Principles and a Generalized J-integral for
Large Deformation and Fracture Analysis of Viscoelastic Media. International
Journal of Fracture, Vol. 25, pp. 195223.
Shen, S. and S. H. Carpenter. (2005). Application of Dissipated Energy Concept in
Fatigue Endurance Limit Testing. Transportation Research Record, TRB, No.
1929, pp. 165173.
Si, Z., D. N. Little, and R. L. Lytton. (2002). Characterization of Microdamage and
Healing of Asphalt Concrete Mixtures. Journal of Materials in Civil Engineering,
Vol. 14, No. 6, pp. 461470.

21

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ZHANG, SABOURI, GUDDATI, KIM

Underwood, B. S., Y. R. Kim, and M. N. Guddati. (2006). Characterization and


Performance Prediction of ALF Mixtures Using a Viscoelastoplastic Continuum
Damage Model. Journal of the Association of Asphalt Paving Technologists, Vol.
75, pp. 577636.
Underwood, B. S., Y. R. Kim, and M. N. Guddati. (2010). Improved Calculation
Method of Damage Parameter in Viscoelastic Continuum Damage Model.
International Journal of Pavement Engineering. Vol. 11, pp. 459476.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

22

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Effect of Particle Mobility on Aggregate
Structure Formation in Asphalt Mixtures
Nima Roohi Sefidmazgi*, Pouya Teymourpour, and Hussain U. Bahia

Department of Civil & Environmental Engineering, University of Wisconsin-Madison,


Madison, WI, 53706.

ABSTRACT: During compaction of asphalt mixtures, aggregate structure starts building up by


proximity and direct contact of aggregates. In previous studies it has been shown that the
aggregate structure directly affects service performance. However, the mechanisms of aggregate
structure formation are not clearly understood. This study is focused on the mechanisms affecting
aggregate mobility during compaction and the effect of material properties on aggregate structure
formation. At the initial stages of compaction there is a relatively thick layer of mastic (i.e., mix of
binder and filler) between aggregates which allows for a shearing mobility in the mix if the mastic
viscosity is sufficiently low. However, as compaction proceeds, the mastic layer at the proximity
zone of aggregates becomes thinner due to high stress intensity and the higher viscosity of thin
mastic film or the aggregates dry contact effect increases the shearing resistance against
compaction (i.e., mix becomes locked). In this study mixes are compacted at different
temperatures using one base binder and three different modified binders. The quality of the
aggregate structure and packing throughout the compaction is characterized using 2-Dimensional
imaging of mixture sections and the total aggregate on aggregate proximity length is measured as
an indication of aggregate packing level. It is shown that for mixtures to obtain the maximum
packing, the compaction temperature should be picked based on mastic viscosity. The viscosity of
the mastic should be low enough for lubrication, but high enough to provide sufficient film
thickness at proximity zones and prevent locking of the mixture at the early stages of compaction.

KEYWORDS: Compaction, Imaging, Asphalt Mixture, Viscosity, Film Thickness, Mastic,


Proximity Zone.

The oral presentation was made by Mr. Roohi.

This is a reproduction of an article whose final and definitive form has been published in Road
Materials and Pavement Design 2013 Taylor & Francis. The article is available online at:
http://dx.doi.org/ 10.1080/14680629.2013.812844

23

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

1. Introduction

In construction of asphalt pavements, workability is defined as the ease of placement and


consolidation (i.e., compaction) of the asphalt mixture in the field (Bennert et al., 2010).
Several studies have characterized workability of asphalt mixtures from diverse
perspectives (Bahia et al., 1998; 2001, Delgadilo and Bahia, 2008; Austerman et al.,
2009; Bennert et al., 2010). In the Superpave mix design procedure, mixing and
compaction temperatures are selected based on the rotational bulk viscosity of the
binders (AASHTO T 316, 2007). Relying on the selection of mixing and compaction
temperatures based on binder viscosity alone has been expanded upon in other studies
(Delgadilo and Bahia, 2008). Bennert et al. (2010) suggested that bulk viscosity may not
be sufficient and thus investigated the application of thin film rheology to ranking
asphalt mixture workability and binder selection. Hanz et al. (2011) also investigated the
role of viscosity, and his group reported that viscosity alone cannot explain the effect of
compaction temperatures on density growth. Hanz et al. suggested lubricity as a possibly
better or complimentary binder property to explain density growth and introduced a
lubricity testing fixture for measurement of coefficient of friction between steel balls
with a thin film of binder in between to quantify the effect of warm mix additives on
workability.
Review of tribology literature, however, indicates that the lubrication phenomenon is
not simple and depends on interaction of lubricant film thickness, pressure in the
interface zone, viscosity, and surface texture (Lu et al., 2006). Stribeck developed the
concept of lubrication and the effect of surface roughness and lubricant properties on the
lubrication mechanism. The "Stribeck curve", or "StribeckHersey curve" (named after
Richard Stribeck and Mayo D. Hersey), used to categorize the friction properties
between two surfaces, was developed in the first half of the 20th century. According to
the Stribeck curve, friction regimes for sliding lubricated surfaces have been broadly
categorized into three main regimes: 1) Boundary lubrication, 2) Mixed lubrication, and
3) Hydro-dynamic lubrication (Figure 1). Boundary regime is defined as a lubrication
mechanism in which two surfaces are mainly in direct contact with maximum interlock
of asperities providing a high coefficient of friction. However, in a mixed lubrication
regime, increased lubricant film thickness between two surfaces decreases the
interlocking of surface asperities and, as a result, the coefficient of friction in between.
The hydro-dynamic regime starts with complete separation of two surfaces when the
lubricant film is thick enough compared to surface texture. The coefficient of friction
increases again in the hydro-dynamic regime as depicted in the Stribeck curve (Figure 1)
due to a viscous drag phenomenon (Lu et al., 2006; Johnson et al., 1971).
Johnson et al. (1971) demonstrated that the applied load on a contact of solid-liquid-
solid is carried by a pressure distribution on the contact which is dependent on the

24
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

distribution of asperities, asperity heights (i.e., texture), liquid film thickness, and
mechanical properties.

Figure 1. Stribeck curve and lubrication regimes: Z, N and p are lubricant viscosity,
shearing speed, and normal pressure.

Therefore, it can be inferred from the literature that explanation of the asphalt
mixture compaction process and workability ranking using binder rheological properties
as a single factor can be misleading, especially for polymer modified binders.
Another challenge to using binder viscosity as the controlling factor is the inclusion
of mineral dust, which is a requirement in mixture design. Adding mineral filler changes
the rheological properties of asphalt binders significantly due to particle interaction and
absorption of asphalt binder (Faheem, 2009). The change in rheological properties, such
as viscosity, is dependent on several factors: filler type and gradation (i.e. surface area),
binder type, filler concentration, etc. (Richardson, 1915). A fundamental challenge in
characterizing mixture workability in terms of mastic viscosity is that a direct correlation
between binder viscosity and mastic viscosity may not exist, although limited research
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

has been conducted on the subject.


To quantify workability and study the effects of temperature, several researchers
have focused on understanding the compaction process using mixture testing (Bahia et
al., 1998; Gudimettla et al., 2003). Bahia et al. (1998) quantified workability of asphalt
mixtures using indices that can be obtained from the densification curve (i.e., Gmm versus

25

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

number of gyrations) with an energy concept which successfully characterized different


mixtures based on the required applied energy during compaction to achieve a target
density. Gudimettla et al. (2003) used a bucket mixer to measure the torque required for
mixing of material at different temperatures to select the appropriate mixing and
compaction temperatures. Although the general belief is that higher temperature should
provide better workability for asphalt mixtures (since the viscosities of binder and mastic
decrease with an increase in temperature), several studies have shown that the shear
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

resistance and density are not a simple nor consistent function of temperature. In a
number of studies it is reported that the profile of density versus compaction temperature
follows a parabolic trend (Bennert et al., 2010; Hurley and Prowell 2005a, 2005b, 2005c;
Gudimettla et al., 2003; Anderson, 2002). In these studies it is clear that increasing the
compaction temperature causes an increase in density to a specific level as shear
resistance of the mixture decreases, after which higher temperatures could cause shear
resistance to increase, and thus density to decrease. Therefore, the results shown in these
studies indicate that there is, for many mixtures (if not all), a zone of minimum shear
resistance at which density is highest.
The zone of minimum internal shear resistance of mixtures is commonly referred to
as the Tender Zone; difficulties in achieving the desired density in the field are
reported for compaction within this range of temperatures (Marker, 1977; Crawford
1986; Brown et al., 2000; Buchanan and Cooley, 2002). Buchanan and Cooley
demonstrated that for some materials the Tender Zone may occur in the typical range
of compaction temperatures commonly experienced in the field, and recommended to let
the mat cool down and use heavier rollers at lower temperatures. This phenomenon
could not be explained with binder viscosity testing at different temperatures.
The majority of mix designs and compaction literature consider a specific range of
density or air void content (i.e. 3.5-4.5 % for Superpave mix design, AASHTO M 323-
07) as a target for compaction and quantify workability based on that metric alone. The
basic assumption is that performance of the mixture can be indicated using density.
However, several studies have demonstrated that materials with the same density may
perform significantly different under service loading (Coenen et al., 2012; Roohi et al.,
2012; Olard, 2012). Based on the contact mechanism analysis, Zhu and Nodes (2000)
demonstrated that the transmission of load in the asphalt mixture is mainly determined
by the interaction of aggregates and binder at the proximity zones of adjacent aggregates.
According to this concept, Roohi et al. showed that specific microstructural indices, such
as total aggregate proximity length in the aggregate skeleton (i.e., in 2-Dimensional
section images of mixtures), can be a better indicator of mixture load bearing aggregate
structure and performance. The aggregate structure indices show sensitivity to
compaction conditions such as compaction temperature, method, and mix design
properties (Tashman et al., 2001; Bahia and Roohi, 2012).

26

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

Therefore, there is a need for a more comprehensive understanding of the mechanics


governing the compaction process to better understand the aggregate on aggregate
lubrication, referred to here as aggregate mobility. The mechanisms resisting
aggregate mobility control the formation of the aggregate skeleton, the main load
bearing structure in asphalt mixtures, thus a deeper understanding of this process can be
invaluable for the control and optimization of the construction process and enhancement
of pavement performance.

2. Mechanisms Controlling Particle Mobility during Compaction

When the compaction load is initially applied on loose mix, packing of aggregates
increases rapidly as the applied stress can overcome the flow resistance of the mastic. In
this stage, mastic viscosity could be the dominant factor since aggregate interlock is
minimal. However, with subsequent aggregate skeleton formation, the rate of bulk
material deformation decreases significantly and tends toward zero. Li and Gibson
(2011) used the gyratory compactor to show that the number of gyrations needed for
locking of mixtures is dependent on the material properties and mix design. In addition,
Li and Gibson reported that most of mixes compacted in the study locked at gyrations of
60 or less and the continued compaction to design gyrations of 100 did not change the
density of mixes significantly. However, there is not a clear fundamental explanation for
the difference in locking point of different mixes.
This study is focused on mixture compaction mechanisms based on the geometrical
and mechanical properties of aggregate on aggregate proximity zones. For the purposes
of this study, aggregates are determined to be in proximity if the perimeters of two
adjacent aggregates are within a predefined spacing in a 2-Dimensional section. It is
assumed that when the shearing stress applied on the mixture is higher than the shearing
strength of aggregate proximity zones (i.e., contact stiffness) or the internal friction of
the lubricant in between, the mixture plastically deforms.
Contact friction of two surfaces with a lubricant in between is known to be a function
of normal load applied, film thickness, viscosity of the lubricant, and morphological
properties of the surfaces, such as texture (Lu et al., 2006). In a contact, given other
factors to be constant, when decreasing the viscosity of the lubricant (such as increasing
temperature of asphalt binder), it is hypothesized that two phenomena can happen
simultaneously:
Reduction of friction between the two surfaces as a result of reduction in viscosity of
the lubricant becasuse the energy needed to overcome the internal friction of the
lubricant is lower.

27
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

Reduction of film thickness and increase in proximity of the two surfaces due to the
lubricant being squeezed out in the case of low or zero confinement. When the surfaces
of the particles are smooth, the reduction of film thickness to a very thin layer (i.e., in
the nanometer order of magnitude) causes the effective viscosity of the lubricant to
increase, which increases the shearing resistance between two surfaces (Luengo et al.,
1997). In case of rough surfaces such as natural crushed aggregates, high stress
concentrations at the asperities remove the lubricant and the solid surfaces with a high
coefficient of friction will be directly in contact. This process is called dry contacts
in this study. This condition is also termed as boundary lubrication in the tribology
science studies (Lu et al., 2006, Figure 1). The rate of squeezing out of lubricant is
directly related to the viscosity of the material (Macosko, 1994). Therefore, if the
viscosity of the lubricant is not sufficiently high to resist squeezing out, the friction
between two surfaces may increase significantly as a result of this dry contact effect.
Therefore, it can be hypothesized that the effective friction between two surfaces in a
proximity zone is a parabolic function of lubricant viscosity as shown in Figure 2. This
trend is caused by the competing mechanisms of decreasing film thickness and an
increase in viscosity. The initial reduction in friction is caused by increased film
thickness and the final increase in friction is caused by an increase in lubricant viscosity,
as shown in Figure 1.

Figure 2. Proposed schematic of change in friction at proximity zones as a function of


temperature.

Compaction of asphalt mixtures can be envisioned as packing of aggregates that are


lubricated by the asphalt mastic. Therefore, for a better packing of aggregates, the

28

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

viscosity of the mastic should be low enough to allow sufficient workability, but high
enough to maintain adequate film thickness at proximity zones to avoid dry contact or
the thin film viscosity effect.
The focus of this study is on the packing of aggregate particles during compaction.
However, there are some other concerns in compaction temperature selection, such as
binder drain down at very high temperatures, aggregate segregation and low mastic film
thickness (which may cause durability problems). These concerns are outside the scope
of this study.
In this study, it is hypothesized that viscosity of mastic, assuming other factors to be
constant, controls the mobility and structure formation of aggregate skeleton. Therefore,
as shown in Figure 2, changing the compaction temperature will result in different
aggregate packing due to the change in friction at aggregate proximity zones. This
implies the existence of an optimum packing at the lowest level of proximity zone
friction.

3. Study Objectives

The main objectives of this study are to determine the possible mechanisms of aggregate
structure formation during compaction of asphalt mixtures and the effect of mastic
rheology and film thickness on aggregate mobility.

4. Materials and Test Methods

4.1. Mixture Sample Preparation and Testing

To isolate the effect of mastic as the lubricant phase, mixes with significantly different
mastic viscosities at a specified compaction temperature were produced. The gradation
(Figure 3) has been designed for the purposes of this study from aggregate sources
commonly used for HMA production in Wisconsin (i.e., granite).

29

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

100

80
% Passing

60

40

20

0
0 0 .0 7 5 0 .6 1 .1 8 2 .3 6 4 .7 5 9 .5 1 2 .5 1 9 .0
Sieve Size (mm)
Coarse MMaxaxDDenenisstiyty Control Points

Figure 3. Aggregate gradation used in this study for mix design.

To cover a wide range of mastic viscosities, four different binders were used for
compaction: a base neat binder and three modified binders with three different modifiers
(i.e., a styrene butadiene styrene based elastomer modifier (S), a polyethylene based
plastomer (C), and the mix of elastomer and plastomer modifiers (H)). The level of
modification was selected to obtain high true grade of 771C for all modified binders.
The binder designations and performance grades are shown in Table 1.
Table 1. Binder designations and Performance Grades (PGs).

Binder PG
Ne a t 64-22
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

S 76-22
C 76-22
H 76-22

The reference mixing and compaction temperatures were selected as 155C and
145C, respectively, based on the viscosity-temperature profile (i.e., using the Superpave
mix design procedure) for the neat binder and was kept consistent for all mixes.
Additionally, samples were mixed and compacted at several temperatures providing
significantly different orders of magnitude of mastic viscosity. The mixing temperature

30

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

was always kept 10C higher than the desired compaction temperature. The compaction
temperatures are shown in Table 2. All mixes are compacted with 100 gyrations, which
is the required Wisconsin E-10 mix design level of gyrations (Ndes) with 5.5% asphalt
content by mass.
Table 2. Compaction temperatures.

Mix Compaction Temperatures (C)


Ne a t 65, 115, 145
S 115, 145, 165
C 65, 95, 145, 165
H 65, 115, 145, 165

After compaction, the samples were cut vertically in three sections leading to
attainment of six 2-Dimensional images (Figure 4), with one cutting section at the
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

middle of the sample and two at 25.4 mm from the middle. All mixes were scanned
according to the procedure of Roohi et al. (2012), and image analysis was performed for
aggregate structural characterization (i.e., using Image Processing and Analysis Software
(IPAS2), Roohi et al, 2012). The index as a result of the image analysis procedure to
represent the packing condition in mixtures is the aggregate total proximity zone length
in 2-Dimensional images. In the imaging software, proximity is defined when two
aggregate perimeter pixels are within a distance specified by the user (i.e., 0.1 mm in this
study) and all of the pixels of the two aggregate perimeters within this distance are
captured. These pixels form a line referred to as a contact line in Roohi et al. (2012) and
proximity line here (Figure 5). Higher total contact length (i.e., proximity length)
represents higher aggregate interlock and packing in the mixture and better rutting
performance as a result. Based on image analysis results, the total aggregate proximity
length profile for different compaction temperatures (i.e., different mastic viscosities)
was obtained.

31

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

Figure 4. Cutting sections (Roohi et al. 2012).


--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Figure 5. Proximity line.


In addition to compaction at different temperatures, mixes were compacted at
different levels of effort (i.e., 10, 20, 40, 80 and 100 gyrations) at the temperature of
145C and image analysis was performed. The objective was to investigate the
difference in trend of aggregate packing attainment for materials with different
viscosities during compaction. The air void content of mixes at different gyrations is
shown in Table 3.

32

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

Table 3. Air voids of mixes compacted at different number of gyrations (compaction


temperature = 145C).

Air Voids, %
Mix
10 gy 20 gy 40 gy 80 gy 100 gy
Ne a t 1 2 .1 8 .8 6 .5 6 .1 4 .5
C 1 0 .0 7 .0 5 .8 3 .8 3 .5
S 1 1 .5 7 .8 6 .3 3 .8 4 .0
H 1 1 .3 7 .2 4 .9 3 .8 3 .0

4.2. Binder and Mastic Preparation and Testing

It is hypothesized that the viscosity of an unconfined/semi confined lubricant between


two surfaces controls the internal frictional behavior of asphalt mixtures through the two
aforementioned mechanisms before the boundary lubrication phase (i.e., dry contact) at
which the locking of the mixture occurs. In order to test this hypothesis and study the
effect of lubricant viscosity on aggregate packing, viscosity testing of binders and
mastics was performed.
Four binders were aged using the Rolling Thin-Film Oven (RTFO) for short term
aging during mixing and compaction. The viscosity testing was performed at all
compaction temperatures with shear rates of 0.01, 0.1, 1, 10 and 100 (sec-1) to cover a
wide range of shear rates that were proposed in different studies (Bahia et al., 2001;
Delgadillo and Bahia, 2008) with rheometer limitation considerations. Binders were
tested using a Bob and Cup fixture (Figure 6) with a binder film thickness of 1 mm. The
samples were conditioned at the compaction temperatures for 15 minutes in the fixture.
The viscosity measurements were performed for three minutes for each shear rate
starting with the lowest shear rate (i.e. 0.01 sec-1).
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

33

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

Figure 6. Bob and cup fixture.

Many of the previous studies on asphalt mixture workability and compaction


mechanisms are focused on rheological behavior (e.g., viscosity) of binder as the most
prominent mix constituent that affects workability. Yet, several studies have shown that
adding filler (i.e., passing sieve the No. 200) to binder may significantly change the
mechanical and rheological behavior of the binder (Faheem, 2009; Clopotel, 2012),
since filler particles reinforce the binder and absorb specific molecular components of
the binder (e.g., asphaltenes) depending on the type, gradation, concentration, and
specific surface area of the filler particles. Therefore, for the purposes of this study,
mastics with the same fillers used in mixture production were prepared for viscosity
testing. The binder content of the mastic samples was determined based on the specific
surface area of the aggregates. With a constant binder film thickness assumption for
aggregates of different sizes in the mix, the portion of binder content of mixture that
coated the filler particles was determined (i.e., depending on the ratio of surface area of
filler particles to the surface area of the whole aggregate blend). Based on such
calculations, the binder content of the mastics was 55.3 % by weight. All the mastics
were mixed using un-aged binders and conditioned at the same mixing and compaction
temperatures as the mixtures.
The viscosity testing procedures performed on mastics were the same as binder
viscosity testing using the bob and cup fixture with the same shear rates. Considering
that the size of particles in mastic is finer than 0.075 mm, which is less than 10% of the
sample film thickness in the bob and cup fixture, the same fixture used for binder testing
can be used for mastic testing since there is no interference expected with sample
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

geometry.

34

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

5. Results and Discussion

A previous study (Roohi et al., 2012) demonstrated that mixtures with the same
gradation design but different binders may have significantly different aggregate
structures. It is hypothesized in this study that the difference in structures can be
explained with mastic viscosity. Therefore, mixtures were produced at the same
temperature (i.e., 145C) and different compaction efforts (i.e., 10, 20, 40, 80, and 100
gyrations) to monitor the aggregate structure formation for different binders. All the
mixes were cut and scanned, and image analysis was performed. The total aggregate on
aggregate proximity zone lengths, as an indication of aggregate internal structure and
packing for images of each mix, were determined and are depicted in Figure 7.

900
Total Proximity Zone Length (mm/100cm2)

800

700

600

500

400

300

200

100

0
0 20 40 60 80 100 120
Gyrations

Neat C H S

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Figure 7. Total proximity zone length for mixes at different number of gyrations.

The results show that the aggregate packing in the neat mixture is mostly obtained
during the first 10 gyrations and the change in aggregate structure with application of
more compaction effort (i.e., up to 100 gyrations) is insignificant. However, for C and
S mixes, the aggregate structure formation took up to 20 gyrations. The number of
gyrations for aggregate structure formation in the mixture designated as H was 40 and

35

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

the structure was formed in a more gradual manner and locked later compared to the
other mixtures.
To help explain the trends of Figure 7, all four binders and mastics were tested for
viscosity using the bob and cup fixture at the temperature of 145C (i.e., compaction
temperature) and at different shear rates. The results are shown in Figure 8. It should be
noted that the average standard deviation of the viscosity measurement in the rheometer
using the bob and cup geometry is +/- 0.005 Pas.

Binder
10
Ne a t C S H
Viscosity (Pa.s)

0 .1
0.01 0 .1 1 10 100
Shear Rate (sec-1)
( a)
Mastic
1000
Ne a t C S H
100
Viscosity (Pa.s)

10

0 .1
0.01 0 .1 1 10 100
Shear Rate (sec-1)
(b )
Figure 8. Viscosity at 145C: a) Binder, b) Mastic.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

36

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

The viscosity results show different rankings for binders compared to mastics.
According to the binder results at 145C, Neat binder showed the lowest viscosity and
C, H, and S have higher viscosities, in respective order. However, mastics are shear
rate dependent and the ranking at low shear rates is different than the binder ranking:
Neat, S, C, and H from low to high viscosity.
As noted, the lubricant phase in a mixture is mastic; therefore, it is more reliable to
consider mastic viscosity for workability properties of mixtures. In addition, it was
shown in the NCHRP 9-10 study (Bahia et al., 2001) that the average shear rate
occurring in mixtures after initial gyrations is low according to the axial strain rate. The
mastic viscous behavior is not easily predicted from the binder viscosity alone at lo w
shear rates. Therefore, for a better characterization of aggregate structure formation
during compaction, the low shear rate viscosity values of mastics (i.e., 0.01 to 0.1 (sec-1))
are chosen.
Mastics with higher viscosity can resist the stress condition at proximity zones of
aggregates and resist the squeeze out from proximity zones longer (more gyrations) than
soft mastics, which provide aggregate mobility for a longer period during compaction. In
other words, locking of mixtures due to dry contacts (i.e., direct contact of aggregate
asperities) or thin film high viscosity happens sooner in mixes with very soft mastics.
However, the mastic viscosity should be low enough to allow for lubrication and
shearing deformation between aggregates with the compaction effort applied. This
theory is in agreement with the results. The H mastic showed the highest viscosity
among others at the shear rate of 0.01 and 0.1 (sec-1) and the neat mastic was the lowest,
which explains the gradual formation of structure in the H mixture and rapid locking in
the neat mixture.
It is further believed that the film thickness of mastic (or mortar) between coarse
aggregates decreases rapidly in the neat mix relative to the other mixtures as it is a
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

function of mastic or mortar viscosity (Macosko, 1994). To prove this hypothesis,


mortar film thickness in four mixtures is estimated at the same number of gyrations
depicted in Figure 7. (Mortar is defined here as binder mixed with aggregates passing the
No. 16 sieve, since aggregates finer than 1.18 mm are not clearly visible in the images
used for analysis in this study.)
Aggregates are naturally rough particles. Therefore, the film thickness of material
(e.g., binder, mastic, or mortar) between surfaces of two aggregates is not constant
(Figure 9). In addition, the friction between two aggregates increases significantly when
asperities on the two surfaces begin contacting each other as the coefficient of friction
between bare aggregates is considerably higher than the internal friction of material in
between (simply meaning that more shearing load is needed to drag an aggregate on
another aggregate when there is no lubricant between them). Therefore, the minimum

37

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

mortar film thickness between each pair of aggregates (i.e., at proximity zones) is
calculated as an index to represent the trend of aggregate contact formation during
compaction.

Figure 9. Black and white image of a mixture representing minimum film thickness
between aggregates 1 and 2.

Histograms of mortar minimum film thickness distribution between aggregates for


the H mix during compaction are shown in Figure 10. The histogram for the 10
gyrations level of compaction is flatly distributed, however by increasing the number of
gyrations, the film thickness of mortar between aggregates was decreased and the
histogram is distributed in a narrower range of film thickness (i.e., less than 0.1 mm
which is considered as proximity zone in image analysis).

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

38

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

20
10 gy
18
20 gy
16
40 gy
14
80 gy
% Occurance

12
10 100 gy

8
6
4
2
0
0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6
Film Thickness (mm)

Figure 10. Evolution of mortar film thickness index at proximity zones of aggregates for
different number of gyrations.

The same analysis shown in Figure 10 was performed on the other three mixtures.
For a more appropriate comparison of different mixtures in terms of film thickness at
proximity zones evolution during the compaction process, a weighted average of film
thickness for each histogram (i.e., at each number of gyrations) is calculated. Results
show that (Figure 11) the film thickness of mortar between aggregates gradually
decreases for the H mix during compaction up to 40 gyrations; however, the neat mix
lost its mortar film between coarse aggregates before 10 gyrations, which is in
agreement with the total proximity zone length and mastic viscosity data.

39
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

2 .5
H C S Ne a t
Film Thickness (mm) 2 .0

1 .5

1 .0

0 .5

0 .0
10 20 40 80 100
Gyrations

Figure 11. Mortar film thickness index.


Mastic viscosities at 145C and shear rates of 0.01 (sec-1) and 0.1 (sec-1) are plotted
in Figure 12 against film thickness at each gyration level. There is a correlation between
film thickness and low shear viscosity of mastics at compaction efforts of 40 gyrations
or less, that is considered the aggregate structure formation stage at which the
controlling lubrication mechanisms are hydro-dynamic and mixed lubrication. This
confirms the hypothesis that mastics (or mortars) with higher viscosities keep thicker
mastic layers between aggregates and it takes a higher number of gyrations (i.e., longer
time) to squeeze mastic out of proximity zones under the stress conditions.
To further validate the hypothesis, mastic viscosity values at different shear rates are
plotted versus total proximity length of the four mixes compacted to 100 gyrations in
Figure 13. The results show that there is a good correlation between total proximity
length as a representation of aggregate packing with mastic viscosity at low shear rates
(i.e., 0.01 and 0.1 sec-1). However, there is no significant correlation at high shear rates
(i.e., 10 and 100 (sec-1)).

40
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

0.01 (sec-1) shear rate - 145 C


1000

R = 0.78 R = 0.90 R = 0.95


100
Viscosity (Pa.s)

10 10 Gyrations
20 Gyrations
1 40 Gyrations
80 Gyrations
100 Gyrations
0 .1
0 .0
1 .0 1 .5 0 .5
2 .0 2 .5
Film Thickness (mm)
( a)
Figure 12a. Low shear viscosity versus average proximity zones film thickness at
different compaction levels: a) 0.01 sec-1 shear rate, b) 0.1 sec-1 shear rate.

0.1 (sec-1) shear rate - 145 C


100

R = 0.80 R = 0.88 R = 0.95


Viscosity (Pa.s)

10
10 Gyrations
20 Gyrations
40 Gyrations
80 Gyrations
100 Gyrations
1
0 .0 0 .5 1 .0 1 .5 2 .0 2 .5
Film Thickness (mm)
(b )
Figure 12b. Low shear viscosity versus average proximity zones film thickness at
different compaction levels: a) 0.01 sec-1 shear rate, b) 0.1 sec-1 shear rate.

41

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

1 (1/sec) 10 (1/sec)
121 0.1 (1/sec) 0.01 (1/sec)

101
Viscosity (Pa.s)

R = 1.00
81
61
41
R = 0.99
21
R = 0.75 R = 0.13
1
500 600 700 800 900 1000 1100
Total Proximity Zone Length (mm/ 100 cm2)

Figure 13. Viscosity versus total aggregate proximity zone length at 145C.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
It can be concluded that among the four binders used for this study, H provided the
best workability and packing of aggregates since the viscosity of mastic H at the
compaction temperature was high enough to provide sufficient film of mastic between
aggregates and consequently aggregate mobility to a higher number of gyrations.
However, in the neat mixture with the lowest mastic viscosity, the film of mastic at the
aggregate proximities is squeezed out at less than 10 gyrations, which possibly resulted
in dry contacts or a very thin layer of mastic with high viscosity that increased friction
between aggregates significantly. The C and S mixes showed a transition from the
H mix (i.e., high mastic viscosity, gradual locking of mix) to the neat mix (i.e., low
mastic viscosity, rapid locking of mix) from the mastic viscosity and aggregate packing
point of view.
In contrast, using mastics with extremely high viscosities is expected to result in
lower workability and lower mobility of particles. Thus, there should be a high and low
limit for viscosity of mastics within which sufficient aggregate mobility during
compaction is achieved. Compaction outside this range of viscosities should be avoided.
To investigate this concept, mixes were produced and compacted at different compaction
temperatures as shown in Table 2. Compaction of mixes at temperatures from 65C to
165C provided a wide range of mastic and binder viscosities.
The image analysis results of mixes compacted in different temperatures is depicted
in Figure 14. The analysis for the S mix compacted at 65C was not performed since it
had significant aggregate coating problems.

42

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

As shown in Figure 14, there is a maximum proximity zone length (i.e. Optimum
packing) at a unique temperature for each mixture. The maximum values of total
proximity length for different mixes are approximately the same. This is possibly due to
the fact that the same gradation is used and thus according to the mastic viscosities at
different temperatures, ultimately a maximum packing of particles is achieved. It is
possible that different mastics provided similar viscosity ranges at different temperatures
and at that specific range for the gradation used in the four mixes optimum packing has
been achieved. It has been shown in a previous study that there is a good relation
between total proximity length and rutting performance (Roohi et al., 2011). Therefore,
this finding shows that the structure needed to resist the service loading can be obtained
by compaction of any of the four mixes only by selecting the correct compaction
temperature.
4,500
Total Proximity Zone Length (mm/100cm2)

4,000
3,500
3,000
2,500
2,000
1,500
1,000
60 80 100 120 140 160
Compaction Temp. (C)

Control C S H

Figure 14. Total proximity zone length - compaction temperature profile.

To determine the effect of viscosity on total proximity zone length at different


compaction temperatures, binders and mastics were tested using the bob and cup
geometry in the DSR. The viscosity results for the binders at different temperatures and
shear rates are shown in Figure 15. According to the results, all binders show Newtonian
behavior (shear rate independent viscosity), specifically at high temperatures, although
for H and C at 65C the materials showed slight shear rate dependency of viscosity.
For all practical purposes, the viscosity ranking of binders at temperatures above 115oC
can be considered independent of shear rate.

