QUANTIFICATION OF QUALITY OF FOAMED WARM MIX ASPHALT BINDERS
AND MIXTURES
By
Hande Isik Ozturk

A DISSERTATION
Submitted to
Michigan State University
in partial fulfilment of the requirements
for the degree of
Civil Engineering- Doctor of Philosophy

2013

ABSTRACT
QUANTIFICATION OF QUALITY OF FOAMED W ARM MIX ASPHALT
BINDERS AND MIXTURES
By
Hande Isik Ozturk
Warm Mix Asphalt (WMA), which is the general term used for the asphalt pavements
produced and placed at lower temperatures, is introduced to the pavement industry to overcome
the environmental and economic challenges of Hot Mix Asphalt (HMA). WMA technologies
reduce the overall mixture viscosity at lower temperatures to increase the workability of the
loose mixtures, provide improved (better) compaction and facilitate aggregate coating at low
temperatures. However, there are still many unknowns related to their long term performance.
Foam-based WMA’s are the most commonly used techniques; however there is no
specification or method to evaluate the quality of foam generated by different techniques.
Although producing foamed binder is relatively simple process, where hot binder is mixed with a
limited amount of water (typically 2-3% by weight of the binder), the rheology of the foamed
binder is not very simple. The quality of the foamed binder depends on various factors such as
the binder type, grade and modification, the foaming technology used, amount of water, and
temperature. Moreover, the quality of the binder plays a crucial role during the mixing, laying
and compaction stages of WMA pavement production. Asphalt foams used in base stabilization
applications were typically characterized using following three parameters: Expansion Ratio
(ER), Half-life (HL) and Foam Index (FI). However, there is no available method to measure
these parameters precisely. Therefore, an accurate and repeatable procedure is needed for the

measurement of initial expansion and reduction in height of foamed asphalt in order to calculate
the foam binder quality parameters.
Asphalt Foam Collapse Test (AFCT), an automated test to measure the reduction in the
height of the foam binder over time via image analysis, is developed during this study and
validated with nondestructive 3D imaging methods (i.e., x-ray microtomography). The height
reduction data obtained from AFCT is used to accurately calculate the commonly used foam
quality parameters. In addition, two new parameters, Bubble Size Distribution (BSD) and
Surface Area Index (SAI), are introduced as quality parameters in this study. It is found that
these parameters are strong candidates for evaluating the workability and coating, as well as the
performance of the pavements.
A nozzle-based laboratory foamer was utilized in this study to determine the effect of
water content and air pressure on the foam quality individually and in combination. Results
revealed that the water content and air pressure have significant influence on ER, HL, FI, BSD,
and SAI. It was observed that the low water content and low pressure produced foams with
relatively small bubbles as compared to foams made with high water content and pressure. The
current WMA pavement design procedures are based on limited empirical data and
recommendations of the WMA technology suppliers. WMA design procedures do not consider
the foam quality since its importance has not been fully understood. Therefore, the long term
performance of the WMA mixtures prepared with foamed binders prepared with various injected
water content and air pressures was evaluated via laboratory performance tests and compared
with the foamed binder quality parameters. It was concluded that a WMA mix design should
consider the foam quality, which is currently ignored.

Copyright by
HANDE ISIK OZTURK
2013

TO MY PARENTS & MY GRANDMOTHER’S SOUL

v

ACKNOWLEDGEMENTS

I would like to express my gratitude to Dr. M. Emin Kutay for his guidance and support
throughout my study at Michigan State University. I would have never completed this work
without his encouragement and his look for the positive.
I would like to extend my appreciation to Dr. Lalita Udpa for her valuable course. I also
would like to thank Dr. Karim Chatti and Dr. Neeraj Buch for reading my thesis and their
valuable comments.
I would like to acknowledge Craig Burck for his technical assistance and Gerrit Littrup
for his support throughout the laboratory work.
I would also like to thank the graduate and undergraduate students in Advanced Asphalt
Characterization Laboratory (AACL) at Michigan State University.
I would like to express my gratitude towards Dr. Denis Keane, the DND-CAT Director,
for his help while scanning the images at the Advanced Photon Source (APS) of the Argonne
National Laboratory (ANL). Use of the APS, an Office of Science User Facility operated for the
U.S. Department of Energy (DOE) Office of Science by ANL, was supported by the U.S. DOE
under Contract No. DE-AC02-06CH11357.
I deeply thank my parents for their love, support and understanding throughout my life. I
would like to acknowledge my grandmother, who encoraged me to go to the graduate school.

vi

TABLE OF CONTENTS

LIST OF TABLES .........................................................................................................................x
LIST OF FIGURES ................................................................................................................... xiii
CHAPTER 1 ...................................................................................................................................1
INTRODUCTION..........................................................................................................................1
MOTIVATION ..........................................................................................................................3
OBJECTIVE AND SCOPE OF THE RESEARCH ...............................................................4
OUTLINE OF THE DISSERTATION ....................................................................................4
CHAPTER 2 ...................................................................................................................................6
LITERATURE REVIEW AND BACKGROUND .....................................................................6
HISTORY OF WMA PAVEMENTS .......................................................................................6
BENEFITS OF WMA ...............................................................................................................8
Engineering Aspects...............................................................................................................8
Aging of WMA binders ...................................................................................................... 9
Workability and Compactability of WMA pavements ....................................................... 9
Usage of RAS in WMA pavements .................................................................................... 9
Usage of High Percentage RAP in WMA pavements ...................................................... 10
Usage of Crumb Rubber in WMA pavements .................................................................. 14
Cold Weather Paving ........................................................................................................ 14
Environmental Aspects ........................................................................................................15
Economical Aspects .............................................................................................................16
Reduced Fuel Usage in WMA Pavements ........................................................................ 16
Long Haul Distances and Fast Construction/Traffic Opening .......................................... 17
COMMON FIELD AND LABARATORY WARM MIX ASPHALT (WMA) FOAMING
TECHNOLOGIES ...................................................................................................................18
Foaming Nozzle Based Methods .........................................................................................20
Synthetic Zeolite Based Methods ........................................................................................21
Indirect Foaming Based Methods .......................................................................................22
Shear Based Mixing Methods .............................................................................................22
WMA MIX DESIGN AND SPECIFICATIONS...................................................................23
FOAMED BASED WMA PAVEMENT APPLICATIONS.................................................26
Foamed binder in base/subbase applications ....................................................................26
Foamed Binder in the Surface Layer Applications ..........................................................30
CHARACTERIZATION OF FOAMS IN VARIOUS DISCIPLINES ...............................35
FOAMED WMA BINDER PROPERTIES BASED ON THE CONVENTIONAL
BINDER TESTS ......................................................................................................................39

vii

SYNTHESIS OF PREVIOUS RESEACHES AND MOTIVATION OF CURRENT
STUDY ......................................................................................................................................43
CHAPTER 3 .................................................................................................................................46
FOAMED BINDER PARAMETERS FOR WMA ...................................................................46
EXPANSION RATIO, HALF-LIFE AND FOAM INDEX .................................................48
The practice of the foam height measurements of the foamed asphalt ...........................51
An exercise to observe the effect of ER and HL on morphology of bubbles ..................52
BUBBLE SIZE DISTRIBUTION...........................................................................................55
SURFACE AREA INDEX ......................................................................................................59
CHAPTER 4 .................................................................................................................................61
DEVELOPMENT OF THE ASPHALT FOAM COLLAPSE TEST (AFCT) AND
VERIFICATION USING X-RAY MICROTOMOGRAPHY .................................................61
AFCT SETUP ...........................................................................................................................61
AFCT PROCEDURE ..............................................................................................................64
AFCT IMAGE ANALYSIS ....................................................................................................65
AFCT TEST RESULTS ..........................................................................................................67
Effect of Injected Water Content on the Foam Properties ..............................................68
Effect of Injected Air pressure on the Foam Properties ..................................................70
Analysis of Foamed Binder Properties under Different Injected Air Pressure and
Water Content Combinations .............................................................................................73
VERIFICATION of AFCT USING X-RAY MICROTOMOGRAPHY IMAGING.........78
X-Ray Microtomography Sample Preparation Procedure ..............................................78
3D Imaging using Synchrotron-Based X-ray Microtomography ....................................79
X-Ray CT Microtomography Image Analysis ..................................................................80
Ring Artifact Removal ...................................................................................................... 80
3D XRM Analysis Procedure ........................................................................................... 83
3D XRM Analysis and Comparision with Respect to AFCT ........................................... 84
CHAPTER 5 .................................................................................................................................91
LABORATORY PERFORMANCE TESTS ON WMA MIXTURES PREPARED WITH
FOAMED BINDER .....................................................................................................................91
WMA ASPHALT MIXTURE DESIGN ................................................................................91
PERFORMANCE TEST SAMPLE PREPARATION .........................................................93
STATISTICAL EVALUATIONS OF LABORATORY TEST RESULTS........................96
COATING OF AGGREGATES.............................................................................................97
COMPACTABILITYOF WMA MIXTURES ....................................................................100
MIXTURE PERFORMANCE TESTS ................................................................................103
Dynamic Modulus Test ......................................................................................................103
Dynamic Modulus Test Procedure.................................................................................. 104
Brief Summary of Dynamic Modulus Master Curve ...................................................... 105
viii

Unconfined Dynamic Modulus Test Results and Discussions ....................................... 107
Confined Dynamic Modulus Test Results and Discussions ........................................... 111
Flow Number (FN) .............................................................................................................115
Flow Number Test Procedure and Analysis ................................................................... 115
Flow Number Test Results and Discussions ................................................................... 116
Push-Pull (Compression-Tension) Fatigue Test ..............................................................121
Push-Pull Test Procedure ................................................................................................ 122
Brief summary of Viscoelastic Continuum Damage (VECD) concept .......................... 123
Push-Pull Test Result and Discussions ........................................................................... 125
Tensile Strength Ratio (TSR) ............................................................................................132
TSR Test Procedure ........................................................................................................ 132
TSR Test Result and Discussions ................................................................................... 134
CHAPTER 6 ...............................................................................................................................142
INVESTIGATION OF FOAM DISIPATION USING SYNHROTRON-BASED X-RAY
MICROTOMOGRAPHY .........................................................................................................142
MATERIALS AND METHODS ..........................................................................................142
3D IMAGING USING SYNCHROTRON BASED X-RAY MICROTOMOGRAPHY 144
BINDERS PREPARED USING DIRECT FOAMING ......................................................144
Temporal change in the total volume of the bubbles ......................................................146
Temporal change in the size distribution of the bubbles................................................148
BINDERS PREPARED USING SYNTHETIC ZEOLITE ................................................150
INVESTIGATION OF FOAMED MASTICS ....................................................................153
CHAPTER 7 ...............................................................................................................................156
CONCLUSION AND RECOMEDATIONS ...........................................................................156
SUMMARY ............................................................................................................................156
CONCLUSIONS ....................................................................................................................157
RECOMMENDATIONS.......................................................................................................162
APPENDICES ............................................................................................................................164
APPENDIX A .........................................................................................................................165
APPENDIX B .........................................................................................................................190
APPENDIX C .........................................................................................................................192
APPENDIX D .........................................................................................................................198
BIBLIOGRAPHY ......................................................................................................................204

ix

LIST OF TABLES
Table 1: Reported reductions in plant emissions (percent) with WMA ................................ 15
Table 2: Common field and laboratory foaming tests ............................................................. 19
Table 3: Comparison of surface areas of bubbles with different sizes................................... 53
Table 4: Overall cost estimate of AFCT setup ......................................................................... 64
Table 5: Comparison of Average Bubble Size from X-Ray Microtomography and AFCT
imaging ......................................................................................................................................... 89
Table 6: Comparision of Average Bubble Size from X-Ray Microtomography and AFCT.
....................................................................................................................................................... 90
Table 7: Performance test sample descriptions........................................................................ 94
Table 8: Statistical analysis for percent of uncoated aggregates: a) Illustration of Kendall’s
Tau Correlation Coefficient calculation b) Correlation and Linear Regression Summary
..................................................................................................................................................... 100
Table 9: Statistical analysis for number of gyrations and compactability ratio ................. 103
Table 10: Statistical analysis between unconfined dynamic modulus at 10 Hz and foam
binder quality parameters ........................................................................................................ 111
Table 11: Statistical analysis between confined dynamic modulus at 10 Hz and foam binder
quality parameters .................................................................................................................... 115
Table 12: Statistical analysis between the permeate strain at 50 cycles and foam binder
quality parameters .................................................................................................................... 121
Table 13: Statistical analysis between Number of Cycles to failure at 300 microstrain, 10
Hz, 20°C and Foam Binder Quality Parameters .................................................................... 131
Table 14: Statistical analysis of unconditioned strength, conditioned strength and TSR with
respect to water content and air pressure ................................................................................ 137
Table 15: Description of the specimens utilized in this study. ............................................. 143
Table 16: Performance ranking of mixtures based on the laboratory tests ....................... 159

x

Table 17: Unconfined Dynamic Modulus for the first replicate of 1% - 10psi WMA mixture
..................................................................................................................................................... 166
Table 18: Unconfined Dynamic Modulus for the second replicate of 1% - 10psi WMA
mixture ....................................................................................................................................... 168
Table 19: Unconfined Dynamic Modulus for the first replicate of 3% - 15psi WMA mixture
..................................................................................................................................................... 170
Table 20: Unconfined Dynamic Modulus for the second replicate of 3% - 15psi WMA
mixture ....................................................................................................................................... 172
Table 21: Unconfined Dynamic Modulus for the first replicate of 5% - 20psi WMA mixture
..................................................................................................................................................... 174
Table 22: Unconfined Dynamic Modulus for the second replicate of 5% - 20psi WMA
mixture ....................................................................................................................................... 176
Table 23: Confined Dynamic Modulus for the first replicate of 1% - 10psi WMA mixture
..................................................................................................................................................... 178
Table 24: Confined Dynamic Modulus for the second replicate of 1% - 10psi WMA mixture
..................................................................................................................................................... 180
Table 25: Confined Dynamic Modulus for the first replicate of 3% - 15psi WMA mixture
..................................................................................................................................................... 182
Table 26: Confined Dynamic Modulus for the second replicate of 3% - 15psi WMA mixture
..................................................................................................................................................... 184
Table 27: Confined Dynamic Modulus for the first replicate of 5% - 20psi WMA mixture
..................................................................................................................................................... 186
Table 28: Confined Dynamic Modulus for the second replicate of 5% - 20psi WMA mixture
..................................................................................................................................................... 188
Table 29: TSR test of the WMA mixtures prepared with foamed binder- 1% water content
-10 psi air pressure .................................................................................................................... 199
Table 30: TSR test of the WMA mixtures prepared with foamed binder- 2% water content
-12.5 psi air pressure ................................................................................................................. 200
Table 31: TSR test of the WMA mixtures prepared with foamed binder- 3% water content
-15 psi air pressure .................................................................................................................... 201

xi

Table 32: TSR test of the WMA mixtures prepared with foamed binder- 4% water content
-17.5 psi air pressure ................................................................................................................. 202
Table 33: TSR test of the WMA mixtures prepared with foamed binder- 5% water content
-20 psi air pressure .................................................................................................................... 203

xii

LIST OF FIGURES
Figure 1: Typical mixing temperatures for asphalt pavements (For interpretation of the
references to color in this and all other figures, the reader is referred to the electronic
version of this dissertation.) ......................................................................................................... 2
Figure 2: Working principle of a) The Foamer b) WLB 10 Laboratory Foaming Device .. 21
Figure 3:The structure of foams a) Type 1: Kugelschaum or wet foam b) Type 2:
Polyederschaum or dry foam. .................................................................................................... 35
Figure 4: (a) Static and (b) dynamic foam tests (Source: Schramm 2005) and (c) NIBEM-T
foam stability system................................................................................................................... 37
t

Figure 5: Typical Expansion Ratio (ER) versus time graph. (V = overall foam volume at t)
....................................................................................................................................................... 48
Figure 6: Illustration of comparison of surface areas of bubbles with different sizes ......... 54
Figure 7: Illustration of the reduction of foamed binder volume with time.......................... 57
Figure 8: A sketch of the main component of the AFCT setup a) side-view b) front-view .. 62
Figure 9: Picture of the AFCT test setup .................................................................................. 62
Figure 10: Foam quality analysis of the binders prepared with constant air pressure and
different water contents .............................................................................................................. 69
Figure 11: Bubble size distribution of the binders prepared with constant air pressure and
different water contents .............................................................................................................. 70
Figure 12: Foam quality analysis of the binders prepared with constant water content and
different air pressures................................................................................................................. 72
Figure 13: Bubble Size Distribution of the binders prepared with constant water content
and different air pressures ......................................................................................................... 73
Figure 14: Foam quality analysis of the binders prepared with different water content and
air pressure combinations .......................................................................................................... 75
Figure 15: Bubble Size Distribution of the binders prepared with different water content
st
nd
and air pressure combinations a) 1 Trial, b) 2 Trial ......................................................... 77

xiii

Figure 16: Picture of (a) the foamed asphalt where bubbles are visible at the surface, (b)
illustration of freezing of asphalt binder using liquid nitrogen, c) PP tubes w and w/o
foamed binder.............................................................................................................................. 78
Figure 17: Picture and illustration of Synchrotron-based X-ray Microtomography setup
used in this research. This setup is at Advanced Photon Source (APS) located in Argonne
National Lab (ANL). ................................................................................................................... 80
Figure 18: Ring artifact removal algorithm ............................................................................. 82
Figure 19: 3D temporal view of bubbles in the foamed binder prepared with 1% water
content and 10psi air pressure a)1min, b)15min, c)30min ...................................................... 84
Figure 20: Bubble size comparison of X-Ray Microtomography images foamed binder
prepared with 1% water content and 10psi air pressure a) Average, b)Maximum, c)
Minimum, d) Number of bubbles, e) Bubble Size Distribution .............................................. 86
Figure 21: Bubble size comparison of X-Ray Microtomography images foamed binder
prepared with 2% water content and 12.5psi air pressure a) Average, b)Maximum, c)
Minimum, d) Number of bubbles, e) Bubble Size Distribution .............................................. 87
Figure 22: Comparison of Bubble Size Distribution from X-Ray Microtomography and
AFCT. ........................................................................................................................................... 89
Figure 23: Comparison of Bubble Size Distribution form X-Ray Microtomography images
and AFCT. ................................................................................................................................... 90
Figure 24: Aggregate gradation of WMA mixtures................................................................. 92
Figure 25: Flow chart of the WMA performance evaluation study ....................................... 95
Figure 26: Percent (%) of uncoated aggregates ....................................................................... 99
Figure 27: Compactability of WMA mixtures: a) number of gyrations, b) compactability
ratio ............................................................................................................................................ 102
Figure 28: Asphalt Mixture Performance Tester (AMPT): a) Unconfined test sample, b)
Confined test sample ................................................................................................................. 104
Figure 29: Illustration of shifting |E*| data at different temperatures to obtain the |E*|
master curve .............................................................................................................................. 106
Figure 30: Unconfined Dynamic Modulus: a) log-log, b) linear –log. .................................. 109

xiv

Figure 31: The comparison of Unconfined Dynamic Modulus at 10 Hz and Foam Binder
Quality Parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index (FI), d)
D50, e) Surface Area Index (SAI) ............................................................................................ 110
Figure 32: Confined Dynamic Modulus: a) log-log, b) linear –log. ...................................... 113
Figure 33: The comparison of Confined Dynamic Modulus at 10 Hz and Foam Binder
Quality Parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index (FI), d)
D50, e) Surface Area Index (SAI) ............................................................................................ 114
Figure 34: a) Permanent (plastic) strain with cycles obtained from unconfined FN tests, b)
Permanent strain at 50 cycles .................................................................................................. 119
Figure 35: The comparison of permeate strain at 50 cycles and foam binder quality
parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index (FI), d) D50, e)
Surface Area Index (SAI) ......................................................................................................... 120
Figure 36: Push-Pull Test: a) Custom made gluing jig, b) AMPT fixture, c) Non-accepted
tests (end failure), d) Accepted tests (mid failure) ................................................................. 124
Figure 37: C versus S curves of different WMAs .................................................................. 127
Figure 38: Number of cycles to failure at 300 microstrain. .................................................. 128
Figure 39: Number of cycles to failure at 20°C ...................................................................... 129
Figure 40: The comparison of Number of Cycles to failure at 300 microstrain, 10 Hz, 20°C
and Foam Binder Quality Parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam
Index (FI), d) D50, e) Surface Area Index (SAI) .................................................................... 130
Figure 41: Indirect tensile strength test specimen and stress distribution .......................... 133
Figure 42: Unconditioned and conditioned tensile strength ................................................. 137
Figure 43: Tensile Strength Ratio (%) .................................................................................... 137
Figure 44: The comparison of Unconditioned/Conditioned Tensile Strength and Foam
Binder Quality Parameters: a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index
(FI), d) D50, e) Surface Area Index (SAI) ............................................................................... 138
Figure 45: The comparison of Tensile Strength Ratio (TSR) and Foam Binder Quality
Parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index (FI), d) D50, e)
Surface Area Index (SAI) ......................................................................................................... 140
Figure 46: 3D temporal view of bubbles in two asphalt binder specimens: (a) & (b) PG5828 unmodified binder and (c) & (d) PG70-22 Elvaloy polymer modified binder. .............. 145
xv