43
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

Neat - binder
10000

1000
Viscosity (Pa.s)

100

10

0 .1

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
0.01 0 .1 1 10 100
Shear Rate (sec-1)
Ne a t 6 5 C Ne a t 1 1 5 C Ne a t 1 4 5 C

( a)

C - binder
10,000

1,000
Viscosity (Pa.s)

100

10

0
0.01 0 .1 1 10 100
Shear Rate (sec-1)
C 65 C C 95 C C 145 C C 165 C

(b )

44

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

S - binder

10,000

1,000
Viscosity (Pa.s)

100

10

0
0.01 0 .1 1 10 100
Shear Rate (sec-1)
S 65 C S 115 C S 145 C S 165 C

( c)

H - binder
10,000

1,000
Viscosity (Pa.s)

100

10

0
0.01 0 .1 1 10 100
Shear Rate (sec-1)
H 65 C H 115 C H 145 C H 165 C

(d )
Figure 15. Bob and cup binder viscosity results at different temperatures and shear
rates: (a) Neat binder (b) C binder (c) S binder (d) H binder.

45
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

Mastics were tested at the same shear rates and temperatures as binders using the bob
and cup fixture (Figure 16). The test results for modified mastics show higher shear rate
dependency (Specifically for C and H mastics) compared to binders at different
temperatures. The other important difference of mastic behavior compared to binders is
that the viscosity ranking of mastics does not follow that of binders. Based on the tests
performed on the four mastics, H is the most viscous material and the neat binder
showed the lowest viscosity at different temperatures. Except for the neat mastic,
viscosity measurements at 65C were out of the torque limit of the rheometer.

Neat - mastic

100000
10000 65 115 145
Viscosity (Pa.s)

1000
100
10
1
0 .1
0.01 0 .1 1 10 100
Shear Rate (sec-1)

( a)
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

46

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

C - mastic

1000
95 145 165
100
Viscosity (Pa.s)

10

0
0.01 0 .1 1 10 100
Shear Rate (sec-1)

(b )

S - mastic

100000
115 145 165
10000
Viscosity (Pa.s)

1000

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
100

10

0 .1
0.01 0 .1 1 10 100
Shear Rate (sec-1)

( c)

47

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

H - mastic

100000
115 145 165
10000
Viscosity (Pa.s)

1000

100

10

0 .1
0.01 0 .1 1 10 100
Shear Rate (sec-1)

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
(d )
Figure 16. Bob and cup mastic viscosity results at different temperatures and shear
rates: (a) Neat mastic (b) C mastic (c) S mastic (d) H mastic.
As explained in previous sections, low shear viscosity of mastics is proposed as a
factor affecting mixture workability at different temperatures because it correlates better
with aggregate packing. In addition, there is always a blend of binder and filler (i.e.,
mastic) among coarser aggregates acting as a lubricant phase. However, it is shown in
this study that the total aggregate proximity zone length values do not demonstrate a
clear trend with increasing viscosity as caused by reducing compaction temperature. The
total proximity zone length results versus mastic low shear viscosity for mixes
compacted at different temperatures are plotted in Figure 17. In addition to mastic
viscosity, film thickness at the proximity zones (which is also a function of viscosity) is
another important factor affecting friction between aggregates. By increasing the
compaction temperature, viscosity is decreased providing sufficient lubrication between
aggregates and higher total proximity zone length (as long as the viscosity is higher than
a specific range or, in other words, mastic film thickness is still sufficient to avoid dry
contacts). This is demonstrated in Figure 17. However, for viscosities lower than a
specific range, film thickness of mastic becomes very thin within the proximity zones of
aggregates, as shown in Figure 18. Therefore, due to the dry contacts and/or the thin film
viscosity effect, workability of the mixture and aggregate mobility decreases, leading to

48

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

a decrease in total proximity zone length (i.e. packing of aggregates) after a maximum
value for mixes compacted at higher temperatures.

4,500
Total Proximity Zone Length (mm/100cm2)

4,000
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

3,500

3,000

2,500

2,000

1,500

1,000
0 .1 1 10 100 1000 10000 100000
Mastic Viscosity (Pa.s)

Control C S H

Figure 17. Total proximity zone length as a function of mastic viscosity at different
compaction temperatures at shear rate of 0.01 sec-.1

The data points from the film thickness plots at which the maximum total proximity
zone length is attained are circled in Figure 18.

49

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

Mortar Film Thickness


5 .0
H C S Ne a t
Film Thickness (mm)

4 .0

3 .0

2 .0

1 .0

0 .0
60 80 100 120 140 160
Compaction Temp (C)

Figure 18. Mortar average film thickness at different compaction temperatures.

In this study, limited numbers of compaction temperatures are selected for proof of
the concept. However, to determine the viscosity range providing sufficient film
thickness and lubrication for optimum workability, compaction of mixes at additional
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

temperatures is needed to accurately determine the compaction temperature (or mastic


viscosity as a material property) leading to maximum packing. In addition, a change in
aggregate gradation design changes the stress distribution in the mix and at the aggregate
proximity zones. Therefore, with the results of this study, it can be concluded that,
different aggregate gradations may lead to different ranges of viscosity to resist the
squeezing out due to the contact normal stress condition. This needs more verification,
but is in agreement with recent findings in the literature that the viscosity as a single
factor is not enough to predict workability of different mixtures (Hanz et al., 2011).

6. Conclusions

This study was focused on understanding the factors that can critically control aggregate
mobility and packing during compaction. The study was limited to one gradation and
four different binders. This was done to challenge the concept that binder viscosity as a
function of temperature is a sufficient parameter to recommend minimum limits for
compaction temperatures. In addition to measuring the density of mixes compacted at
various temperatures, viscosities of binders and mastics were measured, and estimates of
film thickness at aggregate proximity zones using advanced imaging tools were

50

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

performed. Although many previous studies had considered a desired density as a


compaction process target, this study was focused on the evolution of aggregate structure
during compaction and effect of lubricant rheology (i.e., viscosity) on the aggregate
packing as a more fundamental indication of final product performance. The aggregate
total proximity zone length as result of 2-Dimensional image analysis was used to
characterize aggregate structure of asphalt mixtures. The main findings of this study are:
Most design specifications use binder viscosity as an indication of mixture
workability at the compaction temperature. This study demonstrated that binder
viscosity is not sufficient and that workability of the mixture is more dependent on
viscosity and film thickness of the actual lubricant, which is the binder and filler
mastic. The film thickness that is important is not the overall average but the
thickness within the proximity zones of aggregates. This finding is supported by
many studies in the field of tribology and lubrication mechanics.
Two competing mechanisms are proposed as controlling factors of the compaction
process and aggregate structure formation. Lubricants with lower viscosity provide
easier sliding between the lubricated surfaces. On the other hand, lower viscosity
results in a higher squeezing out rate of lubricant for a constant normal force within
the proximity zones between two surfaces, which increases the chance of dry
contact interaction, or thin film high viscosity effect. The result is an increase in
friction between two surfaces. Therefore, by increasing the compaction temperature,
the aggregate mobility and structural packing usually begin to increase to a peak
value and then decrease as dry contact or thin film viscosity effects become
significant at higher temperatures.
Mixes were compacted to different number of gyrations with the same gradation but
different binders to demonstrate the effect of mastic viscosity on aggregate structure
formation. The rates of increase in total aggregate proximity zone lengths were
significantly different for different mixes during compaction. Additionally, binders
and mastics were tested at different shear rates. The viscosity behavior of mastic at
low shear rates could not be predicted from binder viscosity, and since mastic is the
lubricant phase in asphalt mixtures, mastic viscosity was used to explain the
behavior of the mixtures. With the aid of image analysis, the film thickness of
mortar at proximity zones of aggregates at different compaction stages (i.e., number
of gyrations) was determined. The mastic viscosity at low shear rates in addition to
mortar film thickness between coarse aggregate results show that the mastic with
higher viscosity at compaction temperatures squeezed out of aggregate proximity
zones in a more gradual manner compared to mastics with lower viscosity.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Therefore, compaction of mixes with higher viscosity mastics resulted in better


packing of aggregates, and the mix with the lowest mastic viscosity was locked at

51

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

the initial stages of compaction and the total proximity zone length was unchanged
for higher number of gyrations.
In order to study the effect of mastic viscosity on aggregate structure, mixes were
compacted at temperatures from 65C to 165C using four binders providing
significantly different viscosity levels. The results showed that with an increase in
compaction temperature or decrease in mastic viscosity, total proximity zone length
in the aggregate skeleton increased up to a certain temperature and then began to
decrease. The temperature at which the optimum total proximity zone length was
achieved varies for different materials. However, the value of optimum total
proximity zone length was approximately the same for all mixtures. This finding
shows that the particle arrangement and packing level for the same compaction
pressure and aggregate gradation is related to mastic viscosity as a fundamental
material property which can be the same for different materials at different
temperatures. However, the number of mixes and compaction temperatures in this
study were not sufficient to define a meaningful range of mastic viscosity for
optimum packing.

7. Acknowledgment

As members of the Asphalt Research Consortium, funding and support from the FHWA
and WRI is acknowledged. The help and comments from Dr. Raul Velasquez, Dr.
Andrew Hanz, and Mr. Dan Swiertz are greatly appreciated. The testing and sample
preparation of Trevor Schultz is also acknowledged. Partial funding from Honeywell

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Inc., USA is acknowledged.

8. References

AASHTO. (2007). Designation T 316, Standard Method of Test for Viscosity


Determination of Asphalt Binder Using Rotational Viscometer. Standard
Specifications for Transportation Materials and Methods of Sampling and Testing,
27th Ed. Washington, D.C., American Association of State Highway and
Transportation Officials.
AASHTO M 323-07, Superpave Volumetric Mix Design, 2007.
Anderson, R. M., Relationship Between Superpave Gyratory Compaction Properties on
the Rutting Potential of Asphalt Mixes, Journal of the Association of Asphalt Paving
Technologists, Vol. 71, 2002, pp. 725738.

52

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

Austerman, A., Mogawer, W., and Bonaquist, R., Evaluating the Effects of Warm Mix
Asphalt Technology Additive Dosages on the Workability and Durability of Asphalt
Mixtures Containing Recycled Asphalt Pavement, Transportation Research Board
88th Annual Meeting Compendium of Papers CD-ROM. Washington D.C.:
Transportation Research Board of National Academies, 2009.
Bahia, H., Friemel, T., Peterson, P. and Russell, J., Optimization of Constructability
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

and Resistance to Traffic: A New Design Approach for HMA Using the Superpave
Compactor, Journal of the Association of Asphalt Paving Technologists, Vol. 67,
p.189232, 1998.
Bahia, H. U., Hanson, D. I., Zeng, H., Zhai, M., Khatri, M. A., and Anderson, R. M.,
NCHRP Report 459: Characterization of Modified Asphalt Binders in Superpave Mix
Design, TRB, National Research Council, Washington, D.C., 2001.
Bahia, H.U., and Roohi, N., Effect of Compaction Method, Temperature and Pressure
on Aggregate Structure of Hot Mix Asphalt, International Journal of Pavement, in
press, 2012.
Bennert, T., Reinke, G., Mogawer, W., and Mooney, K., Assessment of
Workability/Compactability of Warm Mix Asphalt,- Transportation Research Board
89th Annual Meeting Compendium of Papers. CD-ROM. Washington, D.C.:
Transportation Research Board of National Academies, 2010.
Brown, E. R., Lord, B., Decker, D., and D. Newcomb, Hot Mix Asphalt Tender Zone,
Report No. 2000-02, National Center for Asphalt Technology Auburn, AL, 2000.
Buchanan, M.S., and Cooley, L.A., Case Studies of the Tender Zone in Coarse-Graded
Superpave Mixtures, Report No.: NCAT Report 02-01.: National Center for Asphalt
Technology, Auburn, AL 2002.
Clopotel, C., Filler Reinforcement Mechanisms in Asphalt Mastic, PhD thesis,
University of Wisconsin-Madison, 2012.
Coenen, A.R., Kutay, M.E., Roohi Sefidmazgi, N., Bahia, H.U., Aggregate Structure
Characterization of Asphalt Mixtures Using 2-Dimensional Image Analysis, Road
Materials and Pavement Design, accepted for publication, 2012.
Crawford, C. Tender Mixes - Probable Causes, Possible Remedies, National Asphalt
Pavement Association Quality Improvement Series 108/86.
Delgadillo, R., and H. U. Bahia, Effects of Temperature and Pressure on Hot Mixed
Asphalt Compaction: Field and Laboratory Study, Journal of Materials in Civil
Engineering, v20, no. 6, p. 440448, 2008.

53

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

Faheem, A.F., Modeling of Asphalt Mastic in Terms of Filler-Bitumen Interaction, PhD


thesis, University of Wisconsin-Madison, 2009.
Gudimettla, J., Cooley, L. A., and Brown, E. R., Workability of Hot Mix Asphalt, NCAT
Report 03-03. National Center for Asphalt Technology, Auburn University, Auburn,
2003.
Hanz, A.J., Mahmoud, E., and, Bahia, H.U., Asphalt Lubricity Test Evaluation and
Relationship to Mixture Workability, Transportation Research Board 90th Annual
Meeting DVD Compendium of Papers, 2011.
Hurley, G., Evaluation of New Technologies for Use in WMA, Masters Thesis,
Auburn, AL: Auburn University, 2006.
Hurley, G.C., and Prowell, B., Evaluation of Aspha-min Zeolite for use in Warm-Mix
Asphalt, NCAT Report 05-04, National Center for Asphalt Technology, Auburn, AL,
2005.
Hurley, G., and Prowell, B., Evaluation of Evotherm for Use in Warm-Mix Asphalt,
NCAT Report 06-02, National Center for Asphalt Technology, Auburn, AL, 2005.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Hurley, G., and Prowell, B., Evaluation of Sasobit Zeolite for Use in Warm-Mix Asphalt,
NCAT Report 05-06, National Center for Asphalt Technology, Auburn, AL, 2005.
Johnson, K.L., Greenwood, J.A., Poon, S.Y., A Simple Theory of Asperity Contact in
Elasto-hydro-dynamic Lubrication, Wear, Volume 19, Issue 1, January 1972, pp.
91-108.
Li, X., and Gibson, N., Mechanistic Characterization of Aggregate Packing to Assess
Gyration Levels during HMA Mix Design, Journal of the Association of Asphalt
Paving Technologists, Vol. 80, pp. 33-64, 2011.
Lu, Xiaobin, M. M. Khonsari, and E. R. M. Gelinck , The Stribeck Curve:
Experimental Results and Theoretical Prediction, J. Tribol. 128, 789,
DOI:10.1115/1.2345406, 2003.
Luengo, G., Schmitt, F.J., Hill, R., Israelachvili, J., Thin Film Rheology and Tribology
of Confined Polymer Melts: Contrasts with Bulk Properties, Macromolecules, Vol.
30, No. 8, 1997.
Macosko, C.W., Rheology - Principles, Measurements and Applications, John Wiley &
Sons, 1994.
Marker, V. Tender Mixes: The Causes and Prevention, Asphalt Institute Information
Series No. 168 (IS-168), June 1977.

54

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

Olard, F., GB5 Mix Design: High-Performance and Cost-Effective Asphalt Concretes
by Use of Gap-Graded Curves and SBS Modified Bitumens, Road Materials and
Pavement Design, Vol. 13, Suplement 1, pp. 234259, 2012.
Richardson, C., The Theory of the Perfect Sheet Asphalt Surface, Journal of
Industrial. Engineering and Chemistry, pp 463-165, 1915.
Roohi Sefidmazgi N., Tashman L., Bahia H.U., Internal Structure Characterization of
Asphalt Mixtures for Rutting Performance Using Imaging Analysis, Journal of the
Association of Asphalt Paving Technologists, pp.109138, Vol. 81, 2012.
Tashman L., Masad, E., Peterson, B., Saleh, H., Internal Structure Analysis of Asphalt
Mixes To Improve The Simulation of Superpave Gyratory Compaction to Field
Conditions, Journal of the Association of Asphalt Paving Technologists, Vol. 70, pp.
605-645, 2001.
Zhu , H., Nodes, J.E., Contact Based Analysis of Asphalt Pavement with the Effect of
Aggregate Angularity, Mechanics of Materials Vol. 32, Issue 3, pp. 193202, 2000.

9. Discussion

PROFESSOR DAVID ANDERSON: A lot of interesting concepts here, and I like the
approach and things you're doing, but I do have a couple of comments. I think we need a
more realistic understanding of what the mastic is in the mix. Its not minus 200 filler
plus asphalt binder. It probably is and if you do a little bit of looking in the literature,
you will find instances where people talk about minus 3 micron, minus 20 micron. Its
the mineral filler that floats in the asphalt binder. The minus 200 mesh doesnt do that.
The minus 200, minus 375, whatever, is part of the aggregate framework. I think you
really need to revisit your concept of the mastic and do a little bit of homework there,
and I think maybe youll make more progress towards implementing what youve put
together here.

MR. NIMA ROOHI SEFIDMAZGI: Thank you for the comment. What we tried to do
was basically the gradation design in Superpave. Thats absolutely right, and I think we
are going to probably work more on that. As I said, this is just to introduce the concept,
but basically making it more fundamental, as you said, we have to.

PROF. ANDERSON: Also, please tell us how you measure the specific surface. Thats
pretty important, and its not in the paper.

MR. SEFIDMAZGI: The specific surface of the fillers? I dont remember the name of
the method

55

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
ROOHI, TEYMOURPOUR, BAHIA

PROF. ANDERSON: Dont do it now because youve got more questions here.

MR. SEFIDMAZGI: Okay. But its based on a gas being exposed.

DR. JOHN DANGELO: Again, I would just like to say interesting concepts here, but
two items. One is you're doing all your analysis on a gyratory compactor, and that
compaction process is completely different than the roadway. In the roadway, you're
allowing dilation of the aggregate particles. Youve got very thin films, so you have
rather large reorientation of the aggregate particles during the compaction process that
you dont have in the gyratory.

You talk about the structure that we design mixes for and put up things like 7% air
voids. We dont design mixes for 7% air voids, but we do make specimens at this void
level to do performance testing. The 7% air voids is trying to simulate what is typically
achieved in the field. Theres been a lot of work that has been shown that density does
relate to performance, but they have to be done over a wider range.

MR. SEFIDMAZGI: Thank you for the comments, but I have something to say about the
gyratory compactor. Thats exactly right. I mean the mechanism is different, and were
trying actually to verify that with some failed mix samples. The point is, that we are
designing with gyratory compactors, so if it is a problem, we have to understand it and
solve it in a way finally. People are going to design using that, so although the
mechanism is different, understanding of that probably can help to some extent.

DR. DANGELO: Your own data all converges to identical performance when you get
to little bit higher gyrations and thats what we typically design to.

PROFESSOR REYNALDO ROQUE: My background formerly is in soil mechanics,


geotechnical engineering. When I look at a lot of the work that you're presenting, the
basic principles and associated mechanisms very much relate to the principles that come
out of that very large body of work that is associated with compaction of soil mixtures.
If I take a clayey gravel, for example, where the clay is representing the binder and the
gravel part is well graded and can represent the mixture, we have lots of information on
density curves and optimum moisture contents, which you can translate perhaps to
optimum binder contents. I think the use of things like dry density and porosity
parameters that are well defined and have been used extensively could be very helpful in
making this much more straightforward and clearer and I think more meaningful than
using a lot of new terminology. I was just wondering how much you have looked at that
literature compared to the tribology literature, which I am looking at myself for other
purposes, which primarily addresses flat surfaces on flat surfaces with little grainy
particles. I think thats very different than soil mechanics, which is a field that we have
been studying for decades and decades.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

56

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Particle Mobility on Aggregate Structure Formation in Asphalt Mixtures

MR. SEFIDMAZGI: Right. Thats a very good point. I honestly did some literature
review in soil mechanic compaction. I started with soil mechanic compaction and,
honestly, that was the starting point to think about it. But what happened later was I want
to understand what is happening in lower scales which basically led me toward this
tribology. But the point is that that is exactly right. I mean what we tried to do here was
proof of concept and understanding what is going on, but to make it useful, thats
absolutely right. There are a lot of concepts in soil mechanics and granular mechanics
that can be applied. The indices or characteristics that can be used are probably more
user friendly for designers or producers.

PROFESSOR CARL MONISMITH: May I suggest, if you have not done so, to look at
the work by Nijboer from 1948. Also the work of Schmidt and Santucci published by the
Highway Research Board (now the TRB) in 1962. These investigations used rolling
wheel compaction either in the laboratory or in the field. So it would be a good idea for
you to look at the results of these investigations.

N.B. A list of these publications was sent to Prof. H. Bahia following the meeting. These,
which I consider to be classics in asphalt technology, are as follows:

1. Nijboer, L.W.J., "Plasticity as a Factor in the Design of Dense Bituminous Road


Carpets", Elsevier Publishing Company, 1948. A paper in the AAPT Symposium on
HMA Compaction in 1970 by Kor Wester makes reference to Nijboer's work with an
application to pneumatic tire rolling in the Netherlands. Nijboers work was based on
steel wheel rollers.
2. Two Papers by R.J. Schmidt and L. E. Santucci, California Research Corporation
(now Chevron Research)
a. Schmidt and Santucci, "Influence of Asphalt Type on Pavement Setting Rate", Bulletin
333, Highway Research Board, 1962, p. 10.
b. Santucci and Schmidt, "Setting Rate of Asphalt Concrete", Bulletin 333, Highway
Research Board, 1962, p.1.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

(The work by Schmidt and Santucci was a very comprehensive study using controlled
aggregate gradings, a small hot mix plant, a replica of the HMA paver screed, and full
scale rollers at the site at what is now the Chevron Research Corporation in Richmond,
California.)

MR. SEFIDMAZGI: Thank you very much. Of course I will. Thank you.

57

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional
Resistance as a Function of Aggregate
Physical Properties
John E. Haddocka* and Joan P. OBrienb
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

a
School of Civil Engineering, Purdue University, West Lafayette, IN
b
Parsons Brinckerhoff, Perth, Australia

ABSTRACT: Aggregates were tested for acid insolubility, differential wear, and frictional resistance
and along with physical data were correlated to aggregate frictional resistance. Physical data
included bulk specific gravity, percent absorption, percent loss from impact, and percent loss from
brine freeze/thaw. It was observed that the aggregates used had a negligible amount of insoluble
material and therefore had an insignificant effect on the frictional resistance. Terminal frictional
resistance increased with decreasing bulk specific gravity and increased with increasing percent
absorption and percent loss from impact. Correlation between terminal frictional resistance and
brine loss was moderate. Differential wear measurements were taken on stone-slag aggregate blends
to determine the effectiveness of blends used in pavement mixtures. Differential wear results
indicated that the high friction component of each blend could be improved by blending with softer
carbonate aggregates, which also may be more cost effective.

KEYWORDS: Friction, skid resistance, polish resistance, aggregate, hot mix asphalt.

The oral presentation was made by Professor Haddock.

This is a reproduction of an article whose final and definitive form has been published in Road
Materials and Pavement Design 2013 Taylor & Francis. The article is available online at:
HTTP://DX.DOI.ORG/ 10.1080/14680629.2013.812842

59

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

1. Introduction

Pavement frictional resistance is important for the safe performance of pavements and
ranks second among pavement performance considerations in the United States and many
other countries according to a study by the National Cooperative Highway Research
Program (NCHRP) (Henry, 2000). Frictional, or skid, resistance is defined as the ability of
a traveled surface to prevent the loss of tire traction (ASTM, 2002b) and is expressed as a
friction or skid number (FN or SN), defined as the ratio between the frictional resistance
acting along the plane of sliding and the load perpendicular to this plane (Jayawickrama et
al., 1996).

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
The frictional performance of a pavement is dependent on both the micro- and macro-
texture of a pavement, which are partly controlled by the aggregates used (Jayawickrama et
al., 1996; Skerritt, 1993). On a macro scale, aggregates make up approximately 95 percent
of hot mix asphalt (HMA) by weight and therefore greatly contribute to the texture of an
HMA pavement (Henry, 2000). On a micro scale, the physical properties of the aggregate
also contribute to the texture of a pavement.
Like many highway agencies, the Indiana Department of Transportation (INDOT)
assesses frictional resistance of coarse aggregates before use in HMA pavements through a
series of laboratory tests that measure micro-texture. In 1997 INDOT adopted the
Superpave mixture design method and attendant specifications for HMA mixtures. These
specifications control coarse aggregate characteristics key to HMA performance;
angularity, flat and elongated particles, clay content, toughness, soundness, and deleterious
materials (Asphalt Institute, 1995), without specific regard to macro- and micro-texture,
and therefore may have ramifications on the frictional characteristics of an HMA
pavement. Gradation requirements for Superpave-designed HMA mixtures may also affect
pavement friction characteristics.
After several years of using the Superpave mixture design method for HMA, INDOT
decided to re-examine the methods used to screen coarse aggregates for frictional
resistance to ensure adequate pavement friction. Therefore, the objective of this study is to
determine if current tests used by INDOT adequately screen coarse aggregates for
frictional properties.
In order to achieve the objective, pavements constructed using Superpave designed
HMA wearing courses containing coarse aggregates that have a history of consistent
physical properties were identified, and coarse aggregate samples were collected from the
source quarries and tested for total insoluble residue and laboratory frictional
characteristics. Physical property data for these aggregates were also collected from the
INDOT quality assurance (QA) inventory. The QA and total insoluble residue data were

60

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

correlated with laboratory frictional resistance data of each aggregate. In-service field
friction data were also collected and compared to the laboratory results.

2. Background

Skid resistance as a factor in pavement performance began to garner attention in the early
20th century when researchers became interested and began to develop equipment to study
the pavement frictional resistance. Since a pavement can be structurally sound and still
have poor skid resistance (Dillard and Alwood, 1957), one of the early research objectives
was to increase skid resistance without affecting the structural soundness of a pavement.
Implementation of the Federal-Aid Highway Act in 1952 stressed the awareness and
importance of skid resistance partly because of the increased speeds and number of
vehicles on the road.
Skid resistance of a pavement is greatly affected by its texture. Pavement texture has
two components important to skid resistance, micro- and macro-texture. A pavement is
considered polished when it loses its micro-texture, and worn when it loses its macro-
texture. In a general sense, Webb (1970) describes micro-texture as the finer intra-particle
asperities and macro-texture as the coarser, inter-particle texture. Generally, it is believed
that micro-texture has a greater effect than macro-texture on the frictional properties of a
pavement.
According to Webbs definition of micro-texture, aggregates are the most influential
portion of pavement micro-texture. Micro-asperities on aggregate surfaces form the micro-
texture of a pavement and determine the resistance of a pavement surface to skidding
(Rogers and Gargett, 1991). The micro-asperities are a function of aggregate mineralogy
and how it reacts to climate and traffic loading (Rogers and Gargett, 1991). Polishing of an
aggregate occurs when the micro-texture is removed by vehicular traffic. The ability of an

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
aggregate to maintain micro-texture is dependent on its petrographic properties such as
mineralogy, texture and fabric (Webb, 1970), and can be related to skid resistance
performance under various traffic volumes (Gramling and Hopkins, 1974). The largest
aggregate size in the HMA mixture is thought to govern skid resistance (Skerritt, 1993).

2.1 Mineralogy

Coarse aggregates provide both micro-texture by the properties of the aggregates


themselves, such as differential hardness of the minerals and irregularity of the surface, and
macro-texture through spacing and gradation. However, mineralogy of an aggregate is
possibly the most important factor in the ability of a pavement to maintain micro-texture,

61

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

and thus skid resistance. The minerals themselves are not what provide frictional
resistance, but it is the differential hardness between minerals in a particular rock type that
offer sufficient skid resistance (Mullen et al., 1974). Although harder minerals are
preferable, even the hardest rock types will polish if a differential hardness between
minerals does not exist (Dahir and Mullen, 1971). It is the difference in polishing rates
between minerals with different hardness that gives aggregates a rough, skid resistant
texture (Furbush and Styers 1972). However, a harder mineral will provide a longer rate of
polishing (Webb, 1970).
Limestone and dolomite tend to have a fairly uniform texture with low hardness, which
makes these rock types susceptible to rapid polishing. In a study by Burnett, et al. (1968),
limestone and dolomite had a coefficient of friction below 0.32 after 5 million vehicular
passes, but some retained coefficients high enough to remain skid resistant after ten million
vehicular passes. It was observed that the limestone and dolomite aggregates that had
quartz grains within the matrix had a sand paper-like texture and higher skid resistance.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Therefore, it is possible to have a skid resistant limestone or dolomite aggregate provided
some percentage of the constituents consists of harder minerals.
West and Cho (2000) indicate that skid resistance of a pavement with this aggregate
type is partly dependent on the impurity of the limestone or dolomite aggregate, which can
be determined by elemental magnesium content, specific gravity, and total acid insoluble
residue. Dolomite is considered more impure with lower elemental magnesium content and
higher total insoluble residue, and limestone is more impure with both higher elemental
magnesium content and total insoluble residue (West and Cho, 2000). West et al. (2001)
emphasize that this relationship can be seen in the specific gravity of the aggregate. A
dolomite mineral has a specific gravity of 2.800 to 2.900 (Chesterman, 1979). It is
speculated that an aggregate of the dolomite rock type will have a higher frictional
resistance with specific gravities further away from the 2.800 to 2.900 value (West et al.,
2001).
Consequently, for limestone and dolomite aggregates, testing for insoluble residue is
important. Limestone and dolomite are made up of mostly calcium, magnesium carbonate
or both, which react with dilute hydrochloric acid. In the insoluble residue test, the
aggregate is dissolved in dilute hydrochloric acid, and the remaining insoluble portion is
usually made up of quartz (silica), feldspar (clays), or other insoluble minerals. As
discussed previously, the presence of quartz or other insoluble minerals in a limestone or
dolomite is significant in giving aggregates of this rock type sufficient skid resistance.
Gray and Renninger (1966) were the first to report that insoluble residue correlated
with an aggregates skid resistance. Burnett, et al. (1968), Furbush and Styers (1972),
Skerritt (1993) and West and Cho (2000) later confirmed this concept, which is now
widely accepted by most researchers. However, there are conflicting reports on the amount

62

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

or the size of insoluble residue that is needed in order to retain sufficient skid resistance.
Gray and Renninger (1966) recommend at least ten percent insoluble residue; Furbush and
Styers (1972) and Kearney et al. (1972) recommend at least twenty percent. Kearney et al.
(1972) also propose that at least ten percent sand-sized impurities are needed instead of a
total percentage. Skerritt (1993) contends that dolomites with greater than fifteen percent
total acid insoluble residue consistently produce friction numbers above forty even at high
accumulated traffic. It is unclear which theory is more correct, but it appears to be widely
accepted that the grain size of the remaining insoluble residue should at least be retained on
the 0.075 mm (No. 200) sieve (Furbush and Styers, 1972; Gray and Renninger, 1966;
Kearney et al., 1972).
In a study by West and Cho (2000), it was observed that the percent of insoluble
residue smaller than the 0.075 mm (No. 200) sieve increased frictional resistance. The
authors speculated that the tiny clay particles that make up the portion smaller than the
0.075 mm (No. 200) sieve break out of the carbonate matrix creating an irregular surface
and providing the needed micro-texture for good skid resistance (West and Cho, 2000).
West and Cho also report that although the total percentage of acid insolubility correlates
well with terminal polish value (BPN10) and wear index (WI), this correlation is stronger
with limestone aggregates than dolomite aggregates. Wear index is the difference between
the initial friction value (BPN0) and the final polish value (WI = BPN0 BPN10), and is
indicative of polish resistance (West et al., 2001).
Additionally, Abdul-Malak et al. (1993) identified that the presence of dolomite in

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
carbonates provides a higher level of skid resistance. This may be because dolomite is a
harder mineral than calcite. Calcite is the most common mineral found in carbonate rocks
(Chesterman, 1979). Dolomite and calcite have a similar chemical composition, but
dolomite includes magnesium [CaCO3 and Ca(Mg)CO3, respectively]. INDOT specifies
dolomite with a minimum of 10.3% elemental magnesium for all pavements with medium
to high equivalent single axle load (ESAL) counts (INDOT, 2012). The dolomite aggregate
is considered more pure as the percent elemental magnesium increases. As indicated earlier
by West et al. (2001), impure limestone and dolomite aggregates have better skid
resistance. This is most likely partly due to the differential in hardness between the calcite
and dolomite in such carbonate aggregates, in addition to the insoluble portion.

2.2 Aggregate Blending

Blending low skid resistant aggregate with high skid resistant aggregate is common
practice in states where aggregates with inherently low skid resistance are locally available.
Transportation of aggregates is costly and therefore local aggregates are most commonly
used. Steel slag and air-cooled blast furnace slag are commonly blended with dolomites in

63

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

Indiana to improve HMA pavement skid resistance. Although the terminal polish value of
slag varies, the skid resistance of a pavement constructed with slag remains relatively
constant (Stock et al., 1996). Optimum blending consists of finding an effective blend of
low and high skid resistant aggregates that provides for safer roads and is also cost
effective (Fu, 2004).
Some studies suggest that the frictional resistance of a blend is directly proportional to
blend percentages (Gramling and Hopkins, 1974; Liang, 2003). Therefore, an increased
terminal polishing value can be obtained by increasing the percentage of aggregates with
higher frictional resistance and lower polish susceptibility. However, it is recommended
that a blend be subjected to polish testing to determine whether or not the terminal polish
value is at an acceptable level according to traffic or ESAL count (Liang, 2003).
According to Fu (2004), differential wear (DW) is an important factor in determining
whether a blend will perform well in the field and whether it will be cost effective.
Differential wear is the difference in particle height measurement between the low and
high friction component before and after polishing (Fu, 2004). These values are obtained
by taking measurements with a height-measuring device before and after polishing tests.
Higher DW values maximize both micro and macro-texture and are indicative of blends
that will perform well in the field (Fu, 2004). In other words, if a blend consists of two
aggregates with similar terminal polish values, the differential wear value will be low and
subsequently the terminal polish value of the blend will be lower. Therefore, it is expected
that it will not effectively enhance the skid resistance of the pavement nor will it be a cost
effective blend.

3. Study Methodology

As shown in Table 1, INDOT divides HMA aggregate frictional requirements by ESAL


categories. The experimental methodology consisted of six steps:
1. Identify consistent aggregate sources;
2. Identify pavements with Superpave-designed HMA surface courses containing
the aggregates identified in Step 1;
3. Collect the friction data from the pavements identified in Step 2;
4. Collect samples and quality assurance (QA) data from the aggregate sources
identified in Step 1;
5. Complete laboratory testing of the aggregate samples; and
6. Analyze the data.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

64

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

Table 1. INDOT wearing course aggregate specification (INDOT 2012).


Equivalent Single Axle Load
Coarse Aggregate Type < 3,000,000 < 10,000,000 10,000,000
Air-Cooled Blast
Yes Yes Yes
Furnace Slag
Steel Furnace Slag Yes Yes Yes
Sandstone Yes Yes Yes
Crushed Dolomite Yes Yes Note 1
Polish Resistant
Yes Yes Note 1
Aggregates
Crushed Stone Yes No No
Gravel Yes No No
NOTE: Polish resistant aggregates or crushed dolomite may be used when blended with
ACBF or sandstone but cannot exceed 50% of the coarse aggregate by mass (weight), or
cannot exceed 40% of the coarse aggregate by mass (weight) when blended with steel
furnace slag.

3.1 Identification of Aggregate Sources

The stone aggregate sources were selected through reviewing laboratory quality assurance
data conducted by INDOT and also by historical skid performance of Indiana pavements.
The physical property data were assessed for aggregate consistency and competency.
Reviewed test data included Los Angeles abrasion, brine freeze/thaw soundness,
absorption, elemental magnesium, and specific gravity. The purpose of identifying
aggregates having consistent physical properties and performance over many years was to
attempt to control aggregates as a variable. The list of selected aggregates and their
corresponding physical property data are presented in Table 2. All aggregates are dolomite
per Indiana Test Method (ITM) 205, Acceptance Procedures for Dolomite Aggregates
and have been approved by INDOT for use in medium and high ESAL count pavements
per ITM 214, Acceptance Procedures for Polish Resistant Aggregates. The D-prefix for
each source represents dolomite.
As seen in Table 1, pavements with dolomite or crushed stone aggregates that are in the
equal to or greater than ten million ESAL category require blending with slag or sandstone.
Pavements used in this study that contain aggregate blends were blended with slag. The
slag sources are presented in Table 3 and are represented by an S-prefix. The steel slag
specific gravity was not available, but typical values are approximately 3.000. Blends are
denoted by concatenation of the stone and slag aggregate codes that are listed in Tables 2
and 3.

65
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

Table 2. Stone aggregate source list.


LA Freeze/
Bulk Elemental Abrasion Brine Thaw
Source Specific Magnesium Absorption Loss Loss Loss
ID Gravity (%) (%) (%) (%) (%)
D1 2.749 12.1 1.03 23.34 0.01 -
D2 2.442 11.1 4.53 30.74 11.11 -
D3 2.727 12.5 1.19 19.25 0.36 -
D4 2.554 10.8 3.07 28.54 N/A 1.76
D5 2.489 11.5 3.3 26.98 0.33 -
D6 2.658 10.7 1.89 23.14 26.48 1.08
NOTE: Test procedures are further discussed in subsequent sections.

Table 3. Slag aggregate source list.


Source ID Slag Type Specific Gravity
S1 Steel N/A
S2 Blast Furnace 2.463
S3 Blast Furnace 2.365
S4 Blast Furnace 2.510
NOTE: N/A indicates not available.

3.2 Identification of Pavement Sections

Once consistent aggregate sources were selected, a search was conducted that identified
projects or contracts that placed pavements using the consistent aggregate sources with a
Superpave design mixture formula (DMF). Each district within INDOT was given the list
of consistent aggregate sources. The INDOT district personnel conducted a search of their
records where DMF for contracts that used the selected aggregate sources were obtained.
Since the Superpave mixture design method was not implemented until 1997, all
pavements used for this research were placed after that date.

3.3 Collection of Friction Data

To collect and review the skid data from the INDOT friction inventory database, the road
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

name and mile marker for each contract were obtained from the DMF. If either piece of
information was missing from the DMF, the data were obtained from the INDOT contract
(project) database. Using the road name and mile marker for each contract, a search was
conducted in the INDOT friction inventory database to obtain skid numbers for the
corresponding mile markers or reference posts.

66

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

Upon review of skid data, it was noticed that there were gaps in skid data resulting in
the narrowing of the list of selected pavements. Pavements selected for this study are listed
by contract number and are presented in Table 4. It was not possible to use all pavements
on this list for all analyses.
Skid testing of pavements is independent of the age of the pavement. In other words, if
a pavement on a state road was placed in October of 1997, the first skid resistance testing
could be as early as April of 1998 or as late as September 2000. Since secondary roads are
skid tested once every three years, the amount of data available was limited for these types
of pavements. A final review resulted in 32 pavement sections that were identified for use
in the research. Road numbers are prefixed with an I-, US-, or SR-, indicating an interstate,
federal, or state road, respectively.

3.4 Collection of Aggregate Samples and Quality Assurance Data


--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Once pavements and aggregate sources were selected, the aggregates used in the
pavements were determined and samples for testing were obtained from each quarry. To
initiate collection of samples, each stone aggregate source was contacted and a gradation
was obtained. The aggregate gradation was used to determine the approximate quantity
needed from each source. Subsequently, each of the six selected stone quarries was visited,
and samples were collected. Samples were obtained in accordance with ITM 207,
Sampling Stockpiled Aggregates.
Aggregate producers in Indiana have the option of producing aggregates in accordance
with the INDOT Certified Aggregate Producers Program (CAPP). By doing so an
aggregate producer becomes certified and thus may market aggregate for INDOT
projects without the need for INDOT to individually test and certify each stockpile. As part
of the CAPP program, the aggregate producer generates quality control data while quality
assurance data are generated by INDOT. These data include AASHTO T 96, Resistance
to Degradation of Small-Size Coarse Aggregate by Abrasion and Impact in the Los
Angeles Machine, AASHTO T 85, Specific Gravity and Absorption, and ITM 209,
Soundness of Aggregates by Freezing and Thawing in a Brine Solution. Aggregates that
do not meet the INDOT specifications for ITM 209 are also subjected to AASHTO T 103,
Test for Soundness of Aggregates by Freezing and Thawing, Procedure A. INDOT
considers AASHTO T 103 to be the most accurate method, but the test has a long duration.
Therefore, ITM 209 is used as a quick check of the aggregate. If the aggregate passes the
ITM 209 method it is deemed acceptable. If it fails the method, but passes AASHTO T
103, it is acceptable for use. For this project, the INDOT QA data was gathered for each of
the aggregate sources identified.

67

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

Table 4. Pavement list.


Aggregate
Contract Number Road Source ID ESAL ADT
R-23365 US-31 D5S1 12,000,000 35,000
R-23800 I-265 D6S2 11,000,000 40155
R-23924-A SR-46 D1S1 3,700,000 16240
R-24551 I-65 D6S2 25,000,000 30000
R-24555 I-65 D6S2 44,000,000 114000
R-24561 I-65 D6S2 88,000,000 44390
R-24847 I-65 D6S2 53,000,000 45000
R-24947B US-35 D3S3 5,200,000 10000
RS-23321 US-31 D5S1 3,500,000 8000
RS-23321 US-31 D5S1 3,500,000 17000
RS-23492 SR-44 D5 1,200,000 8000
RS-23600 US-24 D2S4 3,200,000 4870
RS-23611 SR-135 D5 1,400,000 10000
RS-23611 SR-135 D5 1,400,000 19000
RS-23787 US-31 D6S2 11,000,000 5000
RS-23790 US-150 D6S2 3,500,000 12000
RS-23792 SR-60 D6 2,000,000 10000
RS-23793 SR-46 D5 3,500,000 8000
RS-23794 SR-46 D5S1 2,500,000 13500
RS-23794 SR-46 D5S1 2,500,000 5000
RS-24107 SR-56 D6 2,500,000 9000
RS-24107 SR-56 D6 2,500,000 3000
RS-24108 SR-11 D5 3,000,000 3500
RS-24109 SR-45 D6 2,000,000 10000
RS-24109 SR-45 D6 2,000,000 4000
RS-24209 SR-129 D4 2,000,000 7000
RS-24256 SR-46 D4 2,000,000 4000
RS-24389-A SR-46 D5S1 7,500,000 12000
RS-24534-A SR-37 D6S1 7,500,000 22000
RS-24536-A SR-43 D1 2,000,000 2000
RS-24540-A SR-62 D6S1 7,500,000 9000
RS-24769-A SR-135 D6 2,000,000 7000
RS-24771-A US-150 D6 2,000,000 5000
RS-24823 SR-135 D5 500,000 800
RS-24832-A SR-45 D1 2,000,000 1000
RS-25200-A US-35 D3S3 4,700,000 17000
RS-25251 SR-3 D4 2,000,000 7000

68

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

High volume or high ESAL count pavements require an aggregate blend of both
crushed stone and slag aggregate. Therefore, steel and blast furnace slag aggregate samples
were obtained and tested along with the stone aggregate samples.

3.5 Laboratory Testing of Aggregate

Laboratory testing was performed on each slag, stone aggregate, and aggregate blend. Each
aggregate and aggregate blend was subjected to polishing using the British polishing wheel
for ten hours. Initial and terminal polish values were taken using the British Pendulum
Tester in accordance with ITM 214 and Texas test method TEX-438-A, Accelerated
Polish Test for Coarse Aggregates for comparison purposes. Stone aggregates were also
tested for total acid insoluble residue.

4. Experimental Procedures

4.1 Insoluble Residue in Carbonate Aggregates

Testing for total acid insoluble residue in carbonate aggregates was performed in
accordance with ASTM D3042, Standard Test Method for Insoluble Residue in
Carbonate Aggregates. The purpose of the test is to determine the percentage of non-
carbonate material in carbonate rock types and the gradation of insoluble material.
Hydrochloric acid is used to dissolve the carbonate constituents, and the resulting data are
indicators of the differential hardness in carbonate aggregates.
Hydrochloric acid was added to a sample of approximately 500 g (1.1 lb) of aggregate
and the sample agitated until effervescence ceases. The solution was then decanted and 300
mL (1.3 oz) of hydrochloric acid added and again examined for additional reaction. When
a reaction was no longer observed, the sample was heated to approximately 110C (230F)
for one hour to completely react the carbonate portion. If a reaction occurred upon heating,
the procedure was repeated, otherwise the solution was decanted and water added to dilute
the solution until a pH of more than 5.5 was reached.
The sample was next thoroughly washed over a series of nested sieves, ranging in
opening size from 0.075 mm (No. 200) to 4.75 mm (No. 4), and a receiving tank. The
sieves and residue were dried at 110C (230F) and the mass difference between the dry
sieves and the dry sieves plus the residue determined. The decanted solution in the
receiving tank was then poured through a 0.075 mm (No. 200) sieve and filtering paper.
When filtration was complete, the remaining residue was rinsed with clean water, and the
filter paper and residue transferred to an evaporating dish and dried along with the 0.075
mm (No. 200) sieve and residue at 110C (230F). The mass of the remaining residue was

69
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

determined, the mass of the residue from each sieve and the filtering paper summed, and
the percentage of insoluble residue calculated.

4.2 British Pendulum Tester

The test method for measuring surface frictional properties determines the effects of
aggregate polishing by tire action. According to Kummer and Meyer (1962), the British
Pendulum Tester (BPT) (Figure 1) is a simple machine that provides highly consistent
readings. It works by releasing a pendulum with uniform force by gravity from a given
height. A rubber slider is attached to the end of the pendulum, which comes in contact with
the pavement or test specimen surface upon release. When the pendulum is released and
swings down making contact with the pavement or specimen surface, it pushes a pointer up
along a calibrated measuring device and leaves it at the highest point reached by the
pendulum. The less friction that is encountered by the rubber slider, the higher the
pendulum will reach on the calibrated dial resulting in a lower value. One main advantage
is its insensitivity to operator effects (Dillard and Mahone, 1963; Giles et al., 1962).

Figure 1. British Pendulum Tester (ASTM, 2002a).

The BPT can be used to take initial, intermediate, or terminal polishing readings.
Readings are reported as British polishing numbers (BPN) or as polished stone value (PSV
or PV). Initial readings are reported before a specimen undergoes polishing in a polishing
machine and are designated with a zero subscript (BPN0). Interval readings are taken at
given times during the polishing process and are designated with a subscript of the number

70

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

of hours of polishing that the specimen has undergone. The physical meaning of a BPN or
PSV value is the coefficient of friction multiplied by 100. Values in this study are reported
as BPN0 for initial readings before polishing, and BPN10 for terminal readings after ten
hours of polishing.
Testing was performed in accordance with ASTM E303, Standard Test Method for
Measuring Surface Frictional Properties Using the British Pendulum Tester, ITM 214,
and TEX-438-A. The difference between ITM 214 and TEX-438-A is the stiffness of the
rubber slider. ITM 214 uses the ASTM E303 specified rubber slider, which is less stiff
than the TEX-438-A rubber slider. Since testing was conducted on samples prepared for
and subjected to accelerated polishing, the test specimens were curved to fit the British
polishing wheel. Five test specimens were prepared and tested for each crushed stone, slag,
and stone-slag aggregate blend.
Each test specimen was prepared in a mold and had a test surface of 44 by 89 mm (1.76
by 3.56 in.) and an arc to fit a 406 mm (16.2 in.) circle. Coarse aggregate passing the 9.5
mm (3/8 in.) sieve and retained on the 4.75 mm (No. 4) sieve was tested. This grain size
was in accordance with the INDOT Standard Specification. The aggregate was thoroughly
washed and dried to a constant mass at a temperature of approximately 105C (221F).
The mold was coated with a release agent, followed by a bonding agent. A single layer of
densely packed aggregate was placed by hand into the mold and the interstices between the
aggregates filled with sand from one-fourth to one-half the aggregate depth. After the
bonding agent cured, the specimen was removed from the mold. A test coupon and mold
are shown in Figure 2.

Figure 2. British Polishing Wheel test coupon and mold (Liang, 2003).

Particle arrangement has been identified as affecting polish values and splitting the
sample is recommended over handpicking each aggregate (Won and Fu, 1996). For this
study, the split method was used for coupons containing aggregate blends.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

71

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

Before taking the initial friction value (BPN0), the pendulum was leveled, the drag
pointer zeroed, and the slide length adjusted. The test specimen was cleaned to be free of
loose particles and was held firmly in place in order to resist movement by the force of the
pendulum. Water was applied to thoroughly coat the test surface and one test swing
performed without recording the reading. Additional test swings were continued and
recorded until four consecutive identical readings are observed, as recommended by Won
and Fu (1996). The test surface was rewetted with each swing, and the results recorded.
Since frictional resistance of most dry surfaces is sufficient for safe driving conditions,
the BPT is performed with wetted specimens. The BPT frictional resistance therefore more
closely resembles the frictional resistance measured in the field by the full-scale, locked-
wheel method using a ribbed tire rather than a smooth tire (Giles et al., 1962). The width of
the rubber slider acts like the grooving of the ribbed tire allowing water to escape around
the slider when it comes in contact with the specimen. Thus, the BPT is a good measure of
micro-texture rather than macro-texture (Henry et al., 2000).

4.3 British Polishing Wheel

The British Polishing Wheel (Figure 3) was mounted onto the British Polishing
Machine (Figure 4) and used to simulate polishing of coarse aggregate in HMA pavements
by vehicular traffic according to ASTM D3319, Standard Practice for the Accelerated
Polishing of Aggregates Using the British Polishing Wheel. Fourteen specimens, prepared
as described in the previous section, were clamped along the outside edge of a wheel to
create a continuous surface of aggregate. The wheel was then mounted onto the machine
and rotated at a speed of approximately 320 revolutions per minute. A smooth surfaced,
pneumatic, rubber-tired wheel was brought into contact with the surfaces of the specimens
in order to apply a load of 390 N (87.75 lbf). Number 150 silicon carbide grit was fed onto
the specimen surfaces at a rate of 6 grams (0.21 oz) per minute for the entire testing time.
Water was also fed onto the surface at a rate of about 50 to 75 milliliters (0.21 to 0.32 fluid
oz) per minute.

Figure 3. British Polishing Wheel (Liang, 2003).

72
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

Figure 4. British Polishing Machine (Liang, 2003).


In the tests reported herein, after ten hours of polishing, the wheel was removed and
cleaned to remove the silicon carbide grit. Two BPN10 polish values were then determined
in accordance with test methods ITM 214 and TEX-438-A.

5. Results and Analyses

5.1 Insolubility of Carbonate Aggregates


Acid insolubility test results are presented in Table 5. The results reveal that the stone
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

aggregates used in this study contain a negligible amount of insoluble material, and
therefore the insoluble portion would have an insignificant effect on the frictional
resistance of the aggregate. As previously discussed, the general consensus of the amount
of total insoluble residue needed to affect the frictional resistance of an aggregate is about
ten to twenty percent. The total acid insoluble residue for all stone aggregates in this study
ranged from none to 2.8%, far below this value range.
Gradations of the insoluble portions of the aggregates were also performed and are
included in Table 5. Less than one percent of the total insoluble residue for all stone
aggregates was smaller than the 0.075 mm (No. 200) sieve, which is silt and clay-sized
particles. Aggregate D3 did not contain any insoluble materials. Aggregate D4 had the
most total insoluble residue at 2.8 percent. Of the 2.8% insolubles in aggregate D4, 97.9%
was sand-sized particles, which are generally considered to provide increased frictional
resistance. The remaining four aggregates, D1, D2, D5 and D6, contained less than one
percent total insoluble residue.

73

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

Table 5. Acid insoluble residue test results for stone aggregates.


Source ID D1 D2 D3 D4 D5 D6
Sieve Size Cumulative Percent Retained
No. 4 80.0 87.0 - 97.9 42.8 0.0
No. 8 80.0 93.5 - 99.3 54.2 33.3
No. 16 80.0 95.7 - 100.0 59.9 83.3
No. 30 80.0 95.7 - 100.0 71.3 83.3
No. 50 80.0 97.9 - 100.0 82.7 83.3
No. 100 100.0 97.9 - 100.0 88.4 100.0
No. 200 100.0 100 - 100.0 99.9 100.0
Total Acid
Insoluble 0.1 0.9 0.0 2.8 0.7 0.1
Residue %

5.2 Laboratory Frictional Resistance

British pendulum data are presented in Table 6. The BPNavg values are representative of the
average of the five coupon specimens prepared for each aggregate and aggregate blend.
The hardness, frictional resistance, and polishing resistance of each aggregate are given in
Table 7. BPN values are indicative of the frictional resistance of an aggregate before and
after polishing. Higher BPN0 and BPN10 values represent higher frictional resistance. The
difference in BPN0 and BPN10 is indicative of the resistance to polishing of the aggregate,
and is known as the wear index (WI). Smaller differences in BPN0 and BPN10 correspond
to a greater resistance to polishing.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

For the stone aggregates, aggregate D2 had the highest initial and terminal BPN values,
and D3 had the lowest initial and terminal BPN values. Both aggregates showed sufficient
field skid resistance. However, data available for aggregate D3 only included HMA
wearing courses that are one to two years in age, which is when frictional resistance tends
to be higher. Aggregate D6 showed the greatest loss in frictional resistance after polishing,
but still retained a relatively good level of frictional resistance, which was also seen in the
field data. Although aggregate D3 had the lowest frictional resistance of all the stone
aggregates, it appeared to retain a similar level in frictional resistance after polishing using
the TEX-438-A rubber slider.

74

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

Table 6. Polishing test results.


ITM 214 TEX-438-A BPN0-BPN10
Aggregate
BPN0avg BPN10avg BPN0avg BPN10avg WIITM 214 WITEX-438-A

Stone Aggregates
D1 38 30 50 31 8 19
D2 47 36 52 38 11 14
D3 36 25 33 29 11 4
D4 45 33 48 37 12 11
D5 44 34 46 37 10 9
D6 48 33 55 35 15 20
Slag Aggregates
S1 40 34 46 40 6 6
S2 46 33 50 38 13 12
S3 42 35 51 39 7 12
S4 48 33 53 39 15 14
Stone/Slag Blends
D5S1 40 35 44 38 5 6
D1S1 38 34 43 39 4 4
D6S1 40 36 45 41 4 4
D6S2 41 37 45 40 4 5
D3S3 39 30 42 32 9 10
D2S4 46 40 51 45 6 6

Table 7. Summary of aggregate hardness, frictional and polishing resistance.

Frictional
Aggregate Hardness Polishing Resistance
Resistance
Stone Aggregates
D1 Moderate Low to Moderate Low to Moderate
D2 Hard High Moderate
D3 Moderate Low Low to Moderate
D4 Hard High Low to Moderate
D5 Hard High Low to Moderate
D6 Moderate to Hard Moderate to High Moderate --`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Slag Aggregates
S1 Hard High High
S2 N/A High Moderate
S3 N/A High Moderate
S4 N/A High Moderate

75

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

A two-way analysis of variance (ANOVA) for the stone aggregate BPN10 data
indicates there are statistically significant differences (=0.05) between the mean BPN10
values for the six stone aggregates and between the two test methods. The interaction
between aggregate type and test method is not statistically significant.
Given the indicated differences between the mean BPN10 values for the six stone
aggregates, the Tukey HSD test for multiple mean comparisons was used to determine
which mean pairs are considered statistically equivalent and which are not (=0.05). The
method indicates the mean pairs of D2-D4, D2-D5, D4-D5, D4-D6, and D5-D6 are
considered statistically equivalent. All other mean comparisons are considered statistically
different.
All slag aggregates showed a high degree of frictional resistance before and after
polishing, but the blast furnace slag aggregates (S2, S3 and S4) appear to have higher
initial frictional resistance. Although aggregate S1, a steel slag, had the lowest initial
frictional resistance of all the slag aggregates, it showed the smallest loss in frictional
resistance after polishing with a difference in initial and terminal BPN values of only six
units. Therefore, S1 is considered highly polish resistant. The difference in initial and
terminal BPN values for the blast furnace slag aggregates indicates that these aggregates
are moderately polish resistant. However, terminal (BPN10) values for blast furnace slag
indicate a relatively good terminal frictional resistance.
A two-way ANOVA for the slag aggregate BPN10 data indicates there are no
statistically significant differences (=0.05) between the mean BPN10 values for the four
slag aggregates. However, as with the stone aggregates, there is again a significant
difference between the two test methods. The interaction between aggregate type and test
method is not statistically significant.
The BPN10 data for aggregate blends (Table 6) was also subjected to a two-way
ANOVA. The result indicates there are statistically significant differences (=0.05)
between the mean BPN10 values for the six blends and between the two test methods. The
interaction between aggregate blend and test method is not statistically significant. Again,
the Tukey HSD test for multiple mean comparisons was used to determine which mean
pairs are considered significantly equivalent and which are not (=0.05). The method
indicates that the means of blends D1S1-D5S1, D1S1-D6S1, D1S1-D6S2, D5S1-D6S1,
D5S1-D6S2, and D6S1-D6S2 are statistically equivalent. All other mean comparisons are
considered significantly different.
While initial, terminal, and difference in BPN values for aggregate blends were
presented in Table 6, Table 8 is a summary of the differences in terminal BPN10 values of
the stone and slag components of each blend. This table illustrates the changes in frictional
resistance of the stone and slag aggregates from blending. Terminal BPN10 values for the
stone and slag aggregate are subtracted from the terminal BPN10 value of the blend.

76
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

Positive differences indicate an increase in frictional resistance, whereas negative


differences indicate a loss. Table 9 summarizes the rating of effectiveness of the blend
along with additional comments.

Table 8. Summary of differential wear measurements and differences in BPN10 values


between blends and each blend component.
BPN10 (ITM 214) BPN10 (TEX 438-A) DW
Stone Blen Slag B- Stone Blen Slag B- x0.001
Blend (A) d (B) (C) B-A C (A) d (B) (C) A B-C (in.)
D1S1 30 34 34 +4 0 31 39 40 +8 -1 0.3
D2S4 36 40 33 +4 +7 38 45 39 +7 +6 -0.1
D3S3 25 30 35 +5 -5 29 32 39 +3 -7 -0.2
D5S1 34 35 34 +1 +1 37 38 40 +1 -2 -0.2
D6S1 33 36 34 +3 +2 35 41 40 +6 +1 0.5
D6S2 33 37 33 +4 +4 35 40 37 +5 +3 0

All terminal BPN values for stone aggregates were improved by the blend. Although an
improvement of the stone aggregate terminal BPN value is an objective of blending, highly
effective blends improve the frictional resistance of both components. Therefore, there
could be a more effective blend that is also more cost effective, especially for blend D3S3.
Differential wear (DW) measurements are also presented in Table 8. DW is the
difference in particle height measurements of the low and high friction components. Larger
DW measurements indicate a compatible and effective blend. Small DW measurements do
not necessarily mean the blend does not provide sufficient skid resistance, but it does
indicate that the blend could be improved with some alternative that also could be cost
effective. It is suspected that DW measurements are low because the stone aggregates used
in this study are considered skid resistant per ITM 205, and therefore have terminal polish
values that are too similar with the high friction component. All blends in this study could
be improved by blending the high friction component with a softer, carbonate aggregate.

77
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

Table 9. Summary of improvement in frictional resistance.


Improvement
in Frictional
Blend Resistance Comments
D1S1 effective Although the DW is negligible, the frictional resistance
of the blend is slightly higher than that of the high
friction component (i.e. the slag). Blend could be
improved by use of a softer carbonate.
D2S4 highly Frictional resistance of the blend significantly exceeded
effective that of the high friction component even though the
DW indicates equal wear rates of the two components.
Frictional resistance could be further improved by
blending either component with a softer carbonate
source.
D3S3 effective Although the frictional resistance was improved, it still
is not sufficient for use in high traffic roads. DW of -
0.0002 indicates the aggregates are incompatible and
have equal wear resistance.
D5S1 marginal Marginal improvement due to similar frictional
characteristics between the slag and stone aggregate.
Frictional resistance could be improved by blending
either component with an aggregate with lesser wear
resistance.
D6S1 effective Although DW is negligible, frictional resistance was
increased from the dolomite low frictional resistance of
the dolomite to the slag moderate frictional resistance.

D6S2 highly The friction level of the blend significantly exceeded


effective that of the high friction component. The frictional
resistance could be further increase with the use of a
softer carbonate. The DW value of 0 indicates the equal
wear rates of the two aggregates.

5.3 Comparison of Terminal BPN and Aggregate Physical Properties


BPN10 values from both test methods (ITM 214 and TEX-438-A) were plotted as a
function of bulk specific gravity, elemental magnesium, absorption, LA abrasion loss,
brine loss, and total acid insolubility results for each stone aggregate.
Figure 5 shows the relationship between bulk specific gravity and terminal BPN10
values with the ITM 214 and TEX-438-A test methods, respectively. The relationship is
moderate to strong with R2 values of 0.68 and 0.83 for the ITM 214 and TEX-438-A test

78
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

methods, respectively. The Standard Error of the Estimate (SEE), a measure of the
accuracy of the estimates from the linear regression models, are 2.46 and 1.67 for the ITM
214 and TEX-438-A test methods, respectively. The plot demonstrates that the terminal
BPN10 value increases with decreasing bulk specific gravity. Review of the literature
revealed that the impurity of a dolomite aggregate is related to the frictional resistance,
which can be determined by the specific gravity, elemental magnesium and total acid
insoluble residue. As stated previously, the specific gravity of the dolomite mineral is
approximately 2.800 to 2.900. Impurity of the dolomite aggregate increases as the
difference between the specific gravity of the dolomite aggregate and of the dolomite
mineral increases. Figure 5 supports this claim.

50

45
y = -26.37x + 103.13
R = 0.83
Average BPN10

40

35
ITM 214
30 TEX-438-A
y = -25x + 96.91
25 R = 0.68

20
2.2 2.4 2.6 2.8
Bulk Specific Gravity

Figure 5. Relationship between BPN10 and bulk specific gravity of stone aggregates.

The literature review also revealed that the purity of a dolomite is reflected in higher
percent elemental magnesium. Figure 6 shows the relationship between percent elemental
magnesium and terminal BPN10 values with the ITM 214 and TEX-438-A test methods,
respectively. The relationship is excellent with R2 values of 0.97 and 0.91 for the ITM 214
and TEX-438-A test methods, respectively; SEE for the ITM 214 is 0.76% while the SEE
for the TEX-438-A is 1.31% . The maximum BPN10 value occurs at 11.3% and 11.2% for
the ITM and TEX-438-A methods, respectively. The current INDOT minimum allowable
value of 10.3% elemental magnesium for dolomite aggregates results in a slightly lower

79
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

BPN10 value. Also, increasing the percent elemental magnesium beyond approximately
11.2% does not necessarily increase the frictional resistance of the aggregate.

40
y = -5.03x2 + 112.21x - 589.02
R = 0.91
35
Average BPN10

30
ITM 214

25
y = -6.48x2 + 145.81x - 785.18
R = 0.97
20
10 11 12 13 14
Elemental Magnesium, %

Figure 6. Relationship between BPN10 and percent elemental magnesium of stone


aggregates.

50

45
y = 2.42x + 28.45
40 R = 0.81
Average BPN10

35 ITM 214
TEX-438-A
30
y = 2.33x + 25.99
R = 0.68
25

20
0 2 4 6
Absorption, %

Figure 7. Relationship between BPN10 and percent absorption of stone aggregates.