Figure 47: Reduction in the volume of bubbles with time; (a) in linear x-y scale and (b)
logarithmic x-y scale. ................................................................................................................ 147
Figure 48: Moisture Dissipation Index (MDI) values for different binders. ....................... 148
Figure 49: (a) 2D slice XRM image of PG70-22CRMF and (b) 3D visualization of crumb
rubber particles. ........................................................................................................................ 148
Figure 50: The change in size distribution of bubbles over time for: (a) & (b) PG58-28F and
(c)&(d) PG64-22F...................................................................................................................... 149
Figure 51: Change in size distribution of the bubbles with time for binder (a) & (b) PG7022 and (c)&(d) PG70-22CRMF................................................................................................ 150
Figure 52: 3D XRM image of PG70-22CRMA binder foamed using Zeolite additive. ...... 151
Figure 53: Reduction in the volume of bubbles with time in specimens prepared with
Zeolite. ........................................................................................................................................ 152
Figure 54: Moisture Dissipation Index (MDI) comparison of binders prepared with foam
and Zeolite binders. .................................................................................................................. 152
Figure 55: 2D XRM image slice of PG58-28A, which was prepared with Zeolite. ............. 153
Figure 56: (a) & (b) 2D slices from XRM images and (c) & (d) 3D visualization of pores of
..................................................................................................................................................... 154
Figure 57: Change in (a) size distribution and (b) mean & median size of the bubbles in
PG58-28SANDF ........................................................................................................................ 155
Figure 58: Change in the overall volumetric percentage of the bubbles with time in
specimen PG58-28SANDF........................................................................................................ 155
Figure 59: Raw Flow Number Data ........................................................................................ 191
Figure 60: Push Pull Tests for WMA mixtures prepared with foamed binder - 1% water
content -10 psi air pressure ...................................................................................................... 193
Figure 61: Push Pull Tests for WMA mixtures prepared with foamed binder - 2% water
content -12.5 psi air pressure ................................................................................................... 194
Figure 62: Push Pull Tests for WMA mixtures prepared with foamed binder - 3% water
content -15 psi air pressure ...................................................................................................... 195
Figure 63: Push Pull Tests for WMA mixtures prepared with foamed binder - 4% water
content -17.5 psi air pressure ................................................................................................... 196
xvi

Figure 64: Push Pull Tests for WMA mixtures prepared with foamed binder - 5% water
content -20 psi air pressure ...................................................................................................... 197

xvii

CHAPTER 1

INTRODUCTION

Global warming and increase in fuel prices have made researchers to consider alternative
construction techniques to conventional Hot Mix Asphalt (HMA). According to Environmental
Protection Agency (EPA), more than 500 million tons of HMA were produced at the 3,600
(estimated) active asphalt plants in United States in 1996 to construct and maintain thousands of
miles of roads (EPA-454/R-00-019, 2000). It is reported that average asphalt plants emits
approximately 2500 tons of CO2. Overall, it makes emission of 8.75 megaton of CO2 only in the
USA, which is equivalent to approximately 0.5% of total gas emissions (Bahia and Miller, 2009).
It is more than to be neglected for the sustainability of the asphalt pavements. Meanwhile,
approximately 1 billion gallons of fuel is used annually for HMA production, material extraction
and usage. About 90% of total life cycle energy is solely spent in production (Bahia and Miller,
2009).
Recent developments in asphalt production technologies lead to decrease in the
production costs and greenhouse emissions. One way of classifying the technologies is by the
degree of temperature reduction. Figure 1 shows a classification of various application
temperatures for asphaltic concrete. Cold Mix asphalt has been used as an alternative to HMA,
reducing mixing and compaction temperatures by emulsifying the asphalt in water prior to
mixing with the aggregate. The asphalt emulsion is less viscous and the mixture is easy to work

1

with and compact at relatively low temperatures. However, insufficient aggregate coating, higher
air void contents, limited workability duration and reduction of strength for long term
performance has made it unsuitable for all conditions. It is commonly used as a patching material
and on low volume traffic service roads (Blades and Kearney, 2004). To overcome the limitations
of cold mix asphalt, Warm Mix Asphalt (WMA) has been introduced, which is produced
between cold and hot mix asphalt temperatures (Prowell, 2008).
WMA is the general term used for the asphalt pavements produced and placed at lower
temperatures as compared to conventional Hot Mix Asphalt (HMA) (D’Angelo et al., 2008;
Hurley, 2006; Diefenderfer et al., 2007; Prowell, 2008). Lower production temperatures are
achieved by means of various WMA technologies including foaming, wax-based additives,
emulsion-based products and surfactants.

160°C
100°C

20°C

16°C

132°C

120°C

121°C

160°C
HMA

138°C

140°C

WMA

COLD MIX

Figure 1: Typical mixing temperatures for asphalt pavements (For interpretation of the
references to color in this and all other figures, the reader is referred to the electronic
version of this dissertation.)
WMA mixing temperatures are about 20 to 30°C lower than conventional HMA and
slightly above 100°C as shown in Figure 1 (Prowell, 2007). This amounts to about 20-30%
reduction in paving temperatures, which leads to numerous benefits including: (i) economical
2

aspects (reduced fuel usage, increased paving season), (ii) environmental aspects (reduced green
house emissions), (iii) engineering aspects (improved (better) field compaction, less aging (i.e.,
oxidation) of the mixture, less permanent deformation etc.) and (iv) worker exposure (decrease
in exposure to fumes for the paving crew) (De Groot et al., 2001; Larsen et al., 2004; Cervarich,
2003; Prowell et al.; 2007).

The fuel saving due to usage of WMA instead of HMA is

approximately 30 to 35% (Prowell, 2008). WMA also improves the compaction in cold weather,
extends the allowable compaction period, increases the haul distance and potentially more
suitable for using high percentage of reclaimed asphalt pavement (RAP) (Kristjánsdóttir et al.,
2007). Aged RAP binder compensates for the softness of virgin binder in WMA and helps
towards rutting susceptibility. WMA’s advantages (e.g., reduced fuel cost and greenhouse
emissions) over traditional HMA have made it very appealing to state and local roadway
agencies as well as the industry (Hassan, 2009; Mallick et al., 2009). In 2010, just six years after
the first trial in the U.S., 13% of the pavements were constructed with WMA (Prowell et al.,
2012). This corresponds to ~47 million of tons of WMA, which amounts to approximately 30
million gallons of fuel savings that is worth $80 million (Nadeau, 2012). The WMA lessens the
carbon dioxide release around 800,000 tons. To date, all 50 states in US constructed WMA trial
sections (Prowell et al., 2012).
MOTIVATION
Most of the current design methodology of WMA foamed asphalt pavements are based
on the WMA plant manufacturers’ recommendations and past experience of the contractors.
Such design approach has lead to premature WMA pavement failures (Kim et al., 2011).
Recently, an NCHRP project produced recommended WMA mixture design specifications
3

(Bonaquist, 2011), which is primarily based on limited empirical data. Most of the current
knowledge on WMA is based on empirical data and there is a significant lack of understanding
of the fundamental behavior of the foamed binder used in WMA pavement. Understanding the
WMA foamed binder characteristics that affect the mechanical behavior of pavements is crucial
to accurately predict and improve its long term performance.
OBJECTIVE AND SCOPE OF THE RESEARCH
The main objectives of this study are listed as follows:
(1) To develop new parameters to assess the quality of the foamed binder (i.e., bubble
size distribution, surface area, etc.)
(2) To develop an accurate and repeatable laboratory/ field equipment and protocol to
measure the reduction in the height of the foam binder over time.
(3) To develop links between the foam binder quality parameters (i.e., expansion ratio,
half-life, foam index etc.) and asphalt pavement performance (i.e., rutting, fatigue
cracking, moisture susceptibility etc.).
(4) To investigate the characteristics of foamed binders using different types of binders
(Performance Grade (PG), Crumb Rubber (CR), etc.)
OUTLINE OF THE DISSERTATION
The thesis divided into seven chapters including CHAPTER 1, which is the introduction.
In CHAPTER 2, the literature and background on WMA pavements and foamed based binders is
presented. In CHAPTER 3, foamed binder quality indicators (i.e., Expansion Ratio, Foam Index,
Half-Life) are discussed and new parameters (i.e., Bubble Size Distribution and Surface Area
Index) are introduced to assess the quality of the foamed binder. In CHAPTER 4, a novel testing
4

methodology called as Asphalt Foam Collapse Test (AFCT) is introduced to measure the height
reduction of the foamed binder as it collapses in order to precisely calculate the foamed binder
quality indicators. In CHAPTER 5, the long term performance of the WMA mixtures prepared
with different foamed binders in the laboratory is evaluated via performance tests and compared
with the foamed binder quality parameters presented in CHAPTER 3. In CHAPTER 6, X-ray
Microtomography (XRM) imaging technique is used to analyze the binder type and foaming
technology on the generation and evolution of the foam. The findings and the impact of research
are summarized in CHAPTER 7.

5

CHAPTER 2

LITERATURE REVIEW AND BACKGROUND

HISTORY OF WMA PAVEMENTS
WMA technologies are developed to significantly reduce the mixture viscosity at lower
temperatures to facilitate better coating of the aggregates, increase the workability of the loose
mixtures and provide improved (better) compaction. These technologies have been extensively
used in Europe since the first laboratory experiments starting with Aspha-min zeolite by
Mitteldeutsche Hartstein-Industrie AG in 1995 and WAM foam by Shell Bitumen and Kolo
Veidekke in Norway in 1996. In 1997, the first WMS pavement section was constructed with
Sasobit in Hamburg, Germany. The earliest WAM foam and Aspha-min zeolite field trials were
in Norway and Germany in 1999.
In 2002, National Asphalt Pavement Association (NAPA) organized a study tour to
Denmark, Germany and Norway to investigate the following warm mix technologies: (i)Asphamin, (ii)WAM foam, and (iii) Sasobit (Prowell, 2008). Brief descriptions of these technologies
are given below:
(i) Aspha-min is a foaming additives that is a kind of synthetic zeolite, which is
composed of aluminosilicates and alkalimetals that contain about 20% crystallized
water. The water is released by increasing the temperature above the boiling point of
water, creating a controlled foaming effect. This leads to an increase in binder volume

6

and reduction in binder viscosity, which provides 6-7 hours of workability period till
the temperature drops below approximately 100°C (212°F).
(ii) WAM-Foam is a foaming process that includes two-stage mixing of soft and hard
binders, in order to obtain the desired performance grade (PG) binder. The soft binder
controls the coating of coarse aggregates while the stiffer binder increases the overall
stiffness, reducing susceptibility to rutting. Water is added for foaming approximately
2-5% by weight of the hard binder (i.e. about 1.6 lb water per ton of mix assuming
5% total asphalt content, 80% of which is hard binder) at approximately 175 -180°C
(347-356°F).
(iii) Sasobit is a synthetic paraffin wax, which combines hot coal and natural gas with
steam with the aid of a catalyst. Sasobit has higher viscosity than binder below its
melting point and lower viscosity than binder above its melting point. They harden in
asphalt from 65 to 155°C into regularly distributed, microscopic stick-shaped
particles.
The study tour by NAPA is followed by a research, which was initiated at National
Center for Asphalt Technology (NCAT) and sponsored by NAPA, FHWA, Eurovia and Sasol in
2003 (Prowell, 2008). Interests in WMA technologies in the U.S. have significantly increased
after the first field trials with Aspha-min in 2004 in Florida and North Carolina (D’Angelo et al.,
2008) and continued with field trials in Florida, Indiana, Maryland, New Hampshire, Ohio and
Texas in 2005. In 2007, AASHTO and FHWA organized a scan tour to Belgium, France,
Germany and Norway. D’Angelo et al. (2008) reported that WMA performance in Europe is
7

same or better than HMA. In 2008, it was documented that 32 U.S. states has WMA trial
sections. Moreover, Texas DOT introduced the first WMA specification, which allows WMA to
be used statewide. The number of states having trial sections and specifications drastically
increased by the end of 2011, where 45 U.S. states had trial sections and 30 U.S. states had
specifications. It was reported that all 50 states in U.S. conducted trial WMA sections by the end
of 2011 (recent record available). FHWA survey reported that WMA usage was 19.2 millions of
tons in 2009 and 47.6 million tons in 2010, which was equivalent to 13% of overall asphalt
production (Prowell, 2008).
There are currently over 30 different WMA technologies being used in U.S., more
available worldwide. These technologies are based on various chemical additives/surfactants,
foaming

methods

and

non-foaming

additives.

The

most

well-known

chemical

additives/surfactants in U.S. markets are Cecabase, Evotherm, HyperTherm and Rediset.
Common Non-foaming additives in U.S. can be listed as BituTech, Leadcap, Sasobit,
SonneWarmix, and Thiopave (Prowell, 2008). However, the two thirds of the technologies are
based on foaming methods, which are explained in detail at Section 0 (Prowell et al., 2012).
BENEFITS OF WMA
Engineering Aspects
There are several engineering benefits that could be gained from adopting WMA
technologies in asphalt pavement construction, which are discussed below.

8

Aging of WMA binders
The WMA binders are not exposed to elevated temperatures during the production and
construction. Therefore, the WMA mixtures are less susceptible to aging and cracking, which can
lead to longer pavement service life (Hossain et al., 2009; Rubio et al., 2012).

Workability and Compactability of WMA pavements
WMA technologies reduce the overall binder viscosity, which improves the workability
of asphalt mixtures at lower temperatures. Thus, the mixtures can be compacted with less
number of roller passes to reach the targeted density (D’Angelo et al., 2008; Hossain et al.,
2009). Stiff mixes (i.e., mixes with high percent of Reclaimed Asphalt Pavement (RAP) and
Recycled Asphalt Shingles (RAS)) have compactability problems, which lead to lower in-place
densities. In order to suppress the compaction densities to target levels, the compaction
temperatures or efforts may be increased, which can result aggregate breakdowns and damage in
the pavement, even during construction. Therefore, WMA technologies help in the compactabilty
of the relatively stiff asphalt mixtures in conventional compaction temperatures (Prowell, 2008).
Usage of RAS in WMA pavements
Usage of Recycled Asphalt Shingles (RAS) in asphalt pavements has four major
advantages: (i) high asphalt binder content, 20% to 30% by weight of the shingle, (ii) high grade
frictional aggregate, (iii) fiberglass fibers that promote flexibility in the asphalt mixes and, (iv)
lime dust that is a natural anti-strip for asphalt aggregates. RAS used in paving mixtures range
from 3% to 5% by weight of HMA or WMA mixture. WMA enables the incorporation of both
RAP and RAS in many mixtures not possible with HMA. In St. Louis, Missouri, an off-ramp at
9

the exit 249 interchange on Interstate 70 was constructed with RAP, RAS and WMA mix in
2010. The mixture was a 12.5mm Superpave surface mix containing limestone and traprock plus
17% RAP and 3% RAS. (Jackson, 2011). Jackson (2011) reported that the pavement has been
performing well, although there were concerns when it was initially placed.
Usage of High Percentage RAP in WMA pavements
WMA technologies potentially allow use of less virgin materials, by utilizing high
percentage of RAP in asphalt mixtures at lower temperatures. The decreased aging of the binder
due to lower WMA production temperatures helps in rejuvenating the RAP binder, particularly
in regard to low-temperature cracking (D’Angelo et al., 2008).
High RAP asphalt pavements are frequently constructed in Europe, though it is not
common in U.S. In Germany, a trial base course section containing 45% RAP was successfully
placed at a range of -1°C to 3°C ambient temperature as the mixture temperature during the
placement varied from 102°C to 139°C (D’Angelo et al., 2008). It was reported that better
compaction was achieved with WMA as compared to HMA with the same and fewer roller
passes. Additionally, mixtures prepared with low energy asphalt (LEAB) WMA method are
commonly used with 50% unfractionated RAP in Netherlands. Moreover, trial sections have
been constructed with 90 to 100 % RAP using Aspha-min zeolite and Sasobit in Germany
(D’Angelo et al., 2008).
In the U.S., majority of State DOT’s are hesitant to use of high percent RAP in the
surface layer of the asphalt pavements. It was reported in 2011 that 10% of the states (Alaska,
Washington, North Dakota, Iowa and Connecticut) do not permit more than 25 % RAP in any of
10

HMA layers. Another 10% of states such as California, Nevada, Michigan, Louisiana and New
York only permit more than 25 % RAP in the base layer. Using high RAP content in both base
and intermediate layers is allowed by 32% of the states (e.g., Colorado, Arizona, New Mexico,
Oklahoma, Texas).

About 48% of the states (e.g. Illinois, Montana, North Carolina, and

Alabama) allow high percent RAP in all layers. Meanwhile, about 55 % of the states are
experimenting the use of high RAP in the surface layer (FHWA-HRT-11-021, 2011).
Mississippi DOT studied the use of high RAP (50 % to 100%) mixtures containing warm
mix additives as a base layer on low and medium level traffic highways (FHWA/MS-DOT-RD-09200, 2009). Compactability and indirect tensile (IDT) strength were evaluated on samples with
varying RAP contents and sources, virgin asphalt content, Sasobit content and temperature. It
was concluded that compactability of the WMA samples was not challenging if moderate
amount of binder used. IDT strengths of samples with higher RAP content were much greater,
which could indicate cracking potential (FHWA/MS-DOT-RD-09-200, 2009).
Mallick et al. (2008) studied mixtures with Sasobit at 75% RAP level at lower
construction temperatures. In that study, HMA control mixes were prepared at 150°C with
PG64-28 binder and at 135°C with PG52-28 binder. WMA mixtures were prepared at 125°C
with both PG52-28 and PG42-42 binders. The performance tests conducted were rutting and
indirect tensile strength. It was concluded that the air void of mixtures prepared with Sasobit had
low variance. Similar air voids were achieved at lower temperatures. The mixtures with lower
PG grade (42-42) had respectively better performance. However, long term durability and fatigue
properties of these mixtures were not evaluated, which was the significant dilemma in this
research.
11

NCAT placed two full-depth 50% RAP sections with HMA and foamed WMA, and one
control HMA section with no RAP to NCAT Pavement Test Track in 2009. After the application
of 10 million equivalent standard axle loads (ESAL), the high RAP WMA section performed as
well as the control. No cracking, excellent rut resistance and lower texture changes were
observed in the high RAP WMA section as compared to HMA control section. The performance
of as the high RAP WMA was as good as high RAP HMA section (NCAT Asphalt Technology
E-news, 2012).
Florida DOT placed foamed WMA section with 45 % RAP to State Route 11 in Deland,
FL (Copeland et al., 2010). Additionally, a high RAP HMA mix was placed and its performance
was evaluated to compare with the high RAP WMA. Performance tests included PG
determination of binders, dynamic modulus and flow number. It was observed that high RAP
WMA mix is softer than the control. Therefore, the high RAP WMA mix had a lower flow
number than the high RAP HMA. In addition, dynamic modulus tests also confirmed that the
high RAP WMA mix is slightly softer than the high RAP HMA control mix, especially at
intermediate temperatures (Copeland et al., 2010).
Pennsylvania DOT initiated a study to evaluate the usage of high percent RAP in WMA
pavements due to poor performance of a field section produced with 35% RAP content. The
laboratory experiments were limited with one aggregate and RAP source, one virgin binder
(PG64-22) with two RAP contents; 15% and 35%. In addition, the WMA technologies used in
this study were limited to foaming, Sasobit and Evotherm. Although the rutting and moisture
susceptibility of the mixtures were evaluated, fatigue performance and low-temperature cracking
resistance were not evaluated as part of this research. It was concluded that use of anti-stripping
12

agent is mandatory with the foaming mixtures at 15 % and 35% RAP levels. Mixtures with
Evotherm and Sasobit at 15% RAP level yielded a higher or similar TSR value compared to
HMA. On the other hand, mixtures with both additives at 35% RAP level yielded lower TSR
value, though they passed the minimum TSR criteria. Rutting resistance of foaming mixtures at
15% and 35% RAP levels is better than conventional HMA. On the contrary, mixtures with
Evotherm and Sasobit at 15% and 35% RAP performed poor rutting resistance (Solaimanian et
al., 2011).
Zhao et al. (2012) studied the performance of the WMA foamed mixtures with 30% RAP
as compared to HMA with 0% and 30% RAP. It was concluded that WMA with high percentage
of RAP exhibited higher rut resistance, better moisture damage resistance, and better fatigue
performance.
Shu et al. (2012) studied the moisture susceptibility of plant produced WMA containing
high percent RAP in Tennessee. The loose mixtures were compacted and subjected to one
Freeze-Thaw cycle in AASHTO 283 and Moisture Induced Stress Tester (MIST). Their
performance was evaluated through tensile strength ratio, indirect tension, dynamic modulus, and
Asphalt Pavement Analyzer (APA) Hamburg wheel tracking tests. It was concluded that IDT and
dynamic modulus tests are capable of accurately characterizing the moisture susceptibility.
Foamed WMA with high RAP content performed as well as HMA in terms of moisture
susceptibility.

13

Usage of Crumb Rubber in WMA pavements
Scrap tire rubber has been used in asphalt pavements since 1950s. There are numerous
laboratory and field studies that showed superior performance of CR modified asphalt pavements
over traditional HMA (Heitzman, 1992). However, CR modified pavements are mixed and
compacted at higher temperatures as compared to conventional HMA due to their poor
workability at conventional temperatures. Zhao and Amirkhanian (2009) proved that WMA
technologies (Aspha-min and Sasobit) in conjunction with CR modified mixtures have the
potential to decrease the construction temperature requirements while maintaining the highperformance characteristics of the pavement. In addition, Massachusetts DOT has been working
on a specification to specify WMA for CR modified gap-graded asphalt mixtures (Prowell,
2009).
Cold Weather Paving
The paving season in colder regions can be extended by utilizing WMA technologies due
to WMA’s ability to maintain workability at lower temperatures. However, the production
temperatures of WMA at cold weather applications depend on the WMA technology, ambient
conditions and haul distance (D’Angelo et al., 2008).
Kristjansdottir (2006) studied the performance of WMA pavements produced with WAM
Foam, Aspha-min zeolite and Sasobit wax under cold weather conditions. It was stated that
dense graded mixtures are more stable and have low permeability, which is especially important
when compacting in cold weather. In addition, the mixtures prepared with Sasobit and WAM
Foam was found to be more resistant to rutting. Furthermore, it was shown that achieving
14

adequate moisture damage resistance may be a challenge when using warm mix methods and
using anti-stripping agents are mostly desirable. Therefore, it is very crucial to determine the
moisture susceptibility of WMA pavements in cold weather conditions. In Germany, paving was
completed in various case studies using different WMA technologies when ambient temperatures
were between -3°C and 4°C (D’Angelo et al., 2008).
Environmental Aspects
Environmental advantages of WMA technologies include reduced emissions and better
working conditions (Anderson et al., 2008). Increased environmental awareness in Europe and the

emission regulations in Kyoto protocol motivate the European researchers on developing and
improving various WMA technologies. Therefore, HMA industry in Europe has been actively
investigating the ways of reducing CO2 emissions. Plant emission data from the WMA suppliers
from Netherlands, Norway, Italy and France are given in Table 1 (D’Angelo et al., 2008).
Futhermore, the worker exposure while placing WMA is lower than HMA, was also stated in the
French, German, and Italian researches (D’Angelo et al., 2008).
Table 1: Reported reductions in plant emissions (percent) with WMA
Emission Norway Italy

Netherlands France

CO2

31.5

30-40 15-30

23

SO2

NA

35

NA

18

VOC

NA

50

NA

19

CO

28.5

10-30 NA

NA

NO

61.5

60-70 NA

NA

NO2

NA

NA

18

Dust

54

25-55 NA

NA

NA
15

In the U.S., the provision of Clean Air Act in 1990 manifested the emission regulations.
Many states in the U.S. have been struggling to achieve national air-quality standards set by
Environmental Project Agency (EPA). On the other hand, the gas emitted by asphalt plants is
only equivalent to 0.5% of total gas emissions and EPA does not consider the asphalt plants as
main concerns of the air pollution as the emission levels are lower than the regulations. Though,
it is more than to be neglected by the environmentalists. In 2001, the National Institute for
Occupational Safety and Health (NIOSH) in USA published a hazard review on Health Effects
of Occupational Exposure to Asphalt. In this review, NIOSH evaluated the potential health
effects of occupational exposure to asphalt. Lange and Stroup-Gardiner (2007) showed that the
asphalt plant emission is dependent on the plant temperature. It is reported in the German
Bitumen Forum that there is no emission measured below 80°C and emission is about 1 mg/h at
approximately 150°C (Ruhl and Lindemeier, 2006). On the other hand, the emission drastically
increases above 180°C (D’Angelo et al., 2008). Moreover, Hossain et al. (2009) stated that
WMA plant emissions are about 30% to 98% of HMA plant emissions.
Economical Aspects
The economical advantages of using WMA technologies include the reduced fuel usage,
long haul distance, rapid compaction and early traffic opening, which are summarized below.
Reduced Fuel Usage in WMA Pavements
The operation temperatures of WMA mixtures are generally lower that the HMA
mixtures. For instance, a temperature reduction of 28°C in an average asphalt plant corresponds
to about 11% fuel savings (Cervarich, 2007). As part of the NCHRP 9-47 project, it was

16

reported that the fuel savings in different WMA field trials varies from 15.4% increase to 77%
reduction. Prowell (2009) reported the average fuel saving as 23% in WMA applications. In
addition to the reduction in temperature, there are significant operational factors affecting the
plant emissions such as plant design, aggregate moisture, RAP/RAS content in the mixtures and
fuel type. D’Angelo et al. (2008) reported the burner fuel savings vary from 20% to 35%. In
addition, Hossain et al. (2009) indicated that the overall energy consumption ranges from 20% to

75% between HMA and WMA based on the utilized technology.
Long Haul Distances and Fast Construction/Traffic Opening
The rate of cooling of WMA mixes is lower than that of the conventional HMA.
Therefore, the haul distance of the mixes can be longer as the cooling rate is reduced. Thus, the
proper use of WMA may result in reduced overall paving costs. In Norway, Kolo Veidekke
reported that WAM foam mixture still had the ability to be placed after 48 hours. In another
study, a mixture containing Sasobit was hauled up to 9 hours in Australia (D’Angelo et al.,
2008).
The road constructed or maintained by WMA can be opened to traffic faster than
conventional HMA. This is very promising for high maintenance roads and intersections and
airports (Zaumanis, 2010). Hurley and Prowell (2006) studied the curing time of the laboratory
mixtures prepared with Aspha-min, Evotherm and Sasobit. It was concluded that the WMA
mixtures did not gain strength with time as compared to conventional HMA and WMA
pavements do not require curing time before opening to traffic. A field section in Italy was
opened to traffic five hours after paving began by Schumann Sasol (Hurley and Prowell, 2005).

17

Twenty four inches of WMA (Sasobit) were placed in 7.5 hours while repaving the Frankfurt
airport. The runway was opened to jet aircrafts at a temperature of 85°C (Hurley and Prowell,
2005). Texas DOT opened Loop 368 at San Antonio to traffic in 2 hours after laying the WMA
pavement with Evotherm (Rand, 2008).
COMMON FIELD AND LABARATORY WARM MIX ASPHALT (WMA) FOAMING
TECHNOLOGIES
WMA foaming technologies in the common ground introduce small amounts of water to
hot binder, either via a foaming nozzle or using a hydrophilic material such as zeolite, or through
wet fine aggregates. The field and laboratory foaming technologies, that are commonly used in
U.S. are listed in Table 2. Foaming is reported to be the most cost effective WMA technology as
far as the long term production is concerned and also due to less material cost and ease of the
production (Bennert, 2008; Middleton and Forfylow, 2009). Technologies involving waterbased foaming techniques require additional equipment installed at the plant to measure and
deliver the additives.
As shown in Table 2, the field foaming technologies can be divided into four major
categories: (i) foaming nozzle-based methods, (ii) synthetic zeolite-based methods, (iii) indirect
foaming via mixing hot aggregates with asphalt and wet fine aggregate, and (iv) shear-based
mixing. All of these technologies utilize significantly different methods. Because of a wide
variety of methods, WMA mixtures are produced at very different conditions (i.e., temperature,
water content, asphalt absorption by aggregates, etc.). As a result, the degree of coating, amount
of trapped moisture, asphalt binder absorption of aggregates may exhibit great variation. These
parameters play a crucial role on the performance of the WMA pavements. However, there is no
18

clear understanding of the effects of different WMA technologies on the quality of the foam
generated, which can significantly affect the overall global performance of the mixture. This is
partly because there is no standard test method for measuring the characteristics (i.e., the quality)
of foamed binders and evaluating the characteristics of the WMA mix designs.

Type

Table 2: Common field and laboratory foaming tests
Name (Manufacturer)

Warm Mix Additive / Process

Almix WMA System
Foaming Nozzle
AquaFoam (Reliable Asphalt
Foaming Nozzle
Products)
Eco-Foam II
Foaming Nozzle
Meeker Warm Mix
Foaming Nozzle
Tri-Mix Warm Mix injection System Foaming Nozzle
(I) WAM-Foam (Kolo Veidekke, Shell Aggregate coated with soft binder then hard foamed
Bitumen)
binder is added.
AQUABlack (Maxam Equipment Inc.) Foaming unit installed in an existing A/C line
Double Barrel Green (ASTEC)
Foaming Nozzle
Terex Warm Mix Asphalt System
Foaming Nozzle
(Terex Roadbuilding)
Ulrafoam GX (Gencor)
Foaming Nozzle
Aspha-min (Eurovia)
Synthetic zeolite (~0.25% by weight of mixture)
(II)
Advera (PQ Corporation)
Synthetic zeolite (~0.3% by weight of mixture)
Hot coarse aggregate mixed with wet sand. Also
Low Energy Asphalt (LEACO)
coating & adhesion additive added (~0.5% by
weight of binder)
(III)
Low Emission Asphalt
Sequential coating using wet fine aggregate and
(McConnaughary Technologies)
unspecified additive
(IV) Accu-ShearTM (Stansteel)
Mixing asphalt and water via shearing action.
The Foamer (Pavement Technology
Inc.)
Laboratory foamer (Foaming Nozzle)
WLB 10 Laboratory Foaming Device
(Wirtgen America, Inc.)
Laboratory foamer (Foaming Nozzle)
(Lab) Accu-Foamer(D&H Equipment Ltd.) Laboratory foamer (Foaming Nozzle)
Asphalt Hydro-FoamerPP (D&H
Equipment Ltd.)
Laboratory foamer (Foaming Nozzle)
WAM Foam Laboratory Foaming
Device (Kolo Veidekke)
Laboratory foamer (Foaming Nozzle)
19

Foaming Nozzle Based Methods
The working principle of the nozzle based foaming methods is to inject air and water
into hot binder (>100°C) through a spraying nozzle. The water is turned into steam at
atmospheric pressure, creating moisture bubbles in the binder, and significantly decreasing the
overall viscosity. Yunus and Boles (1994) stated that the binder expands by a factor of 1.673.
This aids in aggregate coating, mixture workability and compaction at lower temperatures.
Currently, there are numerous different field and laboratory nozzle based foaming techniques. In
order to illustrate the difference in working principles, a few of the methods are discussed in the
following paragraphs.
AquaFoam is a nozzle based foaming method by AquaFoam, LCC (Prowell, 2008). The
system consists of two fan nozzles mounted 180° apart, which are mounted perpendicular to the
asphalt stream, just before it enters the drum. Addition of 1. % water by weight of the total mix is
recommended by plant manufacturers.
Double Barrel Green by Astec Industries Inc. is a multi-nozzle system that
microscopically foam the binder with water (Prowell, 2008). The air/water mixture (at different
concentrations, typically 2% by weight of the binder) is sprayed into binder to create foam, and
then the foam is forced through a narrow nozzle before mixing with aggregate. Approximately
one pound of water is used per ton of mix and the mixture temperature is recommended to be
from 121°C to 135°C.
In order to simulate the nozzle based foaming in the field, laboratory foaming devices are
developed by several companies. As the nozzle design and foaming methods differ from
technology to technology in the field, the same difference is evident in the laboratory foamers.
20

WLB 10 Laboratory Foaming Device by Wirtgen America Inc. is one of the devices where, air
and water are individually sprayed to pressurized binder in the expansion chamber as illistrated
in Figure 2b. On the other hand, “The Foamer” by Pavement Technology Inc. has a totally
different nozzle system as sketched in Figure 2a. The binder flows down in a pipe (via gravity)
around an air/water injection nozzle that creates conical water spray. The binder hits the injected
air/water mixture, which creates the steam bubbles. In this dissertation, The Foamer is used with
major modifications to increase the control on the water content and pressure as well as air
pressure.
Time Controlled cleaning
and closure of the nozzle

Pipe
Walls
Binder flowing
Down the pipe

Other
Nozzles

Hot
Bitumen
Nozzle
Air/
Water

Air

Water

Foamed Asphalt

Foamed Asphalt

Figure 2: Working principle of a) The Foamer b) WLB 10 Laboratory Foaming Device
Synthetic Zeolite Based Methods
WMA foaming additives, Advera by PQ Corporation and Aspha-min by Aspha-min
GmbH are the most common synthetic zeolites. They are composed of aluminosilicates and
alkalimetals that hold about 20 % crystallized water. The water is released by increasing the
temperature above the boiling point of water and creating a controlled foaming effect. It leads a

21

slight increase in binder volume and reduces the binder viscosity as well as providing 6-7 hours
workability period till the temperature drops below approximately 100°C (212°F). It was
documented that the mixing and compaction temperatures of the WMA are approximately 30oC
less than conventional HMA. However, foamed binder with these additives is prepared and
added in different ways. Advera is added to the binder shortly prior to mixing with aggregates
and generally added at 0.20% to 0.25% by weight of the total mix. Lower dosage of the additive
is used as a compaction aid. On the contrary, higher dosage of the additive is used if the binder
content of the mix exceeds 7% by weight of the total mix. However, Aspha-min is mixed with
the aggregates and the binder at the same time and added about 0.30% by weight of the total mix.
Indirect Foaming Based Methods
The most well-known indirect foaming method through wet and hot aggregates are Low
Emission Asphalt (LEA) by McConnaughay Technologies and Low Energy Asphalt (LEACO)
by Lea-Co (Prowell, 2008). In LEA and LEACO technologies, coarse aggregates are initially
heated to about 150°C, mixed with (unfoamed) binder along with a coating/adhesion additive.
Then, cold wet fine aggregate containing approximately 3% to 4% water and recycled asphalt
pavement (RAP) is added. While mixing, the moisture in the wet fine aggregate turns into steam
and creates the foam.
Shear Based Mixing Methods
Accu-Shear is a shear based method by Stansteel Asphalt Plant Products that applies
mechanical shear to mix the water and/or WMA additives by a colloidal mill (Prowell, 2008).
The binder is foamed dynamically by adjusting the rate of shear, which is assumed to increase

22

the life of the foam. Typically, one gallon of water is used for one ton of mixture (personal
communication with Oldcastle Materials, a nationwide paving contractor).
WMA MIX DESIGN AND SPECIFICATIONS
In Europe and U.S., WMA pavements are constructed with a range of layer thickness, in
a wide variety of traffic levels, utilizing different aggregate skeletons such as dense graded, stone
matrix, porous and mastic asphalt pavement. In Norway, WAM foam process is allowed by
Norwegian Public Roads Administration. In Germany, a bulletin called as “Merkblatt” is
released for the construction of WMA pavements utilizing five modifiers, i.e. Fischer-Tropsch
wax, Montan wax, fatty acid amides, a blend of Montan wax and fatty acid amides, and zeolite
(Prowell, 2008). In France, Service of Technical Studies of the Roads and Expressways
(SETRA) certified Aspha-min zeolite in 2007 (Prowell, 2008).
Asphalt pavement mix design is typically different in Europe as compared to U.S. In
addition, materials used in the asphalt pavements are also different. For instance, the water
absorption of aggregates is less than 2% in most regions of the Europe, though it is much higher
in the U.S.

Although WMA technologies significantly drop the pavement production

temperatures, the production temperatures are above the suggested temperatures in U.S. than
Europe in order to ensure the coarse aggregate dryness.
NAPA and FHWA formed WMA Technical Working Group (WMA TWG) in 2005
(Prowell, 2008). The group had representatives from FHWA, NAPA, NCAT, NIOSH, State
Asphalt Pavement Associations (SAPA), American Association of State Highway and
Transportation Officials (AASHTO) and the Hot Mix Asphalt Industry. Their mission was to
evaluate and validate WMA technologies and foster the environmental concerns. The WMA
23

TWG submitted three high priority research needs statements that were combined into two
projects by the NCHRP: i) Project 09-43 "Mix Design Practices for Warm Mix Asphalt
Technologies" for fiscal year 2007, ii) Project 09-47 "Engineering Properties, Emissions, and
Field Performance of Warm Mix Asphalt Technologies" for fiscal year 2008.
The objective of NCHRP 09-43 was to develop a WMA mixture design and performance
analysis procedure for a wide range of WMA technologies in the U.S. Therefore, a procedure
similar to AASHTO R35 was developed for the design of dense graded WMA mixtures based on
limited empirical data. The first phase of this project included analyzing: (i) the effect of sample
reheating on the properties, (ii) the binder grade selection, (iii) the conditions for short term
conditioning, (iv) the degree of mixing that occurs between the RAP and new binder, (v) the
workability of mixtures, and (vi) the continuum damage fatigue analysis. The second phase of
the project was to evaluate the first phase and included: 1) the mix design study, 2) the validation
study to compare the laboratory and field produced WMA mixes, 3) the study to assess the
fatigue properties of mixes produced at lower temperatures. NCHRP Report 691: Mix Design
Practices for Warm Mix Asphalt was published as the final report of NCHRP 9-43 in 2011. The
NCHRP Report 691 suggested that the compactability, moisture sensitivity and rutting resistance
of WMA are significantly different than HMA and should be evaluated according to draft
Appendix of AASHTO R35 for WMA. The findings of the project in binder selection can be
listed as followings: The stiffness of the recovered binder from the mixtures can drastically
decrease when the production temperatures are extremely low. However, low WMA production
temperatures enable minor improvements in the low temperature grade of the binder. On the

24

contrary, if the production temperatures are extremely low, the high temperature grade of the
binder needs to be increased in order to meet the flow number (rutting resistance) requirements.
The foamed WMA binder properties (such as viscosity) continuously change with time.
Therefore, mixing and compaction temperatures cannot be determined from the binder viscosity.
It was suggested in NCHRP 9-43 that the coating and compactability of the mixtures should be
determined directly. In addition, WMA pavements can be more prone to moisture damage than
HMA pavements. Therefore, the effect of anti-stripping agents was also studied during the
NCHRP 9-43 project. It was stated that 67% of the mixes with the help of the agents had the
same or improved TSR value than WMA mixtures prepared without agents.
The mixture tests indicated that short-term aging (2 hours) is adequate to condition WMA
mixtures. It was observed that the virgin binder coats the virgin aggregates and RAP during the
mixing period, and then the RAP and virgin binder continue to mix during the storage at elevated
temperatures. Thus, the short term conditioned samples better represent the field performance of
WMA pavements because of the less aging at lower mixing/compaction temperatures. The
performance test sample preparation criteria for WMA mixtures are set as 2 hours of short-term
conditioning at the compaction temperature since this duration is adequate the simulate the
binder absorption and stiffening that occurs during the construction.
Recent national research projects on WMA are: NCHRP 9-47A: Properties and
Performance of Warm Mix Asphalt Technologies, NCHRP 9-53: Properties of Foamed Asphalt
for Warm Mix Asphalt Applications, and NCHRP 9-55: Recycled Asphalt Shingles in Asphalt
Mixtures with Warm Mix Asphalt Technologies. In both Europe and U.S., there are still lots of

25

unknowns and concerns about the WMA technologies and their long term performance. In
addition, there is no standard preparation, production, evaluation and lay down specifications on
the WMA pavements.
FOAMED BASED WMA PAVEMENT APPLICATIONS
While there are numerous advantages of WMA over HMA, WMA is still a relatively new
technology and several concerns about WMA technology such as long-term performance and
moisture susceptibility still need to be addressed (D’Angelo et al., 2008). Foamed binder was
first developed by Dr. Ladis Csanyi to improve the properties of poor quality Iowa aggregates in
Bituminous Research Laboratory of the Engineering Experiment Station at Iowa State University
about fifty years ago (Jenkins, 2000). The steam was injected to heated binder using a nozzle to
produce controlled foam. Cold and wet aggregates or soils were easily coated with the foam
binder due to its reduced viscosity and the increased surface energy (Lee, 1980). Mobil of
Australia got the patent rights of Csanyi’s foamed binder process in 1968. They developed and
patented the expansion chamber and nozzle system, called as Foamix, to inject 1-2% cold water
with steam in 1971.
Foamed binder in base/subbase applications
Foamed binder has long been used in the soil and base/subbase stabilizations. In these
applications, typically wet unheated aggregates and asphalt cement are mixed while the asphalt is
in foamed state to form fairly stiff (mortar like) material. In the 1970’s, over 16 countries were
using foamed binder for the base/subbase stabilization in Europe and South Africa. Although
foamed binder was first developed in U.S., the first documented application was by the Georgia
Department of Transportation, in Ware County in 1982 (Raffaelli, 2004).
26

Nataatmadja (2001) summarized the advantages of foamed binder stabilization as: (i)
increase in strength over unbound materials, (ii) quick construction, (iii) lower cost than
reconstruction, (iv) immediate traffic and (v) improved durability. In addition, past researches
indicated that a wide range of aggregates from crushed stone to silty sands can be used with
foamed-bitumen in the stabilization. However, the fine content of mixes is crucial and should
preferably be above 5% by weight of the mix (Ruckel et al, 1983). Sakr and Manke (1985) and
Bissada (1987) showed that the mixes with high percentage fines (around 10%) were more stable
and had high tensile strength since the binder primarily concentrated in the finer fraction of the
aggregates.