80
--`,`,,,``,,```,,`,,,,,`

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

The relationships between terminal BPN using both test methods (ITM 214 and TEX-
438-A) and percent absorption of the stone aggregate are illustrated in Figure 7. The plot
shows that terminal frictional resistance increases with increasing percent absorption. The
R2 values are 0.68 and 0.81 for ITM 214 and TEX-438-A test methods, respectively; SEE
values for the regression models are 2.45% and 1.80% for ITM 214 and TEX-438-A test
methods, respectively. An aggregate with a higher percent absorption may have a higher
porosity (Lee et al., 1990), which may provide an irregular surface. As the aggregate
surface is worn down, the irregular surface is renewed providing greater frictional
resistance. This finding is consistent with the percent absorption and specific gravity
relationship. Percent absorption increases with decreasing specific gravity, which indicates
that there is a loss of density with increasing porosity.
Figure 8 is a plot of BPN10 as a function of percent LA abrasion loss for both test
methods. The relationship between terminal frictional resistance and results from the LA
Abrasion test is strong with R2 values of 0.80 (SEE = 1.95%) and 0.83 (SEE = 1.67%) for
ITM 214 and TEX-438-A test methods, respectively. The plot shows that terminal
frictional resistance of the stone aggregate increases with increasing LA abrasion loss.
Blast furnace slag provides good frictional resistance by renewal of the surface with
abrasion; however, blast furnace slags tends to fail requirements for the LA abrasion loss
test and therefore are not tested. For the same reason, frictional resistance for stone
aggregates increases with a higher percent LA abrasion loss.

50

45
y = 0.80x + 14.24
R = 0.83
40
Average BPN10

35 ITM 214
TEX-438-A
30
y = 0.82x + 11.0
25 R = 0.80

20
15 20 25 30 35 40
Los Angeles Abrasion Loss, %

Figure 8. Relationship between BPN10 and Los Angeles Abrasion loss of stone aggregates.

81
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

The relationship between terminal BPN10 and brine loss and is shown in Figure 9.
Values of R2 are nearly identical for the ITM 214 and TEX-438-A methods at 0.44 and
0.43 respectively; SEE for ITM 214 and TEX-438-A test methods are 3.72% and 3.51%,
respectively The brine loss test simulates freezing and thawing of an aggregate, whereas
LA abrasion loss is an impact test. It appears that impact, like load, has a greater effect on
friction number than mechanical weathering. It is suspected that increased loads (higher
ESAL values) assist in retaining frictional resistance.

50

45 y = -0.03x2 + 0.84x + 32.13


R = 0.43
40
Average BPN10

35
ITM 214
30
y = -0.03x2 + 0.91x + 29.47

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
R = 0.44
25

20
0 10 20 30
Brine Loss, %

Figure 9. Relationship between BPN10 and percent brine loss of stone aggregates.

Terminal polish values were plotted as a function of total acid insoluble residue for
both test methods and are shown in Figure 10. The relationship between the two is good
with R2 values of 0.86 (SEE = 1.81%) and 0.75 (SEE = 2.09) for ITM 214 and TEX-438-A
test methods, respectively. However, as pointed out previously, the amount of total
insoluble residue for the aggregates used in this study was negligible and therefore would
have an insignificant effect on the frictional resistance; the results in Figure 10 should be
viewed with this in mind.

82

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

50

45

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
y = 36.24x0.021
40 R = 0.75
Average BPN10

35
ITM 214
y = 33.85x0.026
30 R = 0.86 TEX-438-A

25

20
0 1 2 3
Total Acid Insoluble Residue, %

Figure 10. Relationship between BPN10 and total acid insoluble residue of stone
aggregates.

5.4 Correlation of Laboratory Tests and Field Friction Tests


Frictional resistance values of in-service pavements were obtained from the INDOT
database. INDOT measures pavement friction using a standard smooth tire in accordance
with ASTM E274, Standard Test Method for Skid Resistance of Paved Surfaces Using a
Full-Scale Tire, ASTM E524, Standard Specification for Standard Smooth Tire for
Pavement Skid-Resistance Tests, and guidelines set forth (Li et al., 2003). Testing is
conducted at a speed of 64 km/h (40 mph) and reported as a friction (skid) number, FN.
Although not a part of the work reported in this paper, the in-service pavement frictional
number data indicate that pavements older than three years tend to reach equilibrium FN
values and these equilibrium values are approximately the same as before Superpave
mixture designs were implemented.
An attempt was made to correlate the in-service equilibrium FN values to BPN10
measured in the lab. The BPN10 for both ITM 214 and TEX-438-A test methods were
plotted as a function of the equilibrium FN values. As seen in Figure 11 the R2 values for
BPN10 obtained from both tests methods strongly indicate that there is not a relationship.

83

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

However, scatter in FN values from skid tests could be a source of difficulty in observing
any strong relationships. Also, the trailer test speed of 64 km/h (40 mph) may not represent

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
the actual in-service speed of the facility. Additionally, as was stated earlier, the BPN
frictional resistance as measured in the BPT more closely resembles the in-service
frictional resistance measured by the full-scale, locked-wheel method using a ribbed tire
rather than a smooth tire (Giles et al., 1962). Thus, BPN is a good measure of micro-
texture rather than macro-texture (Henry et al., 2000). Since INDOT uses the locked-
wheel, smooth tire, more a measure of macro-texture, it stands to reason that the BPN
would not correlate well with the in-service FN numbers obtained by the INDOT testing.

50
y = -0.10x + 40.09
R = 0.08
40
Average BPN10

30

y = -0.085x + 36.43 ITM 214


20 R = 0.11 TEX-438-A

10
10 20 30 40 50
Average Friction Number

Figure 11. Comparison of field and laboratory friction data.

6. Conclusions and Recommendations

Laboratory testing was conducted to determine total insoluble residue and frictional
characteristics of aggregates. Physical property data collected from INDOT quality
assurance records and total insoluble residue data were correlated with aggregate frictional
resistance results. Findings and conclusions are summarized and recommendations made in
the following subsections.

84

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

6.1 Relationship Between Physical Properties and Frictional Resistance

Physical property characteristics of each aggregate were plotted against corresponding


laboratory BPN10 results from both the ITM 214 and TEX-438-A test methods. The TEX-
438-A test method had higher coefficients of determination with the Los Angeles abrasion
results, and coarse aggregate bulk specific gravity and water absorption; the ITM 214 test
method had higher coefficients of determination with the amount of elemental magnesium
in coarse aggregates, brine loss and acid insoluble residues.
Comparison of BPN10 values to specific gravity showed that BPN10 increases with
decreasing bulk specific gravity, indicating that impure dolomites have greater frictional
resistance. Additionally, the terminal polish value for dolomite aggregates is higher when
percent elemental magnesium is approximately 11.2%. This also confirms that impure
dolomites have greater frictional resistance. BPN10 of coarse aggregates also increases with
higher percent water absorption and percent LA abrasion loss. The relationship between
BPN10 and brine loss is moderate and suggests that a coarse aggregate brine loss
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

percentage of 14-15% may be desirable. While the relationship between BPN10 and total
acid insoluble residue appears to be moderate, the total acid insoluble residue for all
dolomite aggregates used in this study was less than three percent and is not expected to
have an effect on frictional resistance. Currently, INDOT does not measure acid aggregate
insoluble residue for frictional specification purposes.
Each of the relationships between BPN10 and the aggregate physical property tests used
in the research show a bias between the ITM 214 and TEX-438-A test methods. This bias
is most certainly a result of the different rubber sliders used for each method. The ITM 214
test method uses the ASTM E303 specified rubber slider, which is less stiff than the TEX-
438-A rubber slider. At first glance it may seem that since the bias is consistent, it will not
matter which test an agency chooses to use. However, since each test method is better
correlated to certain aggregate physical property tests, an agency should choose the
laboratory polishing test method that best correlates to the aggregate properties being used
by the agency. For example, for INDOT, the percentage of elemental magnesium is an
important property in identifying dolomite aggregates for use in HMA pavement surfaces.
Thus the use of ITM 214 is appropriate since it correlates to the BPN10 values better than
does the TEX-438-A test method.
British polishing results indicate that harder stone aggregates tended to have higher
frictional resistance but not necessarily high polishing resistance. The frictional resistance
of all stone aggregates was improved by blending. However, a few blend combinations
used in this study show a reduction in frictional resistance of the slag component. Blends
could be improved by blending the high friction component (slag) with a softer carbonate
aggregate. A blend is considered effective if both components increase in frictional
resistance. Blends are considered compatible if differential wear measurements are large.

85

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

However, small differential wear measurements do not necessarily indicate insufficient


skid resistance, but it does indicate that the blend could be improved.
Finally, when the BPN10 data were correlated to in-service equilibrium FN values, no
relationship was observed. This may be due to the fact that INDOT performs locked-wheel
friction testing with the smooth tire, measuring macro-texture of the surface rather than the
micro-texture. If some locked-wheel friction testing were performed with the ribbed tire,
the results might better correlate to the BPN values measured on the coarse aggregates
used in the HMA surface mixture.
Overall, the current testing INDOT uses does appear to adequately screen coarse
aggregates for frictional properties. However, this does not mean that improvements
cannot be made.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
6.2 Recommendations

From the analyses conducted in this study, the following is recommended:


1. Raising the minimum requirement for percent elemental magnesium in dolomite
could help to insure the use of dolomites that offer higher friction resistance. The
relationship between laboratory measured aggregate friction properties and
percent elemental magnesium in aggregate established in this research suggests
that were the required minimum elemental magnesium percentage raised to 11%,
the average BPN10 for the aggregates in this study would be raised from 30.2 to
34.7. This recommendation should be weighed against the number of dolomite
sources that would be disallowed were the specification raised. Two of the six
dolomites used in this study have elemental magnesium values below 11%.
2. Use tests that identify the impurity of carbonate aggregates to identify skid
resistant aggregates. This could include the total acid insoluble residue test along
with a grain size distribution that includes material smaller than the 0.075 mm
(No. 200) sieve.
3. Continue using the ASTM E303 specified rubber slider for BPT as specified in
the ITM 214 test method.
4. Conduct a study to investigate the effect of blending high friction aggregates with
softer carbonate aggregates.
5. Measure at least some in-service pavement friction using the ribbed tire in order
to better correlate the results with laboratory measured aggregate properties.

86

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

7.0 Acknowledgment

The authors wish to acknowledge the support of Indiana Department of Transportation, the
Joint Transportation Research Program at Purdue University, and the Federal Highway
Administration. The contents of this report reflect the views of authors who are responsible
for the facts and accuracy of the data presented herein. The contents do not necessarily
reflect the official views or policies of the Indiana Department of Transportation or the
Federal Highway Administration at the time of publication.

8.0 References

Abdul-Malak, M.U., Fowler, D.W. and Meyer, A.H., "Mineralogy of Aggregates in


Relation to the Frictional Performance of Seal Coat Pavement Overlays: a
Petrographic Study." Transportation Research Record, No. 1418, 1993. p. 3542.
Asphalt Institute, Superpave Level 1 Mix Design. Superpave Series No. 2 (SP-2),
Lexington, KY, 1995.
ASTM, "Standard Practice for Accelerated Polishing of Aggregates or Pavement Surfaces
Using a Small-Wheel, Circular Track Polishing Machine." Book of ASTM Standards.
ASTM International, West Conshohocken, PA, 2002a.
ASTM, "Terminology Relating to Vehicle Pavement Systems." Book of ASTM Standards.
ASTM International, West Conshohocken, PA, 2002b.
Burnett, W.C., Gibson, J.L. and Kearney, E.J., "Skid Resistance of Bituminous Surfaces."
Highway Research Record, No. 236, 1968, p. 4960.
Chesterman, C.W., National Audubon Society Field Guide to North American Rocks and
Minerals. Alfred A. Knoph, Inc., New York, 1979.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Dahir, S.H.M. and Mullen, W.G., "Factors Influencing Skid-resistance Properties."
Highway Research Record, No. 376, 1971, p. 136148.
Dillard, J.H. and Alwood, R.L., "Providing Skid-resistant Roads in Virginia." Proceedings
of the Association of Asphalt Paving Technologists, Vol. 26, 1957, p. 122.
Dillard, J.H. and Mahone, D.C., "Measuring Road Surface Slipperiness." 66th Annual
Meeting Papers, ASTM 366, Atlantic City, NJ, 1963.
Fu, C.N., "Aggregate Properties and Blending versus Frictional Performance of
Bituminous Pavements." Raba-Kistner Consultants, Inc., 2004.

87

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

Furbush, M.A. and Styers, K.E., "Relationship of Skid Resistance to Petrography of


Aggregates." Pennsylvania Department of Transportation, Harrisburg, PA, 1972.
Giles, C.G., Sabey, B.E. and Cardew, K.H.F., "Development and Performance of the
Portable Skid-Resistance Tester." ASTM Special Technical Publication No. 326:
Symposium on Skid Resistance - 65th Annual Meeting Papers. American Society for
Testing and Materials, New York, NY, 1962, p. 5074.
Gramling, W.L. and Hopkins, J.G., III, "Skid Resistance Studies: Aggregate-Skid
Resistance Relationship as Applied to Pennsylvania Aggregates." Report No. 65-4,
Pennsylvania Department of Transportation, Harrisburg, PA, 1974
Gray, J.E. and Renninger, F.A., "The Skid-Resistant Properties of Carbonate Aggregates."
Highway Research Record, No. 120, 1966, p. 1834.
Henry, J.J., NCHRP Synthesis 291: Evaluation of Pavement Friction Characteristics.
Transportation Research Board - National Research Council, Washington, D.C., 2000.
INDOT, Standard Specifications, 2012.
Jayawickrama, P.W., Prasanna, R. and Senadheera, S.P., "Survey of State Practices to
Control Skid Resistance on Hot-Mix Asphalt Concrete Pavements." Transportation
Research Record (1536), 1996, p. 5258.
Kearney, E.J., McAlpin, G.W. and Burnett, W.C., "Development of Specifications for
Skid-Resistant Asphalt Concrete." Highway Research Record, No. 396. 1972, p. 12
20.
Kummer, H.W. and Meyer, A.H., "Measurement of Skid Resistance." ASTM Special
Technical Publication No. 326: Symposium on Skid Resistance - 65th Annual
Meeting Papers. ASTM Special Publication No. 326. American Society for Testing
and Materials, New York, NY, 1962.
Lee, D., Guinn, J.A., Khandhal, P.S. and Dunning, R.L., "Absorption of Asphalt into
Porous Aggregates." SHRP-A/UIR-90-009, Strategic Highway Research Program,
Washington, D.C., 1990.
Liang, R.Y., "Blending Proportions of High Skid and Low Skid Aggregate". FHWA/OH-
2003/014, Ohio Department of Transportation and US Department of Transportation,
Federal Highway Administration, Akron, OH, 2003.

88

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Hot Mix Asphalt Pavement Frictional Resistance as a Function of Aggregate Physical Properties

Mullen, W.G., Dahir, S.H.M. and El Madani, N.F., "Laboratory Evaluation of Aggregates,
Aggregate Blends, and Bituminous Mixes for Polish Resistance." Transportation
Research Record, No. 523, 1974, p. 5664.
Rogers, M.P. and Gargett, T., "Skidding Resistance Standard for the National Road
Network." Highways and Transportation, Vol. 38, No. 4, 1991, p. 1016.
Skerritt, W.H., "Aggregate Type and Traffic Volume as Controlling Factors in Bituminous
Pavement Friction." Transportation Research Record, No. 1418, 1993, p. 2229.
Stock, A.F., Ibberson, C.M. and Taylor, I.F., "Skidding Characteristics of Pavement
Surfaces Incorporating Steel Slag Aggregates." Transportation Research Record, No.
1545, 1996, p. 3540.
Webb, J.W., "The Wearing Characteristics of Mineral Aggregates in Highway Pavements."
VHRC-70-R7, Virginia Highway Research Council, Washington, D.C., 1970.
West, T.R. and Cho, K.H., Development of a Procedure to Identify Aggregates for
Bituminous Surfaces in Indiana. FHWA/IN/JHRP-2000/28, Joint Transportation
Research Program, West Lafayette, IN, 2000.
West, T.R., Choi, J.C., Bruner, D.W., Park, H.J. and Cho, K.H., "Evaluation of Dolomite
and Related Aggregates Used in Bituminous Overlays for Indiana Pavements."
Transportation Research Record, No. 1757, 2001, p.137147.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Won, M. and Fu, C.N., "Evaluation of Laboratory Procedures for Aggregate Polish Test."
Transportation Research Record, No.1547, 1996, p. 2228.

9.0 Discussion

DR. MICHAEL HEITZMAN: Good presentation. I have a couple of comments. One,


when you were trying to put your British pendulum coupons together with blended
aggregates, how did you do the blending? How did you do the coupons for blended
aggregates when typically you only have one aggregate in that coupon?
PROFESSOR JOHN HADDOCK: We actually put them in blended to the desired
percentage.
DR. HEITZMAN: You are individually putting in particles one next to another?
PROF. HADDOCK: Yes.

89

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
HADDOCK, OBRIEN

DR. HEITZMAN: Okay. I think thats interesting when you're doing blended aggregates.
You observed that two different test methods for friction are going to give you different
values. We are seeing that, as well. We have to be very careful about comparing studies
with respect to absolute friction values because not everyone is running the same tests. My
second comment, I am concerned about the focus on the coarse aggregate properties for
pavement friction when, in fact, the mixture properties on the surface are whats going to
tell us more about the friction. In fact, your one graph (Figure 11 in the reprint) shows that
there was really no correlation between the coarse aggregate properties and the friction
measurements in the field. We have to step beyond coarse aggregate properties and get into
the mixtures. I know Becky McDaniel with the North Central Superpave Center and here
at NCAT, we have that ability now to start looking at those mixture properties on the
surface.
PROF. HADDOCK: Thank you. I agree with the points you have made.
DR. REBECCA MCDANIEL: Well, with that opening, Im Becky McDaniel with the
North Central Superpave Center at Purdue, and I would add to what Mike said. We too
recognize that it is the mixture properties that are key. The Indiana DOT recognizes that as
well, and theyve funded some additional research looking at frictional properties of
mixtures. Weve looked at the contributions of the coarse and the fine aggregate, and the
blending of softer carbonates with high-friction aggregates, which, as John mentioned, is

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
something additional to look at. INDOT has funded research along those lines, and perhaps
I should write a paper based on that. Thank you.
PROF. HADDOCK: Thank you, Becky. I agree with Becky, as well. It is the pavement
that must really be looked at, and I know that Becky has some work going on at the
Superpave Center looking at the pavement. I know that NCAT is doing some things on it,
as well. In Indiana, I might just mention, we have a procedure that we go through. If you
have an aggregate that you would like to use as a friction aggregate, we actually go through
an entire process of lab testing, and then it gets put in the field, and that friction gets
monitored in the field to see what happens. So, I look at these aggregate physical properties
as more of quick checks to make sure that the aggregate might be suitable as friction
aggregates. But in the end, one must really get a measurement of the pavement to tell what
kind of skid resistance the pavement has, Thank you.

90

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model
for Asphalt Pavements
Eshan V. Davea*, William G. Buttlarb, Sofie E. Leonb,
Behzad Behniab and Glaucio H. Paulinob
a
Department of Civil Engineering, University of Minnesota Duluth, Duluth, MN 55812
b
Department of Civil and Environmental Engineering, University of Illinois, Urbana, IL
61801

ABSTRACT: Low temperature cracking (LTC) is a major distress and cause of failure for asphalt
pavements located in regions with cold climate; however, most pavement design methods do not
directly address LTC. The Thermal Cracking Model (TCModel) utilized by the AASHTO
Mechanistic-Empirical Pavement Design Guide (MEPDG) relies heavily on the phenomenological
Paris law for crack propagation. The TCModel predictions are primarily based on tensile strength
of the asphalt mixture and do not account for quasi-brittle behavior of asphalt concrete.
Furthermore, TCModel uses a simplified one-dimensional viscoelastic solution for determination
of thermally induced stresses. This paper describes a newly developed comprehensive software
system for low temperature cracking prediction in asphalt pavements. The software system called
IlliTC utilizes a user-friendly graphical interface with a stand-alone finite-element based
simulation program. The system includes a preanalyzer and data input generator module that
develops a two-dimensional finite element pavement model for the user and which identifies

The oral presentation was made by Professor Dave.

This is a reproduction of an article whose final and definitive form has been published in Road
Materials and Pavement Design 2013 Taylor & Francis. The article is available online at:
http://dx.doi.org/ 10.1080/14680629.2013.812838

91
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

critical events for thermal cracking using an efficient viscoelastic pavement stress simulation
algorithm. Cooling events that are identified as critical are rigorously simulated using a
viscoelastic finite element analysis engine coupled with a fracture-energy based cohesive zone
fracture model. This paper presents a comprehensive summary of the components of the IlliTC
system. Model verifications, field calibration and preliminary validation results are also
presented.

KEYWORDS: Asphalt, thermal cracking, fracture, performance, simulation, cohesive zone,


transverse cracking, viscoelasticity, model, pavement, IDT, DC(T), IlliTC.

1. Motivation and Introduction

One of the main advantages of asphalt concrete over Portland cement concrete (PCC) is
the smoothness and cost savings afforded by continuous paving, i.e., without the need
for transverse joints. Unlike PCC and other infrastructure materials, asphalt concrete is
generally able to undergo thermal cycling without the need for expansion or contraction
joints due to its viscoelastic nature. Under imposed strain, which is constantly occurring
in pavements due to temperature change, viscoelastic materials are able to relax stress
over time. In addition, asphalt is generally a fracture resistant material, owing to its
flexible mastic matrix and particulate composite morphology. Significant energy is
required to initiate and propagate a crack through asphalt concrete, as the asphalt mastic
is tough, strain tolerant, and viscoelastic (stress relaxing), and the aggregates add
strength, crack bridging, and crack surface tortuosity. However, improper selection of
asphalt grade, excessive aging of the asphalt binder, and/or a weak asphalt mixture
(weak aggregates, low cohesion, low adhesion) can all contribute to poor mixture
fracture resistance. Poor mixture fracture resistance can lead to the development of
thermal cracks, which are typically transversely oriented with traffic and periodic in
nature.
Thermal cracking is very serious pavement distress, as it can significantly increase
pavement roughness and because it creates a permanent discontinuity in the pavement
structure. In a recent study by Islam and Buttlar (2012), pavements allowed to reach a
rough condition were shown to increase user costs (vehicle repair, tire, and fuel cost) by
over 5 million dollars per lane mile over a 35 year life cycle. In contrast, by investing
just 1/50th of this cost in additional maintenance, the pavement could be kept in smooth
condition over its lifespan, avoiding these additional user costs. However, in order to
achieve this result, the pavement would need to be properly designed to avoid the
development of medium or high severity thermal cracking, since an improper mixture

92

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

design could lead to the development of thermal cracks prior to the application of the
first maintenance or rehabilitation treatment.
Asphalt technologists have long recognized the need to control thermal cracking in
asphalt pavements, and the tests and models available to assist in this endeavor have
continuously evolved. Early efforts to control asphalt behavior at low temperatures were
focused on the asphalt binder. A comprehensive review of early binder tests in the US
has been documented by Brown et al. (2009). The penetration test, especially if run at
two temperatures (generally 25C and 4C), provided some control over binder
consistency and temperature susceptibility. The ductility test provided a simple
measure of binder stretch or strain tolerance. However, neither of these tests were
applicable to temperatures below 0 C, where thermal cracking is likely to occur. In
Europe, the Fraass breaking point test is used (as specified by EN 12593:2007), which is
a torture-type binder test designed to determine the temperature at which a thin film of
binder bonded to a small rectangular brass plate becomes intolerant to a bending strain
arising from flexing the brass plate to a specified curvature. Although this test allowed
direct mechanical testing of the binder at temperatures below 0C, its direct relation to
thermal cracking is questionable due to the very high strain level imposed. Superpave
(AASHTO M320) addressed thermal cracking with modern binder tests run at low
temperatures and in fundamental testing configurations. A bending beam rheometer
characterized the stiffness and m-value (a measure of the ability of the binder to relax
stress). An optional direct tension test was also specified, which was designed as a
second referee test to address certain polymer modified binders that possessed higher
stiffness but also high strain tolerance and fracture resistance. However, the system was
never intended to directly control thermal cracking; rather, it was developed as a binder
purchase specification.
Although binder tests are convenient, practical, and important from the perspective
of binder selection, purchase, and quality control, asphalt technologists have
acknowledged the need to address thermal cracking more directly through testing of the
asphalt mixture and modeling of the pavement structure. Canadian researchers
developed limiting mixture stiffness recommendations based upon a comprehensive field
investigation at Ste. Anne (Deme and Young, 1987). During the Strategic Highway
Research Program (SHRP), a mixture-based testing and analysis scheme was developed
to validate the binder test and specification system being developed. A low temperature
mixture creep and strength test was developed, later called the Superpave Indirect
Tension Test (the acronym IDT was developed in early Superpave publications, and is
still commonly used), as specified in AASHTO T-322. Master creep compliance curves,
shift factors, tensile strength, and optionally, mixture coefficient of thermal expansion
and contraction at low temperatures can be obtained with the Superpave IDT. SHRP
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

researchers, working under project A-005 also developed a computer-based thermal

93

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

cracking model, called TCMODEL, as part of the effort to validate the Superpave binder
specification. The first version of TCMODEL was completed in 1992, near the end of
the SHRP program.
TCMODEL made great strides in modelling some of the key physics underlying the
thermal cracking mechanism. In particular, TCMODEL had a highly sophisticated
viscoelastic pavement response model, which predicted pavement tensile stress versus
depth on an hourly basis throughout the life of the pavement based upon principles of
linear viscoelasticity. Due to limitations in computational power, a phenomenological
pavement cracking model was used for distress prediction. Change in crack length was
predicted using a power-law type model reported by Paris et al. (1961), which uses
change in stress intensity (which was calculated in an approximate manner by
interpolating presolved 2D elastic finite element runs) to predict change in crack length.
In TCMODEL, Paris law parameters were empirically linked to IDT strength and to the
slope of the log mixture compliance vs. log time relationship at long loading times
(mixture m-value). TCMODEL was selected for inclusion in the AASHTO Mechanistic
Empirical Pavement Design Guide, and was improved and streamlined as part of the
NCHRP 1-37A project. For instance, an automated mixture master curve generation
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

program was bundled with TCMODEL. In addition, additional field data, including that
obtained from MnROAD test sections, was used to recalibrate TCMODEL.
With changes in asphalt binder and mixture designs and materials over the past 20
years, such as the increased use of polymers and other additives and the increased use of
recycled asphalt pavement (RAP), it became apparent that the heavy reliance on mixture
tensile strength in cracking predictions was limiting the prediction accuracy of
TCMODEL. New mixture fracture tests were developed to address these new materials,
including the disk-shaped compact tension test, or DC(T), which provided a convenient
means to obtain mixture fracture energy using a fracture mechanics based approach.
Recognizing the shortcomings of TCMODEL in light of modern mixture fracture tests
and computational power, a new thermal cracking model was developed at the
University of Illinois Urbana-Champaign, called IlliTC, as part of a national Pooled
Fund Study on Low Temperature Cracking. IlliTC improves the manner in which
fracture is handled in the simulation scheme, namely; the 1D Paris-Law
phenomenological modeling approach was replaced with a 2D, cohesive zone fracture
modeling approach implemented within a viscoelastic finite element modeling
framework. The cohesive zone approach considers both material strength and fracture
energy in computing crack initiation and propagation using fundamental fracture
mechanics principles. In summary, the new approach used in IlliTC has the following
improvements over TCMODEL:
A 2D model is used instead of 1D.

94

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

The physics of cracking in a quasi-brittle, heterogeneous particulate composite


are more correctly captured by using a cohesive zone approach, where softening
and fracture have a distinct length scale that is captured.
Asphalt mixtures may have unique combinations of strength and ductility (as
characterized by mixture fracture energy). For instance, some polymer-
modified mixes portray moderate tensile strength and high fracture energy;
some have high strength and lower fracture energy, and some have both high
strength and high fracture energy. Mixtures with higher recycled material
content may have high strength, but low fracture energy. IlliTC can capture all
of these combinations in a direct manner, while TCMODEL could only capture
these effects in an indirect manner.
A user-friendly graphical interface (GUI) has been provided for IlliTC. The
GUI program module within IlliTC is referred to herein as Visual-LTC.
This paper presents the IlliTC model components, model verification, model
calibration, and preliminary validation results. Ongoing research, aimed at adding
additional software capabilities and modeling features to the IlliTC program, are also
described.

2. IlliTC Framework
The software program (IlliTC) provides an intuitive and user-friendly graphical user
interface (GUI), as a convenient gateway to the rigorous viscoelastic finite element /
cohesive zone modeling engine. The overall flow of the IlliTC program along with
various inputs and outputs are graphically illustrated in Figure 1. The code consists of a
GUI, which is represented as a box on the left in Figure 1, and four analysis modules,
which are encased in dotted boxes in Figure 1. The analysis modules include a
preprocessor, an input file generator, a preanalyzer, and a finite element based thermal
cracking prediction engine. An overview of the GUI and analysis modules is presented
in the remainder of this section and further implementation details are presented in
sections 3-5.
To initiate an IlliTC thermal cracking simulation, the user is queried to enter
information pertaining to the project location, design life, pavement structure, and
material properties into the GUI. The GUI handles cumbersome tasks like data
organization and unit conversion, then automatically assembles additional data needed
for analysis from internal databases.
The GUI passes the raw creep compliance data to the preprocessor, which returns the
thermo-viscoelastic material properties in the form of Prony series parameters
(generalized Maxwell model) and time-temperature shift factors. The preprocessor

95
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

consists of two modules, the first one is based on the code Master developed by
Buttlar et al. (1998) and second one is based on the TCModel (Lytton et al., 1993, Roque
et al., 1995a, 1995b). The first module generates the creep compliance master curve
from the raw creep compliance data and fits a generalized Voigt-Kelvin model to the
master curve. The Voigt-Kelvin model for creep-compliance is then converted to a
generalized Maxwell model in the form of a relaxation modulus.
The GUI reads the Maxwell model parameters from the preprocessor output,
organizes it along with other data, and passes it to the input file generator. The input file
generator conducts two main tasks: 1) to develop a finite element mesh for the pavement
geometry specified by the user, called the Geometric Data File, (*.mesh) shown hashed
box in Figure 1, and; 2) to create a material data file, based on information provided by
the user, called the Material Data File (*.mtr), also shown in the hashed box in Figure 1.

Figure 1. Flowchart of new thermal cracking model IlliTC. --`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

The GUI collects the climatic data necessary to perform the analysis in the chosen
location for the specified duration from internal databases. This data is stored in the
Climatic Information File (*.poly), shown in the hashed box in Figure 1. Then the GUI
passes all input files to the preanalyzer.

96

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

The preanalysis module (or preanalyzer) serves to minimize the analysis time of
the more rigorous finite element engine. Typically, low temperature cracking analysis is
focused around critical cooling events. Hence, critical cooling events are identified by
the preanalyzer so that the full model can be focused on critical cooling events, typically
reducing computational time by more than 90% (i.e., in a 365 day year, less than 36 day
would typically need to be simulated). The GUI reads the results of the preanalysis
module and extracts the data pertinent to the critical cooling events. These data are then
passed to the finite element model.
The user is not required to have direct interaction with the viscoelastic and cohesive
zone finite element simulation; instead, results are sent back to the GUI, which are
interpreted for the user.

3. Graphical User Interface (GUI)

The GUI, called Visual LTC, collects and compiles the input conditions provided by
the user, executes various analysis modules to conduct finite element analysis, and
interprets and displays the results. Visual LTC was written with the object-oriented
programming language C# (pronounced see-sharp) under Microsofts .NET
framework, and which is intended for the development of deployable software. Visual
LTC was designed to be intuitive to use by practitioners or researchers. A series of
windows are used to query the user for required model inputs. For example, the window
that collects the asphalt layer material properties is shown in Figure 2. User inputs and
options shown in Figure 2 are discussed in more detail in subsequent sections. For
further details on conducting analysis using Visual LTC the reader is referred to Dave et
al., 2011.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

97

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Figure 2. Visual LTC window asphalt layer material properties.

3.1. Communication with Analysis Modules

Data is passed between Visual LTC and the analysis modules via input/output files.
Visual LTC reads the user input then performs the series of converting data, writing
files, executing programs, and reading output shown in Table 1.

98

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

Table 1. Visual LTC steps.

1. Read and store user input


2. Extract and store climatic information for user specified analysis period
3. Write input files for preprocessor
4. Run preprocessor
5. Read and convert preprocessor output
6. Write input files for input file generator
7. Run input file generator
8. Run preanalyzer
9. Process preanalyzer output to identify critical events and generate finite element
temperature input
10. Run finite element analysis engine
11. Read finite element analysis output
12. Convert crack depth to amount of cracking
13. Display results

3.2. User Types

Visual LTC is intended for use by practitioners and researchers alike. Therefore, two
user types are supported: Standard User, and Advanced User. Both users have
access to all functionality previously described. However, Advanced Users have the
additional capability of adding new asphalt mixes and modifying properties of existing
asphalt mixes. The distinction between these user types is present so that existing
properties are protected from inadvertent user error. The user can easily change from one
user type to the other.

3.3. Visual LTC User Inputs

The main user inputs that are required for low temperature cracking analysis are the
analysis location, analysis duration and the pavement material properties. A series of
Integrated Climatic Model (ICM) simulations were conducted to create a library of
pavement temperature profiles available to the user in Visual LTC. Sets of temperature
profiles were generated for one cold, one intermediate and one warm location in each
participating state of the pooled fund low temperature cracking study, as shown in Table
2. Eventually, IlliTC will be coupled with the ICM model, so that climatic data from
thousands of geographic areas are available within the model. The data required to
generate these libraries are the same as those used in the AASHTO MEPDG system.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

99

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

Temperature profiles at each location were generated for the following asphalt concrete
thicknesses: 75, 100, 125, 150, 175, 200, 225, 250, 300, 350, and 400 mm.
Table 2. Climatic locations available to user in Visual LTC.

State Climate City Air PG


Temperature Grade
Cold Norfolk -29.5C -28
Connecticut Intermediate Hartford -26C -22
W ar m New Haven -20.5C -22
Cold Elizabeth -37C -34
Illinois Intermediate Urbana -31.5C -28
W ar m Anna -27C -22
Cold De c o r a h -40.5C -34
Iowa Intermediate Des Moines -32C -28
W ar m Fort Madison -30.5C -28
Cold Massena -39C -34
Ne w Yo r k Intermediate Albany -33.5C -28
W ar m New York City -19.5C -16
Cold Westhrope -44C -40
North Dakota Intermediate Bismarck -41.5C -40
W ar m Wahpeton -38C -34
Cold International Falls -43.5C -40
Minnesota Intermediate St. Cloud -41.5C -34
W ar m Worthington -34.5C -34
Cold Minong -46C -40
Wisconsin Intermediate Stevens Point -36.5C -34
W ar m Milwaukee -32C -28

In Visual LTC, the user selects a location that is the most climatically similar to the
analysis location. The user also provides the pavement cross-section. The Visual LTC
extracts the appropriate data from the temperature profiles associated with the location
and pavement cross-section. These data are passed to the preanalyzer and finite element
engine where nodal temperatures are computed. The pavement material property related
user inputs are summarized in Table 3 and discussed below.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

100

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

Table 3. Summary of pavement material user inputs for IlliTC.


Property Units Test
Tensile strength MP a AASHTO T 322
or Extracted from
DC(T) Test
Fracture energy J/m2 ASTM D7313 1
Option 1 Unit weight g/cm3 AASHTO M323
Mixture VMA 2 % AASHTO M323
Aggregate coefficient mm/mm/C No standardized
of thermal expansion test
and contraction
(CTEC) 2
Option 2 Mixture coefficient mm/mm/C No standardized
of thermal expansion test
and contraction
(CTEC) 3
Creep compliance test data (100 or 1000 1/GPa AASHTO T-322
seconds for 3 temperatures)
Creep compliance test temperatures C AASHTO T-322
1
Fracture energy may be obtained with different test geometry; however the model is
calibrated for the ASTM D7313 (disk-shaped compact tension, DC(T) test procedure
2
Mixture VMA and aggregate CTEC do not need to be entered if Mixture CTEC is provided
3
Mixture CTEC will be calculated if mixture VMA and aggregate CTEC are provided

The tensile strength of asphalt concrete can be determined using the Superpave
indirect tension test (IDT), as specified in AASHTO T 322. Recently, a procedure was
developed to extract mixture tensile strength from DCT test data, which will be
incorporated into IlliTC in the near future. The fracture energy can be determined using
a variety of test geometries, such as disk-shaped compact tension (DC(T)), semi-circular
bend (SCB) and single-edge notched beam (SENB) test, although the model has been
calibrated and validated based on the DCT test geometry. Furthermore, the fracture test
is expected to be performed at a crack mouth opening displacement (CMOD) rate of
0.0167 mm/s and at temperature of 10C above the 98% reliability Superpave PG low
temperature grade, as dictated by the project location.
The user can either directly input the coefficient of thermal expansion and
contraction (CTEC) or provide asphalt mixture volumetric properties. If volumetric
properties are provided, the CTEC is estimated using the approximation equation used
by the AASHTO MEPDG software. The researchers at the University of Wisconsin have
proposed experimental procedures to measure the CTEC of asphalt mixtures (Bahia and

101
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

Nam, 2004, 2009, Marasteanu et al., 2007); the use of their procedure is recommended
for added accuracy in prediction. The work by the same research group has shown a
bilinear trend in the volumetric changes that occur in asphalt binder and mixes, with
significantly different CTEC values above and below the glass transition temperature.
This feature is not currently implemented in the IlliTC system; however it is recognized
as one of the tasks for implementation in future versions.
The user can directly enter laboratory measured 100 or 1000 second creep test data
from three temperatures following the AASHTO T-322 test procedure. These data are
passed to the preprocessor, which converts the data into thermo-viscoelastic material
properties in the form of Prony series parameters (Generalized Maxwell model) and time
temperature shift factors.

3.4. Data Storage

A simple and intuitive class structure is employed to store and maintain data required for
low-temperature cracking analysis, i.e. material properties, climatic data, pavement
structure, and project information. The data should be easily accessible by the user and
should not require installation of additional software. A working directory containing
input files stores all of the data necessary for Visual LTC to conduct the analysis.
Furthermore, the user is not required to directly access the files, as Visual LTC creates
and modifies files automatically. The project input file stores general information (i.e.
project name, description, date, etc.), climatic information, and the pavement structure.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Asphalt concrete input files store all material properties associated with the mix. A
working directory can contain many project files, thus giving the user the option of
creating a new project by modifying an existing one. Similarly, the working directory
can contain as many asphalt concrete input files as necessary, which creates a library of
mix designs for the analyst or designer to investigate.

4. Input Generator and Preanalyzer

The IlliTC prediction system is designed to be practitioner friendly and hence all the
necessary inputs for the finite element simulation are generated by the software. The
input generator module handles the creation of finite element mesh and corresponding
data file with all geometric information and also generates a material data file in the
necessary format for the finite element analysis code. The first task of this module is to
develop a finite element mesh for the pavement geometry selected by the user. The
details on the mesh generation were previously discussed by Dave et al. (2011). Briefly,
the code generates a finite element mesh using four node quadrilateral elements (Q4) and

102

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

it automatically generates a transition mesh with element sizes increasing as the distance
from potential thermal crack increases. A single line of cohesive zone elements are
inserted into the mesh. These elements allow for simulation of a single thermally
induced crack that can span across the pavement thickness. The crack is simulated in the
transverse direction. The input generator takes the material data provided by the user as
well as the viscoelastic parameters determined by the preprocessor and generates the
material data file (.mtr).
The preanalysis module (or preanalyzer) was developed to optimize analysis times of
the finite element engine. A simplified one-dimensional (1D) problem is solved by this
module to identify critical events that are then analyzed by the finite element analysis
engine; hence only critical cooling events are analyzed with the full model. The
preanalyzer module solves the stresses on a restrained 1D viscoelastic body that is
imposed with temperature boundary conditions representative of the temperatures at the
pavement surface. The body is assumed to have same properties as the thermo-
viscoelastic properties of asphalt concrete. The one dimensional viscoelastic solution for
thermal stress can be found in Apeagyei et al. (2008); this solution is implemented in
IlliTC using a recursive-incremental numerical integration method.
The results from the preanalyzer were verified with the analytical solution. The
stresses obtained with the preanlayzer (VE1D) and the analytical solution for a 1D body
imposed with thermal loading shown in Figure 3(b) are compared in Figure 3 (a). The
results show excellent agreement between the preanalyzer results and analytical solution.

( a) (b )
Figure 3. Verification of preanalyzer. (a) Comparison of stresses obtained with the
preanalyzer (VE1D) and the analytical solution. (b) Thermal loading.

103
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

The critical cracking events are identified when the thermal stress in the 1D model
exceeds 80% of the tensile strength of the asphalt mix. The thermal stress from the 1D
analysis is assumed to represent the stresses on the pavement surface. The threshold of
80% was selected based on previous experience of researchers in determining the stress
threshold corresponding to the onset of damage. The full analysis with the finite element
engine is performed for the 24 hours surrounding the critical event.
Figure 4 shows an example of the results of the preanalyzer. The winter time surface
temperatures during the five-year analysis period are shown in Figure 4(a) and the
resulting thermal stresses are shown in Figure 4(b). Only the time duration between
October 1st and March 31st are simulated, due to significantly greater potential for
thermal cracking during these months. Four critical events were identified by this
analysis; the full finite element simulation is performed on these events accordingly.

( a)
80% of Tensile Strength 2 4
1
3

(b )
Figure 4. Results from preanalyzer. (a) Winter time surface temperatures for five-year
analysis period. (b) Resulting thermal stresses where four critical events were identified.

104
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

5. Finite Element Analysis (FEA) Engine

The finite element analysis (FEA) method is used extensively in everyday design and
analysis of civil infrastructure. The biggest strengths of FEA are its ability to simulate
challenging geometries, such as pavements, and the relative ease in which it is possible
to incorporate complicated material behavior. For situations involving these
characteristics, analytical solutions would be challenging and often unfeasible. The
IlliTC system uses a finite element code that was developed in-house for simulation of
thermo-viscoelastic problems. It has the capability of simulating cracking in asphalt
concrete through use of a powerful yet computationally efficient cohesive zone fracture
model. Traditional modeling approaches have not provided a direct means for the study
of crack initiation and propagation in asphalt materials. The cohesive zone fracture
approach provides a rational means for modeling cracking in quasi-brittle materials such
as asphalt concrete, as the length scale associated with the fracturing process is
accounted for. The following subsections briefly describe the cohesive zone model and
the thermo-viscoelastic implementation in IlliTCs FEA engine. This is followed by
selected verification examples and brief description of the post-processing methodology
to extract results from the FEA.

5.1. Cohesive Zone Fracture Model

In order to correctly replicate the complex mechanisms underlying cracking in asphalt


concrete, a standard strength of materials type analysis is not sufficient, due to: 1) the
highly non-linear behavior in the vicinity of the crack tip, and 2) the importance of the
crack in the overall structural response (i.e., the need to model the crack as a moving
boundary value problem). For simulation of the crack initiation and propagation, a
cohesive zone model was selected because of its accuracy and efficiency in accounting
for material response ahead of the crack tip in the fracture process zone (region of
micro-cracking, crack pining, crack branching, material softening, etc.). Several
researchers have used this type of approach for simulation of cracking in asphalt
materials, for example Soares et al. (2004), Song et al. (2006), Dave and Buttlar (2010),
Baek et al. (2010), and Kim et al. (2010).
The cohesive zone model provides the relationship between the displacement jump
or the opening along the crack path and the total capacity of material to transfer traction
(or load) across that crack path. In the case of brittle materials, the capacity to carry load
across the crack path is either 100% (uncracked) or 0% (cracked). The threshold of stress
at which this capacity goes from 100% to 0% is the material strength and, in the case of
mode I (opening) type of cracking, it is tensile strength. For quasi-brittle materials the
capacity to transfer load across the crack path also reduces once the material has incurred

105
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

damage. However, unlike brittle materials, for quasi-brittle materials this capacity
reduces gradually. The stress threshold at which the damage begins to occur is still
tensile strength. Finally, when the crack separation (displacement jump) is greater than a
critical value, the material no longer has bearing capacity and the traction is zero. The
relationship between displacement jump and traction capacity across crack can be
defined by different geometric shapes. Various shapes have been proposed for use with
asphalt concrete, such as, bilinear (Song et al., 2006), power law (Song et al., 2008), and
exponential (Dave and Buttlar, 2010).
In this study, the bi-linear CZM described by Song et al. (2006) is being employed;
the graphical representation of the model is shown in Figure 5. Note the unloading and
reloading behavior of the model is also shown on the figure. This behavior assumes that
the damage incurred by the material is permanent and is present when it is reloaded. The
area under this traction and displacement jump curve is the fracture energy of the
material. Thus, the CZM implementation allows for fracture representation of asphalt
concrete through use of two material properties, namely, tensile strength and fracture
energy. An intrinsic cohesive zone modeling approach is used in this work; hence a
penalty stiffness (i.e. initial ascending slope) is introduced in the computational
implementation. The initial penalty stiffness is determined on the basis of the numerical
stability associated with the finite element implementation (Roesler et al., 2007). The
implementation of CZM in IlliTC is limited to mode I cracking. Since the thermally
induced stresses in conventional asphalt pavements are along the longitudinal direction
the predominant failure occurs through tensile failure in mode I. Very small quantities of
shear stresses are generated due to the thermal only loading case and thus mode II or
mixed mode cracking is not expected.
Traction (Capacity) across crack (MPa)

Tensile strength

Area = Fracture Energy

Unloading

Critical displacement

Reloading

Displacement jump across crack (mm)

Figure 5. Bi-linear cohesive zone model (CZM) used in IlliTC.

106
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

5.2. FE Implementation and Verification

The thermo-viscoelastic finite element code was implemented in the C programming


language. The code is based on the incremental-recursive formulation proposed by Yi
and Hilton (1993) and Zocher et al. (1997). The details on the implementation and
formulation have been presented by Dave et al. (2012). Due to the non-linear nature of
the CZM, the FE analysis was implemented using a modified Newton-Raphson solution
scheme. A simple adaptive time increment scheme was utilized that automatically
increases the time increment as long as the convergence error is below the threshold.
The FE implementation was verified using analytical solutions and commercial
programs. Two verification examples are presented herein. The first example involves
the simulation of a boundary value problem which resembles the thermal stress retrained
specimen test (TSRST). The simulation was conducted to include both cooling and
warming temperature boundary conditions to ensure the rigor of IlliTC (some cohesive
zone models have convergence problems upon crack closure). Figure 6 presents the
thermal stresses predicted by IlliTC against those predicted by the commercial FE
program ABAQUS. The figure also shows the temperature boundary condition that was
used in this simulation. The results show very good agreement between IlliTC and
ABAQUS results, providing verification for the thermo-viscoelastic stress prediction
capabilities of IlliTC.
Thermo-viscoelastic predictions coupled with CZ fracture simulation were also
verified, as shown in Figure 7. The verification was conducted on a solid rectangular
body as shown on the plot, where a temperature drop of 20C over duration of 150
minutes was imposed. Once again, very good agreement was observed between the
IlliTC and ABAQUS predictions. Note that there were minor deviations in the predicted
stresses, especially during the damage and cracking process (post-peak). For this
verification example the stress variations between two programs ranged between 1 and
2%. This level of discrepancy is not unexpected, due to differences in the
implementation of the CZM and use of different numerical solvers in the two FE
engines.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

107

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

Figure 6. Comparisons for thermo-viscoelastic stress predictions using IlliTC and


commercial finite element program ABAQUS.

Figure 7. Comparisons of stress for a thermo-viscoelastic body with CZ fracture for


predictions using IlliTC and commercial program ABAQUS.

108
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

5.3. Post Processing

The output generated by the FEA contains displacements at nodal points and stress and
strain predictions as element integration points. For the average user, the results require
post-processing to generate a simple thermal cracking versus time output. The IlliTC
screens the outputs generated by the FEA and produces a series of text files that contain
information regarding crack opening widths, which is the deformation experienced by
CZM elements in the longitudinal direction. By comparing the deformation of cohesive
zone elements against the critical displacement in the CZM, the location of the crack tip
and location of the point beyond which the asphalt has started to soften is computed. For
example, in a pavement with an asphalt concrete thickness of 150 mm, if the cohesive
zone elements in the top 25 mm of the pavement have experienced deformation in excess
of the critical deformation, the thermal crack has formed to a depth of 25 mm.
Furthermore, if the cohesive zone elements along the upper 75 mm of the asphalt layer
have undergone deformations exceeding the deformation corresponding to the tensile
strength in the local CZM (c.f. Figure 5), the upper 75 mm of asphalt concrete has

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
undergone softening damage. In the case of this example, the upper 25 mm of the
pavement is fully cracked; while the pavement between a depth of 25 and 75 mm has
undergone softening damage. In the later section on field validation an example is
provided containing more details on this topic. In addition, the field calibration section
describes how the post-processed information is used to predict the extent of thermal
cracking using a probabilistic crack distribution model.

6. Probabilistic Crack Distribution Model

To avoid the modeling complexities and computation expense needed to simulate


multiple thermal cracks, the scheme used in the original TCMODEL to translate a single
thermal crack depth prediction to thermal crack density (spacing) was adopted in the
current version of IlliTC (IlliTC v1.0). The modelling of multiple thermal cracks, while
more exact, was not deemed as being worth the added computational expense, since the
point at which crack interaction occurs is well within the range of severe cracking. Since
it is unlikely that a designer would use a high cracking level as a design target, it was
decided that multiple cracks would not be considered in this version of IlliTC. Rather,
the model completes its execution once a high level of cracking is reached (200 m of
transverse cracking per 500 m of pavement, which corresponds to a crack spacing of 10
m).
This probabilistic crack distribution model converts the computed crack depth of a
single modeled crack (viewed as a representative thermal crack, having a crack depth

109

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

representing an average crack depth) to an amount of thermal cracking (crack frequency)


with the following expression:

C=
f 1 PR ( log C > log hac ) [1 ]

log C / hac
C=
f 1 N [2 ]

Where:
C f = Predicted amount of thermal cracking (m/500m) at a given simulation time
1 = Multiplier representing maximum thermal cracking level
N ( x ) = Standard normal distribution evaluated at x
= Standard deviation of the log of the depth of cracks in the pavement
C = Depth of crack predicted by IlliTC at a given simulation time
hac = Thickness of asphalt layer being simulated (generally taken as the thickness
of all asphalt layers)
Crack amount (m/500m) can be converted to thermal crack spacing by dividing
predicted crack amount, C f , by lane width (typically assumed to be 4 m), and taking
the inverse of this quotient and multiplying by the unit section length (500 m). Citing
the example provided earlier in this section, a crack amount of 200 m corresponds to:
(1/(200/4))*500 or 10 meters. This corresponds to 1000/10 or 100 full-lane-width cracks
per km, which corresponds to approximately 161 thermal cracks per mile. This
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

corresponds to the maximum thermal cracking level predicted by IlliTC in the current
version. Note, similar to the approach taken in the development of TCMODEL, the
parameters 1 and were taken as model calibration parameters. Since thermal
cracks are difficult to detect until they propagate completely through the pavement, it
would be extremely difficult to directly measure and assess the parameter. Thus, its
selection as a model calibration parameter is a practical means to circumvent the need to
directly measure .

110

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

7. Field Calibration

The MnROAD full scale pavement test sections were used to calibrate IlliTC, namely;
sections 03, 19, 33 and 34. Details about these sections can be found elsewhere
(Marasteanu et al., 2007). A decision needed to be made with regards to the climatic
files used in model calibration, since two approaches were possible: (1) use the actual
time ranges corresponding to the field thermal cracking data for each section simulated,
or; (2) use the climatic files available in IlliTC. The argument for using the actual time
ranges that correspond with the field data is that predicted critical cooling events would
match actual critical events in the crack history data files, leading to more accurate
thermal cracking predictions for model calibration. The argument for using the climatic
files included in IlliTC is that future pavement simulations conducted using IlliTC would
be expected to utilize these climatic files (unless the user takes the effort to modify
IlliTC to utilize alternate climatic files, which is a cumbersome process in the current
version of the software). Considering that most users will likely use the climatic files
provided in IlliTC, and also considering that the model should be re-calibrated to local
conditions rather than rely on the calibration provided herein, it was decided to conduct
model calibration using the climatic files provided in the current version of IlliTC.
The subsequent subsections present the results from the preanalyzer and the finite
element analysis engine for each of the calibration sections. This is followed by a brief
description of actual calibration of parameters in the probabilistic crack distribution
model.

7.1. Preanalyzer Runs

Selected outputs from IlliTCs preanalyzer are provided in Figures 8 (a) through (e), and
summarized in Table 4. By comparing Figure 8 (a) with Figure 8 (b) through (e), it is
clear that the days with the coldest temperatures correspond to the events with the
highest surface tensile stress. Table 4 shows that one critical cooling event was
computed for MnROAD section 03 during the simulated 5-year analysis period, while
four, one, and zero critical cooling events were predicted for sections 19, 33, and 34
respectively. Comparing the number of computed critical cooling events with field
cracking behavior indicates the correlation between mixture viscoelastic behavior (as
captured by the creep compliance master curves) and cracking behavior. The correlation
between mix creep compliance and fracture behavior was also demonstrated via a
statistical analysis in the previous phase of this study (Marasteanu et al., 2007).

111
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
(a) Pavement surface temperature using default climatic files in IlliTC for MnROAD site
(in the Category of Moderate Climate within the State of Minnesota).

(b) Thermal stress on pavement surface for MnROAD03 from preanalyzer (Bold
horizontal line indicates 80% of tensile strength).

(c) Thermal stress on pavement surface for MnROAD19 from preanalyzer (Bold
horizontal line indicates 80% of tensile strength).

(d) Thermal stress on pavement surface for MnROAD33 from preanalyzer (Bold
horizontal line indicates 80% of tensile strength).

112

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

(e) Thermal stress on pavement surface for MnROAD34 from preanalyzer (Bold
horizontal line indicates 80% of tensile strength).

Figure 8. Results from preanalyzer for calibration field sections.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Table 4. Preanalyzer results (number of critical events) compared to field cracking.

Number of Critical
MnROAD Field Cracking
Events (as predicted Binder Grade
Cell (m/500 m)
by Preanalyzer)
PEN 120/150
03 1 182
(PG58-28)

19 4 AC20 (PG64-22) 547

33 1 PG 58-28 91

34 0 PG 58-34 6

7.2. Finite Element Runs

Sample finite element modeling results from the calibration phase of the study for
MnROAD section 19 is provided in Figure 9 (a) through (d). The various aspects of the
cohesive zone based finite element modeling approach can be seen in these stress and
(exaggerated) deformed structure plots, which show the elevation view of the asphalt
layers in the vicinity of the modeled crack. The progression of stress build-up, crack
initiation, and crack propagation can be tracked as follows:

Figure 9(a) shows high surface tensile stress (as indicated by the dark color
contours), and a slight disruption in the contours at the crack interface caused

113

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

by the early stages of damage (post-peak softening when tensile stress exceeded
material strength at the surface of the pavement) at -23.3C surface
temperature. Mild compression is still present in the lower regions of the
pavement, due to time-lag effects of heat flow.
Figure 9(b) shows that a thermal crack has propagated partially downward
through the pavement at a temperature of -24.3C, and that a fracture process
zone of about 15% of the pavement thickness exists ahead of the current crack
tip location, illustrating one of the features of the cohesive zone modelling
approach (length scale of fracture is directly considered). A compression zone
still exists near the bottom of the asphalt layer.
Figure 9(c) shows a later stage of crack propagation, where the fracture process
zone has grown in size, and demonstrates that a compression zone no longer
exists (which may partially explain the expansion of the fracture process zone)
at a temperature of -25.4C.
Figure 9(d) shows a fully formed crack, occurring around -29.6C. In reality,
IlliTC considers the pavement section as fully cracked prior to this analysis
step, as described below.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

114

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

Thermal Stress in Longitudinal


Thermal Stress in Longitudinal
Direction near the Crack Path (MPa)
Direction near the Crack Path (MPa)

(a) Thermal stress built-up along (b) Partial depth softening damage
longitudinal direction (Surface temperature = -24.3C)
(Surface temperature = -23.3C)

Thermal Stress in Longitudinal Thermal Stress in Longitudinal


Direction near the Crack Path (MPa) Direction near the Crack Path (MPa)
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

(c) Partial depth crack (d) Fully formed crack


(Surface temperature = -25.4C) (Surface temperature = -29.6C)

Figure 9. Results from finite element analysis for MnROAD section 19 (calibration field
section). (Tensile strength = 4.22 MPa)

7.3. Model Calibration Discussion

Model calibration in pavement studies acknowledges the significant complexities


associated with pavement materials, construction, climatic effects, traffic loading and

115

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

performance. Pavement performance model calibration is almost always needed as a


result. In the case of thermal cracking, factors such as construction variability, inability
to model aging and aging gradients with accuracy, approximate nature of tests and
material models, approximate nature of climatic records and pavement temperature
predictions, presence of load-associated effects and damage on pavement, etc., exist and
result in the need for model calibration.
A number of factors were available to be used for model calibration, including:
fracture energy multiplier, tensile strength multiplier, thermal coefficient multiplier,
crack tip definition, and beta ( 1 ) and sigma ( ) parameters from the probabilistic
crack distribution model were readily available for use in model calibration. As a
preliminary approach, it was decided to leave the material property factors as
uncalibrated, and focus on the following three factors for model calibration: crack tip
definition, beta ( 1 ) parameter, sigma ( ) parameter. Crack tip definition refers to the
fact that more than one material state can be considered as the point of crack initiation in
the cohesive zone modeling technique. For instance, in Figure 5, any point along the
post-peak softening curve (declining linear function in the case of the bi-linear cohesive
zone model, which represents the gradual accumulation of material damage and loss of
load carrying capacity across the forming crack as the material separates) could be
selected as the arbitrarily chosen location of the crack tip. The point at where the
softening curve reaches zero traction (the right hand limit of the plot shown in Figure 5)
is arguably the point where the material no longer possesses the ability to heal.
However, it can also be argued that an intermediate point along the softening curve may
be a realistic choice for the crack tip. After examining the finite element results from the
MnROAD calibration finite element runs, it was decided that the crack tip would be
defined as the point in the pavement along the line of cohesive zone elements where a
softening threshold of 75% post-post peak decay of material strength (25% traction
remaining) is reached. In addition, an identical beta factor ( 1 = 400 m of cracking per
500 m of pavement section) as used in the original TCMODEL as calibrated in the
MEPDG would be used. Finally, the sigma parameter in the probabilistic crack
distribution model was calibrated to a value of = 1.1.
The results of the calibrated IlliTC model, using MnROAD pavement sections, are
presented in Table 5. As can be seen, reasonable modeling predictions were achieved
for three out of the four sections evaluated. For instance, MnROAD section 19, which
experienced very high pavement cracking due to the use of an AC-20 binder (PG 64-22)
in a PG XX-34 climate, was predicted to have a maximum level of cracking. Recall that
when the maximum predicted crack depth is reached (crack depth = thickness of
pavement), this implies that the average crack is equal to the pavement thickness. Thus,
half of other pavement cracks will be less than the thickness of the pavement, and
therefore, not yet counted as thermal cracks. Stated otherwise, the probabilistic crack
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

116

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

distribution model has a maximum cracking level of 200 m of cracking per 500 m
section, when 1 = 400. MnROAD section 33 was found to have a cracking level of 94
m of cracking, as compared to a measured level of 91 m (this was the section that drove
the calibration of = 1.1), and MnROAD section 34 was found to have a cracking
amount of 0 m as compared to a measured cracking level of 6 m. The only poor
prediction that resulted was for MnROAD section 03, where zero cracking was predicted
as compared to 182 m of measured cracking. It should be noted, however, that IlliTC
did indicate that softening damage had begun to occur in this section (although not
enough to reach the 75% softened threshold). It is also acknowledged that the time
period for the IlliTC simulation was shorter than the period of field performance
reported for Section 03. In addition, the master curve data used in the calibration was
less-than-optimal, with data from only two test temperatures being available (three is
preferred). Rather than add additional calibration factors to IlliTC, it was decided that
the aforementioned calibration parameters were sufficient for the calibration of IlliTC.
However, it is recommended that IlliTC be recalibrated to local conditions to arrive at
better model accuracy. Model validation using an independent data set is provided in the
next section of the paper.

Table 5. IlliTC model calibration results.


MnROAD Measured Field Predicted Field
Binder Grade
Cell Cracking (m/500 m) Crackinga (m/500 m)
PEN 120/150
03 182 0
(PG 58-28)
AC-20 > 200 (max. allowable
19 > 200 (547)
(PG 64-22) cracking)
33 PG 58-28 91 94
34 PG 58-34 6 0
a
Predictions are made using non-synchronized climate files

8. Field Validation

Five pavement sections were constructed in Olmsted County, Minnesota during the 2006
construction season, which were used for the analysis portion of the validation process.
The mixes were sampled during the construction process and were characterized
extensively in the lab. Detailed information about the pavement sections and mixture
properties obtained experimentally are documented elsewhere (Marasteanu et al., 2012).

117
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

The five Olmstead Co. pavement sections were simulated using IlliTC, and the results
from the preanalyzer and finite element analysis are briefly presented in the following
section.
Table 6 presents the results from the IlliTC preanalyzer routine. As can be seen,
only Section 4 experienced a critical tensile stress level in the five years analyzed. The
preanalyzer thermal stresses for this section are presented in Figure 10. Since this was
the worst section in terms of field cracking, this indicates that the IlliTC program has
correctly ranked the five field sections.

Table 6. IlliTC preanalyzer results and field cracking for the validation sections.
Number of
Field Performance
Critical Events
Validation Section (Transverse
(Predicted by
cracking m/500m)
Preanalyzer)
1 0 23 (low)
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

2 0 2 (very low)
3 0 29 (low)
4 1 53 (moderate)
5 0 25 (low)

Figure 10. Preanalyzer results for validation section 4 (Bold horizontal line indicates
80% of tensile strength).

Based on the preanalyzer results, the validation section 4 was selected for detailed
presentation herein. The thermal stresses and damage predicted by IlliTC for Section 4 is
presented in Figure 11. These results show that while zero cracking was predicted,
softening was activated along the cohesive zone fracture elements. As discussed
previously, the IlliTC system uses the probabilistic crack distribution model to predict
field cracking from the finite element analysis. Based on the preanalyzer and finite

118

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

element analysis the IlliTC field cracking predictions for validation sections are
presented in Table 7.

Softening near the


top of pavement

Thermal Stress in Longitudinal Direction near the Crack Path


(MPa), Tensile strength = 4.37 MPa

Figure 11. Thermal stresses at the end of the critical event for validation section 4.

Table 7. IlliTC predictions and field cracking for the validation sections usaing
calibrations discussed in previous section.
Field Performance
Validation Predicted Crack Predicted Field
(Transverse
Section Depth (mm) Cracking (m/500 m)
cracking m/500m)
1 0 0 23
2 0 0 2
3 0 0 29
4 0 0a 53
5 0 0 25
a
softening was predicted, indicating that thermal cracking would likely result if a longer
analysis period was used.

As evident from Table 7 under the current calibration parameters established in the
previous section, zero cracking was predicted for all sections. Given the fact that most
of the sections have experienced low cracking to date, it can be concluded that IlliTC
under its current calibration is slightly under-predicting the cracking behavior for these
sections. It should also be noted that a limited amount of creep compliance data were
available for these sections (testing at two temperatures instead of the preferred three), so

119

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

errors caused by incomplete compliance data could also have contributed to the under
prediction observed. Given the fact that one of the five sections in the calibration data
set was also under-predicted, the validation trials here may suggest that IlliTC should be
recalibrated to produce higher levels of cracking. However, given the limited validation
data available and since local calibration is recommended before implementing IlliTC in
a given region, further calibration of IlliTC using the current field data was not pursued
herein.