However, there were many concerns in these applications since there are no

complete specifications and standards. The most extensive design guides are: (i) South African
Interim Technical Guideline: The Design and Use of Foamed Bitumen Treated (2002), (ii)
Wirtgen Cold Recycling Manual (2004).
Moisture is crucial to soften and break the agglomerations between the aggregates and to
diffuse the binder in the mixture during the mixing and compaction. However, one of the
primary concerns in the stabilization is the trapped moisture in the pavement layers. It was
indicated by Csanyi (1960) and Ruckel at al. (1983) that the quantity of the water is not critical.
However, if inadequate amount of water is used, it prevents the dispersion of foam, workability
and compaction of the mixture. On the contrary, if excessive water is used, it extends the curing
time and reduces the density and strength of the mixture (Brennen et al., 1983). Bowering (1970)
indicated that the mixture gains full strength only after certain period of time where curing
(water evaporation/dissipation) takes place. Ruckel et al. (1983) stated that the failure (rutting
and raveling) of foamed asphalt mixtures occurs usually weeks after the construction not years.
27

This is because of insufficient curing (water dissipation) of the foamed asphalt mixture. Clarke
(1976) confirmed this when he found out that the engineering properties of the mixture improve
with age and temperature. Because of these findings, Lee (1981) recommended that effects of
curing on the strength development of specific mixtures should be evaluated locally and the
specifications should be created accordingly. Engelbrecht et al. (1985) suggested using low
stockpiles (i.e., store in relatively small quantities to increase exposed surface area) after mixing
so that the moisture dissipates more rapidly and mixture gains full strength faster. However, care
should be taken not to “over-cure” the mixtures in order not to hinder the workability during
placement and compaction.
Initial foam binder base stabilization was performed using virgin materials (Acott, 1979;
Bowering, 1970; Bowering, 1976; Lee, 1981; Ruckel et al., 1980). Due to the shortage of virgin
materials, foamed asphalt begun to be implemented in the form of full depth reclamation (FDR)
to efficiently utilize old asphalt pavement (Brennen et al., 1983; Engelbrecht et al., 1985; Van
Wijk et al., 1983). In the 1980s, several foamed asphalt projects were placed in Colorado and
Wyoming. In Maine, the first foamed asphalt full depth reclamation project was constructed in
June 2001 (Marquis et al., 2003). In Louisiana, the foamed asphalt treated RAP was utilized for
the first time as a base material in lieu of a crushed lime stone base underneath a concrete
pavement layer in January 2002. This study consisted laboratory foamed asphalt treated RAP
mix designs, field test sections and field evaluation of strength/stiffness of foamed asphalt base
course. It was concluded that foam asphalt treated RAP mixtures had higher in-situ stiffness
than the lime stone base. In addition, it was indicated that there was no significant stiffness

28

changed between the 100% RAP mix and thee mix with the combination of 75% RAP and 25%
crushed stone (Mohammad et al., 2003).
The foamed asphalt improves the workability, stiffness, and strength, and reduces
moisture sensitivity with the inclusion of active (portland cement, lime) and/or inert (fly ash,
mineral fines) fillers. FDR is generally used to rehabilitate the crack pavements and to eliminate
the effects of reflective cracks in California (Jones et al., 2008). Jones et al. (2008) studied the
identification of the properties affecting the performance and distress mechanisms of materials
recycled with foamed asphalt and the determination of the acceptable ranges of the properties of
FDR foamed asphalt materials in California. It was concluded that foamed asphalt material with
a cementitious filler can be used as a rehabilitation option on thick and crack asphalt pavements
on highways, in which the traffic is less than 20,000 vehicles per day. This method is also
applicable to pavements where multiple overlays have been placed on a weak base and where
cracks reflect through overlay. In addition, the recycled layer can be used as a subbase
underneath a new base layer. However, the performance of FDR should be assessed for each
project, mix design and construction.
Cold in-place recycling (CIR) with foamed binders is another common way of
rehabilitation of existing pavements (Kim and Lee, 2006; Kim et al., 2006). In 2002, two
different stabilizing agents (asphalt emulsion and foamed binder) used at a section of Route 20 in
Iowa. Foamed and emulsion based stabilization differs in many ways such as aggregate coating.
As indicated before, the foam binder tends to initially coat the fine and small aggregates. The
coarse aggregates adhered with the asphalt mastic. On the contrary, emulsions coat the coarse
aggregates and they bind the uncoated fine aggregate. Moreover, foamed asphalts have shorter
29

curing time and results early opening to traffic due to the lower water contents in foamed asphalt
stabilization compared to the water in emulsion treatments. It was concluded that indirect tensile
strength (IDT) of foamed CIR is critical and should be evaluated. However, IDT is not adequate
by itself and it was recommended that further tests such as dynamic modulus and dynamic creep
tests should be conducted. Ramanujam and Jones (2007) also indicated that foamed asphalt
stabilization performed better than emulsion-treated stabilization.
Foamed Binder in the Surface Layer Applications
WMA pavements in surface layer applications are relatively new and there are similar
unknowns and questions to be answered as stabilization (Diefenderfer and Hearon, 2008). These
questions are primarily related to their long-term performance. One of the primary concerns is
their potential moisture susceptibility (Kvasnak et al., 2009). In most of the WMA products,
water (steam) is introduced into the mixture to reduce the viscosity at low temperatures. It is still
unclear if additional moisture added to the mix can cause long-term problems such as stripping.
Incomplete drying of the aggregates at lower temperatures may further accelerate the moisture
damage in WMA pavements (especially with absorptive limestones) (Diefenderfer and Hearon,
2008). Current knowledge on diffusion and evaporation of moisture in foamed WMA is limited.
The foaming process decreases the over viscosity of the asphalt mixture so that it is workable
during construction. However, after construction, viscosity increases rapidly as the foam
disappears and temperature drops. During the process of dissipation of foam, if the temperature
decreases rapidly, the moisture may not escape and may be trapped inside the mixture. This
trapped moisture can cause detrimental failures by breaking the adhesive bonds between the
aggregates and the asphalt binder (through diffusion and because of freeze/thaw cycles). It can
30

also destroy the cohesive bond within the binder over time and during application of the traffic
load. Therefore, it is crucial to know how the moisture escapes from the asphalt mixture as the
specimen cools down and foam disappears. However, the knowledge on evidence of moisture
susceptibility is limited and mostly depends on the empirical studies.
Hurley and Prowell (2005) reported that WMA mixtures made of Aspha-min exhibited
low Tensile Strength Ratio (TSR). A laboratory research performed by Bhusal (2008) indicated
that TSR values for WMA mixtures made with Zeolite were lower than the TSR of control
(HMA) mixtures. Hurley and Prowell (2005) showed that Hamburg wheel tracking tests showed
evidence of moisture damage, with lower number of cycles to stripping inflection point. Field
cores in that project also showed low TSR. On the other hand, Powers (2008) indicated that there
is no visible difference in TSR tests between WMA and HMA samples on 7 different WMA
projects in Ohio. In all of these projects, ASTEC Double Barrel Green (DBG) foaming
technology was used.

Kvasnak et al. (2009) analyzed the moisture susceptibility of two

laboratory and plant prepared mixtures with Evotherm DAT in Birmingham, AL. The study
indicated that laboratory prepared WMA mixtures had more tendency to moisture susceptibility
than plant produces mixtures. Additionally, the control (HMA) mixtures performed better than
WMA, though most of the WMA samples met the moisture criteria. Xiao et al. (2010)
investigated the moisture damage in WMA mixtures containing moist aggregates. The
conventional moisture susceptibility tests such as indirect tensile strength (ITS), TSR, toughness
and percentage of toughness loss were run to laboratory prepared samples with Aspha-min and
Sasobit. ITS values and deformation resistance decreased for mixtures containing moist
aggregates. WMA modification method did not show significant effect on toughness values.
31

Kavussi and Hashemian (2011) investigated the WMA foam mixes based on their moisture
susceptibility and rutting potential by utilizing ITS and wheel tracking tests. These studies
indicate that there is no common ground regarding to the moisture susceptibility of different
WMA technologies. Wielinski et al. (2009) assessed the field and laboratory performance of
WMA utilizing ASTEC DBG and HMA sections composed of same aggregate skeleton.
Production temperature was the only property changed in the design. Therefore, the initial
stiffness of the WMA section was lower than HMA due to lower production temperatures. As
expected, the performance tests indicated lower Hveem stability, Marshall Stability and flow,
and higher Asphalt Pavement Analyzer (APA) rut tests.
Rutting is the other main concern in WMA mixture design (NCHRP Report 891 and
714). WMA pavements have early rutting failure potential, right after construction under traffic
load. Rutting potential increases due to reduced aging of the binder (Su et al., 2009). Goh and
You (2007) studied the performance of WMA prepared with Aspha-min in the laboratory. Lower
rutting depth and a higher indirect tensile resilient modulus were observed at higher compaction
temperatures. On the other hand, there was no significant difference on the resilient modulus of
samples compacted at low (100°C) and relatively high temperatures (120°C). Dynamic modulus
of samples compacted at high temperatures (120°C) was significantly higher than WMA
compacted at 100°C. Hurley and Prowell (2008) analyzed WMA field sections in St. Louis,
Missouri, and Milwaukee, Wisconsin by utilizing the following WMA technologies: Sasobit,
Evotherm, and Aspha-min. The conclusions drawn were: WMA sections had better or equal inplace densities than control sections, 6 months after the construction. It was observed that WMA
sections are slightly more susceptible to rutting based on the results of APA rutting test. West
32

(2009) indicated that lower compaction temperatures make the WMA mixtures more susceptible
to moisture damage. Kanitpong (2007) showed that Sasobit improved the compactability of
asphalt mixtures at significantly reduced temperatures and these mixtures performed better in
rutting and fatigue resistance and achieved higher shear complex modulus. You et al. (2011)
evaluated the mixtures prepared with Aspha-min. Dynamic modulus tests (|E*|) were performed
and resulted that WMA with Aspha-min had no significant effect. It was also found that
increasing amount of Aspha-min significantly improved the rutting resistance of the mixture.
Xiao et al. (2012) studied the effect of compaction temperature on the moisture susceptibility and
rutting of foamed WMA mixtures based on foamed aggregates. The experimental matrix
consisted two aggregate moisture levels (0% and ~0.5% by weight of the dry mass of the
aggregate), one lime content (1% lime by weight of dry aggregate), two foaming water contents
(2% and 3%) with control, and two aggregate sources. Compactability, ITS, rut depth of both
conditioned and unconditioned and flow number of each samples was assessed. It was concluded
that the aggregate source drastically affects the ITS and rutting resistance regardless of the
foaming water content, aggregate moisture content, and compaction temperature. The samples
with moist aggregates were more compactable regardless of aggregate type, foaming technology
and compaction temperature. In addition, the ITS values of foamed mixtures containing moist
aggregate increased as the compaction temperature increased. Moreover, the rut depths of all the
conditioned and unconditioned mixtures slightly increased when their compaction temperatures
decreased, regardless of aggregate moisture content, foaming water content, and aggregate type.
Ayman et al. (2012) studied the performance of laboratory foamed WMA and compared with the
conventional HMA. The test matrix was composed of two aggregates (crushed limestone and
33

natural gravel) and two asphalt binders (neat PG64-22 and polymer modified PG70-22M) and
following tests were utilized: ITS, |E*|, TSR and APA. Foamed WMA mixes had lower ITS
values and more susceptible to moisture damage than HMA. Meanwhile, there was no significant
difference in the dynamic modulus of WMA and HMA, though WMA mixes were more
susceptible to rutting failure. It was proved that the aggregate and binder type had a significant
effect on the performance WMA pavements.
Fatigue life of WMA pavements is another issue to be investigated. WMA pavements
may have longer fatigue life and higher fatigue characteristics because of the reduced aging of
the binder during mixing/compaction (NCHRP 374). However, Jones et al. (2009) stated that
WMA technologies do not affect the fatigue performance. Additionally, mixture stiffness
decrease due to the limited aging of the binder. Therefore, it provides WMA mixtures to
incorporate more recycled asphalt pavement than HMA mixtures. It was observed that some of
the WMA additives can increase the potential for low temperature cracking based on the binder
tests (Jones et al., 2009). Xiao et al. (2009) investigated the fatigue performance of rubberized
WMA pavements. The test beams (for Four Point Bending Beam (FBBB) test) were prepared
with one rubber size (#40 mesh), two aggregate sources, two WMA additives (Aspha-min and
Sasobit) and tested at 20°C. It was indicated that compaction and mixing temperatures of the
crumb rubber asphalt concretes significantly reduces with the WMA additives. Additionally,
inclusion of the WMA additives extended the fatigue life and increased the stiffness of the
rubberized WMA pavements with respect to control (HMA) pavements.
Based on the findings of past research, it is important to perform laboratory experiments
on moisture susceptibility, rutting and fatigue cracking on the WMA specimens, in order to
34

predict the effect of foamed binder characteristics on the long-term performance of WMA
pavements in the field.
CHARACTERIZATION OF FOAMS IN VARIOUS DISCIPLINES
Foam is described as a thermodynamically unstable colloidal dispersion in which a gas is
dispersed in a continuous liquid phase (Schramm, 2005). Foams are investigated in various
disciplines because of its widespread application in life. For instance, food industry (champagne,
soda heads, whipped cream etc.), detergent industry (dishwashing and clothes-washing), personal
care products (shaving cream, bubble bath foam, hair shampoo suds), process industry (foam
blankets, fire extinguishing foams, mineral and oil flotation froths) are the common foam
application areas in everyday life (Schramm, 1994).

Lamellae
b) Kugelschaum or wet foam

a) Polyederschaum or dry foam

Figure 3:The structure of foams a) Type 1: Kugelschaum or wet foam b) Type 2:
Polyederschaum or dry foam.

Foams in different disciplines are typically classified as Type-1 and Type-2 due to their
structures (shapes) as shown in Figure 3 (Sebba, 1987; Schramm, 2005). Type-1 foams have well
separated spherical bubbles in the liquid, where the liquid volume is the same or larger than the
bubbles. This type is also known as Kugelschaum or wet foam, gas emulsion or ball foam. Type35

2 foams have non-spherical (polyhedral) bubbles separated by almost flat liquid films called
lamellae. This type of foams is referred as Polyederschaum or dry foam. Jenkins (2000) indicated
that the foamed asphalts used in base stabilization applications are more close to Type-2 foam.
However, Kutay and Ozturk (2012) showed that the bubbles in foamed WMA binders made
using a laboratory foamer are spherical and similar to Type-1. This study is discussed in Chapter
6.
The characterizations of the foams are commonly assessed in terms of their stability. The
foam stability is evaluated against: (i) film thinning and (ii) film rupturing (Schramm and
Wassmuth, 1994). In the film thinning process, the bubbles are separated with thin films (liquidfilms) and no change in the total surface area of the foam is observed when the bubbles touch
each other. On the other hand, in the film rupturing process, the bubbles are turned to one single
larger bubble when the bubbles coalesce. Therefore, the total surface area decreases with time.
As referred in the description, foams are thermodynamically unstable and it is crucial to
quantify their evolution with time. Typical tests for characterization of foam stability are divided
into three major categories: (i) lifetime of individual bubbles, (ii) static foam tests and (iii)
dynamic foam tests (Schramm, 2005). The measurement of bubble lifetime is not commonly
used since small contaminations and vibrations influence the results and reproducibility is not
good (Schramm, 2005). In static foam tests, typically a constant volume of foaming solution in a
pipette is allowed to fall a specified distance with recording the time into a separate volume of
the same solution that is contained in a vessel. Decay of the volume of the foam as well as the
initial volume (right after draining of the fluid above) is measured (ASTM D1173-53: Ross and
Miles Test) as shown in Figure 4a. There are simpler tests such as ASTM D3601-88 (Bottle
36

Test), D3519-88 (Blender Test) and DIN 53902 Part 1 (Perforated disk test), which are also
variations of the static foam tests where reduction in height of the foam over time are measured.
The main difference is how the foam is generated such as poring, shaking, beating and stirring
(Barel et al., 2005). In dynamic foam tests, foam is generated by flowing gas through a porous
orifice into the fluid as shown in Figure 4b. The volume of the foam is measured when steadystate flow is achieved. Examples of such tests include ASTM D892-92, D1881-86 (Diffuser
Stone Test) and D3427-86 (Gas Bubble Separation Test). The difference between the tests is the
foam generation method such as air injection and circulation (Barel et al., 2005). It should be
noted that unlike foams in many other disciplines, there is typically no static foam layer on the
surface of the binder in foamed WMA applications. The bubbles rising to the surface typically
collapse and disappear. Therefore, dynamic foam tests are probably not applicable to foamed
WMA binders.
a)
Foam Pipet
Foaming
Solution
Fine
Orifice

b)

Regulated
Gas Injection

c)

Graduated
Cylinder

Foam
Receiver
Static Foam
Height

Dynamic
Foam Height
Porous Stone

Foaming
Solution

Test Solution

Figure 4: (a) Static and (b) dynamic foam tests (Source: Schramm 2005) and (c) NIBEM-T
foam stability system.

37

In most of the ASTM tests described above, the foaming characteristics of the fluid are
measured for individual foaming method (e.g., pushing air through the porous stone). However,
there are various methods of making foam in WMA applications. Therefore, the test method for
foamed WMA should be such that the foamed WMA specimen sampled at the plant or
laboratory device can be measured and compared to a baseline method. As a result, some kind of
static foam test is needed, where the reduction in height of the overall foamed asphalt is
measured.
The NIBEM-T foam stability instrument (Haffmans, 2012) is an equipment typically
used to measure the reduction in foam height of beers (Figure 4c). In this system, a movable
plate with three electrodes is lowered to make contact with the surface of the foam. As the foam
collapses, contact between the electrodes and the foam is lost. The instrument continuously
moves the plate down to restore contact. The measured rate at which the plate is lowered
quantifies the rate of collapse of the foam. However, in its present configuration, this system has
operation temperature range between -5 and 40oC, which will not allow testing of foamed binder.
Guillerme et al. (1993) developed an apparatus to analyze the formation and stability of foams
that is coupled with a video camera. The system evaluates the foam texture as well as the
physical characteristics, such as foam volume and the liquid in the foam. However, the overall
system is not applicable for WMA because of its opaque nature.
The other methods to determine foam stability also include the physical properties of the
foam such as density and viscosity. German and McCarthy (1989) utilized MRI imaging to
measure the dissipation rate of the foam. The variation in the proton signal across the imaging
plane with time meets the change in density and indirectly the dissipation pattern for every point
38

in the foam. However, this method cannot be applied to foamed binder since the rate of the
dissipation with time is variable due to temperature decrease of the foam and rapid density
change. Hutzler et al. (1995) used AC Capacitance method to measure the density of the nonionic detergent solution (foam). A trial was done to measure with the capacitance of foam binder
in Advanced Characterization Asphalt Laboratory at Michigan State University. However, it was
observed that the foam dissipates before the capacitance stabilizes. Therefore, it is not applicable
and practical for foamed binder.
Neu (1960) presented various methods of foam measurement and defined several
parameters to characterize foam. The focus of this research was on the foams generated by
shampoo and toothpaste. He developed a method using Sunbean mixer to assess the foaming
profile. He defined parameters such as (i) Specific Foam Volume (ml/g), (ii) Density= Mass of
Foam/Volume, (iii) Viscosity = measured using Modified Techne Viscometer, (iv) Light
Transmission: The loss of light transmission through a layer of foam is a function of the degree
of dispersion of air, (v) Photomicrography: The particle size distribution and specific surface
area of the foam measured on a 100 cm2 area, and (vi) Foam Drainage. Among these methods,
most appropriate parameters that may apply foamed binder are probably the particle (bubble)
size distribution and specific surface area.
FOAMED WMA BINDER PROPERTIES BASED ON THE CONVENTIONAL BINDER
TESTS
As stated before, there are various foamed binder technologies. Therefore, the foam
characteristics are scattered in a wide range because of the technology, temperature, water
content, air pressure, additive content, etc. The preliminary studies and discussions stated that the
39

test methods for the foams summarized in section 0 are not applicable to the WMA foamed
binders because of their temperature, opaque nature etc. On the other hand, many researchers in
the pavement area have attempted to characterize foamed binder properties by modifying or
utilizing traditional HMA binder tests.
Saleh (2006) used Brookfield viscometer to measure the change in viscosity of the
foamed asphalt with time. As an alternative measure of the quality of the foamed asphalt, he
suggested that the average foam viscosity over the first 60 seconds of foaming could be used. It
should be noted that foams are typically non-Newtonian fluids (Schramm 2005). Therefore, their
viscosity depends on the applied shear rate. Also, steady-state shearing motion is needed to
measure a correct viscosity. It is important to evaluate the repeatability and accuracy of viscosity
measurements in highly dynamic (unsteady) foamed asphalt, which is typically, collapses
quickly. While the method may be promising, care should be taken while defining and
interpreting the viscosity measurements.
Gandhi and Amirkhanian (2007) used Brookfield viscometer and Dynamic Shear
Rheometer (DSR) to analyze the variation in the binder properties right after binder modification
and after aging to simulate the plant shutdown. The samples were prepared by two different
WMA modification methods were used foaming additive (Aspha-min) and non-foaming additive
(Sasobit). The initial viscosity tests were run at 135 and 120°C at 30, 60 and 90 minutes of
adding the WMA additives. The |G*|/sinδ at the PG temperature and at the failure temperature
were measured. Then, the binders were aged at 120°C for 3 days. Their viscosities were
measured at 135°C and PG grade of the aged binders were determined. It was clearly observed
that the effect of WMA additives significantly is dependent on the chemical properties of the
40

asphalt binders. Additionally, it was noted that Aspha-min increased the viscosity of the base
binder. On the other hand, Sasobit significantly decreased the viscosity of the binder.
Wasiuddin et al. (2007) studied the rheological properties of two binders (PG64-22 and
PG70-28) with Sasobit and Aspha-min. The additives were added 2%, 3% and 4% by weight of
the binder. It was concluded that the change in the amount of Sasobit had no effect on the mixing
temperature of PG64-22. On the other hand, the mixing temperature of PG70-28 decreased by
10°C, 12°C, 13°C for added 2%, 3% and 4%, respectively. Moreover, it was observed that the
amount of Aspha-min had no effect on the mixing temperatures. The G*/sin(δ) measurement
indicated that both Aspha-min and Sasobit had no negative effect on binder grading due to high
temperature viscosity reduction. In addition, the authors performed rutting measurements on the
WMA mixtures. It was stated that that rutting potential decreases with decreasing mixing and
compaction temperatures.
Binder lubricity test for DSR was introduced to evaluate the workability of asphalt binder
(Hanz et al., 2010). It was stated that internal friction of asphalt binders decreased by the
inclusion of additives. Hence, this provided the aggregates to compact at lower temperatures.
The setup that Hanz et al. (2010) used is composed of three balls mounted on a ball assembly
and a fourth ball placed on the tip of rotating chuck. A thin film of lubrication fluid is spread in
between the ball assembly and chuck .Then, the chuck is rotated in one direction. The testing
temperatures are selected in between 80-100°C due to the limitation of the DSR apparatus. The
normal forces are suggested to be 20 and 30 N to maintain the contact between the assembly and
the chuck under 50 revolutions per minute (rpm). The measured coefficient of friction is used to
evaluate the effect of binder grade and WMA additive. It was stated that the viscosity reduction
41

is not the only factor to reduce the production temperatures since no dependence on the shear
rate was observed.
Nazzal and Qtaish (2013) utilized Atomic Force Microscopy (AFM) to evaluate the
moisture susceptibility and healing characteristics of WMA pavements. Various AFM
technologies such as tapping mode imaging and force spectroscopy were performed

on two

types of binders using four WMA technologies (Advera, Evotherm, Sasobit, and foamed WMA).
AFM images introduced the dimensional changes in the foam structure (bee-like) within the
virgin and polymer modified asphalt binders. Sasobit additive decreased the size of the bee-like
structures within the neat and polymer modified asphalt binders. On the other hand, no
significant difference in the structure dimensions was observed in other WMA technologies.
Therefore, the stiffness of binders with Sasobit measured to higher than the other binders as the
higher shear modulus values obtained in the DSR test. In addition, it was observed that nanoscale adhesive forces increased in all binders in the utilization of WMA technologies before the
moisture conditioning. The forces measured to be highest in Advera and foamed WMA and
lowest in Sasobit. Therefore, Sasobit mixture can have lower indirect tensile strength value than
the other mixtures. In addition, the AFM analysis showed that these adhesive forces significantly
decrease after moisture conditioning in both the control (HMA) and mixtures produced with
different WMA technologies. However, the stress reduction was the least in the Evotherm WMA
and the control mixtures and the highest in the Advera WMA. Therefore, it was expected that
Advera WMA can have the least TSR value in overall mixtures. In addition, the AFM force
spectroscopy experiments stated that TSR value depends on the adhesive forces between the
aggregates and the binder. Moreover, AFM healing experiments indicated that WMA
42

technologies except the Sasobit improved the micro-crack closure rate. It was concluded that
AFM measurement is feasible to study the moisture damage and healing phenomena in WMA
mixtures.
Huang et al. (2012) used neutron scattering technology to determine the microscopic
structure of asphalt and for determining the presence of moisture and its distribution in foamed
binder. The resolution varies from 200 nm to 1 nm under the small angle neutron scattering
(SANS) in the vector transfer range from 0.003 Å-1to 0.5 Å-1. Two types of asphalt binder
(PG64-22 and one from Korea) and ordinary and heavy water (deuterium oxide, D2O) at 4% by
weight of binder were used to make samples at 150°C via a laboratory foaming device. However,
the sampling of the foamed binder was poor and water dissipated through the sampling.
Therefore, the authors’ conclusions were debatable since no water entry less than 0.1 µm was
observed in the foamed asphalt and it was claimed that if water greater than 0.1 µm exists, no
micro-structural changes were detected less than 0.1 µm.
SYNTHESIS OF PREVIOUS RESEACHES AND MOTIVATION OF CURRENT STUDY
While the WMA technologies are appealing, the long-term performance of WMA as
compared to HMA is not well-known in U.S and Europe. WMA’s disadvantages are mainly
related to rutting and moisture susceptibility issues as referred. Premature rutting failure has been
reported for surface asphalt concrete in different studies. This has been mostly related to
decreased ageing at lower production temperatures and increased moisture content for foaming
technologies. In addition, it is early to rely on the long term performance of WMA pavements
based on the laboratory studies. It should be recalled that the first field sections constructed in
U.S. are less than nine years old. The first sections constructed in Europe (Germany and
43

Norway) are about fourteen years old. Since water is injected in most of the WMA foaming
technologies during the initial mixing process. Residual water in the mixture can cause
premature rutting and stripping due to incomplete vaporization. Therefore, the moisture
susceptibility of mixtures should be carefully evaluated. Moreover, most of the current design
methodologies of WMA foamed asphalt pavements are based on the WMA plant manufacturers’
recommendations and past experience of the contractors. Such design approach has lead to
premature WMA pavement failures (Kim et al., 2011).
NCHRP 09-43 Final Report is the most recent study in U.S. This report suggests a WMA
mixture design specification which is primarily based on limited empirical data (Bonaquist,
2011). In addition, most of the current knowledge on WMA is based on empirical data and there
is a significant lack of understanding of the fundamental behavior of the foamed binder used in
WMA pavement. Understanding the WMA foamed binder characteristics that affect the
mechanical behavior of pavements, is crucial to accurately predict and improve its long term
performance. Therefore, there is a growing need for understanding the WMA pavements from
binder production to mixture performance.
There are four major challenges in understanding the foamed WMA pavement. First, the
foamed binder quality indicators should be determined such as expansion ratio, half-life and
foam index etc. Secondly, a repeatable testing method should be developed to repeatedly and
precisely measure and calculate these quality indicators. Thirdly, the foam structure should be
observed with time to understand the workability of the binder as well as the physical properties
of the binder such as residual water. Finally, the performance of the WMA pavements should be
consistent with the foam binder quality indicators.
44

To respond to these needs, this research; (1) identified various parameters as foam binder
quality indicators, (2) developed a practical laboratory device to measure these parameters, (3)
validated the device by comparing with nondestructive 3D imaging methods (i.e., x-ray
microtomography), (4) investigated the relationship between the binder quality indicators
proposed and the mixture performance tests, (5) investigated foaming characteristics of different
kinds of binders, (6) investigated the effects of air pressure and water content on the foamed
binder and mixture performance.

45

CHAPTER 3

FOAMED BINDER PARAMETERS FOR WMA

Making foamed binder is relatively simple process where hot binder is mixed with a
limited amount of water (typically 2-3% by weight of the binder). However, the rheology of the
foamed binder is not way simple. The quality of the foamed binder depends various factors such
as the binder type, grade and modification, the foaming technology used, amount of water,
temperature etc. Moreover, the quality of the binder plays a crucial role during the mixing, laying
and compaction stages of warm mix asphalt pavement production.
Typically, the mixing and compaction temperatures of the asphalt pavements are
determined by the viscosity of the binder, which is an indicator of the mixture workability.
However, measuring the viscosity of the foamed binder leads to disturbance to bubbles and
unreliable values. It should be also noted that the viscosity of the foamed binder is time
dependent, in which the viscosity of the foamed binder varies with both time and temperature.
Therefore, researchers developed parameters such as expansion ratio, half-life and foam index,
as a measure of the quality of foamed binders. These parameters are discussed in detail in the
fallowing section. In addition, the factors influencing the foam quality have been studied by
various researchers.
Abel (1978) concluded that the silicone content in bitumen reduces the foaming abilities
of binder. Binders with relatively low viscosity had a tendency to foam more than those with

46

high viscosity. Thus, they had more expansion and shorter half-life. On the other hand, it was
observed that the higher viscosity binders resulted better aggregate coating due to longer halflife. Abel (1978) also suggested that the temperature should be above 147°C for acceptable
foaming quality.
Barinov (1990) proved that the increase in the concentration of asphaltenes increases the
expansion ratio and half-life of the binders. Asphaltenes act as surfactants reducing the surface
tension in the bubbles, which result in the delay of foam collapse. On the other hand, Lesueur et
al. (2004) concluded that bitumen composition did not significantly influence foam
characteristics. It should be noted that very limited research data is available on the effects of
binder composition on foam characteristics. Castedo-Franco and Wool (1983) indicated that any
binder independent from its type, grade and source could be foamed with the proper
combinations of the nozzle type, water content, air and bitumen injection pressure.
Namutebi et al. (2011) studied the affect of the foaming process on the chemistry of the
binders. It was hypothesized that the injected air and water may cause oxidative aging to the
binder. Fourier Transform Infrared spectrometry (FTIR) method was utilized to analyze the
different components of the foamed and unfoamed binder. The carbonyls and sulphoxides
compounds are the major indicative of binder aging. It was concluded that the foaming caused no
change in the binder chemistry because of the short term exposure to water and air. In addition, it
was concluded that the foam characteristics were affected by the penetration grade rather than
the source of bitumen.

47

EXPANSION RATIO, HALF-LIFE AND FOAM INDEX
Asphalt foams used in base stabilization applications were typically characterized using
following three parameters: Expansion Ratio, Half-life and Foam Index. Expansion ratio (ER) is
the ratio of the expanded volume to the initial volume of the binder (Brennen et al., 1983). Figure
5 shows an illustration of reduction in ER with time, which is equivalent to reduction in height of
a foamed binder with time.
The ER can be defined as:

ER =

V
V

0

0

[1]

f

f

where V and V are overall foam volume at time t=0 and final binder volume after all foam
dissipates, respectively (Figure 5). The ER is a measure of relative volume of the steam bubbles.

t

f

t=0

t+Δt

t

t=inf.

ER=V /V
0

f

V /V

t f
V
t+Δt/V f

V

/V

0

f

0.5V /V

0

V

t

t+Δt

V

f

V

V

1
t t+Δt thalf-life

Time=t

t

Figure 5: Typical Expansion Ratio (ER) versus time graph. (V = overall foam volume at t)
48

The rate of dissipation of moisture, on the other hand, is typically quantified using the
parameter called half-life (HL). HL is defined as the elapsed time between the time at which the
foamed binder reaches its maximum volume and the time it reaches to half of the maximum
volume (Brennen et al., 1983). HL can be determined as follows in Equation 2:
HL = t

where t
0 .5 V

0

/V

f

0 f
f
0 . 5 ( V -V ) / V

[2]

is the time at which the overall foam volume is reduced by half, as shown in

Figure 5.
Brennen et al. (1983) stated that the expansion ratio and the half-life are affected by the
amount of foamed asphalt prepared, water content used and the temperature at which the
foaming took place. Ruckel et al. (1983) indicated that the foam parameters are affected by the
size of the container, in which the binder is foamed.
Typically expansion ratio increases with foaming temperature and water content;
whereas, half-life decreases with increasing temperature and water content (Kim et al., 2006).
Moreover, the decrease in the half life causes the film thinning of the foam as well as the
reduction in the viscosity as temperature increases (Wirtgen, 2004). As the viscosity decreases,
the surface tension of the bitumen films decreases, and the steam pressure within the bubbles
exceeds the surface tension of the bitumen and bubbles and the bubbles collapse.
Researchers have been working to determine optimum combinations to increase both the
quality of the foamed binder and foamed based WMA asphalt pavements. For instance, Brennen
et al. (1983) reported that a foaming temperature of 160°C and water content of 2% were
measured to be the best combination for optimum expansion ratio and half-life. Ruckel et al.
49

(1983) recommended the range of the expansion ratio from 8 to 15 and at least 20 seconds for
half-life. Maccarrone et al. (1995) proved that highly expandable and stable binder has optimum
quality with expansion ratio greater than 15 and half-life greater than 60 seconds by adding
certain surface active additives. Maccarrone et al. (1995) also suggested that the high expansion
of the foamed binder improved the aggregate coating and mix properties. The optimum condition
achieved with 2.6 % of water content and 0.7% of surface active additives. Similarly, Bowering
and Martin (1976) showed that the cohesion and compressive strength of stabilized base mixes
were significantly greater when high expansion (15:1) foamed bitumen was used. Muthen (1998)
suggested the minimum expansion ratio to be 10 and the minimum half-life to be 12s.
Nataatmadja (2001) suggested the water content to be in the range of 2% to 2.5%. On the other
hand, Mohammad et al. (2003) used an optimum water content of 2.75% for PG58-28 binder at
160°C. Marquis et al. (2003) used an optimum water content of 3.0% for PG64-28 binder, which
measured an expansion ratio of 11 and half-life of 8.5 s at 160°C. Kim and Lee (2006)
determined the optimum foaming for the PG52-34 with 1.3 % water content at 170°C under air
pressure of 400 kPa and water pressure of 500 kPa. Leek ad Jameson (2011) recommended the
ER to be between 8 and 20 and the half-life to be minimum 6 seconds for foams used in base
stabilization applications. Jenkins (2000) related the half life of foamed asphalt to the
binder/mixture temperature (i.e., higher temperatures cause shorter half-life).
Foam Index (FI), the area under the ER versus time curve, is another parameter
introduced by Jenkins (2000) and it is a measure of a combination of ER and half-life. The FI
can simply be calculated via a discrete integration as follows:

50

t ( ER = 1)
∑1 / 2 * (E R + E R + 1 ) * (t t + 1 -t t )
FI =
t = 0

[3]

where ERt and ERt+1 are the expansion ratios at times t and t+1 respectively. The original FI
equation, presented by Jenkins (2000), was developed for base stabilization applications where
the binder content in the asphalt mixture is much less (2-3%) than the binder content of WMA
(4-6%) pavements (Namutebi, 2011). As a result, its applicability to WMA needs to be
investigated. Attempts by researchers to apply the foam index in foam characteristics
optimization has been partially succeeded (Sunarjono, 2008). Even though it is not mentioned
and investigated in Jenkins (2000), FI is an indirect indicator of the total surface area of the
bubbles. The bubble size distribution as well as the total surface area of the bubbles can be
computed from the ER versus time data.
As summarized from the literature, there is no common ground between the researchers
to select optimum ER, HL and FI values, and there is no accurate method to measure these
parameters. In addition, the recommendations are based on the performance of stabilized based
not on WMA. Therefore, these parameters are investigated and new parameters are introduced to
better relate the foam quality to asphalt performance in this study.
The practice of the foam height measurements of the foamed asphalt
There are challenges in the conventional method for measuring the foam height
reduction, as a result, accurately calculating the foamed binder parameters. During asphalt foam
testing in base stabilization applications (for ER and HL), measurements were facilitated by use
of a fudicial marker (e.g., ruler) attached to the side of a container. This container is then filled
51

with foamed binder and the height of the binder is recorded over time (Muthen, 1998; Maine
DOT, 2004). Although this method is simple and practical, it is inaccurate due to the foamed
binder opacity and results can be highly dependent on the operator (He and Wong, 2006).
Jenkins (2000) proposed to measure the foam height reduction in 10 second intervals to
calculate the FI. However, typically only two measurements, volume at the beginning of the
foaming for the calculation of ER and the time when the height reduced to half of the initial for
the calculation of HL are taken. As a result, the time dependent ER curve, which is needed for
Foam Index (FI) calculation, cannot be obtained. Namutebi (2011) used a video camera to
capture the images of the foamed asphalt in a container during collapse of the bubbles with the
aid of a dipstick with marks. However, the method was only used to collect data to measure the
ER and HL and the entire ER versus time data was not measured. He and Wong (2006) utilized a
video camera to record the foam generation and dissipation process. Then, they visually
determined the height values at different times from the recorded video and generated ER versus
time graphs.
An automated procedure is needed for repeatable and accurate measurement of reduction
in height of foamed asphalt. A novel and practical method based on image analysis is developed
in this study and explained in CHAPTER 4.
An exercise to observe the effect of ER and HL on morphology of bubbles
In WMA applications, HL and ER may be important parameters that indirectly relate the
workability and coating, respectively. For example, if the half-life is long (i.e., the foam
collapses in a long period of time), overall viscosity of foam will remain relatively low and good
workability can be expected. On the other hand, the expansion ratio is an indicator of total
52

volume of bubbles. It can be claimed that as the bubble volume increases, the surface area will
also increase. However, ER cannot provide the size distribution of the bubbles, which is very
important for surface area calculation. High surface area is desirable because more surfaces will
be available for fine and coarse aggregates for better coating. Table 3 shows a comparison of
surface areas of several bubbles with different sizes, as well as their illustrations in Figure 6. All
3

have the same total bubble volume (i.e., 523.6 mm ), which would lead to same expansion ratio
if they were in a foamed binder. However, since surface area is inversely related to the radius,
when the radius reduced from 5 mm to 0.25 mm, the total surface increased 20 times (even
though volume, i.e., ER is same).
It should be emphasized that the small bubbles will collapse much slower than large
bubbles, which may affect the long term performance of the pavement if encapsulated small
moisture bubbles exist after the pavement construction. Therefore, an optimum size range should
be specified in foamed WMA applications.
Table 3: Comparison of surface areas of bubbles with different sizes
Number of bubbles (n)