9. Summary, Conclusions and Future Research Tasks

A new thermal cracking model called IlliTC was developed as part of a recently
completed pooled fund study on low temperature cracking. Various components of the
IlliTC model and their verification along with model calibration and preliminary
validation were presented, including: a user-friendly graphical interface called Visual
LTC; a finite element modeling engine involving viscoelastic bulk material and cohesive
zone fracture elements, and; a probabilistic crack distribution model, identical to the one
used in the original TCMODEL program. Model calibration strategies, including a
discussion regarding crack tip definition in the cohesive zone modeling scheme was also
presented.
The model calibration was conducted using four MnROAD sections, three of which
(19, 33, and 34) were found to have very good model predictions after calibration. The
only unreasonable prediction that resulted was associated with MnROAD section 03,
where zero cracking was predicted as compared to 182 m of measured cracking. It
should be noted, however, that IlliTC did in fact indicate that softening damage had
begun to occur in this section (although not enough to reach the 75% softened
threshold). The time period for the IlliTC simulation in this case was shorter than the
period of field performance reported for Section 03. In addition, the master curve data
used in the calibration was less-than-optimal, with data from only two test temperatures
being available (three is preferred). Very limited model validation data was available,
with only limited cracking observed to date in the four validation sections studied
(Olmstead Co., Minnesota). IlliTC predictions were in general agreement with the
observed field cracking (zero cracking predicted vs. very low to low cracking observed
in most sections). The calibration and validation sections indicated that IlliTC should
probably be recalibrated to produce slightly larger cracking predictions; however, it is
recommended that such calibration be performed by highway agencies, designers, or
researchers using local material properties and local field performance data.

120
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

Over the course of this study several future research tasks were identified that would
further enhance the capabilities and applications of the IlliTC system. Some of the future
implementation tasks for IlliTC are as follows:
The current model assumes a constant coefficient of thermal expansion and
contraction; the future version can include the bilinear thermo-volumetric
trend observed by researchers from University of Wisconsin at Madison.
The effects of oxidative aging are not included in the current simulations;
the aging will have very pronounced effect on the thermal cracking
performance due to stiffening and embrittlement of asphalt mixtures near
the pavement surface. A material aging model, such as one used in
AASHTO MEPDG, can be utilized in the IlliTC system to include the effect
of aging. Furthermore, at low temperatures asphalt binders may also exhibit
significant non-oxidative stiffening (steric hardening), the effect of steric
hardening on cracking performance should be evaluated.
The IlliTC system does not account for multiple asphalt mix types, the
extension to multiple material types requires minor modifications to the
software without need for new development in finite element analysis code.
It is anticipated that the future version will have this capability.
The current IlliTC utilizes single fracture energy input at one temperature;
the system should be modified to allow users to input fracture properties at
multiple temperatures (or in functional form).
The current program is currently calibrated using global fracture energy for
finite element simulations; however, researchers have shown that use of
local fracture energy improves the prediction accuracy. Local fracture

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
energy can be determined using the raw data from fracture tests and the
viscoelastic properties of the mix, along with inverse analysis by modeling
the test specimen. This capability should be added to a subsequent version
of IlliTC.
An algorithm recently developed to extract mixture tensile strength fro m
DC(T) test data should be added to IlliTC, so that users can avoid the need
to directly input mixture tensile strength data.
The effects of variability in material property inputs on the predicted
thermal cracking performance should be evaluated.
A much more comprehensive calibration and validation of IlliTC is needed;
the current calibration was limited to four pavement sections at one location.

121

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

With more extensive calibration, the prediction capability of IlliTC should


be improved, which can be evaluated using additional validation sections.

10. Acknowledgement

The authors are grateful for the support provided by the sponsors and partners of Federal
Highway Administration (FHWA) Pooled Fund Study TPF-5(132). Any opinions
expressed herein are those of the writers and do not necessarily reflect the views of the
sponsors.

11. References

Apeagyei, A.K., Dave, E.V., Buttlar, W.G., Effect of Cooling Rate on Thermal
Cracking of Asphalt Concrete Pavements, Journal of Association of the Asphalt
Paving Technologists, Vol. 77, p. 709738, 2008.
Baek, J., Ozer, H., Wang, H., Al-Qadi, I.L., Effects of Interface Conditions on
Reflective Cracking Development in Hot-Mix Asphalt Overlays, Road Materials
and Pavement Design, Vol. 11, No. 2, p. 307335, 2010.
Brown, E.R, Kandhal, P.S., Roberts, F.L., Kim, Y.R., Lee, D.-Y., Kennedy, T.W., Hot
Mix Asphalt Materials, Mixture Design and Construction, Lanham, MD, National
Asphalt Pavement Association, 2009.
Buttlar, W.G., Roque, R., Reid, B., Automated Procedure for Generation of Creep
Compliance Master Curve for Asphalt Mixtures,, Transportation Research Record,
No. 1630, p. 2836, 1995.
Buttlar, W.G., Sahu, R., Behnia B., Dave, E.V., Determination of Asphalt Concrete
Tensile Strength from the Disk-shaped Compact Tension (DCT) Fracture Test, To
be submitted for journal publication, 2012.
Dave, E.V., Buttlar, W.G., Low Temperature Cracking Prediction with Consideration
of Temperature Dependent Bulk and Fracture Properties, Road Materials and
Pavement Design, Vol. 11-SI, p. 3359, 2010.
Dave, E.V., Leon, S.E., Park, K., Thermal Cracking Prediction Model and Software for
Asphalt Pavements, Proceedings of the First T&DI Congress 2011, ASCE
Conference Proceedings, Vol. 398(41167), p. 6472, 2011.
Dave, E.V., Paulino, G.H., Buttlar, W.G., Viscoelastic Functionally Graded Finite
Element Method with Recursive Time Integration and Applications to Flexible
Pavements, Int. J. Numer. Anal. Meth. Geomech., Vol 36, No. 9, p. 11941219,
2012.

122
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

Deme, I.J., Young, F., Ste. Anne Test Road Revisited 20 Years Later, Proc. of
Canadian Technical Asphalt Association, p. 254283, 1987.
Islam, M.S., Buttlar, W.G., "Effect of Pavement Roughness on User Costs,"
Transportation Research Record, In Press, 2012.
Kim, Y.-R., Aragao, F.T.S., Allen, D.H., Little, D.N., Damage Modeling of Bituminous
Mixtures Considering Mixture Microstructure, Viscoelasticity, and Cohesive Zone
Fracture,, Canadian Journal of Civil Engineering, Vol. 37, No. 8, p. 11251136,
2010.
Lytton, R.L., Roque, R., Uzan, J., Hiltunen, D.R., Fernando, E., Stoffels, S.M.,
Performance Models and Validation of Test Results, Final Report to Strategic
Highway Research Program; Asphalt Project A-005, SHRP Report A-357, 1993.
Marasteanu, M., Zofka, A., Turos, M., Li, X., Velasquez, R., Li, X., Williams, C.,
Bausano, J., Buttlar, W., Paulino, G., Braham, A., Dave, E., Ojo, J., Bahia, H.,
Gallistel, A., and McGraw, J., Investigation of Low Temperature Cracking in Asphalt
Pavements, Report No. 776, Minnesota Department of Transportation, Research
Services MS 330, St. Paul, MN 55155, 2007.
Marasteanu, M, Moon, K.H., Teshale, E.Z., Falchetto, A.C., Turos, M., Buttlar, W.,
Dave, E., Paulino, G., Ahmed, S., Leon, S., Bahia, H., Arshadi, A., Tabatabaee, H.,
Ojo, J., Velasquez, R., Mangiafico, S., Williams, C., Buss, A., Investigation of Low
Temperature Cracking in Asphalt Pavements National Pooled Fund Study -Phase II,
Report No. MN/RC 2012-23, Minnesota Department of Transportation, St. Paul MN,
2012.
Nam, K., and Bahia, H.U., "Effect Of Binder and Mixture Variables on Glass Transition
Behavior of Asphalt Mixtures," Journal of Association of Asphalt Paving
Technologists, Vol. 73, p. 89119, 2004.
Nam, K., and Bahia, H. U., "Effect of Modification on Fracture Failure and Thermal-
Volumetric Properties of Asphalt Binders," J.Mater.Civ.Eng., Vol. 21, No. 5, p. 198-
209, 2009.
Paris, P.C., Gomez, M.P., Anderson, W.E., A Rational Analytic Theory of Fatigue,
The Trend in Engineering, Vol. 13, p. 914, 1961.
Roseler, J.R., Paulino, G.H., Park, K., Gaedicke,C., Concrete Fracture Prediction Using
Bilinear Softening, Cement and Concrete Composites, Vol. 29, No. 4, p. 300-312,
2007.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Roque, R., Hiltunen, D.R., Buttlar, W.G., Thermal Cracking Performance and Design
of Mixtures Using Superpave(TM), Journal of the Association of Asphalt Paving
Technologists, Vol. 64, p. 718735, 1995a.
Roque, R., Hiltunen, D.R., Buttlar, W.G., Farwana, T., Engineering Properties of
Asphalt Mixtures and the Relationship to their Performance, STP1265, Engineering

123

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

Properties of Asphalt Mixtures and the Relationship to their Performance, ASTM


International, 1995b.
Soares, J.B., Colares de Freitas, F.A., Allen, D.H., Crack Modeling of Asphaltic
Mixtures Considering Heterogeneity of the Material, Transportation Research
Record, No. 1832, p. 113120, 2004.
Song, S.H., Paulino, G.H., Buttlar, W.G., "A bilinear cohesive zone model tailored for
fracture of asphalt concrete considering viscoelastic bulk material," Eng.Fract.Mech.,
Vol. 73, No. 18, p. 28292848, 2006.
Song, S.H., Wagoner, M.P., Paulino, G.H., Buttlar, W.G., "25 Crack Opening
Displacement Parameter in Cohesive Zone Models: Experiments and Simulations in
Asphalt Concrete," Fatigue and Fracture of Engineering Material and Structures,
Vol. 31, No. 10, p. 850856, 2008.
Yi, S., Hilton, H.H., Dynamic Finite Element Analysis of Viscoelastic Composite
Plates in the Time Domain, International Journal of Numerical Methods in
Engineering, Vol. 37, No. 12, p. 40814096, 1994.
Zocher, M.A., Groves, S.E., Allen, D.H., A Three-Dimensional Finite Element
Formulation for Thermoviscoelastic Orthotropic Media,, International Journal of
Numerical Methods in Engineering, Vol. 40, No. 12, p. 22672288, 1997.

12. Discussion

MR. MOHAMMAD MOLAYEM Prepared Discussion: Thanks Dr. Dave for your
good research and presentation. My question is about the reason for choosing cohesive
zone concepts in your model. Cohesive zones are of course simple meshless
computational frameworks, but the drawback is that the crack path should be known in
advance which is not the case for pavement surfaces. So besides the cohesive zone
methodology, why didn't you incorporate other alternatives such as GFEM/XFEM,
viscoelastic fracture mechanics or even linear elastic fracture mechanics, because as you
are modeling the low temperature cracking, the behavior of asphalt concrete tends to be
brittle irrespective of the nature of the material itself and what happens around the crack
tip at micro-scale level.
PROFESSOR ESHAN DAVE: The first and foremost reason was practicality, to be able
to produce software that runs at relatively faster speeds. When we started working on
this, the extended finite elements had just come into the market and were gaining
popularity. You are correct in saying that has a lot of potential and definitely maybe
someday in the future we would go to that. Some of the other methods that you
mentioned (such as, Viscoelastic fracture mechanics or linear elastic fracture mechanics)
are quite highly dependent on the mesh sizes because you are looking at the strain

124
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
IlliTCLow Temperature Cracking Model for Asphalt Pavements

gradients. In that case, then the challenge is that you have to run a mesh dependency
analysis before you simulate your problem. So that was the reason to go with the
cohesive zone model here with traditional finite elements.
PROF. MOLAYEM: Thank you very much.
PROFESSOR REYNALDO ROQUE: Eshan, I think you definitely attack the right
deficiencies, one of them being, of course, the failure limits that were considered in the
original model. My question is whether there any evidence specifically and maybe it
relates to the previous question that this cohesive zone model is applicable to asphalt
mixtures? Is there any real evidence that this is actual behavior or is it just something
that we have taken and used as another way to interpret data without really knowing
whether it is applicable? Frankly, at low temperatures, I dont see that it is applicable.
PROF. DAVE: Thats a very good question. What we have done, and this is was done
probably about seven or eight years ago, we have shown at least at the lab scale where
with the cohesive zone model you can capture or you can at least match a lab behavior
really well. More recently what we have shown is that if you simulate a test using the
cohesive zone model and get materials properties from that, we are able to simulate a
different test using those properties at different loading rates and different temperatures.
So that is one aspect. The other aspect is through a National Science Foundation GOALI
Study, we were able to do something where we took lab measured responses and were
able to predict reflective cracking in asphalt overlays. So that was a slightly different
application, not necessarily a thermal cracking but closely associated one. In that case
also, especially for accelerated pavement test sections, we were able to match the field
behavior pretty well. So we feel pretty confident in using this model.
PROF. ROQUE: The basic mechanism of cracking is what I am questioning. One can
match data and you can do that with most any model but that does not mean that the
mechanism being modeled is representative. In other words, is the reduction in stiffness
or damage really associated because a part of that crack is hanging on or because you are
getting distributed microdamage in a zone around the crack? You are interpreting it one
way, but it may not be that way.
PROF. DAVE: Thats a good question. Currently, there is a study going on at the
University of Illinois where they are trying to use digital image correlation technique
(DIC), again to try to learn more about cracking. I think we need to go to advanced
mechanisms and Prof. Gabriele Tebaldi is here in audience. He is leading a task group
for the RILEM Technical Committee on Mechanisms of Cracking and Debonding in
Asphalt and Composite Pavements (TC-MCD), which is planning an integrated study to
better understand the microscopic damage in asphalt material during cracking. Maybe
computer tomography will tell us something more or DIC will tell us something more to
learn more about what is really happening in the physical sense. Right now those are all
based on what you describe, essentially trying to see whether we can match responses.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

125

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DAVE, BUTTLAR, LEON, BEHNIA, PAULINO

As far as knowing what is really happening to the material in physical sense, I think
work is still underway.
PROF. ROQUE: Okay. Thank you.
DR. M. EMIN KUTAY: Very nice presentation, Eshan. I have a question about your
preprocessing stage. Right before the final analysis, you do one-dimensional analysis
and you're choosing the critical values. Does that mean that you do not accumulate
damage during those small events (i.e., variations of temperature/thermal stresses)? So
you're only looking at the critical event?
PROF. DAVE: We are only looking at the critical events, and that is why I mentioned
the capability to have thermal fatigue in future extensions, if such a thing exists, where
we are continuously accumulating damage. The objective for this work was: Can we run
the model in 10 minutes and be done with it to make it more practitioner friendly? So
right now we are skipping over time and only simulating critical events. Although, lets
say you have five critical events and you got a certain amount of damage or cracking on
event one, then we go through an equilibrium phase, so it would remember how much
damage you had and then go to the next event. But theres no healing built into this
model.
DR. KUTAY: In your opinion, would the strength that you measure at the undamaged
state be same as the strength of the mixture after several cyclic thermal stresses are
applied?
PROF. DAVE: I think strength would be greater. Thats why I said in future, we need to
add that type of cyclic or fatigue related effects.
DR. KUTAY: Okay, thanks.

126
--`,`,,,``,,```,,`,,,,

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-
Term Aging on Permanent Deformation
Characteristics of Asphalt Mixtures
Haleh Azaria* and Alaeddin Mohsenib

a
AASHTO Advanced Pavement Research Laboratory at National Institute of Standards
and Technology, Gaithersburg, MD 20899-8615
b
Pavement Systems Inc., Bethesda, MD

ABSTRACT: Quantification of the effect of short-term conditioning and long-term aging on the
permanent deformation properties of asphalt mixtures is important for a reasonable estimation of
rutting in the field. As the asphalt industry continues to develop more sophisticated mixtures,
measuring the effect of short-term conditioning and long-term aging on performance becomes
even more critical. Significantly more pavements are being built with Reclaimed Asphalt
Pavement (RAP), Shingles (RAS), and warm mix asphalts, which may age very differently from
mixtures with virgin asphalt. The use of warm mix asphalt in particular is rapidly increasing,
which tends to soften the mixture in the early life of the pavement and impose a complex challenge
for determining appropriate laboratory and field conditioning. For these reasons, it is critical that
more precise determination of the effect of aging on permanent deformation properties of asphalt
mixtures be made. This study proposes a new methodology that could be used in conjunction with
the design of conventional asphalt mixtures and any new asphalt technology. Using the
Incremental Repeated Load Permanent Deformation (iRLPD) test, which is the basis of the new

The oral presentation was made by Dr. Azari.

This is a reproduction of an article whose final and definitive form has been published in Road
Materials and Pavement Design 2013 Taylor & Francis. The article is available online at:
http://dx.doi.org/ 10.1080/14680629.2013.812833
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

127

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

methodology, the effect of short-term conditioning and long-term aging on the permanent
deformation properties of asphalt specimens can be evaluated. The minimum strain rate (MSR)
property from the iRLPD test, measured before and after the aging process, will show the rate of
aging and rate of change in the mixtures rutting performance. In this study, six mixtures were
tested according to the iRLPD procedure and the aging effect on permanent deformation
properties of the mixtures was quantified. It is shown that different asphalt mixtures age in very
different ways and this will significantly affect their short-term and long-term rutting
performances. In addition, it was found that there exists a significant interdependence between
short-term conditioning and long-term aging of asphalt mixtures. Thus, adequate short-term
conditioning is important for acceptable long-term rutting performance. The MSR master curve
from the iRLPD test provided the necessary information for estimating rutting, considering the
effect of temperature, tire pressure, and aging. In this respect, rut depths of the six mixtures were
estimated using the hourly pavement temperatures for a 20-year period.

Keywords: asphalt mixtures, rutting, permanent deformation, repeated load, strain rate, short-
term conditioning, long-term aging

1. Introduction

The permanent deformation of an asphalt mixture is a major distress that could result in
unsafe driving conditions such as hydroplaning, accelerating moisture damage, and in
some cases steering problems. The selection of an appropriate mixture with respect to
the climatic conditions and traffic levels of the intended location could significantly
reduce the damage due to rutting. In this regard, a laboratory permanent deformation test
should be able to accurately evaluate a mixture design and provide properties for reliable
prediction of rut depth in the field, by incorporating the main factors affecting permanent
deformation. The stiffening of asphalt due to short-term conditioning and long-term
aging is one of the important elements that should be considered in laboratory testing for
reliable selection of an appropriate mixture and accurate prediction of field rutting.
The most commonly used laboratory tests for measuring resistance of asphalt
mixtures to permanent deformation are the Flow Number (FN) test, the dynamic
modulus test, and the wheel-track test. While the effect of short-term conditioning is
incorporated in results of these tests, the effect of long-term aging is not reflected in the
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

tests properties. In this study, a new test procedure and analysis method referred to as
Incremental Repeated Load Permanent Deformation Test (iRLPD) (Azari and Mohseni
2012, Azari and Mohseni 2013) is used for determining the resistance of un-aged and
aged mixtures to permanent deformation. Unlike the existing high-temperature
laboratory tests on asphalt mixtures, which are either stiffness-based or failure-based, the

128

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

iRLPD test is damaged-based, meaning that the level and duration of the load is selected
to cause incremental micro-damage to the material without causing it to fail. This would
resemble the damage experienced from load applications in the field without masking
the sensitivity of the test to damage caused by other factors being studied such as aging.
During the iRLPD test, asphalt mixture specimens are subjected to multiple stresses
in several increments. The property of the test is the strain per cycle at the end of each
test increment, also known as Minimum Strain Rate (MSR), which represents a unit
damage in the field from the passing of a single axle. A sweep of MSR values from
different stress levels is used to develop the MSR master curve. The master curve
provides high temperature performance of an asphalt mixture at any temperature and
stress condition and can be used to predict rut depth in the field. To evaluate the effect
of aging on rut depth prediction, the change in the mixture property with short-term
conditioning and long-term aging are quantified by measuring MSR after various short-
term conditioning and long-term aging durations. The iRLPD test is especially suitable
for this evaluation since the test increments are applied on the same test specimens
before and after each aging duration.

2. Effects of Short-Term Conditioning and Long-Term Aging on Rutting

The effect of short-term conditioning and long-term aging on performance of asphalt


mixtures is acknowledged in AASHTO R 30, Standard Practice for Mixture
Conditioning of Asphalt Mixtures (AASHTO, 2012). For performance testing, the
standard prescribes short-term conditioning of the uncompacted mixtures at 135C for 4
hours and long-term aging of compacted mixtures at 85C for 5 days. Although the

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
importance of the conditioning /aging on stiffening of asphalt mixtures is a well known
fact, as addressed in AASHTO R 30, the effect of aging is seldom measured and
incorporated into the mixture design and performance prediction. Several researchers
have been aimed to quantify the amount of laboratory long-term aging; however, the
results have not been very successful. This is partly due to the nature of the performance
tests used, which could not capture the effect of aging. During NCHRP Project 9-23, the
dynamic modulus test was used to measure the stiffening of mixtures due to long-term
aging. However, due to high sensitivity of the dynamic modulus to the slump that
typically occurs during long-term aging, a clear increasing trend in stiffness of the
mixtures in the study was not found (Houston et al., 2005).
In another study, Azari (2011) examined the effect of short-term conditioning and
long-term aging durations on FN test results. Although the trend of increase in FN from
the increase in aging duration was observed, the trend was not significant enough to
conclude an appropriate aging duration. The reason was that due to the destructive nature

129

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

of the FN test, separate sets of specimens were used to determine the properties of the
un-aged and aged specimens and, as a result, the effect of aging was masked by the
sample-to-sample variability.

3. Objectives

The main objective of this study is to incorporate the effect of long-term aging on
pavement rut depth prediction using iRLPD test results. The second objective is to
examine the effect of short-term conditioning duration on long-term performance of
asphalt mixtures. The third objective is to rank asphalt mixtures with respect to rutting
resistance using the aged MSR master curves. These objectives were achieved by short-
term conditioning of six mixtures for various durations and measuring the MSR of the
compacted specimens, before and after various long-term aging durations.

4. iRLPD Testing and Analysis Methodology

4.1 iRLPD Testing Protocol

The iRLPD procedure, developed by the AASHTO Advanced Pavement Research


Laboratory (AAPRL), was used for measuring permanent deformation resistance of the
mixtures (Azari and Mohseni 2012, 2013). The test follows the FN test protocol,
AASHTO TP 79, (AASHTO, 2012) in terms of test apparatus, specimen fabrication,
load pulse requirements, and data quality criteria; however, the iRLPD test offers
variations in the number of load applications, test properties, and method of analysis.
Figure 1 shows the Asphalt Mixture Performance Tester (AMPT) that is used for
performing both the FN and iRLPD tests. For a typical FN test, the test continues for
10,000 cycles, until flow is reached (at minimum strain rate), or when permanent strain
exceeds 50,000 microstrain. Figure 2 provides a typical output of an FN test, where the
test is performed at a single combination of temperature and stress (either confined or
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

unconfined). The FN test parameter is the number of cycles at the test termination. On
the other hand, the iRLPD test includes several test increments since the test is stopped
after 500 cycles and restarted at either the same stress level or at higher stress levels,
depending on the purpose of the test. Figure 3 shows the output of an iRLPD test, where
three replicate specimens are each tested at a confining stress of 69 kPa and four
different deviatoric stress levels of 200 kPa, 400 kPa, 600 kPa, and 800 kPa in four 500-
cycle test increments. For the iRLPD test, temperature is fixed at 60C. The property of
the test is the minimum strain rate (MSR) at the end of each test increment. Therefore,

130

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
from testing each specimen, several MSR values could be calculated, as pointed out in
Figure 3.

Figure 1. iRLPD test setup using AMPT loading machine.

Figure 2. A typical output of an FN test; graphs show permanent strain and permanent
strain rate versus number of cycles; the three stages of a repeated load test (primary,
secondary, and tertiary) are also shown in the figure.

131

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

MSR, 4th increment

MSR, 3rd increment


MSR, 2nd increment

Figure 3. A snapshot of the the iRLPD test output; figure shows the permanent strain
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

and permanent strain rate of the four 500-cycle test increments; Figure also shows
minimum permanant strain rate (MSR) at the end of each test increment.

4.2 iRLPD Analysis Method


The analysis of the iRLPD test results involves developing the MSR master curve. The
MSR values from various stresses are plotted versus a parameter TP, which is the
product of temperature (C) and deviatoric stress (MPa). The MSR versus TP is referred
to as the MSR master curve and explains the rutting damage characteristics of a mixture
at any stress and temperature combination. A power model in the form of MSR=a*(TP)b
is fit to the MSR master curve. The coefficient a of the power curve is fixed at 0.001.
Through a series of statistical analyses of the iRLPD data, it was determined that fixing
the a coefficient at 0.001 would result in a better fit to the upper portion of the master
curve, where MSR values are the most important, without adversely affecting the fit to
the lower portion of the curve. Using this approach, each master curve is reduced to a
single number (power coefficient b), which makes it convenient to use the master
curve for mixture ranking and field implementation. Figure 4 shows a plot of MSR

132

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

versus TP (MSR master curve) for a mixture from Wisconsin (WI-E10). The MSR

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
values were measured at a single temperature of 60C and three stress levels of 400 kPa,
600 kPa, and 800 kPa. The TP values of 24, 36 and 48C-MPa are calculated by
multiplying the test temperatures in C and stress levels in MPa for each test increment
(Azari and Mohseni 2012).

4.3. Logic behind MSR Master Curve

In the initial work by the authors (Azari and Mohseni 2012), the iRLPD test was
conducted at multiple temperatures as well as multiple stresses. From the results of the
tests it was shown that the effect of temperature and stress are interchangeable; meaning
that by increasing the stress, the same effect on MSR is captured as by increasing the
temperature. In this respect, the TP variable was defined to explain both effects of
temperature and stress on MSR. The use of TP made possible performing the test at only
one temperature but multiple stresses to obtain the same master curve as that obtained
from conducting the test at multiple temperatures and stresses and then shifting the MSR
versus stress curves from different temperatures into a single MSR master curve.

MSR Master Curve- WI (E10)


5 0 .0
4 5 .0
MSR (mic r ostr ain pe r c yc le )

4 0 .0
3 5 .0
3 0 .0 y = 0.001x2.5839 Inc r e me ntal
2 5 .0
Powe r Mode l
2 0 .0
1 5 .0
1 0 .0
5 .0
0 .0
0 .0 1 0 .0 2 0 .0 3 0 .0 4 0 .0 5 0 .0 6 0 .0 7 0 .0
TP (C * MPa)
Figure 4. MSR master curve of the WI-E10 mixture (4 hour short-term conditioned,
without long-term aging).

133

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

4.4. Advantages of iRLPD Testing and Analysis Methodology

The iRLPD is a unique approach compared to previous methods (Kenis, 1978) in the
sense that multiple stresses are performed on the same specimens, which would remove
sample-to-sample variability. This is essential in capturing the effect of factors such as
aging, which could be masked otherwise. In addition, the definition of MSR at the 500th
cycle, as opposed to the slope of the secondary stage, which was used by previous
researchers (Goh et al., 2011; Von Quintus et al., 2012) is a new definition. The slope of
the secondary stage is not a constant value and is changing with the increase in cycle
number while using the slope at the 500th cycle (MSR) sets a fixed criterion within the
secondary stage of deformation and therefore, provides a more robust and less variable
test parameter. Furthermore, creating the MSR master curve and fitting a power curve
model is an advancement in the analysis of the repeated load test data from various
temperatures and stresses. The repeated load data that is typically conducted at different
laboratories at different stresses and temperatures could never have been compared
before; however by creating the MSR master curves, the results of tests can easily be
compared across different laboratories.
In this study, the application of the iRLPD testing and analysis methodology for
measuring the effect of long-term aging on asphalt mixture rutting performance is
provided. Using the MSR of the long-term aged mixtures, the rutting performance of the
mixtures in this study would be predicted.

5. Materials and Mix Design


Six materials from pavement projects with different design traffic levels in various
climatic regions across the United States were used in this study for the high temperature
characterization of asphalt mixtures subjected to various short-term conditioning and
long-term aging durations. Table 1 provides the sources of these materials, the mixture
ID, asphalt grade, aggregate nominal maximum aggregate size (NMAS), location of the
paving projects, design traffic level, and the pavement temperature selected based on the
LTPPBind 50% reliability performance grade at a depth of 20 mm (LTPPBind V3.1,
2005).

134
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

Table 1. List of mixtures evaluated in this study.

Design LTPPBind High


Asphalt Mixture Temperature, 50%
Traffic Mixture
Location Binder NMAS, Reliability, C
Level, ID
Gr a d e mm
MESAL

Monroe County,
PG58-28 1 2 .5 WI-E3 4 9 .1
Wisconsin
3 Wilmington
County, North PG64-22 9 .5 NC 5 8 .6
Carolina
Marathon County,
PG64-28 1 2 .5 WI-E10 4 6 .7
Wisconsin
10
Crawfordsville,
PG64-22 9 .5 IN 5 3 .3
Indiana
Hernando County,
PG67-22 9 .5 FL 6 3 .0
Florida
30
Kern County,
PG70-10 1 9 .5 CA 6 2 .0
California

6. Experimental Testing
The six mixtures were prepared at different short-term conditioning periods. A total of
126 specimens were prepared using seven short-term conditioning periods of 0 hours to
6 hours, in 1-hour intervals. The mixtures were then compacted and tested using iRLPD.
Each specimen was subjected to three test increments at three stress levels before the
long-term aging and one increment after the long-term aging to develop the MSR master
curves for un-aged and aged samples. The long-term aging included heating the
specimens at 85C for 2, 5 or/and 9 days. The one increment of iRLPD after long-term
aging was conducted at the highest stress level that was applied to the specimen before
long-term aging.

135

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

7. Experimental Results

7.1. Effect of Long-Term Aging on Power Coefficient b

MSR master curves for various short-term conditioning and long-term aging scenarios
were developed for all six mixtures. Figure 5 shows the master curves for the IN
mixture test specimens, short-term conditioned for total of three hours at 135C prior to
compaction and long-term aged at 85C for 0, 2 and 5 days after compaction. To develop
the un-aged master curve, the power function MSR=0.001* (TP) b was fit into the un-
aged MSR versus TP data. The un-aged b coefficient was determined from the fit. For
the aged master curve, the measured MSR of the aged specimens and their
corresponding TP were substituted into the power function to calculate the aged b
coefficient. Figure 5 shows that exponent b for this mixture is 2.68 prior to the long-
term aging and 2.48 and 2.37 after 2 and 5 days of long-term aging at 85C, respectively.
The reduction in the b coefficient indicates the increase in resistance to permanent
deformation of the IN mixture due to long-term aging.

IN Mix: MSR Master Curves for 3 Hours Short-


term Conditioning and Various Long-term Aging
20

18 No LT Aging

16 2 Days @ 85 C

14 5 Days @ 85 C
MSR, Microstrains

12

10
MSR = 0.001 TP2.68
8 R = 0.9981

6 MSR = 0.001 TP2.48


R = 0.9958
4
MSR = 0.001 TP2.37
2 R = 0.9991

0 --`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

0 5 10 15 20 25 30 35 40
TP, Deg-Mpa

Figure 5. Sample MSR master curve for IN specimens, un-aged and long-term aged for
2 and 5 days (LT in the graph stands for long-term).

136

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

7.2. Effect of Short-Term Conditioning on Power Coefficient b


Figure 6 shows the power exponent b for various short-term conditioned and long-
term aged specimens of the six mixtures. The slope of the fitted lines to b coefficients
reveals important information regarding the susceptibility of the mixtures to short-term
conditioning duration. The slope of the lines for no long-term aging indicate the
susceptibility to short-term conditioning right after construction and the slope of the b
lines corresponding to long-term aging indicates the effect of short-term conditioning
after several years that the road has been in service. Considering the effect of short-term
conditioning right after construction, the E3 mixture is most susceptible to short-term
conditioning as indicated by a slope of -0.079 for the un-aged b line, the NC mixture is
the second most susceptible (slope of -0.0571), and E10 is the third most susceptible to
short term conditioning (slope of -0.0489). This indicates that for these mixtures, the
duration of short-term conditioning is the most critical for right-after-construction rutting
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

performance. The slopes of un-aged b lines for FL and CA are the lowest (-0.010 and
-0.030), indicating the lowest effect of short-term conditioning on the right-after-
construction performance.
The slope of the b values versus short-term conditioning time for the aged mixtures
indicates the susceptibility of the mixtures to the effect of short-term conditioning after
the pavement has been in service for some period of time. As shown in Figure 6, the
slopes of the fitted b lines for the 2-days and 5-days long-term aged WI-E3 mixtures
are the largest (-0.082 and -0.077). Thus, this mixture is considered the most susceptible
to short-term conditioning even after long-term aging. The next most susceptible mixture
to short-term conditioning after long-term aging is WI-E10 with slopes of -0.058 and
-0.041 from 2-days and 5-days of long-term aging, respectively. This indicates that for
WI-E3 and WI-E10, the effect of short-term conditioning remains significant even after
the roads have been in service for several years. FL is the least susceptible to the effect
of short-term conditioning after long-term aging, as indicated by the smallest slopes of
its aged b lines (-0.010 for the specimens aged both 2 days and 5 days). Therefore, the
duration of short-term conditioning has the least effect on the long-term performance of
Florida mixture.

137

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Figure 6. Change in exponent b of master curve with change in short-term con-
ditioning and long-term aging.

7.3. Comparison of Aging Rates using Power Coefficient Ratios


The power coefficient b is used to compare the rate of change in resistance to
permanent deformation of the long-term aged specimens of the six mixtures. In this
respect, the power coefficient b of every mixture for various aging times was divided
by the power coefficient of the un-aged specimen to obtain power coefficient ratios.
Figure 7 shows the power coefficient ratios of the long-term aged specimens of the six
mixtures for the 4-hour short-term conditioning. It is evident from the figure that for
these mixtures, the rate of decrease in power coefficient ratios (aging rate) as a result of

138

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

long-term aging is very different from each other. As shown in Figure 7, the Wisconsin
mixtures E10 and E3 had the highest aging rate. These two mixtures are used in regions
of Wisconsin where the asphalt is not exposed to very high temperatures that cause
excessive aging (the 50% reliability high pavement temperature from LTPPBind is
49.1C and 46.7C for the WI-E3 and WI-E10 mixtures, respectively). Therefore, in
terms of aging, the two Wisconsin mixtures are appropriate for their environmental
conditions, where rapid aging of the mixtures is not facilitated. Figure 7 also shows that
the CA mixture had the lowest aging rate. This mixture is used in a region where
prolonged heat is known to cause excessive aging. Therefore, in terms of long-term
aging effect, the CA mixture is appropriate for its environmental condition.

Normalized Power Coefficent Ratios for 4 Hour


Short-Term Conditioning
1 .0 5

1
Power coefficient Ratio

0 .9 5
CA
0 .9 E1 0

0 .8 5 E3
FL
0 .8
IN
0 .7 5
NC
0 .7
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

0 2 4 6 8 10
Days of Long-term Aging

Figure 7. Normalized power curve coefficients b versus long-term aging for all six
mixtures.

7.4. Reduction in MSR with Short-Term Conditioning and Long-Term Aging


Another means for evaluating the effect of long-term aging on permanent deformation
properties of the mixtures is determining the percent reduction in MSR due to long-term
aging. MSR is a measure of unit damage per load cycle and reduction in MSR represents

139

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

the reduction in permanent deformation damage. The power equations of the MSR
master curves (MSR= a *TPb) were used to calculate the MSR at a reference temperature
of 60C and a tire pressure of 600 kPa for various short-term conditioning and long-term
aging periods. The percent reduction in MSR due to long-term aging was then calculated
using the un-aged and aged MSR values. Figure 8 shows the reduction in MSR due to
long-term aging for various short-term conditioning durations for all mixtures.
Irrespective of the duration of short-term conditioning, the reduction in MSR after 5
days of long-term aging for the FL and CA mixtures was about 65% and 40%,
respectively. The reduction in MSR after 5 days of long-term aging for the two
Wisconsin mixtures (E3 and E10) was around 75%.
The reduction in MSR of the other three mixtures in Figure 8 shows a somewhat
different trend. While for the CA, FL, and the two Wisconsin mixtures, the reduction in
MSR was the same regardless of the duration of short-term conditioning, the reduction
in MSR of the long-term aged specimens of IN and NC are significantly affected by
short-term conditioning duration. The NC mixture had the highest reduction (80%) in
MSR if it was short-term conditioned for 2 hours, but the lowest reduction (45%) if it
was short-term conditioned for 4 hours. The IN mixture also had a lower reduction in
MSR at 4 hour conditioning compared to 2 hour conditioning. The results of IN and NC
mixtures show significant interdependency between the effect of short-term and long-
term conditioning on these mixtures. This observation indicates the importance of
enforcing a specification for the duration of short-term conditioning in asphalt plants
and/or quantification of the effect of short-term conditioning for accurate long-term rut
depth prediction.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

140

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

Reduction in MSR with Long-term Aging


(2 Hour short-term conditioning)
1 0 0% CA
90%
80% E 10
70%
Re duct ion in MSR

E3
60%
50% FL
40%
IN
30%
20% NC
10%
0%
0 2 4 6 8 10
Days of L ong-t e r m Aging

Reduction in MSR with Long-term Aging


(4 Hour short-term conditioning)
1 0 0% CA
90%
80% E 10
70%
Re duct ion in MSR

E3
60%
50% FL
40%
IN
30%
20% NC
10%
0%
0 2 4 6 8 10
Days of L ong-t e r m Aging

Reduction in MSR with Long-term Aging


(6 Hour short-term conditioning)
1 0 0% CA
90%
80% E 10
70%
Re duct ion in MSR

E3
60%
50% FL
40%
IN
30%
20% NC
10%
0%

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
0 2 4 6 8 10
Days of L ong-t e r m Aging

Figure 8. Percent reduction in MSR after long-term aging of asphalt mixtures.

141

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

8. Performance Prediction Using MSR Master Curves


A computer program was developed to estimate 20-year rut depths using the MSR
master curves of the six mixtures. The program includes estimated hourly pavement
temperature data at a depth of 20 mm. The asphalt layer depth was assumed to be 50

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
mm, which is the depth at which rutting is believed to be developed; thus the
temperature at a depth of 20 mm would correspond approximately to the mid-depth of
the surface layer. The hourly pavement temperature was calculated from hourly air
temperature, wind speed and sky cover using estimated pavement parameters (MEPDG,
2004). Figure 9 shows a screen shot of the software. In the material section, master
curves of the IN mixture are chosen for analysis. The weather data from a weather
station located at Indianapolis was used for estimating the temperature. Traffic was
selected at 20 million ESALs and tire pressure was assumed to be 0.6 MPa. The MSR
values for the short-term conditioning of 4 hours were used for the prediction. The top
bar chart shows the number of hours the pavement was at certain temperatures for the
month of July of the first year. It can be seen that the pavement temperature reached
62C in that month. The bottom bar chart shows the calculated rutting for every degree
centigrade. As temperature increases, the rutting damage also increases. It is shown that
although there were only a few hours with 62C, the damage at this temperature was still
significant. Rutting for one month is the sum of the rutting for all temperatures. The
MSR master curve for every month is different due to the effect of long-term aging; as
the pavement ages the permanent deformation rate decreases. The total 20-year rutting
is the sum of all rutting for all months.

Figure 9. Computer screen of the software for prediction of rutting.

142

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

Table 2 shows the estimated 20-year rutting for all mixtures using hourly pavement
temperatures at a depth of 20 mm. The MSR master curve for long-term aging was used
to discount the rutting damage for the absorbed heat. Rutting was calculated for three
different short-term conditioning prior to compaction: 2 hours, 4 hours, and 6 hours. As
shown in the table, there is a significant difference between rutting estimated for the
mixtures with 2 hours and 6 hours short-tem conditioning. The highest difference is for
the E3 mixture (110%) and the lowest is for the CA mixture (47%). Therefore,
uncertainty about the length of short-term conditioning before the mixture is placed and
compacted may result in significant under or overestimation of rutting.
Table 2. Estimated rutting for mixtures at their design location and traffic levels;
STC stands for short-term conditioning.

Rutting after 20 years, mm Percent


Increase
Design between 2
Traffic, 2 hour 4 hour 6 hour and 6
Mix Location MESALs STC STC STC hours STC
WI-E3 Madison 3 4 .4 3 .1 2 .1 1 0 9 .5 %
NC Raleigh 3 3 .5 3 1 .9 8 4 .2 %
WI-E10 Madison 10 8 .8 7 .7 5 .1 72.6%
IN Indianapolis 10 4 .6 3 .6 3 .1 48.4%
FL Tampa 30 8 6 .1 5 .4 48.2%
CA Los Angeles 30 2 5 .2 2 2 .2 1 7 .1 47.4%

9. Summary of Findings
The following provides a summary and the conclusions of this study:

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
1) A method for quantifying the effect of short-term conditioning and long-term
aging on the permanent deformation properties of asphalt specimens was proposed. The
MSR master curve from the iRLPD test provides the key information for estimating rut
depth; considering effects of temperature, tire pressure, and aging.
2) Knowledge of the change in material properties during long-term aging is crucial
in estimating asphalt mixture rut depth. The iRLPD test is particularly useful for this
evaluation since it allows testing the same specimens before and after long-term aging
without sample-to-sample variability. The change in the MSR master curve of a single

143

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

specimen with increase in laboratory long-term aging duration provides the necessary
information for the long-term performance prediction.
3) Different asphalt mixtures may age in very different ways and this will
significantly affect long-term properties of asphalt mixtures. For this reason, it is critical
that the effect of long-term aging be quantified for each asphalt mixture and the
information be used in performance prediction.
4) This study shows that there is a significant interdependency between short-term
conditioning and long-term aging of some asphalt mixtures. The short-term conditioning
of asphalt mixtures affects the long-term aging, and the effect is more significant for
some mixtures than others. This indicates the importance of having a set specification
for short-term conditioning of the mixtures before construction.
5) Since controlling the short-term conditioning that takes place in a silo could be
difficult, it is critical to quantify the effect of short-term conditioning on the permanent
deformation properties of asphalt mixtures. This can be accomplished by conducting the
iRLPD test right after plant mixing and right before mixture lay down. Without the
knowledge of the various aspects of short-term conditioning in a plant and its effect on
material properties, it will be difficult to provide any reasonable estimate of rutting in the
field.

10. Recommendations
The following recommendations are given from the conduct of this study:
1) The effects of long-term aging on other asphalt properties, such as intermediate-
and low-temperature cracking, also need to be determined. Similar to the permanent
deformation application, the iRLPD test method can be used to quantify the effect of
long-term aging on intermediate and low temperature properties by measuring the
change in minimum strain rate (MSR).
2) Duration of short-term conditioning of asphalt mixtures should be controlled in
the field and its effect should be considered in mixture performance. This study has laid
out a method for determining the effect of short-term conditioning on permanent
deformation. In the field, the reduction in MSR of a plant-produced mixture right after
mixing and right before compaction should be monitored and considered in prediction of
long-term performance.
3) Asphalt technology is changing by the day. Significantly more roads are built
using Reclaimed Asphalt Pavements (RAP) and shingles (RAS), which may age very
differently from virgin asphalt pavements. The use of warm mix asphalt is also rapidly
increasing the trend tend to soften the mix from an early age of asphalt pavement. For

144
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

these reasons, it is even more critical that more precise determination of permanent
deformation properties of asphalt mixtures is made. This study proposes a new method
that can be used in conjunction with new asphalt technology.
4) The long-term aging method of asphalt mixtures in AASHTO R30 should be re-
examined and improved to include the environmental temperature and the amount of
heat absorbed by the mixture in the field. Currently, all long-term aging is performed at
a fixed temperature and time and does not distinguish hard and soft binders or the
location where the mixture will be used. For this reason, the laboratory generated data is
only loosely related to the field conditions.

11. Acknowledgement
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

The authors would like to thank Mr. Steve Lenker, the Director of the Construction
Materials Reference Laboratory, for his support throughout this study. The authors wish
to acknowledge Mr. Frank Fee for coordinating the acquisition of the materials for this
study. The efforts of State DOT and private agencies, which provided the materials and
the corresponding mixture designs, are most appreciated. The agencies are Florida
Department of Transportation, Gainesville, Florida; North Carolina Department of
Transportation; Heritage Research Group, Indiana; MTE Construction, Wisconsin; and
California Department of Transportation.

12. References
Azari, H., "Analysis of the Effect of Laboratory Short-Term Conditioning on
Mechanical Properties of Asphalt Mixture," TRB 90th Annual Meeting
Compendium of Papers DVD, Paper No. 11-1427, Transportation Research Board,
Washington, DC. 2011.

Azari, H. and Mohseni, A., "Incremental Repeated Load Deformation Testing of Asphalt
Mixtures," TRB 91st Annual Meeting Compendium of Papers DVD, Paper No. 12-
4381, Transportation Research Board, Washington, DC. 2012.

Azari, H. and Mohseni, A., "Permanent Deformation Characterization of Asphalt


Mixtures Using Incremental Repeated Load Testing," TRB 92nd Annual Meeting
Compendium of Papers DVD, Paper No. 13-5160, Transportation Research Board,
Washington, DC. 2013.

AASHTO Standard Practice. AASHTO R 30: Standard Practice for Mixture


Conditioning of Hot Mix Asphalt (HMA), Thirty-Second Edition, American
Association of State Highway and Transportation Officials, Washington, DC. 2012.

145

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

AASHTO Provisional Standards. AASHTO TP 79: Standard Method of Test for


Determining the Dynamic Modulus and Flow Number for Hot Mix Asphalt (HMA)
Using the Asphalt Mixture Performance Tester (AMPT), Thirty-Second Edition,
American Association of State Highway and Transportation Officials, Washington,
DC. 2012.
Goh S., You Z., Wang H., Mills-Beale J., Ji J., Determination of Flow Number in
Asphalt Mixtures from Deformation Rate During Secondary State, Transportation
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Research Record, No. 2210, Transportation Research Board, Washington, DC., 2011.
Houston W.N., Mirza M. W., Zapata C., and Raghavendra E. S. Environmental Effects
in Pavement Mix and Structural Design Systems, NCHRP Web Document 113,
2005
Kenis W., Predictive Design Procedures, VESYS Users Manual: An Interim Design
Method for Flexible Pavements using the VESYS Structural Subsystem, Federal
Highway Administration, Report No FHWA-RD-77-154., 1978.
LTPPBind V3.1, 2005
http://www.fhwa.dot.gov/research/tfhrc/programs/infrastructure/pavements/ltpp/inst
all.cfm
MEPDG, Guide for Mechanistic-Empirical Design of New and Rehabilitated Pavement
Structures, NCHRP 1-37A Final Report, Appendix DD-4, National Academies,
Washington DC 2004.
Von Quintus H. L., Mallela J., Bonaquist R., Schwartz C. W., and Carvalho R. L.
Calibration of Rutting Models for HMA structural and Mixture Design, NCHRP
Report 719, National Academies, Washington, DC. 2012.
http://onlinepubs.trb.org/onlinepubs/nchrp/nchrp_rpt_719.pdf,

13. Discussion

PROFESSOR MIHAI MARASTEANU : Nice presentation. I have more of a


philosophical question. Aging is good for rutting, so why do we spend so much time on
aging and rutting? Aging negatively affects fatigue and low temperature cracking. The
work you presented is good, but what I would like to understand is what is the ultimate
purpose to come up with a mixture that ages more and is better for rutting?
DR. HALEH AZARI : You want to select the right mixture for a pavement project and
also to be able to predict the performance of a mixture correctly. Rutting and fatigue and
low temperature cracking this method applies for all of them. We are presenting it for
rutting, but, you want to be able to predict performance for any property. Effect of aging

146

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

needs to be considered in combination with all properties. When designing, the effect of
aging should be considered on both rutting and fatigue and on low temperature cracking
if it is a concern for you.
PROF. MARASTEANU: I understand that. It is ok to accurately predict rutting.
However, it is well known more aging improves rutting and significantly decreases

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
cracking resistance. I would have thought it made more sense to study aging effect on
cracking.
DR. AZARI: The material selection should be a balanced process. In an environment
where the mixture is exposed to prolonged heat, you dont want to place mixtures that
age rapidly. Even though it would be good for rutting, it wont be beneficial for fatigue
cracking. You need to know at what rate your material ages in order to select the right
material for the environment.
PROF. MARASTEANU: Sure. All right. Thank you.
DR. ALAEDDIN MOHSENI : If I may answer. The usefulness of the procedure is that it
tells you if you are going to have a problem with cracking when you place a mix thats
too stiff. If a mix is aging rapidly, then maybe we do not want to place it down when its
too stiff, and its going to help with the cracking later on. So the two work together.
Right now, we dont have a way of distinguishing between a high age material or not. As
shown in some of the graphs here, aging can have tremendous effect on rutting and as a
result on cracking performance in the future.
MR. FRANK FEE: Mixes taken from plant mix where did you obtain these mixes that
you're aging?
DR. AZARI: The mixtures are the same mixtures we used as part of the ETG study, so
they were aggregates and asphalts.
MR. FEE: They were lab mixes.
DR. AZARI: Yes, they were laboratory mixes.
MR: FEE: Okay. So do we have any field performance on these mixes or do you kno w
anything about what they would look like coming out of an asphalt plant in terms of
characterization?
DR. AZARI: All we know about these mixtures is the traffic level they have been
designed for, and we know that they have been performing well for their climatic
condition and traffic level. Some of these mixtures, even without considering the effect
of long-term aging, were predicted to perform well in their projects; but by incorporating
aging, we are making better prediction and less rutting has been predicted for them,
which is closer to the actual performance of those pavements.

147

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
AZARI, MOHSENI

MR. FEE: But in terms of the actual performance in the field, they were all equal, they
were all good, so we dont really know if we could rank them against field performance
at this time. In other words, they were all good. Youre saying that these rank differently,
but in the field, can we show that? I guess thats where Im going with this.
DR. AZARI: Yes. I mean this ranking is in laboratory without consideration of the
traffic level and the climatic region. In their own climatic region, they all performed
well. But when you compared them in laboratory, some performed better than the others.
For example, the Wisconsin mixture is not performing as well as California, but in their
own climatic conditions and projects they both performed well.
MR. FEE: I think you have good work here, its potential is there. As you know, one
of the concerns we have with this difference between the plant and lab aging is early
pavement failure. A material that gets out of the plant and is too soft. It doesnt age
quickly enough and traffic will rut it quickly. As time goes on, we dont worry about it
as much because it continues to stiffen. We need to get field performance on these
relationships. Youve selected certain temperatures and times for the lab aging. Now see
if you can get those to match something that actually happens in the field.
DR. AZARI: That is why we need to know at what rate every asphalt ages and as it was
shown different asphalt mixtures age at very different rates. Mixture that age slowly can
be placed stiffer than mixtures that age rapidly to avoid early rutting and long-term
cracking.
The long-term aging that is proposed in AASHTO R30 is five days at 85C regardless of
where these mixtures are going to be placed. The laboratory aging should be in
consideration of the amount of heat the mixtures will be exposed in the field. For those
materials that are not exposed to a lot of sun, five days of aging might be too long. For
those that are exposed to excess sun, five days of aging might not be enough. So these
need to be considered, and we are using the Degree-Day parameter from LTPPBind to
determine for each particular climatic condition what should be the amount of aging in
the laboratory.
PROFESSOR GABRIELE TEBALDI: You show a performance about mixture in
Madison, Tampa, or in California. So my question is: Have you adjusted your aging
process related to the environmental condition or how can you do that? Because of
course the aging is strongly related with the climatic condition. I do not understand how

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
you can calibrate that related to
DR. AZARI: Exactly. Thats what I just mentioned. For this study, we used two days of
aging, five days of aging, nine days of aging. It was constant for all of these mixtures.
But the correct way of doing it is to determine how much aging takes place in each
environmental condition. If the temperature is too high and if the sun exposure is too
much then the laboratory aging process should be adjusted. All of these are
determined by the Degree-Day parameter from LTPPBind. Dr. Mohseni, actually, can

148

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Effect of Short-Term Conditioning and Long-Term Aging

explain more on the climatic part of the data and on how we incorporate that into the
laboratory aging.
DR. MOHSENI: The actual implementation was that we did estimate the rutting. We
used the climatic data for only one location. That was the Indiana location. Now there
were six mixes from different parts of the United States, but they were actually applied
in only one location. Now if we want to go to different location, then we are actually
working on the Degree-Day concept to adjust the MSR master curve for the location. So
for different locations in the United States, there would be different adjustment factors to
the aging ratios.
PROF. TEBALDI: Related to temperature or with the temperature and other
environment and conditions.
DR. MOHSENI: Degree-Days parameter includes both temperature and time. So the
Degree-Day in LTPPBind shows pretty much the heat quantity. If you go to LTPPBind
for any location, you can get an estimate of the heat quantity per year. We are actually
tying the MSR master curve to that heat quantity. The laboratory aging would still be
five days, but then we might not use the whole property after five days of aging. We are
going to correct that based on the Degree-Day. So for Wisconsin, it might be four days
and for Alabama it might be more like six days of aging that is going to be used.
PROF. TEBALDI: It is only temperature based aging?

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
DR. AZARI: Temperature and duration of the temperature.
DR. MOHSENI: Degree-Days is temperature and duration. So say in Florida, the
temperature may not go very high, but its high for like four months. That means the
Degree-Day would be high.
PROF. TEBALDI: Okay.

149

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement
Effectiveness in Real World Environments
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

through Developing Models: A Statistical


and Kinetic Study

Heather Dyllaa, Somayeh Asadib, Marwa Hassana*,


and Louay N. Mohammadc
a
Louisiana State University, Department of Construction Management, Baton Rouge,
LA, 70803;
b
Texas A&M University-Kingsville, Department of Civil and Architectural Engineering,
Kingsville, TX, 78363
c
Louisiana Transportation Research Center, Louisiana State University, Baton Rouge,
LA 70808

ABSTRACT. Photocatalytic asphalt pavements are evaluated in this study as a possible air-pollution
reduction strategy and to comprehend their behavior in the real world environment. Statistical
models from real world experiments or reaction kinetics are two approaches to understand the
photocatalytic reduction in real world environments. The objective of this study was to develop a
statistical model for nitrogen oxide (NO) reduction using data from a field study and to evaluate
the photocatalytic reaction kinetics of NO reduction, which could be used in future theoretical air
pollution model simulations. To achieve this objective, a photocatalytic water-based spray coating
was applied on an existing asphalt pavement site for the field study statistical model and on
laboratory samples for the kinetic study. Based on the field data, the NO reduction was modeled
using statistical regression techniques by creating a model for a non-coated pavement and
photocatalytic pavement. The coefficient of determination was 0.79 and 0.67, respectively. To
improve prediction, other parameters may need to be included into the model and more sampling
time is required. Based on the lab results, the NO reduction was reaction controlled following the
Langmuir-Hinshelwood model. The adsorption equilibrium constant calculated for photocatalytic
asphalt pavements was similar to those of concrete pavements, while the reaction rate constant

The oral presentation was made by Ms. Dylla.

This is a reproduction of an article whose final and definitive form has been published in Road
Materials and Pavement Design 2013 Taylor & Francis. The article is available online at:
http://dx.doi.org/ 10.1080/14680629.2013.812839

151

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

was significantly lower. While humidity has a negative correlation on both L-H constants,
intensity has a positive correlation. However, interaction between these two parameters exists.

KEYWORDS: photocatalytic asphalt; TiO2; statistical modelling; kinetic modelling

1. Introduction

Negative health effects from vehicle pollution are associated with living, working and
going to school near highways from long-term and short-term exposure. As a result,
many technologies have been implemented to reduce vehicle emissions. However,
increasing urbanization, vehicle use, traffic congestion, and desire for larger cars, often
offset the reductions realized by these new vehicle emissions control strategies (HEI,
2010; Menz, 2002). This has led to the practice of continually reducing vehicle
emissions standards. Another technique of pollution abatement is photocatalytic
pavements. Photocatalytic pavements use a semiconductor photocatalyst that, when
irradiated by sunlight, can reduce air pollutants after they are emitted.
Initial interest in environmental photocatalysis began in the 1970s, initiated by
Fujishima and Hondas research in photoelectrochemical solar energy conversion.
Through biomimicy of plant photosynthesis, Fujishima attempted to replicate the photo-
induced redox reactions, oxidizing water and reducing carbon dioxide, by using a
semiconductor irradiated by UV light (Fujishima and Zhang, 2006). To accomplish this,
the semiconductor is used as an electrode connected to a counter electrode to generate
electrical work to drive the redox chemical reactions (Fujishima and Honda, 1972). By
removing the electrode, environmental photocatalytic oxidation occurs completely
decomposing both organic and inorganic compounds. Since then, interest in
environmental photocatalysis has increased and TiO2 photocatalysts have been applied to
glass, tile, paper, and pavements for self-cleaning materials, water purification, air
purification, sterilization, and oil spill remediation (Fujishima and Zhang, 2006).
For abatement of pollution in roadway microenvironments, pavements are an ideal
substrate due to their close proximity to higher concentrations of pollution and large
surface area (Beeldens, 2006). Titanium dioxide is the preferred photocatalyst used
because of its high stability, super-hydrophilicity, relative cheapness, low toxicity, and
commercial availability (Cassar, 2004; Fujishima and Zhang, 2006; Diamanti et al.,
2008; Toma et al., 2009; Yu, 2009). Titanium dioxide can be applied to both asphalt and
concrete pavements. In laboratory studies, Li et al. demonstrated that an emulsified
asphalt blend with 2.5% TiO2 reduced up to 40% of the nitrogen oxides (NOx) (Li et al.
2009). In addition, Hassan et al. achieved as high as 66% reduction of NOx using a

152
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
spray coat on warm mix asphalt laboratory samples (Hassan et al., 2012). However, in
order to grasp the widespread pollution reduction made possible by photocatalytic
pavements, the next major step is to comprehend its behavior in the real world
environment. Statistical models from field experiments or reaction kinetics are two
approaches to understand the photocatalytic reduction in real world environments.
Therefore, the objective of this study was to develop a statistical model for the NO
reduction using data from a field study and to evaluate the photocatalytic reaction
kinetics of NO reduction, which could be used in future theoretical air pollution model
simulations. Not only are these models important for understanding the significance of
the NO reduction expected, they could be included into the State Implementation Plan
(SIP) air quality calculations and models allowing photocatalytic pavements to be
evaluated as a possible pollution reduction strategy (EPA, 2012).

2. Background

As previously mentioned, titanium dioxide is the preferred photocatalyst used in


photocatalytic pavements. When TiO2 is exposed to energy, from photons, that exceeds
the band gap energy of 3.2 eV, an electron is expelled from the valence band to the
conduction band, leaving a hole behind (Zhao and Yang 2003, Fujishima et al., 2000).
For TiO2, this process is initiated by energy from a UV light wavelength determined by
the electromagnetic radiation, hv, where h is Plancks constant and v is the frequency of
light. The wavelength required for TiO2 irradiation is between 300 to 365 nm (Zhao and
Yang 2003). The production of electron-hole pairs, called excitons, results in redox or
oxidation chemical reactions (Fujishima et al., 2000, Zhao and Yang 2003, Hunger et al.,
2008). In the presence of water, these oxidizing holes, h+, and photogenerated electrons,
e-, create hydroxyl radicals and superoxides respectively, as shown in Equations 1 and 2
(Fujishima et al., 2000).

OH - +h+ OH * [1 ]

O2 e- O2- [2 ]

The resulting hydroxyl radicals and superoxides are key for oxidation or reduction
reactions allowing for degradation of pollutants in the oxidation of NOx to water soluble
nitrates, as shown in equations 3 and 4 below (Beeldens 2008):

Ti O2
NO OH * NO2 H * [3 ]

153

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

2
2 3 [4 ]

This allows for two techniques to measure photocatalytic efficiency. The first is to
measure the reduction directly by measuring the ambient air pollution concentration, and
the second is to measure the reduction indirectly by measuring the byproducts created
from the degradation process. For nitrogen oxides, the approved method of ambient air
NOx detection is chemiluminescence, which continuously monitors the NOx
concentrations. This is challenging since environmental parameters are constantly
changing. The indirect technique to measure photocatalytic reductions of NOx, is to
measure the NO3 and NO2 deposited on the surface. Nitrates and nitrites are water-
soluble and therefore washed from the surface with water to be quantified (Beeldens,
2008). Water samples are usually collected daily and analyzed for nitrates and nitrites,
thus it does not capture the environmental variability essential for understanding
photocatalytic reduction of NOx in real world environments. Despite this limitation,
nitrate accumulation provides evidence that the NO reduction measured in ambient air is
indeed from a photocatalytic reduction and not adsorption.

2.1. Photocatalytic Models

Few models exist to describe the performance of photocatalytic pavement materials


reduction of NOx. Statistical techniques or theoretical reaction kinetics are two
modeling methods that could be used to model the reduction of NO from photocatalytic
pavements.
Statistical models can be used to describe the pollution in an area based on various
parameters. Statistical models have been important to identify the contribution of
pollution sources to human exposure for air pollution policy and regulation to mitigate
associated risks (Zeng et al., 2010). Statistical approaches include regression,
multiregression, and artificial neural networks. Estimates are calculated by the statistical
relationship to various factors collected at a particular receptor (Sharma et al., 2004).
Source apportionment models, which attempt to identify different sources contribution to
the pollution at a particular receptor, use many of these statistical methods such as
principal component analysis (PCA) and multilinear regression (Vallius et al., 2008).
These models are easy to use, but they are receptor or location specific, require large
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

amounts of historical data, and do not identify the significance of various physical and
chemical processes (Sharma et al., 2004). Nevertheless, they can be used to create
simple models to characterize pollution in roadway microenvironments.
Photocatalytic oxidation reaction kinetics theory is well-established by researchers
for reactor engineering. The most widely used model for heterogeneous surface

154

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

reactions is the Langmuir-Hinshelwood where at high pressures a zero order reaction


prevails and at low pressures a first order reaction prevails (Wang et al., 2007; Hunger et
al., 2010). Using this model, the reaction rate for NO photodegradation is described as
follows:

k*KCNO
rNO [5 ]
1+KCNO

where: rNO = NO photodegradation rate (mg/m3min),


k = L-H reaction rate constant (mg/m3min),
K = the L-H adsorption equilibrium constant (m3/mg),
CNO = the concentration of NO (mg/m3).
To account for the impact of humidity, competition between the two pollutants for
adsorption exists, thus the equation takes the form presented in Equation 6 (Hunger et
al., 2010):

kKNO CNO
rNO [6 ]
1+KNO CNO +KH2OCH2O

where: CH2O = the concentration of H2O,


KH2O = equilibrium constant for H2O.
Despite knowledge of this relationship, many models found regression techniques to
incorporate the impact of relative humidity more adequate (Hunger et al., 2010;
Bengtsson and Castellote, 2010).
Recently, kinetic studies have been incorporated in photocatalytic pavement studies.
Hunger et al. modeled the kinetics of the photocatalytic degradation of NOx using
photocatalytic concrete paving stones. The impact of the UV-light intensity and
humidity on the kinetic constants were each modeled separately (Hunger et al., 2010).
To incorporate these parameters into one model, Bengtsson and Castellote used non-
linear regression to model the reaction rate constant (Bengtsson and Castellote, 2010).
Nevertheless, to the authors knowledge, the only kinetic studies completed for
photocatalytic pavements have been on concrete pavements.

3. Methodology

To develop the statistical models, a field study was conducted while the reaction kinetics
were determined from laboratory experiments. The same photocatalytic thin film was
applied to both the field study pavement and the laboratory samples. The thin film was
applied as a spray coat in two parts. A primer was first applied as a base coat. Then

155
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

immediately after, an aqueous liquid of suspended anatase TiO2 nanoparticles, 2% by


volume, was applied. The TiO2 nanoparticles are nanorods ranging from average 12 nm
width by 75 nm length diameter.

3.1. Field Study

3.1.1. Field Study Description

A 0.2-mile asphalt pavement located in Baton Rouge, Louisiana, was selected as the
field study location. A distributor sprayed the primer and photocatalytic coat at a rate of
16.1 ml/m2 to 21.5ml/m2 (Figure 1). Electrostatic precipitators were used in the nozzles
to ensure more even distribution of the TiO2 photocatalyst.
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Figure 1. TiO2 photocatalytic coating field application.