1
3

3

2

2

Volume (VB = n*4/3 πR ), mm

Surface area (SB=n*4πR ), mm

Surface area ratio (with respect to
R=5 mm), i.e., SB / SB(R=5mm)

1000

8000

5.00

Radius (R), mm

125
1.00

0.50

0.25

523.6

523.6

523.6

523.6

314.2

1570.8

3141.6

6283.2

1

5

10

20

53

R= 5mm

20

Z (mm)

15

10

5

0
0

20

5

10

10

15
XX (mm)
(mm)

20

0

Y (mm)

R= 1mm
20

Z (mm)

15
10
5
0
20

0
15

10

10
5

0

X (mm)

20
Y (mm)

Figure 6: Illustration of comparison of surface areas of bubbles with different sizes

54

Figure 6 (cont’d)
R= 0.5mm
20

Z (mm)

15
10
5
0
20

15

10
X (mm)

5

0

20

10
15
Y (mm)

5

0

BUBBLE SIZE DISTRIBUTION
Bubble Size Distribution (BSD) is potentially a very important parameter, since it can
directly relate to the ability of the foamed binder to coat the aggregates as well as the
workability. The BSD is also an indicator of the total surface area of the bubbles. It is
hypothesized that as the surface area of the bubbles increase, more interfaces are available for
interaction of binder and aggregates. As a result, better coating can be achieved. It has been
already mentioned that small-size bubbles collapse (or dissipate) much slower than large size
bubbles, which leads to longer half-life. This can potentially aid in workability during placement
and compactability. However, there is a danger that encapsulated moisture bubbles remaining

55

after construction may lead to moisture damage. Therefore, an optimum size range or an
optimum surface area range should be defined.
Theoretically, BSD can be computed from the rate of reduction of the volume of the foam
with time. This can be accomplished by using the Stoke’s law (Lamb, 1932), similar to the
method used in the traditional Hydrometer test, which is commonly used in geotechnical
engineering for measurement of grain size distribution of the fine soils (Das, 2009). One major
difference is that the hydrometer apparatus is not used since it is not practical because of the high
temperature of the foamed binder and binder’s opaque nature (one cannot see the hydrometer). A
method to measure the reduction in the volume of the foamed asphalt is presented in CHAPTER
4. Once the BSD is computed, total surface area of the bubbles can be computed as explained in
the next section.
Figure 7 shows an illustration of rising and collapsing of the bubbles in a cylindrical
container as well as the reduction in overall height of the foamed binder during this process. It is
well known (based on Stoke’s law) that the bubbles with large volume (diameter) will rise to the
surface faster than those with smaller diameter. Stoke’s law for rising bubbles in a fluid can be
expressed in Equation 4 (Lamb 1932):

D=

18 μv

[4]

(ρ f -ρ b )g
2

where v = velocity of bubble (m/s), g = gravitational acceleration (9.81 m/s ), D = diameter of
3

bubble (m), ρf = density of fluid (kg/m3), ρb = density of the bubble (kg/m ), µ= dynamic
viscosity of the fluid (Pa.s=kg/(m.s)). The viscosity of the foamed binder is assumed to be
56

constant through the calculations since the temperature of the container (±3°C) is kept constant
during the measurements. In addition, the viscosity change with time is neglected as the overall
measurements take less than 4 minutes.
Equation 5 shows that if the velocity of a bubble rising in a fluid is known, its diameter is
calculated using the density and viscosity of the fluid (bubble density may be neglected since it is
much lower than the fluid density). In order to calculate the average diameter of bubbles
escaping within a time interval (Δt), average velocity of the bubbles is needed. At any time
interval Δt, the rate of reduction of the height of the foamed fluid is the same as the average
velocity of the bubbles escaped within that time interval. Average velocity can be obtained from
the reduction in height of the foamed fluid in the container as follows:

c) time = t+Δt

d) time =inf

t+Δt

b) time = t

h

f

h

h

h

0

t

a) time = 0

0

0

2

t

V =h pd /4
0

2

t

2

V =h pd /4
0 f

VB = pd /4(h -h )

t

2

V
t f

VB = pd /4(h /h )

t+Δt

VB

t+Δt

=h

t+Δt

2

f

pd /4
2

t+Δt f

= pd /4(h

f

/h ) VB = 0

Figure 7: Illustration of the reduction of foamed binder volume with time.

57

f

2

V =h pd /4

v

t

=

h

t

-h

t + Δt

[5]

Δt

t

t

where v =average velocity of the bubbles escaped at time t, Δt = time interval, h and h

t+Δt

are

the height of the foamed fluid at t and t+Δt, respectively.
For the foamed binder in
Figure 7, percentage of the bubbles escaped (PBE) from the binder (i.e., bubbles that
have risen to the surface) at an intermediate time (Δt) interval can be obtained as follows:

PBE

V

t

=

t
t + Δt
-V
B
B
× 100
0
V
B

[6]

t

0
where PBE = percentage of bubbles escaped at time interval Δt, V = (initial) volume of the
B

bubbles at t=0, V t and V t + Δ t = volume of the bubbles at t and t+Δt, respectively. Equations of
B

V

B

t
t
and V t + Δ t are shown in Figure 7. It is noted that PBE is analogous to percent retained in
B
B
t

each sieve, which is calculated during sieve analysis of aggregates. As a result, PBE can be used
to calculate the percent passing (PP) as follows:

PP

t

= 100

t

where PP is percent passing at time t.

58

t = t
i
t
∑ PBE i
i= 0

[7]

SURFACE AREA INDEX
The number of the bubbles and the total surface area of the bubbles can be computed,
from the bubble size distribution, as follows:
t
B
N =
B
single
V
B
V

S

[8]

single
t
= N S
B
B B

[9]

where NB = number of bubbles, V t = total volume of the bubbles escaped at time t, V
B

sin gle
B

=volume of a single bubble = 4 3 π R 3 , S t = total surface area of the bubbles escaped at time t,
B

S

sin gle
= surface area of a single bubble = 4 π R 2 , R= average radius of bubbles escaped at
B

time t. Combining equations 8 and 9 and plugging the values of V

S

sin gle
3
= 4 3 πR ,
B

sin gle
2
= 4 π R and R=D/2 reveals:
B

t
S =
B

6V
D

t
B
t

[10]

t
where S t = surface area and D = average diameter of bubbles escaped from the foam at time t.
B

Total bubble surface area of all the bubbles can be calculated by adding the S t

B

values at

different times:

BSA =

∞ t
∑ S i
B
i= 0

59

[11]

2

where BSA = total surface area of all bubbles (in mm ) in the foam at time = 0. A dimensionless
parameter can be obtained by dividing the BSA by the surface area of the fluid (i.e., the asphalt
binder) in the container as follows:
BSA

SAI =

πd (h

f

+

d
2

[12]
)

f

where SAI =surface area index (dimensionless), d=diameter of the container and h = final height
of the binder after all the bubbles dissipate (in
Figure 7). The SAI can be a useful dimensionless parameter for quantifying total surface
generated by the foaming action and will influence the effectiveness of aggregate coating in
WMA applications. It is also anticipated to relate to workability since small bubbles will lead to
large SAI. Small bubbles typically do not collapse as fast as the large bubbles therefore the foam
viscosity will stay low longer. This potentially leads to improved workability during placement
and compaction.

60

CHAPTER 4

DEVELOPMENT OF THE ASPHALT FOAM COLLAPSE TEST
(AFCT) AND VERIFICATION USING X-RAY MICROTOMOGRAPHY

An accurate and repeatable testing method is needed for the measurement of the rate of
reduction in the height of foamed asphalt in order to calculate the foam binder quality parameters
precisely. As part of this study, an automated test device (called Asphalt Foam Collapse Test
(AFCT) was developed to measure the reduction in the height of the foam binder over time. As
discussed in CHAPTER 3, once the reduction in height of the asphalt foam with time is
measured, the foamed binder’s quality parameters (i.e., the expansion ratio, foam index, half-life,
bubble size distribution and surface area index) can be calculated. In addition, the absence of an
accurate and repeatable testing method in measuring the foam quality in the current practice can
be filled with AFCT, which is also practical and affordable test for the practitioners.
AFCT SETUP
The major components of the AFCT setup are: (i) a camera, (ii) a light source (preferably
a light table),(iii) two steel rods, (iv) two aluminum custom made pulleys with precision
bearings, (v) fishing line and weights, (vi) a plastic bobber, (vii) a steel floating ball and (viii) a
stopper. The conceptual drawing of the main components of the setup is illustrated in Figure 8
(except the light table and camera). The actual picture of the overall setup, including the camera
and light source (which are not shown in Figure 8) is given in Figure 9. Each component of the
setup was determined after various trials with different designs.
61

150

300
150

a)

b)

300

400
150

150

150
400
400



400

All dimensions are in
mm.

Figure 8: A sketch of the main component of the AFCT setup a) side-view b) front-view
Backlight
Black Painted
Ball (B1)

Laboratory
Foamer

iPhone
Camera

Floating
Ball (B2)

Counter Weight
Dhfoam
Δhfoam

t=0

t=0s

t=20s

t=20s

Figure 9: Picture of the AFCT test setup
62

The costly part of the AFCT setup was the digital camera. However, the trials indicated
that the test could be performed both using non-industrial and industrial cameras to capture
images. In this study, two different cameras (an iPhone camera and a high speed industrial
camera) were used and compared with respect to each other. The details of the image capturing
and analysis were discussed in the later sections. The back light source was a fluorescent light
table.
The other components such as rods, bearings and stopper were all mounted or attached to
the L-shaped white painted wooden block as shown in the isometric view in Figure 8. Initially,
the rods (Rod 1 and Rod 2) were bolted to the wooden block. Then, custom-made aluminum
pulleys with precision bearings were attached to the rods. The precision bearings provided close
to frictionless rotation of the pulleys during the experiment. In addition, a lubricant (i.e. WD40)
was also sprayed to minimize the friction. Finally, a stopper was mounted on Rod 2 in Figure 8,
and used as a triggering mechanism.
The rest of the components of the test setup were glued or tied to the fishing line. The
fishing line was the most convenient, abrasion resistant and almost frictionless material that
provided free rotation of the pulleys during the test. A black painted plastic bobber (B1) with 2
cm diameter was initially drilled with a hand driller (2 small holes slightly larger than the
diameter of the fishing line - 180° apart). Then, B1 was glued to the middle of the fishing line
using superglue. The floating ball (B2) was attached to one side of the fishing line. B2 is a
temperature-resistant steel floating ball that has a diameter of 2” (50.8mm). Before tieing the
fishing weights to the other end of the fishing line, B2 was dipped into hot binder and fully
coated with binder. Thus, the change in the weight of B2 during the test was minimized. Then,
63

the weight of the binder coated bobber (B2) was balanced with the weights as shown in Figure 8.
Finally, an extra weight of 1 -2 grams was attached onto the B2 to ensure the free flow when the
stopper was released at the trigger of the experiment. The test starts by releasing the stopper and
starting the acquisition of the images by the camera. The cost analysis of AFCT setup is given in
Table 4, where the overall cost was approximately $389. The camera price was estimated
based on the cost of a smart phone.
Table 4: Overall cost estimate of AFCT setup
AFCT Setup Components:

Unit

Unit Price

Total

Camera

1 ea

$200

$200

8x Optical Zoom Lens

1 ea

$20

$20

Light Table

1 ea

$52

$52

Aluminum Rods

2 ea

$5

$10

Pully + Precision Bearing

2 ea

$21

$42

Stopper

1 ea

$13

$13

Fishing Line

1 ea

$10

$10

Fishing Weight

2 box

$5

$10

Bobber (B1)

1 ea

$0.5

$0.5

Floating Ball (B2)

1 ea

$13

$13

2000 ml beaker

1 ea

$18

$18
Total $389

AFCT PROCEDURE
The AFCT test is designed to be convenient for both laboratory and field practices.
Initially, the binder is foamed into a beaker, which is heated to the binder temperature in a
conventional oven or in a heating mantle. The size of the binder sampling dish is selected to be

64

2000 ml glass beaker for 200±20 gr. of foamed binder after various trials (i.e., quart aluminum
can, 1000 ml beaker etc). However, the users can change the size of the beaker as well as the
binder amount depending on the properties of the foam binder, if the binder expends excessively
and overflows from the beaker. It is crucial to keep the size of the beaker and the amount of the
binder same while comparing the quality of different foamed binders. As summarized in
CHAPTER 3, the foam quality parameters depend on the amount of the binder and the size of the
beaker (Brennen et al. 1983, Ruckel et al. 1983), except the SAI introduced in this study.
The beaker filled with the foamed binder needs to be rapidly placed under the floating
ball (B2), and B2 is leveled to the surface of the foam, then quickly, the stopper that fixes the
fishing line is released. As foam collapses, B2 goes down with the surface of the foamed binder,
but does not sink into the foamed binder. This is a crucial step during test and can be ensured by
balancing the two sides of the fishing line. If B2 sinks into the foamed binder, the test is
discarded and repeated. Meanwhile, a camera simultaneously captures the movement of the
black-painted ball (B1) as the foam collapses (Figure 9). Since B1 and B2 are on the same line,
they move as the same amount. In Figure 9, sub-images with white background and black circle
(which is B1) are the images captured by the camera and shows the movement of bobber B1
from t=0 sec to t=20 sec towards left (as B2 goes down). The details of the image analysis is
explained in the following section
AFCT IMAGE ANALYSIS
Image analysis is used to calculate the movement of B1 with time, which corresponds to
the change in the foam height. Since both B1 and B2 are attached to the same fishing line, they

65

move the same distance within the equivalent time intervals. However, B1 moves horizontally
while B2 moves vertically.
In order to record the video images, as it was stated previously, a non-industrial (iPhone)
camera and a high speed industrial camera were used (Figure 9). The non-industrial camera
allows the system to be easily transported to the field as well as being less costly. Initially, the
accuracy of the iPhone camera was validated against a high-speed industrial camera. The highspeed industrial camera captures 400 frames per second (fps), whereas the iPhone camera
captures 30 fps. In order to validate the frame rate, a timer was placed at the top of the setup and
included in the video images. It was observed that the non-industrial (iPhone) camera has
adequate accuracy to measure the height reduction of foamed binders. In addition, an 8x Optical
Zoom Lens was mounted to the iPhone camera to capture the foam collapse from approximately
2m distance. This facilitated the overall operation by providing sufficient space to work with the
foaming device and AFCT setup. The frame rate of the camera for the AFCT test can be as low
as 1 fps, which provides sufficient data interval.
Once a video is captured, first, the frame rate is verified using non-commercial video
editing software (e.g., VirtualDub). Then, a sequence of images was extracted from the video file
every 1 second. The centroid of the black painted ball (B1) in each image was computed using an
algorithm developed in MATLAB®. In this algorithm, first, each image is converted to a binary
(black/white) image through a thresholding operation such that only the bobber is black and the
rest of the image (background) is white (see sub-images in Figure 9). It is noted that, since the
fishing line is relatively transparent, with the aid of the background lighting (light table in Figure
9), it disappears when the original image is converted to binary image. As a result, in each
66

image, only a black circle (i.e., B1) is visible. The center coordinate of the black circle is
determined using a morphological labeling operation (Kutay et al. 2010, 2011). This procedure is
repeated for all consecutive images. The change in the x-coordinate of the center of B1 in
consecutive images corresponds to the change in the height of the foam (B2). The displacement
of B2 with recorded time is equal to the reduction of the foam height with time. Thus, the
foamed binder parameters explained in CHAPTER 3 can be easily and precisely calculated with
these data.
AFCT TEST RESULTS
After developing the AFCT procedure, the effects of the injected air pressure and water
content on the binder quality indicators were investigated. A virgin (non-modified) binder
(PG58-28) was utilized in this study. The foaming temperature was 155 °C. The viscosity of
non-foamed binder at this temperature is approximately 300 mPa.s. The laboratory Foamer
utilized in this study had an air injection capacity varying from 0 to 30 psi. However, the trials
indicated that the repeatability of the foaming became poor after 20 psi. It was observed that if a
feedback system is mounted just before the foaming nozzle to better control the injection
process, it can increase the foamer’s repeatability. Due to this deficiency, the maximum injected
pressure in this study was limited to 20 psi. It should also be noted that the pressure levels of
laboratory foamers are significantly lower than the field foamers.
In this study, the water content range was selected to be in the range of 1% to 5%, which
simulates a wide range of the current field applications. The accuracy of the injected water
content was approximately ± 0.5-1% in the laboratory foamer used in this study. Since the
overall water amount is relatively low, even one drop of water may affect the quality of the
67

foamed binder. For example, 1% water content, 2 grams of water is injected to 200 grams of
binder.
Effect of Injected Water Content on the Foam Properties
The foam quality parameters of foams were initially analyzed for the binder prepared
with constant injected air pressure (15 psi) and three different water contents (1%, 2%, and 4%
by weight of the binder). Figure 10 and Figure 11 show the foamed binder characteristics of
these water contents. Based on these figures, the following conclusions can be drawn:


Expansion ratio of the foamed binders increased with the increases in the water
content.



Half-life of the foamed binders decreased with the increase in the water content.
The half-life of the foamed binder prepared with 1% water content was
approximately twice of the half-life of the one with 4% water content.



Foam Index, which is the area under the expansion ratio versus time, decreased
with the increase in the water content.



As the injected water content in the foamed binder increased, the maximum
diameter of the bubbles increased. This leads to the short half-life and high
expansion ratio.



The diameter of foamed binder at 50% of passing (D50), which was calculated
from bubble size distribution, increased with the increase in the water content.



Surface area index, which is dimensionless parameter, is an indicator of the
surface area of the bubbles. The SAI decreased as the water content increases.

68



Bubble size distribution of the foamed binders plotted in Figure 11 clearly shows
the effect of water content on the internal microstructure of the foamed binders.
As the water content increases the bubble size distribution becomes coarser. On
the contrary, the low water content resulted finer bubble gradation.

2.20

12
a)

b)
10
Half Life (HL)

Expansion Ratio (ER)

2.40
2.00
1.80
1.60
1.40

8
6
4

1.20

2

1.00

0
1% - 15psi 2% - 15psi 4% - 15psi

200
180
160
140
120
100
80
60
40
20
0

6.00

c)

d)
5.00
4.00
D max

Foam Index (FI)

1% - 15psi 2% - 15psi 4% - 15psi

3.00
2.00
1.00
0.00

1% - 15psi 2% - 15psi 4% - 15psi

1% - 15psi 2% - 15psi 4% - 15psi

Figure 10: Foam quality analysis of the binders prepared with constant air pressure and
different water contents

69

Figure 10 (cont’d)
250

1.40

f)

e)

1.20

200
SAI

D 50

1.00
0.80
0.60

150
100

0.40

50

0.20

0

0.00
1% - 15psi 2% - 15psi 4% - 15psi

1% - 15psi 2% - 15psi 4% - 15psi

100

Percent Passing (%)

90
80
70
60
50
1% - 15psi -1
1% - 15psi -2
2% - 15psi -1
2% - 15psi -2
4% - 15psi -1

40
30

20
10
0
0

1

2
3
Sieve Size (mm)

4

5

Figure 11: Bubble size distribution of the binders prepared with constant air pressure and
different water contents
Effect of Injected Air pressure on the Foam Properties
In this section, foamed binder quality indicators were studied for constant water content
(1% by weight of the binder) and three different injected air pressures (10psi, 15psi, and 20psi).