The NO, NO2, and NOx concentrations, traffic count and climatic conditions were
monitored and recorded for a control area and for a photocatalytic pavement area over a
period of two months during the winter. The equipment used for field data collection
was housed in an air-conditioned trailer. A Thermo 42i NOx analyzer was used for
monitoring NO, NO2 and NOx concentrations as shown in Figure 2. The NOx analyzer
was calibrated in accordance with EPA calibration procedures using the Thermo 146i
gas calibrator with the gas phase titration (GPT) option. The stainless steel tubing was
placed to sample at the middle of the lane at pavement level in order to measure the
concentration that comes in contact with the pavement surface, Figure 3. The
concentrations were stored as minute averages.
To monitor climatic conditions at the site, a Davis Vantage Pro2 weather station was
installed in the field that recorded and stored meteorological data including humidity,
ambient temperature, wind speed, wind direction, rain, dew point, and solar radiation
continuously every 5 minutes as shown in Figure 3-a. Furthermore, a traffic counter was

156

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

installed to count the number of vehicles per 5 minute intervals per lane as shown in
Figure 3-b.

Figure 2. NOx analyzer and calibrator.

(a ) (b)

( c)

Figure 3. Field Monitoring (a) Weather station, (b) Traffic counter, and (c) NO
collector.

157
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

3.2. Kinetic Study

3.2.1. Laboratory Sample Preparation

Two asphalt pavement samples, 33 x 25.4 x 6.4 cm in size, were prepared using a
conventional hot mix asphalt with a PG 64-16 asphalt binder. The spray coat was
applied to each sample using a crosshatch formation (Figure 4) using a hand spray gun at
a 0.21 mg/cm2 catalyst loading.

Spray 1 Spray 2
Figure 4. Crosshatch spray formation.

3.2.2. Sample Preparation

Heterogeneous photocatalytic oxidation is a reaction that occurs between a gas or liquid


in contact with a photocatalytic solid surface. Thus, in order for the reaction to occur,
the pollutant must transfer to the photocatalytic surface where it is adsorbed. Therefore,
the reaction can be either mass transfer controlled or reaction controlled, determined by
whichever process is slowest. Previous kinetic studies have shown that photocatalytic
oxidation is reaction controlled and are thus modeled by the Langmuir-Hinshelwood. In
order to test whether the photocatalytic degradation of NO was mass transfer limited or
reaction rate limited, the effect of the inlet concentration was varied. The NO
concentration was varied by using inlet concentrations of 150, 250, 350, 550, 1000,
2000, and 3000 ppb, while the flow rate, relative humidity and intensity remained
constant at 1.5 l/min, 20%, and 2.4 mW/cm2 respectively.
In addition, in efforts to create a model for the photocatalytic reduction effective for
various environmental conditions, a kinetic study was completed at three relative
humidity levels (23, 48, 78 5%) and three average light intensities (2.4, 1, 0.5
mW/cm2). A fractional factorial study was completed such that nine different
environmental conditions summarized in Table 1 were tested and the Langmuir-
Hinshelwood (L-H) constants were calculated. In order to construct a robust regression
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`-

158

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

fit and to calculate the L-H constants, the inlet concentration was varied at four levels:
150, 250, 550, and 3000 ppb (1000 ppb for 0.5 mW/cm2 intensity since there was no
reduction at 3000 ppb) resulting in a total of 36 experimental runs.
Table 1. Experimental laboratory cases to determine L-H constants.
Run
Humidity (%) Intensity (mW/cm2) Concentration (ppb)
Scenario

1 23 2 .4 150, 250, 550, 3000

2 23 1 150, 250, 550, 3000

3 23 0 .5 150, 250, 550, 1000

4 48 2 .4 150, 250, 550, 3000

5 48 1 150, 250, 550, 3000

6 48 0 .5 150, 250, 550, 1000

7 78 2 .4 150, 250, 550, 3000

8 78 1 150, 250, 550, 3000

9 78 0 .5 150, 250, 550, 1000

3.2.3. Experimental Setup

The experimental setup used to quantify NOx removal efficiency was modified from the
Japanese Industrial Standard (JIS TR Z 0018 Photocatalytic materials air purification
test procedure) in order to accommodate larger samples and variations of environmental
parameters (JIS, 2009). The photoreactor is a plug flow model with an air space volume
of 16.75 liters. Further details of the setup can be found elsewhere (Dylla et al., 2010).
The sample was housed inside the photoreactor, to create an enclosed controlled
environment for the duration of the test. The photoreactor was irradiated with 5 UV-
fluorescent black tube lamps (20W, Philips) emitting wavelengths within 300-400 nm
(Zhao and Yang 2003). The distance between the sample surface and the photoreactor
determined the UV intensity measured by UV-A intensity meter (OAI Model 306) at 365
nm. The UV profile was measured for each test condition and the average intensity

159
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

observed was calculated. The highest intensity, 2.4 mW/cm2, had the most variance due
to the sample proximity to the lights. This variance is illustrated in Figure 5 showing the
intensity exposed over the sample area at the highest intensity level.

4 3-4
2-3
E (mW/cm2)

3
2 1-2
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

330
1 0-1
220
0 110
0 59 118 177 0 X (c m )
236 295 354
Y (c m )

Figure 5. Profile of UV-A irradiance over sample surface (2.4 mW/cm2).

While the JIS test procedure requires 5 hours of irradiation, it is noted that steady
state equilibrium is achieved within the first hour of irradiation. Therefore, the test
procedure was shortened until equilibrium was achieved, defined by stability of NOx
concentration for 10 minutes. To prevent deactivation by intermediates adsorbed on the
TiO2 active sites, samples were soaked in deionized water for 24 hours and dried for 48
hours before being retested.

4. Model Development

4.1. Field Study Regression Model Development

Two regression models were developed, one for predicting hourly NO concentration
before TiO2 application and the second for predicting hourly NO concentration right
after application using the data collected from the field study. Nitrates were collected
and measured from the field confirming a photocatalytic reduction of NOx. Results can
be found in Hassan et al. (2012). The regression models were used to compare NO
concentrations per hour under different environmental conditions. Comparing the two

160

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

models allowed for quantification of the field hourly NO reduction efficiency (Hassan et
al., 2012). The developed regression models relate the measured NO concentration per
hour linearly to the environmental factors as follows:

NO = f (T, H, V, T Ambient, S) [7 ]

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
NO Before = 0.96T+0.22H-1.33 T Ambient -10.5V+ 0.02S [8 ]

NO After = 0.31T+0.06H-0.1 T Ambient -0.75V + 0.0003S [9 ]

NO Reduction = NO Before - NO After [1 0 ]

Where:
NO Before = average hourly NO concentration before TiO2 application (ppb),
NO After = average hourly NO concentration after TiO2 application (ppb),
NO Reduction = average hourly NO concentration reduced (ppb),
T = number of vehicles per hour,
H = relative humidity,
V = wind speed (m/s),
T Ambient = ambient temperature (C),
S= solar radiation (W/m2).
Statistical goodness of fit was assessed through the coefficient of determination (R2)
and Root-Mean Square Error (RMSE) as shown in Table 2(a). Table 2(b) illustrates the
range of variation for each parameter investigated in this study. The developed models
are only valid in the input ranges shown in Table 2(b).

Table 2. (a) Descriptive statistic of the developed models and (b) Valid input ranges of
variables in the developed models ranges of variables in developed models.
(a )

Statistical Parameters Untreated Treated

Coefficient of Determination (R2) 0 .7 9 0 .6 7

Root Mean Square Error (RMSE) 1 3 .7 3 .7

Normal Root Mean Square Error (NRMSE) 1 .3 1 3 .0 3

161

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

(b)
Range of variation
Parameters
Low Level High Level

Number of vehicle per hour 2 100

Humidity (%) 30 90

Wind speed (m/s) 0 3 .0

Outside temperature ( C) 0 35

Solar Radiation (W/m2) 0 1000

The validity of the statistical models was evaluated by comparing statistical results with
observed data not used to build the models. Figure 6 (a and b) compares the results of
the statistical models with independent measurements. Results are presented for two
cases: before TiO2 application and just after TiO2 application. As shown in these
figures, at some points, a good agreement was observed between the hourly predicted
concentrations of NO and observed concentrations while some points had high errors.
This suggests that more data are needed or other parameters may need to be added to
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

improve the prediction from the model.

( a) (b )

Figure 6. Comparison of measured vs. predicted NO concentrations for (a) before and
(b) after photocatalytic coating.

162

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

4.2. Photocatalytic Reaction Kinetic Model Development

Figure 7 shows the NO reduction per initial concentration. As the concentration


increases, the percentage of NO degradation decreases. Since NO reduction is
dependent on pollutant concentrations, this indicates that the mass transfer is reaction-
kinetics controlled rather than mass-transfer controlled.

100%

80%
y = e-1E-03x
NO Reduction

60%
R = 0.8078
40%

20%

0%
0 500 1000 1500 2000 2500
Initial Concentration NO (ppb)

Figure 7. NO reduction by the influence of initial concentration (1.5 l/min flow, 20%
relative humidity, 2.4 mW/cm2 intensity).

4.2.1. Mass Transfer Limited

Further confirmation can be found by assuming that the reaction is not mass transferred
controlled and diffusion is the limiting step. Assuming instantaneous conversion, the
mass balance can be written. Derivation of the NO mass balance equation for a plug
flow reactor is described elsewhere (Hunger et al., 2010). Integrating the mass balance
equation over the length of the photocatalytic surface the percent reduction can be
calculated by the following equation:


() 2
1 = 2 [1 1 ]
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

()

where: CNO,in = inlet concentration (mg/m3),


CNO,out = outlet concentration (mg/m3),

163

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

Sh = Sherwood number,
= velocity of air (m/s),
h = height of air space in photoreactor (cm),
D = Diffusion Coefficient of NO (m2/s),
L = Length of sample (cm).

Using the variables defined in Table 3, if mass transfer controlled, the percent
reduction would be 92%. This, however, is not the case as shown in Figure 7 where the
reduction was varied by the initial NO concentration and the highest reduction was 73%
when the inlet concentration was at 150 ppb.

Table 3. Mass transfer variables.


Variable Symbol Value
Sherwood Number Sh 4 .4
Length L 3 3 cm
Height h 5 cm
Diffusion Coefficient NO D 1.51x10-5 m2/s
Velocity of Air 0.0015 m/s

4.2.2. Reaction Rate Limited


To solve for the L-H reaction rate constant and the L-H adsorption equilibrium constant,
laboratory experiments were used. For a single pass experimental setup, plug flow is
assumed and the mass balance is written as follows:

dCNO k*KdCNO
rNO -vair * [1 2 ]
dx 1+KdCNO

dCNO
where: = rate of change of concentration per horizontal distance (mg/m2).
dx

Integrating the mass balance, Equation 12, over the length with the boundary conditions
shown in Equation 13, results in the following linear relationship in Equation 14.

CNO CNO,in [1 3 ]

CNO,in V
lnC kKd
NO,out Q 1
- [1 4 ]
CNO,in -CNO,out CNO,in -CNO,out k
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

164

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

where: Q = flow rate (m3/min),


V = volume of air space (m3).

Thus, from linear regression analysis by graphing lnCNOin/CNOout/(CNOin-CNOout) versus


(V/Q)/(CNOin-CNOout) the k and Kd fitting parameters can be determined from the graph;
where k is the reciprocal of the y-intercept and Kd is 1/(mk), where m is the slope of the
line. Furthermore, the linear relationship is evidence that the reaction is a gas-solid
reaction following the L-H mechanism, rather than a gas-phase reaction (Sleiman et al.,
2009).
The results plotted in Figure 7 were used to calculate and plot the
lnCNOin/CNOout/(CNOin-CNOout) versus (V/Q)/(CNOin-CNOout) as shown in Figure 8. As
shown in this figure, there is a strong linearity thus supporting the Langmuir-
Hinshelwood model and that the mass transfer is reaction-rate limited. In addition,
previous studies have shown that the percent degradation of NO is increased by
increasing the amount of TiO2 catalysis, also suggesting reaction-controlled (Hassan et
al., 2012).
From the regression results, the Langmuir-Hinshelwood constants were calculated as
k=0.11 mg/m3min and Kd=4.70 m3/mg. Compared to previous concrete kinetic studies
conducted in the Netherlands, the reaction rate is much lower and the adsorption
equilibrium constant is higher. Hunger et al. reported for concrete a reaction rate of
k=0.42 mg/m3 and adsorption equilibrium of Kd=2.00 m3/mg for 50% relative humidity
and 10 mW/cm2 (Hunger et al., 2010). The higher adsorption equilibrium rate could be a
result of the lower relative humidity used in this study thus having less competitive
adsorption. Meanwhile, the lower reaction rate could be from a chemical interaction
between the asphalt hydrocarbons and the NOx.

165
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

35
y = 1.9394x + 9.091
30
R = 0.9899
25

20

15

10

0
0 5 10 15

Figure 8. Determination of L-H constants for asphalt pavements.

4.2.3. Effect of Relative Humidity and Intensity on L-H Constants

From Figure 9, it is clear that relative humidity impacts both the L-H adsorption
equilibrium constant and the L-H reaction rate constant. The extent of the impact of
relative humidity is largely dependent upon the substrate material. Titanium dioxide can
be both hydrophobic and hydrophilic. Typically, and during irradiation, TiO2 undergoes
a hydrophilic effect. This promotes a water monolayer that inhibits pollutants to adsorb
on the photocatalyst active sites (Hunger et al., 2010). As seen in Figure 9, in general as
the relative humidity increases, the L-H adsorption equilibrium rate decreases
confirming this theory that the relative humidity competes with photocatalytic active
sites (Figure 9a).
Furthermore, the reduction in photocatalytic active sites also leads to lower L-H
reaction rates, which is illustrated in Figure 9b. The lower reaction rates are explained
by less active sites available on the TiO2 photocatalyst thus reducing the reaction rate. In
addition, due to the nonparallel trends illustrated in both Figures 9a and 9b, there seems
to be a slight interaction between the relative humidity and light intensity. For example

166
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

at the lowest light intensity, the reaction rate seems to have an optimum relative
humidity point.

5 2 .4
Kd (m3/mg) 4 1
3
0 .5
2
1
0
0 50 100
Relative Humidity (%)

( a)

2 .4
0.12 1
k (mg/m3min)

0 .1 0 .5
0.08
0.06
0.04
0.02
0
0 50 100
Relative Humidity (%)

(b )
Figure 9. L-H adsorption equilibrium constant (a) and L-H reaction rate (a) versus
humidity at 2.4, 1, and 0.5 mW/cm2 intensities.

Figure 10 illustrates that the light intensity also had an impact on both the L-H
adsorption equilibrium constant and the L-H reaction rate constant. This is contrary to
previous kinetic studies, which illustrated that the light intensity does not have any
significant impact on the equilibrium constant and only influences the reaction rate
constant (Hunger et al., 2010). Similarly to conclusions from Figure 9, both graphs in
Figure 10 support evidence of an interaction effect between the relative humidity and
intensity factors on both the L-H constants.

167

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

5 20

Kd (m3/mg)
4 48
3 78
2
1

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
0
0 1 2 3
Intensity (mW/cm2)

( a)

0.12 20
0 .1 48
k (mg/m3min)

0.08 78
0.06
0.04
0.02
0
0 1 2 3
Intensity (mW/cm2)

(b )

Figure 10. L-H adsorption equilibrium constant (a) and L-H reaction rate (b) versus
light intensity at 20, 48, and 78% relative humidity.

Table 4 is a summary of the regression fits and the resulting L-H constants calculated
for each environmental scenario simulated. Under similar environmental conditions as
the Hunger et al. (2010) kinetic study for concrete, the L-H reaction rate was
significantly smaller at 0.03 mg/m3min compared to 0.42 mg/m3 and the L-H adsorption
equilibrium constant was relatively similar at 1.87 m3/mg compared to 2 m3/mg.
Therefore, a major reason for the lower NO reduction efficiencies in photocatalytic
asphalt pavements compared to concrete pavements is a result of the significantly lower
L-H reaction constants characteristics of photocatalytic asphalt pavements.

168

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

Table 4. Summary of L-H constants and coefficient of determination.

Run Intercept Kd K Coefficient of


Scenario Slope (m) (b ) (m3/mg) (mg/m3min) Determination (r2)
1 2 .1 7 7 .4 7 4 .6 8 0 .1 1 0 .9 8
2 3 .3 5 1 0 .7 1 3 .2 0 0 .0 9 0 .9 4
3 3 .8 7 1 6 .8 5 4 .3 5 0 .0 6 0 .9 6
4 5 .1 6 2 3 .0 8 4 .4 8 0 .0 4 0 .9 4
5 1 5 .4 8 2 8 .9 0 1 .8 7 0 .0 3 0 .9 8
6 7 .4 3 1 6 .6 5 2 .2 4 0 .0 6 0 .8 6
7 1 1 .5 2 3 3 .5 5 2 .9 1 0 .0 3 0 .9 8
8 2 7 .7 2 4 4 .7 7 1 .6 1 0 .0 2 0 .9 9
9 4 1 .8 0 8 7 .6 6 2 .1 0 0 .0 1 0 .9 6

5. Conclusions

Using data from a field study, the NO reduction was modeled using statistical regression
techniques. The coefficient of determination was 0.79 and 0.67 for a non-coated
pavement and photocatalytic pavement, respectively. To improve prediction, other
parameters may need to be included in the model and more sampling time is required.
As a result, using mathematical models may play an intermediate role in understanding
real-world photocatalytic pavements.
A kinetic study was completed for asphalt photocatalytic pavements being the first
step to creating a chemodynamic model to understand the significance of pollution
reduction from using photocatalytic pavements. The results of the study confirmed that
the NO mass transfer was controlled by the reaction rate. These results are similar to
previous photocatalytic kinetic studies for other substrates. Outcomes of the kinetic
study showed that the adsorption equilibrium of NO for photocatalytic asphalt
pavements was similar compared to concrete substrates; however, the reaction rate
constant was significantly lower. Consequently, the lower reduction efficiencies
observed in past studies for asphalt photocatalytic pavements as compared to concrete
photocatalytic pavements are a result of the lower L-H reaction rate constants.
The results of the parametric study identified that the relative humidity and light
intensity had a significant impact on NO reduction efficiencies. The impact of the
relative humidity was attributed to competition in adsorption sites and reduction of

169
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

active sites, evident by the negative correlation of Kd and k, respectively, with


increasing relative humidity. As for light intensity, there was a positive linear
correlation related to the NO reduction efficiencies. There was interaction between the
relative humidity and intensity variables. These results are important for future work to
develop a NO reaction rate model for photocatalytic pavements to understand the
significance of NO reductions in real world environments.

7. References

Beeldens, A. (2006, June). An environmental friendly solution for air purification and
self-cleaning effect: the application of TiO2 as photocatalyst in concrete.Belgian
Road Research Centre. Proceedings of Transport Research Arena, Europe - TRA,
Gteborg, Sweden.
Beeldens, A. (2008). Air purification by pavement blocks: final results of the research
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

at the BRRC. Transport Research Arena Europe, Ljubljana.


Bengtsson, N. and Castellote, M. (2010). Photocatalytic activity for NO degradation by
construction materials: parametric study and multivariable correlations. Journal of
Advance Oxidation Technologies, 13(3), 341349.
Cassar, L. (2004, May). Photocatalysis of cementitious materials: clean buildings and
clean air. MRS Bulletin, 14.
Diamanti, M.V., Ormellese, M. and Pedeferri, M. (2008). Characterization of
photocatalytic and superhydrophilic properties of mortars containing titanium
dioxide. Cement and Concrete Research, 38, 13491353.
doi:10.1016/j.cemconres.2008.07.003.
Dylla, H., Hassan, M., Schmitt, M., Rupnow, T. and Mohammad, L. (2011). Laboratory
investigation of mixed nitrogen dioxide (NO2) and nitrogen oxide (NO) gasses on
titanium dioxide photocatalytic efficiency in concrete pavements. Journal of
Materials in Civil Engineering, 23(7), 10871093. doi:10.1061/(ASCE)MT.1943-
5533.0000248.
Environmental Protection Agency (EPA). (2012). Nitrogen dioxide implementation
Programs and requirements for reducing oxides of nitrogen.
(http://www.epa.gov/air/nitrogenoxides/implement.html).
Fujishima, A., and K. Honda. (1972). Electrochemical Photolysis of Water at a
Semiconductor Electrode. Nature, 238, 3738. doi:10.1038/238037a0.

170

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

Fujishima, A., Rao, Tata N. and Tryk, Donald A. (2000). Titanium dioxide
photocatalysis. Journal of Photochemistry and Photobiology C: Photochemistry
Reviews, 1, 1-21. doi:10.1016/S1389-5567(00)00002-2.
Fujishima, Akira, and Zhang, Xintong. (2006). Titanium dioxide photocatalysis:
present situation and future approaches. C.R.Chimie, 9 750760.
doi:10.1016/j.crci.2005.02.055.
Hassan, M. M., Mohammad L., Asadi, S., Dylla, H., and Cooper, S. (2012).
Sustainable Photocatalytic Asphalt Pavements for Mitigation of Nitrogen Oxide
and Sulfur Dioxide Vehicle Emissions. ASCE: Journal of Materials in civil
engineering. Advance online publication. doi: 10.1061/(ASCE)MT.1943-
5533.0000613.

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
Health Effects Institute (HEI). (2010). Traffic-Related Air Pollution: A Critical Review
of the Literature on Emissions, Exposure, and Health Effects. (Special Report 17).
Hunger, M., Husken, G., and Brouwers, J. (2008). Photocatalysis applied to concrete
products, Part 1: Principles and test procedure. Materials Science, ZKG
International, 61, 7785.
Hunger, M., Husken, G., Brouwers, H.J.H. (2010). Photocatalytic degradation of air
pollutants From modeling to large scale application. Cement and Concrete
Research. 40, 313-320. doi:10.1016/j.cemconres.2009.09.013.
Japanese Industrial Standard (JIS). (2004). Fine ceramics (advanced ceramics,
advanced technical ceramics) Test method for air purification performance of
photocatalytic materials- Part 1: Removal of nitric oxide. JIS R 1701-1, 1-9.
Li, L. and Qian, C. (2009). A lab study of photo-catalytic oxidation and removal of
nitrogen oxides in vehicular emissions and its fieldwork on Nanjin No.3 bridge of
Yangtze River. Journal of Pavement Resource Technology, 2(5), 218-222.
Menz, F.C. (2002). The US experience with controlling motor vehicle pollution:
lessons for China. Int. J. Environment and Pollution, 18(1), 121.
Sharma, N., Chaudhry, K.K., and Chalapati Rao, C.V. (2004). Vehicular pollution
prediction modeling: A review of highway dispersion models. Transport Reviews,
24(4), 409435.
Sleiman, M., Conchon, P., Ferronato, C., and Chovelon, J.M. (2009). Photocatalytic
oxidation of toluene at indoor air levels (ppbv): Towards a better assessment of
conversion, reaction intermediates and mineralization. Applied Catalysis B:
Environmental, 86(3-4),159165. doi:10.1016/j.apcatb.2008.08.003.

171

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

Toma, F.-L.; Berger, L.M.; Jacquet, D.; Wicky, D.; Villaluenga, I; Miguel, Y.R.;
Lindelov, J.S. (2009). Comparative study on the photocatalytic behavior of
titanium oxide thermal sprayed coating from powders and suspensions. Surface
and Coatings Technology, 203(15), 21502156.
Vallius, M., Ruuskanen, J., and Pekkanen, J. (2008). Comparison of multivariate source
apportionment of urban PM2.5 with chemical mass closure. Boreal Environment
Research, 13, 347358.
Wang, Shaobin; Ang, H.M.; and Tade, Moses O. (2007). Volatile organic compounds
in indoor environment and photocatalytic oxidation: State of the art. Environment
International, 33, 694705. doi:10.1016/j.envint.2007.02.011.
Yu, Q.L. and Brouwers, H.J.H., (2009). Indoor air purification using heterogeneous
photocatalytic oxidation. Part I: Experimental study. Applied Catalysis B:
Environmental, 92, 454461. doi:10.1016/j.apcatb.2009.09.004.
Zeng, F., Shi, G-L., Li, X., Feng, Y-C., Bi, X-H., Wu, J-H., Xue, Y-H. (2010).
Application of a combined model to study the source apportionment of PM10 in
Taiyuan, China. Aerosol and Air Quality Research, 10, 177184.
doi:10.4209/aaqr.2009.09.0058.
Zhao, J. and Yang, X. (2003). Photocatalytic oxidation of indoor air purification: a
literature review. Building and Environment, 38, 645654. doi: 10.1016/S0360-
1323(02)00212-3.

8. Discussion

DR. HONGBIN XIE: I have a question, not exactly from this study but it is relevant.
What is the durability of this coating in the real world, in a field application particularly?
Like how often you will be required to repaint it?
MS. HEATHER DYLLA: We havent done complete long-term durability testing. I
know there are future prospects of projects that are going to do that at LTRC. But we did
test it differently, like the loaded wheel tester, we did the rheological test on the asphalt
pavement. But as far as durability on the environmental side, on how much reduction
can you expect, we have not gotten there yet. It has just been so difficult to test. And Dr.
Hassan wants to add something.
PROFESSOR MARWA HASSAN: From a durability perspective, we did do an initial
study that is currently under review by ASCE, so it is going to be published. For that
field study, we measured the deterioration or the reduction of photocatalytic efficiency
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

over time. Then we did a regression analysis prediction and it is very preliminary.
Results suggest that the service life of TiO2 coating in concrete is between 6 to 11

172

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

months and between 10 to 16 months in asphalt pavements. But that is very preliminary
because of the varying weather conditions and all the variability that Heather has been
talking about. To get an accurate durability, what we need is to test under accelerated
pavement testing facility conditions, and thats the next step that were going after.
DR. XIE: Okay, thank you.
PROF. HASSAN: No problem.
DR. JEAN-PASCAL PLANCHE: I do not have a question. It is more of a comment for
information. This type of technology has been used in Europe for about 10 years. It
started, I believe, in the concrete industry, to make walls with the painting to reduce
NOx. There is also a French contractor who put together an asphalt and they have made
a field trial, maybe five years ago or something like that. I dont know exactly the
outcome, but just to let you know that this type of thing is already being applied to some
extent. I think it is limited, but I think its used more in the industry for petrochemical
plants or refineries to capture and reduce some exhaust.
MS. DYLLA: Yes. Thanks for the comment. That is true. I didnt go into all the
literature review. Most of our literature review comes from Europe Europe and Japan.
Its far advanced, and they have lots of different products.
DR. PLANCHE: There are some IR patents behind it.
MS. DYLLA: Yes.
PROFESSOR HUSSAIN BAHIA: Very nice study. My question is about the purpose of
spreading it in the wheel path. If there is a durability issue, can we think about putting it
on the shoulders or buildings along the highway? Is the proximity to the source
important? If not, do we need to spray it where the tires could actually erode it?
MS. DYLLA: The reason why we initially put it on the wheel path is we were looking at
different air pollution modeling studies that show the pollution dispersion and it is higher
concentration the closer you are to the source. But, of course, it changes because some
vehicles have their emission spouts up, some have it down. It is very difficult right no w
to say where the best place to put it is. But initially it was the thought of higher
concentrations or at the source or on the pavement level. We have confirmed that by
measuring pavement level concentrations. Weve hit high spikes around 1000 ppb. Once
it gets to like walls, it has been dispersed some, so that is the main reason.
PROF. BAHIA: Thank you.
MS. DYLLA: You are welcome.
PROF. HASSAN: Just to add a little bit about durability. Where to put this
photocatalytic coating, whether it is asphalt or on concrete is not a question that we can
answer in a day. It has to be correlated with urban planning. So you are really looking

173

Copyright DEStech Publications, Inc.


--`,`,,,``,,``

Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan


No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
DYLLA, ASADI, HASSAN, MOHAMMAD

for interstate or areas that have high traffic volumes so the concentrations of pollution
are high. You are also looking for urban areas that have tall buildings that prevent the
dispersion of the pollutant, which has lots of health impacts. But once you select these
areas, you also want to have it in areas that are away from water bodies so that you are
not producing, for example, nitrates and then worrying about another environmental
issue like eutrophication. Once you decide where these areas are, it can be a combination
of what is on the wheel path and what is on the shoulder of the pavement and maybe the
retaining walls too. What we are saying is we have tested different substrates. Heather is
talking about the field study that chooses the least durable method, which is the spray
coat because, really, you are putting on a spray coat. You can mix it with the asphalt and
the concrete mix, which would actually, if you can create an ultrathin layer or some kind
of an overlay, have a higher durability. These are all expectations that need to be tested
via research and then quantify the long-term durability of it. But it is worth studying
because there is a lot more potential to purifying the outdoors when it is closer to the
pollutant source versus not. Thank you.
DR. ERVIN DUKATZ: Again, compliments on a very fine paper and update of your
research. But one question I have for you is how close is this to practical
implementation? How contractor friendly or DOT friendly is this process and how
available are the materials that if a city, county or state wanted to implement this, ho w
close are you ready to do this?
MS. DYLLA: As far the distributor truck, it was actually really easy. I mean that is no
problem, I think, as far as application for a spray coat. I think the real questions that
remain are how efficient is it in the field? Like if we are looking at sustainability, is it
going to actually have a net benefit? Really understanding that and where do we
implement it? Can we implement it? Yes, we can. Its being done. Roadways? Theyve
done a few roadways, but for sure, if you know the statues in Minnesota where the
bridge fell? They're coated with titanium dioxide. The Super Dome is coated with
titanium dioxide. Not for environmental purposes to clean the air. but it is to keep the

--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---
concrete white.
DR. DUKATZ: Okay. Then the other practical aspect of it is, is the network set up so
that if people try specifying this, are they going to be able to get the titanium dioxide to
actually put in paving quantities? You know, its one thing to do a 200 foot section or
maybe a wall of a building, but when you're paving 10 lane miles or 30 lane miles.
MS. DYLLA: It would definitely create a market. Right now, we have one contractor in
the U.S. that we work with. Most of the manufacturers of titanium dioxide are overseas.
DR. DUKATZ: Thank you.
PROF. HASSAN: To answer your question, what is titanium dioxide? Titanium dioxide
is a very common material that is available. It is available in all kinds of white paint. The
difference is if you use what you put in white paint, it is not photocatalytic because it is

174

Copyright DEStech Publications, Inc.


Provided by IHS under license with DESTECH Licensee=Singapore-Jurong/3219500185, User=T, Kumanan
No reproduction or networking permitted without license from IHS Not for Resale, 05/01/2017 18:27:10 MDT
Evaluating Photocatalytic Asphalt Pavement Effectiveness in Real World Environments

made out of rutile crystals. What makes it photocatalytic is if you change the crystal
structure so it is anatase. Or you can also get photocatalytic efficiency if you mix it with
70% anatase, 30% rutile or something of that combination. So it is about creating more
manufacturers that are producing the photocatalytic grade. Also whether you are making
it a nanosize or an ultrafine size. There are major companies worldwide that are
producing this at a photocatalytic grade, although that production line is limited just
because there is no market for it. They can produce it. They have the capacity; just they
are selling it to very few people. The biggest worldwide manufacturer is known as
Nagosa and then the second is called Cristal Global. And all of these companies,
although they are international, found in Europe, they have offices in the U.S. and we
have dealt with them. There is a local company and a local manufacturer that is present
in the U.S. and the company is known as PURETi, and the CEO is Glen Finkel. That is
--`,`,,,``,,```,,`,,,,,`,``,````-`-`,,`,,`,`,,`---

the manufacturer of the spray coat application. So we can do it. In addition, I have talked
to professors from the University of Minnesota who are working in mineralogy, and they
were talking about the amount of titanium dioxide that they are mining that they have no
market for. So once we create that market, it will actually create more job opportunities
and more manufacturing opportunities, and its a win-win situation for the economy.
DR. RANDY WEST: Im going to ask a question. It may be a really stupid question
because any subject matter that deals with chemistry is certainly out of my expertise. But
if it is a stupid question, it wont be the first time I have asked one. It is really about the
conversion of NO to nitrates and conversion of one source of pollution to another source
of pollution. So could you comment about the runoff issue of nitrates into ground water
and those kinds of things?
MS. DYLLA: We have detected that and what has been actually reported is it is lo w
quantities that actually go in, and there is no eutrophication effect extra than what is
normally out there. It is really small. Smaller than what you would get in drinking water
as for the amount of nitrates. But it is something that is a precaution that needs to be
taken and that is very current. Good question.
PROFESSOR LOUAY MOHAMMAD: This is a comment also to Erv to add to what