70

Figure 12 and Figure 13 show the variation of ER, FI, HL, Dmax, D50 and SAI at different air
pressures. Based on these figures, the following conclusions can be drawn:


The ER increased with increasing the air pressure.



Half-life of the binders decreased exponentially with the increase in the injected
air pressure. As shown in Figure 12b, the half-life of the foamed binder prepared
under 10 psi air pressure was approximately 6 times longer than the one prepared
under 20 psi air pressure.



Foam Index of the binders linearly decreased with the increase in the injected air
pressure.



Dmax and D50, which were calculated from the bubble size distribution curves
given in Figure 13, were clear indicators of how the air pressure influenced the
bubble sizes. It was revealed that the bubble size increased with the injected air
pressure.



As plotted in Figure 13, the variability in the bubble size distribution of the
replicates is relatively high. However, it is clear that the gradations of the bubbles
for different air pressures are significantly different. This bubble size distribution
can affect the workability and coating, as well as the performance of the
pavements.

71

Expansion Ratio (ER)

2.50
a)
Half Life (HL)

2.00
1.50
1.00
0.50
0.00

80
70
60
50
40
30
20
10
0

1% - 10psi 1% - 15psi 1% - 20psi
350

1% - 10psi 1% - 15psi 1% - 20psi
2.50

c)

300

b)

d)

2.00

200

D max

Foam Index (FI)

250
150

1.50
1.00

100
0.50

50
0

0.00
1% - 10psi 1% - 15psi 1% - 20psi

1.40
1.20

1% - 10psi 1% - 15psi 1% - 20psi
200

e)

f)
160
120

0.80
SAI

D50

1.00
0.60

80

0.40
40

0.20

0

0.00
1% - 10psi 1% - 15psi 1% - 20psi

1% - 10psi 1% - 15psi 1% - 20psi

Figure 12: Foam quality analysis of the binders prepared with constant water content and
different air pressures

72

Percent Passing (%)

100
90
80
70
60
50
40
30
20
10
0

1% - 10psi -1
1% - 10psi -2
1% - 15psi -1
1% - 15psi -2
1% - 20psi -1
1% - 20psi -2
0

0.5

1
1.5
Sieve Size (mm)

2

2.5

Figure 13: Bubble Size Distribution of the binders prepared with constant water content
and different air pressures
Analysis of Foamed Binder Properties under Different Injected Air Pressure and Water
Content Combinations
The effects of water content and air pressure were investigated individually in the
previous sections. In this section, results of three combinations (1% - 10psi, 3% - 15psi, and 5% 20 psi) presented. These trials were called as “1

st

Trial” in the Figure 14 and Figure 15. As

shown, there are clear trends, where ER, D50 and Dmax increase with water content and air
pressure. In order to increase the number of combinations of water content and air pressure, tests
were planned to be repeated at 2% - 12.5psi and 4% - 17.5psi. However, because of known
variability of the laboratory foamer, the entire set of combinations was re-tested on the same day.
This set is called as “2

nd

Trial” in the Figure 14 and Figure 15. Based on these figures, the

following conclusions can be drawn:

73



st

Although the magnitudes of the parameters are different when 1 and 2

nd

trials

are compared, overall trends are the same.


ER increased as the injected water content and air pressure increased (Figure
14a).



As shown in Figure 14b, the half-life decreases with increasing water content and
air pressure. The half-life of the foamed binder prepared with injecting 1% water
content and 10psi air pressure varied increased 15 seconds to 76 seconds in
st

between the 1 and 2

nd

trials, which illustrates the variability of the laboratory

foamer.


It can be clearly be observed from Figure 14b that the foam index decreases with
the increase of the injected water content and air pressure.



Bubble size distributions of the foamed binders for both 1

st

and 2

nd

trials are

shown in Figure 15a and Figure 15b. For each trial, the size distribution for each
combination significantly diverged from each other.


Dmax and D50 increased with the increases of the injected air pressure and water
content.



The SAI, which was calculated from the bubble size distribution, decreased with
increasing water content and air pressure combinations.

The AFCT analysis showed that the foam quality indicators significantly depend on the
water content and air pressure. However, it is still unknown how these foam quality indicators

74

relate to aggregate coating, as well as the mixture performance. Therefore, in CHAPTER 5, the
relation between the foam quality indicators and mixture performance are presented.

Expansion Ratio (ER)

3.50
3.00

a)

ER (1st Trial)
Linear (ER (1st Trial))

ER (2nd Trial)
Linear (ER (2nd Trial))

2.50
2.00
1.50
1.00
0.50
0.00
1% - 10psi

2% - 12.5psi

3% - 15psi

4% - 17.5psi

5% - 20psi

120
b)

HL (1st Trial)
Expon. (HL (1st Trial))

1% - 10psi

2% - 12.5psi

Half Life (HL)

100
80

HL (2nd Trial)
Expon. (HL (2nd Trial))

60
40
20

Foam Index (FI)

0
450
400
350
300
250
200
150
100
50
0

3% - 15psi

5% - 20psi

FI (1st Trial)
FI (2nd Trial)
Linear (FI (1st Trial))
Linear (FI (2nd Trial))

c)

1% - 10psi

4% - 17.5psi

2% - 12.5psi

3% - 15psi

4% - 17.5psi

5% - 20psi

Figure 14: Foam quality analysis of the binders prepared with different water content and
air pressure combinations
75

Figure 14 (cont’d)
1.2

d)

1

D50 (1st Trail)
Linear (D50 (1st Trail))

D50 (2nd Trail)
Linear (D50 (2nd Trail))

D50

0.8
0.6
0.4
0.2
0
1% - 10psi
2.50
e)
2.00

2% - 12.5psi

3% - 15psi

4% - 17.5psi

5% - 20psi

Dmax (2nd Trail)
Dmax (1st Trail)
Linear (Dmax (2nd Trail))
Linear (Dmax (1st Trail))

D max

1.50
1.00
0.50
0.00
1% - 10psi

2% - 12.5psi

3% - 15psi

Surface Area Index (SAI)

700
600

4% - 17.5psi

5% - 20psi

SAI (1st Trail)
SAI (2nd Trail)
Linear (SAI (1st Trail))
Linear (SAI (2nd Trail))

f)

500
400
300
200
100
0
1% - 10psi

2% - 12.5psi

3% - 15psi
76

4% - 17.5psi

5% - 20psi

100

st

a) 1 Trial

90

Percent Passing (%)

80
70
5% - 20psi -1

5% - 20psi -2

5% - 20psi -3

5% - 20psi -4

5% - 20psi -5

3% - 15psi -1

40

3% - 15psi -2

3% - 15psi -3

30

3% - 15psi -4

1% - 10psi -1

20

1% - 10psi -2

1% - 10psi -3

1% - 10psi -4

1% - 10psi -5

60
50

10
0
0.00

1% - 10psi -6
0.50

1.00
1.50
Sieve Size (mm)

2.00

2.50

100

nd

b) 2

90

Trial

Percent Passing (%)

80
70
5% - 20psi -1
5% - 20psi -3
4% - 17.5psi-2
4% - 17.5psi-4
3% - 15psi-1
3% - 15psi-3
2% - 12.5psi-2
2% - 12.5psi-4
1% - 10psi -2

60
50
40
30
20
10
0
0.00

0.50

5% - 20psi -2
4% - 17.5psi-1
4% - 17.5psi-3
4% - 17.5psi-5
3% - 15psi-2
2% - 12.5psi-1
2% - 12.5psi-3
1% - 10psi -1
1% - 10psi -3

1.00
1.50
2.00
2.50
Sieve Size (mm)
Figure 15: Bubble Size Distribution of the binders prepared with different water content
st
nd
and air pressure combinations a) 1 Trial, b) 2 Trial
77

VERIFICATION OF AFCT USING X-RAY MICROTOMOGRAPHY IMAGING
In order to verify the AFCT procedure to compute the bubble size distribution, 3D
images of foamed binders were acquired using X-Ray microtomography technique. Then, 3D
image-based bubble size distribution was compared against bubble size distribution computed
from the AFCT. The verification process is described below.
X-Ray Microtomography Sample Preparation Procedure
Careful sampling of the foamed binder was very crucial in order not to disturb the
specimens during the process. Initially, the binder was foamed into a beaker at approximately
same temperature with the binder, as shown in Figure 16a. Then, the binder was poured into
13x75 mm (11 mm inner diameter) polypropylene tubes, which were heat resistant (melting
point ~170°C).

After pouring, each specimen was instantly frozen using liquid nitrogen, as

shown in Figure 16b. The frozen PP tubes are shown in Figure 16c. The foamed binders were
st

th

th

sampled at the 1 , 15 , and 30 minutes. In between the sampling intervals, the beaker with the
binder was kept in the conventional oven at 155°C.

Bubbles

Asphalt binder specimen in
PP tube
13 mm
Liquid
Nitrogen

a)

75 mm
c)

b)

Figure 16: Picture of (a) the foamed asphalt where bubbles are visible at the surface, (b)
illustration of freezing of asphalt binder using liquid nitrogen, c) PP tubes w and w/o
foamed binder

78

The purpose of this study was: (1) to verify the bubble size distribution computed by the
AFCT as soon as the foam is generated, and (2) to monitor the change in microstructure at
different times using 3D XRM images. It was observed that the instant freezing procedure did
not damage the binder samples. This observation was further validated when the 3D XRM
images were analyzed, where there were no visible cracks within the specimens. For the XRM
analysis, the frozen tubes were transported to Argonne National Laboratory (ANL). 3D internal
images were generated using the synchrotron based X-ray Microtomography device (5-BM-C
beam line at Advanced Photon Source (APS)).
3D Imaging using Synchrotron-Based X-ray Microtomography
The 3D image acquisition of the specimens was done at the 5-BM-C Microtomography
beam line at the Advanced Photon Source (APS) facility in ANL. The illustration of the
equipment is given in Figure 17. This XRM system uses a cryo-cooled CCD system and optics,
which permit a range of spatial resolutions from 3 microns to about 100 microns (depending on
the sample size). In this research, 20keV parallel beam was utilized, which provided a volume
scan of 13x13 mm area, 5.5 mm tall cylinders. This size was the maximum scan range for the
camera and optical system (i.e., the X-ray detector). Parallel beam herein is defined as a beam
where photon particles travels through linear accelerator with 20keV energy. The final image
size was 1299 by 1299 by 550 pixels. This corresponded to 12.9 mm/ 1299 pixels = 0.01
mm/pixel (10 micron) image resolution. The PP tubes (10.89 mm inner diameter and 75 mm tall)
were convenient for this study since the maximum sizes of the bubbles were much smaller (~2.5
mm calculated from AFCT analysis).

79

X-Ray
Beampipe

PP
tube

Parallel
X-Rays

CCD
Camera

Figure 17: Picture and illustration of Synchrotron-based X-ray Microtomography setup
used in this research. This setup is at Advanced Photon Source (APS) located in Argonne
National Lab (ANL).
X-Ray CT Microtomography Image Analysis
The scanned images are reconstructed with the inbuilt software of the X-Ray
Microtomography device and saved in “*.img” format. An algorithm was developed in
MATLAB® to convert these images to “*.tif” format as given in Figure 18a. The sequence of
images was then imported to a non-commercial video editing software (ImageJ), in order to crop
the binder medium to improve the computational efficiency during analysis (each image was
initially about 1.6 MB after cropping it was reduced to 760KB). Before beginning the 3D
analysis, it was observed that most images has ring artifact problem, which would affect the
overall analysis. Therefore, a ring artifact removal algorithm was developed in this study.
Ring Artifact Removal
In X-Ray images, the ring artifacts appear to be a number of dark concentric rings or
semi-rings on the scanned surface as shown in Figure 18b. These artifacts may be formed due to
slight movement of the tubes during the scanning or due to the temperature change in the
samples. The samples were kept in a freezer and taken out just before scanning. Each scan took
80

about three hours, which resulted huge temperature variability in between the beginning and end
of the scan. However, this was essential because the binder (PG58-28) used in this study was
very soft and bubbles may be disturbed even before scanning.
The ring artifact removal algorithms available in literature are typically applicable to the
sinograms (a visual representation of the raw data obtained in a computed axial tomography
scan) of the raw XRM data (Prasad et al, 2011; Munch et al., 2009). However, these data was not
accessible in this study. Since most of the ring artifacts in the scanned images were semicircular, initially, the images were transformed from Cartesian to Polar, as shown in Figure 18c.
The semi-circular artifacts became dark lines in the images. Then, a 2D median filter of size 5
pixels by 30 pixels was applied to these images. Median filter was selected since it is a common
noise reduction filter, which preserves the edge structure while smoothing the non-uniform
regions. As shown in Figure 18d, the most of the rings were eliminated with this filter. Then,
Polar images were converted back to Cartesian coordinates (Figure 18e). When it was zoomed in
the bubbles, as shown in the subimage of Figure 18e, it was observed that some noise occurred in
the bubbles. Therefore, a second median filter (4x4) was applied to the image, as shown in
Figure 18f. When it was zoomed in to the bubbles as illustrated in the subimage, it was clear that
the noise was reduced in the bubbles.

Ɵ

81

Ring
Artifact

x

y
a) Original Image (1299x1299)
a) Original Image (1299x1299)

r

Ɵ

b) b) CroppedImage (878x878)
Cropped Image (878x878)
r

Ɵ

c) Cartesianto Polar coordinate
c) Cartesian to Polar Coordinate

d) Median Filtered Image (5x30)
d) Median Filtered Image (5x30)

e) Polar to Cartesian Coordianate
e) Polar to Cartesian Coordinate

f) Median Filtered Image (4x4)
f) Median Filtered Image (4x4)

Figure 18: Ring artifact removal algorithm

82

3D XRM Analysis Procedure
The analysis of 3D XRM images was relatively straightforward since the bubbles had
distinctly darker color in the images as compared to the binder medium, as shown in Figure 18.
In addition, they were occupying separate locations within the binder (i.e., they were not
touching each other). However, the image size was to concern due to the computational
limitations. Initially, the images (878x878) were resized to half (439x439). Then, the resized
images were converted to binary (black/white) images using a thresholding algorithm. Then, a
connected components algorithm was utilized to label individual bubbles. After each bubble was
labeled, the volumes and equivalent diameters (i.e., the diameter of a sphere of equivalent
volume) of the bubbles were computed. In order to quantify the change in the structure of
bubbles, their volume and size distribution were computed using algorithms developed in
Matlab®.
3D temporal view of bubbles in the foamed binder prepared with 1% water content and
st

10psi air pressure at the 1 , 15

th

and 30

th

minutes was given in Figure 19. As shown in the

subimages, there were distorted bubbles at the border of the binder and the PP tube. Therefore,
before further analysis, each bubble was individually visualized and grouped as accepted and
rejected subsets. The accepted bubbles were in sphere or in ellipsoid shape, as shown in Figure
19.

83

6
Z (mm)

Z (mm)

6
4
2
0

4
2
0

0

0
5

10

X (mm)

10 0

10

5

5
Y (mm)

X (mm)

a) 1% -10psi -1min

10

0

5
Y (mm)

b) 1% -10psi -15min

Accepted
Rejected

Z (mm)

6
4
2
0
0
5

10

Rejected

5
X (mm) 10 0
Y (mm)
c) 1% -10psi -30min

Figure 19: 3D temporal view of bubbles in the foamed binder prepared with 1% water
content and 10psi air pressure a)1min, b)15min, c)30min

3D XRM Analysis and Comparision with Respect to AFCT
st

For each foamed binder, two replicates (PP tubes) were prepared at the 1 , 15

th

and 30

th

minutes. It took about 1 minute to fill two replicates at a time. Therefore, some larger bubbles
may collapse during the sampling process. In order to limit the error, two foamed binders were
84

selected for XRM scanning and analyzing: i) 1% water content and 10 psi air void (1%-10psi),
ii) 2% water content and 12.5 psi air void(2%-12.5psi), since it had been already revealed that
both these combinations had the longest half-life, low average bubble size, and low expansion
ratio.
The bubble size comparison of the samples were performed with respect to the equivalent
diameters, as given in Figure 20 and Figure 21. Initially, the average bubble size for each
st

replicate at the 1 , 15

th

and 30

th

minutes were calculated and plotted in Figure 20a and Figure

21a. It was clearly observed that the bubbles collapse with time. In addition, it was also verifed
that the rate of foam collapse was relatively high for the foamed binder prepared with high water
content and air pressure (2%-12.5psi) than the one with relatievly low water content and air
pressure (1%-10psi). In other words, the foam dissipates more rapidly from the binder with high
expansion ratio and short half-life. As plotted in Figure 20b and Figure 21b, the maximum
st

bubble sizes also decreased from the 1 min to 30

th

minute. However, it can be misleading to

compare the maximum bubble size in between 1%-10psi and 2% -12.5psi by analyzing a single
bubble. Similarly, the number of bubbles decreased as the time passes, as given in Figure 20c
and Figure 21c. Since the bubbles were not distributed homegenous within the binder, the
sampling may affect the number of bubbles in each replicate.As it was hypothesized before, the
bubble size distribution of the samples were plotted in Figure 20d and Figure 21d, in which the
gradation became finer as time passes, since larger bubbles dissipates quicker than the smaller
bubbles. This observation validates the use of Stoke’s Law in analyzing the AFCT height
reduction data.
85

0.6

a)

Max. Bubble Size
(mm)

Avg. Bubble Size (mm)

0.23
0.22
0.21
0.20
0.19
0.18

b)

0.5
0.4
0.3
0.2
0.1
0

0.17
15min
80
70
60
50
40
30
20
10
0

1min

30min

15min

30min

c)

# of Bubbles

1min

15min

1min

30min

Percent Passing (%)

100
d)

80
60

1% - 10psi -1min

40

1% - 10psi -15min

20

Finer Gradation
(Smaller Bubbles)

1% - 10psi -30min

0
0

0.1

0.2

0.3
0.4
0.5
0.6
0.7
Sieve Size (mm)
Figure 20: Bubble size comparison of X-Ray Microtomography images foamed binder
prepared with 1% water content and 10psi air pressure a) Average, b)Maximum, c)
Minimum, d) Number of bubbles, e) Bubble Size Distribution

86

0.6

a)

Max. Bubble Size (mm)

Avg. Bubble Size (mm)

0.21
0.20
0.19
0.18
0.17

0.16
0.15
1min

15min

b)

0.5
0.4
0.3
0.2
0.1

30min

0
1min

15min

30min

50
c)
# of Bubbles

1min
45
40
35
30

1min

Percent Passing (%)

100

15min

30min

1min

d)
80

2% -12.5psi -30min

60

2% -12.5psi -15min

40

Finer Gradation
(Smaller Bubbles)

20

2% -12.5psi -1min

0
0

0.2

0.4
0.6
Sieve Size (mm)

0.8

1

Figure 21: Bubble size comparison of X-Ray Microtomography images foamed binder
prepared with 2% water content and 12.5psi air pressure a) Average, b)Maximum, c)
Minimum, d) Number of bubbles, e) Bubble Size Distribution
87

The bubble size distributions calculated from the XRM images at the 1st minute and the
AFCT were plotted for the 1%-10psi and 2%-12.5psi in Figure 22 and Figure 23. The bubble
size distributions of three replicates from AFCT data and two replicates from XRM data were
given in Figure 22. In addition, the average D50 were calculated and given in Table 5. Although
both the analysis had relatively high variability, the gradations from both analyzes matched very
well, except the very fine side of the gradation (smaller size). This variability was also expected
since the bubble size distribution calculated from AFCT analyses takes into the change in first 5
minutes. On the other hand, the XRM samples were taken in the 1 minute and the sample size is
comparably small. Also, after 5 minutes during AFCT test, the rate of reduction becomes very
small and cannot be measured accurately with the camera used in this research. Perhaps a higher
resolution camera and longer measurement of heights would have lead to better match.
Considering all the variability, it can be claimed that the Bubble Size Distribution from AFCT
matches very well with the Bubble Size Distribution from XRM images.
The bubble size distributions of four replicates from AFCT data and two replicates from
XRM data for 2%-12.5psi were given in Figure 23. The gradations from both analyzes were
varied in the same range although the difference was relatively high in the coarse gradations.
This could be because of collapse of larger bubbles during sampling of XRM. It may be occurred
because of the relatively rapid collapse of the large bubbles while sampling for XRM. It was also
clearly observed from the calculated D50 from AFCT and XRM, given in Table 6.

88

100
90
Percent Passing (%)

80
70
60

1% - 10psi -1 (AFCT)

50

1% - 10psi -2 (AFCT)

40

1% - 10psi -3 (AFCT)

30

1-10-1m-1 (XRM)

20

1-10-1m-2 (XRM)

10
0
0

0.2

0.4
0.6
Sieve Size (mm)

0.8

1

Figure 22: Comparison of Bubble Size Distribution from X-Ray Microtomography and
AFCT.
Table 5: Comparison of Average Bubble Size from X-Ray Microtomography and AFCT
imaging
Bubble Size(mm)

Avg. Bubble Size
(mm)

StDev. Bubble Size
(mm)

1% - 10psi -1 (AFCT)

0.198

1% - 10psi -2 (AFCT)

0.279

1% - 10psi -3 (AFCT)

0.239

0.239

0.040

1% - 10psi -1 (XRM)
1% - 10psi -2 (XRM)

0.311
0.404

0.357

0.066

89

100

Percent Passing (%)

90
80
70
60

2% - 12.5psi-1 (AFCT)
2% - 12.5psi-2 (AFCT)
2% - 12.5psi-3 (AFCT)
2% - 12.5psi-4 (AFCT)
2% -12.5psi -1min -1 (XRM)
2% -12.5psi -1min -2 (XRM)

50
40
30
20

10
0
0

0.2

0.4

0.6
0.8
Sieve Size (mm)

1

1.2

1.4

Figure 23: Comparison of Bubble Size Distribution form X-Ray Microtomography images
and AFCT.
Table 6: Comparision of Average Bubble Size from X-Ray Microtomography and AFCT.
Avg. Bubble
Bubble Size (mm) Size (mm)

StDev. Bubble Size
(mm)

2% - 12.5psi -1 (AFCT)

0.340

2% - 12.5psi -2 (AFCT)

0.363

2% - 12.5psi -3 (AFCT)

0.459

2% - 12.5psi -4 (AFCT)

0.412

0.394

0.053

2% - 12.5psi -1 (XRM)
2% - 12.5psi -2 (XRM)

0.328
0.292

0.310

0. 026

90

CHAPTER 5

LABORATORY PERFORMANCE TESTS ON WMA MIXTURES
PREPARED WITH FOAMED BINDER

The current WMA pavement design procedures are based on limited empirical data and
recommendations of the WMA technology suppliers. WMA design procedures do not consider
the foam quality since its importance has not been fully understood. Therefore, the long term
performance of the WMA mixtures prepared with different foamed binders (the ones in
CHAPTER 4) in the laboratory was evaluated via performance tests and compared with the
foamed binder quality parameters presented earlier.
WMA ASPHALT MIXTURE DESIGN
Draft Appendix of AASHTO R 35: Special Mixture Design Considerations and Methods
for WMA, was followed in this study for the WMA mix design. The aggregate gradation of the
WMA mixtures was a dense graded mixture as shown in Figure 24. This design was previously
used in a traditional HMA and crumb rubber modified asphalt (terminal blend) pavement
sections in Lansing, Michigan. The performance of these mixtures was evaluated by the author
of this thesis. All mixtures were prepared with PG58-28 virgin binder, which was provided by a
local petroleum supplier. The properties of the binder were given in CHAPTER 4.
The standard Superpave mix design procedure, as suggested in the draft Appendix of
AASHTO R 35, was utilized in this study to obtain the optimum asphalt content and to evaluate
the volumetrics of the mix design. The optimum binder content of the HMA mix was 4.53% by
91

weight of the mixture. The WMA mixtures were initially prepared with the same binder content
as the HMA to calculate the volumetrics and if needed to adjust the binder content. The first trial
with WMA resulted in the mix properties (i.e., air void at Ndesign, VFA, VMA) that were within
the Superpave limits. The target air void at Ndesign for WMA mix was within the limits 4±0.5%
as suggested in Superpave Mix Design, though it was close to the lower limit, 3.54%. The voids
in mineral aggregates (VMA) at Ndesign was 15%, which is higher than the minimum, 14%. The
voids filled with asphalt (VFA) was 78%, which is in the range of 70-80, as specified by
Superpave specification. Thus, the WMA mix design was almost identical to that of HMA/

100
90

Percent passing (%)

80
70
60
50
40

30
20
10
0
0.45 Power Sieve Size (mm)

Figure 24: Aggregate gradation of WMA mixtures
92

PERFORMANCE TEST SAMPLE PREPARATION
The WMA samples for performance testing were prepared based on the suggestions of
NCHRP Reports 691 and 714, which were the most recent studies on WMA design and
performance analysis.
Flow chart of the WMA asphalt mixture design and performance tests is illustrated in
Figure 25. Initially, the foamed binder temperature (FT) was determined based on the viscosity
of the non-foamed binder. The viscosity of the binder should be approximately 300 mPa.s at the
foaming temperature for successful foaming. The rate of flow of 200 grams of binder took 15
seconds at this viscosity under the gravity (manufacturer’s calibration). FT was 155°C for the
PG58-28 binder used. Then, the aggregate temperature (AT) was determined to be approximately
20°C lower than the FT, which was 135°C. This was also based on the experience of the states
and current practice.
The preparation process for WMA and conventional HMA are very similar. Initially, the
aggregates were batched and heated to AT. The binder is heated about 5°C more than FT
(FT+5°C) in order to compensate the temperature loss of the binder while pouring it to the
chamber of the foamer. It should be noted that it was crucial to calibrate and verify the water
content and air pressure of the foamer, frequently. Once the binder temperature equilibrated, the
aggregates were transferred to the mixing bucket and placed on a scale that is located under the
foamer. The binder was directly foamed into the mixing bucket. If excessive binder was foamed,
it was gently moved with a piece of paper towel. The loose mixture was mixed in a bucket mixer
approximately 2 minutes till a homogenous mix was accomplished. Then, the mixtures were
conditioned for short term (two hours) at compaction temperature (CT) before compaction. CT
93

was determined to be 10°C lower than AT, which was 128°C in this study. The samples were
compacted with a Superpave Gyratory Compactor (SGC). After overnight cooling of the samples
at the room temperature, the samples were cut and cored to the specific sizes for the performance
tests. The cored and cut samples were accepted for performance testing if percent of the air void
was in the range of 7±0.5%.
The performance test samples in this chapter were named based on the foamed binder’s
injected water content and air pressure, as given in Table 7. For instance, a WMA sample
prepared with 1% water content and 10 psi air pressure injected foamed binder was called as
“1% - 10psi”.
Table 7: Performance test sample descriptions
Foamed Binder
Name of the WMA Sample

Water Content
(% of the binder)

Air Pressure (psi)

1%

10 psi

1% - 10psi

2%

12.5 psi

2% - 12.5psi

3%

15 psi

3% - 15psi

4%

17.5 psi

4% - 17.5psi

5%

20 psi

5% - 20psi

94

Foam Temperature
(FT)
Aggregate Temperature
(AT=FT-17°C)
Mix Aggregates and
Binder
Evaluate
Coating

Evaluate
Compactability

Evaluate
Performance

95% Coating
Criteria

Condition @
Compaction Temp. (CT)
(CT=AT-10°C) for 2 hr

Condition @ (CT)
(CT=AT-10°C) for 2 hr

Compact @
CT

Compact @
CT

Compact @
CT-30°C

Ratio<1.25

Dynamic Modulus
(Fatigue Cracking
and Rutting)
Confined
Unconfined
Flow Number
(Rutting)
Push-Pull
(Fatigue Cracking)
TSR
(Moisture Sensitivity)

Figure 25: Flow chart of the WMA performance evaluation study

95

STATISTICAL EVALUATIONS OF LABORATORY TEST RESULTS
This study was designed with the mixtures with identical aggregate gradations with
different foamed binders (i.e., injected air pressure and water content) to understand the
influence of the foam quality on the pavement response (see Table 7). Both correlation and
regression analyses were used in this study since the data set was limited with five different
mixtures. Initially, the measured data was ranked according to the Kendall rank correlation
coefficient (tau (τ) coefficient), which is a distribution free/non-parametric, ranking correlation
parameter for small data sets (Gibson, 2012). The set of data pairs are initially ordered in the
increasing rank of the first column. Then, the observations are specified as concordant and
discordant pairs based on the rank of the second column. The data pairs are compared with
respect to the first row. If the value of the pair is greater than the first row, it is specified as a
concordant pair. On the contrary, if the value of the pair is smaller than the first row, it is
classified as discordant pair. The same process is repeated for each consequent row till the last
row. The tau coefficient is calculated as follows:

τ=

n c -n d

[13]

n ( n- 1)
2

where nc and nd are the total number of concordant and discordant pairs and n is the total
number of points. The tau coefficient ranges from -1 to 1. If the coefficient is equal to 1, it
indicates perfect agreement between two rankings. On the contrary, if the coefficient is equal to 1, it designates the disagreement between two rankings (i.e., one ranking is in the reverse of the
other). A coefficient of 0 signifies the lack of association or complete independence between
96

two different data sets. The significance of correlation is also assessed based on the (nc-nd) and n
using Kendall’s Tau significance table (Kendall and Gibbons, 1996). A detailed Kendall’s Tau
coefficient calculation is given in the next section.
Regression analysis was also used to study the relation between the foamed binder
parameters and performance of mixtures. Linear regression lines were fitted using the least
squares approach. In addition, polynomial regression was fitted for characterizing the nonlinear
nd

relationships as a 2

order polynomial. The regression lines were coupled with the coefficient of

2

determination (R ), in which the goodness of the regression fit is evaluated.

COATING OF AGGREGATES
Coating of the WMA mixtures is a concern due to low production temperatures. The
degree of particle coating is typically determined using AASHTO T195, Standard Method of
Test for Determining Degree of Particle Coating of Bituminous-Aggregate Mixtures. In NCHRP
Reports 691 and 714, the coating criterion for the WMA mixtures was recommended as 95%.
AASHTO T195 requires sieving the loose mixture through 3/8” sieve, right away after
mixing while it is still hot. Then, the aggregates retaining on the sieve are laid on a wax paper in
a single layer without further mixing. The aggregates are separated into two subgroups: as coated
and uncoated aggregates.
Figure 26 shows the percentage of uncoated aggregates at different combinations of water
content and air pressures of foaming. As shown, percent of uncoated aggregates are similar in
1% - 10psi, 2% - 12.5psi, and 3% - 15psi, and then it increases significantly at 4% -17.5psi and
97

5% - 20psi. Statistical analysis of the relation between the injected water content/air pressure of
the foamed binder and the percent of uncoated aggregates are given in Table 8. In addition,
Kendall’s tau coefficient, which is explained in the previous section, is illustrated in Table 8a.
The tau coefficient is relatively low at 0.4 and the (nc-nd) is 4. For single tailed significance test,
the proportion of rankings is 0.242 (24.2%). It can be concluded that the ranking correlation is
significant as a level of 0.758 (75.8%) while statistical significance is typically at 95%. The
coefficient of determination of the linear fit is 0.673 and its significance is 0.911 (91%).
Based on the findings in CHAPTER 4, low water content and air pressure leads to small
bubbles. For example, 1% water content and 10 psi air pressure injected foamed binder was
anticipated to be composed of relatively small bubbles and the collapse of the foamed binder
takes longer, which allowed the binder to be more workable and coat the aggregates better. On
the other hand, the foamed binder prepared with 5% water content and 20 psi injected air was
composed of large bubbles and the foam collapsed relatively quick. The findings indicated that
the time interval of the binder was too short to coat the aggregates. Based on the suggested
criteria in NCHRP reports, only three mixes (i.e., 1% - 10psi, 2% -12.5psi and 3% - 15psi)
passed the coating criteria.
According to the coating tests, following conclusions can be drawn:
(i)

As the expansion ratio of the foamed binder increased and its half-life decreased,
degree of coating decreased.

98

(ii)

As the bubble sizes in the foamed binder increased, the coating of aggregates
became poor. On the contrary, as the surface area of the bubbles increased
because of the small bubbles, the aggregates were coated better.

As it was observed in Figure 26, while the trend is clear, the variability within the same
mixtures was relatively high. The coarse aggregates used in this study had limited number of
ironstones, which had partially coating problems. Clay-ironstone is a widespread a yellowishbrown to dark brow aggregate (siderite) composed of FeCO3 in Michigan. These aggregates are
soft and porous and there are limitations of their usage in asphalt pavements in MI (MDOT
Procedures for aggregate inspection, 2009).

% of Uncoated Aggregates

18
16
14
12

NCHRP
95%
Coating Criteria

10
8
6
4

2
0
1% - 10psi

2% - 12.5psi

3% -15psi

Figure 26: Percent (%) of uncoated aggregates

99

4% - 17.5psi

5% -20psi

Table 8: Statistical analysis for percent of uncoated aggregates: a) Illustration of Kendall’s
Tau Correlation Coefficient calculation b) Correlation and Linear Regression Summary
a) Illustration of Kendall’s Tau Correlation Coefficient calculation
Row
st

1 Row
nd
2 Row
rd
3 Row
th
4 Row
th

5

Row
n

nC
nD
nC -nD

Significance

1
2
3
4

6.15
5.33
4.58
11.70

5
5
7
3
4.00
0.40

13.48

2nd Row

3rd Row

4th Row

5th Row

C*

D**

C

D

C

D

C

D

-

Water
% of
Content/Air Uncoated
Pressure
Aggregates

-

-

-

-

-

-

+

-

+

+
+
-

+

-

+

+
-

+

-

+

-

+

-

C*= Concordant
D**= Discordant

0.76

b) Correlation and Linear Regression Summary
Correlation
Kendall's
Kendall's
Tau
Tau
Significance
(1- pvalue)
% of Uncoated Aggregates vs.
Water Content/Air Pressure

0.400

0.758

Linear Regression
R

R

Regression
Significance
(1-pvalue)

0.821

0.673

0.911

2

COMPACTABILITYOF WMA MIXTURES
Compactability of the WMA mixtures is another concern due to low production and
laying temperatures. Compactability of WMA mixtures was determined using the procedure
described NCHRP Reports 691 and 714. The procedure compares the number of gyrations to
reach 92% relative density (i.e., 8% air voids) at compaction temperature (CT) and 30°C lower
100

than CT, which is designated as the temperature loss in between the production and construction.
Then, the compactability ratio (CR) is calculated using:
CR = (N 92 )
CT- 30 ° C

(

(N 92 )

CT

[14]

)

where N 92 CT- 30 ° C = the number of gyrations to reach 92% relative density at 30°C below

(

)

the CT and N 92 CT = the number of gyrations to reach 92% relative density at CT. The
mixture is deemed compactable if the compactability ratio is less than 1.25 (NCHRP Report 691
and 714).
Three air voids and water content combinations were selected based on the foam quality
parameters: i) 1% water content and 10 psi air pressure, ii) 3% water content and 15psi air
pressure, and iii) 5% water content and 20 psi air pressure. Three replicates were compacted for
each mixture set at 128°C and 98°C for this study. Then, the average number of gyrations was
computed to achieve the 8% air voids as shown in Figure 27a. The statistical analysis of the
relation between the number of gyrations and the injected water content/air pressure of the
foamed binder is given in Table 9. Since the data pairs are limited with three different mixtures
and the ranking is poor and not linear, the tau coefficient is calculated to be 0.333, which is
relatively low. Thus, the significance of the ranking correlation is 0.5, which is significantly
lower than the typical level 0.95. In addition, the linear regression shows similar trend with
respect to ranking correlation. The coefficient of determination and significance are both
relatively low.

101

The compactability ratios for the selected samples were significantly lower than 1.25
(suggested limit) as shown in Figure 27b, no further testing was performed for the other foamed
binder combinations, since compactability was not a concern for the selected mix design and not
directly related with the pavement performance. It appears from Figure 27b that the water
content/air combinations did not indicate a trend. However, different foaming technologies used
in the field may or may not result in such a trend, which is not investigated in this study.
30
a) Number of Gyrations

128 C

98 C

Number of Gyrations

25
20
15

10
5
0
1% - 10psi

3% - 15psi

5% - 20psi

Compactability Ratio

1.25
b) Compactability Ratio

1.20
1.15

NCHRP
1.25
Compactability Criteria

1.10
1.05
1.00
0.95
0.90
1% - 10psi

3% - 15psi

5% - 20psi

Figure 27: Compactability of WMA mixtures: a) number of gyrations, b) compactability
ratio
102

Table 9: Statistical analysis for number of gyrations and compactability ratio
Correlation
Kendall's
Kendall's
Tau
Tau
Significance
(1- pvalue)
Number of Gyrations @
128°C versus Water
Content/Air Pressure
Number of Gyrations @
98°C versus Water
Content/Air Pressure
Compactability Ratio
versus Water Content/Air
Pressure

Linear Regression
R

R

Regression
Significance
(1-pvalue)

2

0.333

0.500

0.832

0.693

0.626

0.333

0.500

0.622

0.387

0.427

0.333

0.500

0.255

0.065

0.164

MIXTURE PERFORMANCE TESTS
After the coating of the aggregates and compactability of the mixtures were assessed, the
next step was evaluating the long-term performance of WMA mixes via laboratory tests. The
performance tests included in this study were: (i) Dynamic Modulus, (ii) Flow Number, (iii)
Compression-Tension Fatigue, and (iv) Moisture Susceptibility (TSR).
Dynamic Modulus Test
Dynamic Modulus (|E*|) test is a non-destructive test to determine the stiffness and the
viscoelastic primary responses (i.e., undamaged, low-strain response) of asphalt mixtures at
different temperatures and loading frequencies. In addition, this test is useful for preliminary
estimation of rutting and fatigue cracking of asphalt pavements at design stage, and also major
input to the Mechanistic Empirical Design Guide (MEPDG) software.

103

Dynamic Modulus Test Procedure
The |E*| tests were conducted according to AASHTO TP79: Determining Dynamic
Modulus and Flow Number of Hot Mix Asphalt (HMA) by using the Asphalt Mixture
Performance Tester (AMPT), shown in Figure 28, though it should be noted that the sample
preparation for WMA mixes is different than HMA mixes. Initially, the short-term (2 hours)
conditioned loose mixtures (long-term conditioning (4 hours) is an obligation for HMA
mixtures) were compacted at a diameter of 150mm to a height of 180mm. Then, the compacted
samples were cored at a diameter of 100mm and the ends were cut to a final height of 150mm.
The final samples were accepted for testing if percent of the air void was in the range of 7±0.5%.
Then, the LVDT tabs were glued using two components high strength epoxy (120° apart from
each other). LVDT gauge lengths were about 70 mm and the top and bottom tabs were about
37.5 mm away from the top and bottom edges of the samples. The |E*| tests were performed at 10, 10, 21, 37, and 54°C at loading frequencies of 0.1, 0.5, 1.0, 5, 10, and 25 Hz for at each
temperature in two different modes: (a) unconfined, and (b) confined.
b)

a)

Figure 28: Asphalt Mixture Performance Tester (AMPT): a) Unconfined test sample, b)
Confined test sample
104

Brief Summary of Dynamic Modulus Master Curve
After determining |E*| values and phase angles at each temperature and frequency, |E*|
master curves are generated using the time temperature superposition principle, in which both the
effects of temperature and frequency on the asphalt mixture is combined. The raw |E*| data at
different temperatures and frequencies and its master curve generated by shifting the data at
different temperatures along the loading frequency are given in Figure 29. The resulting
parameter in x-axis is called the reduced frequency (fR), which is defined as follows:
f

R

= f a (T )

or

log( f

R

) = log( f ) + log ( a ( T ))

[15]

where a(T) is the shift factor coefficient which is a function of temperature (T) and f is the
loading frequency. The temperature dependency of the asphalt mixture is measured by the
amount of the shifting of the raw |E*| data. The shift factor is determined by fitting a second
order polynomial as function of T and the reference temperature (Tref), as follows:

a (T
a ( T ) = 10 1

2

-T

2
) + a (T - T
)
ref
2
ref

[16]

where a1 and a2 are the shift factor coefficients. Finally, the following sigmoid function is fitted
to the measured |E*| data to generate the master curve as shown in Figure 29.

105

1.E+05

|E*| (MPa)

1.E+04

T=-10°C
T=10°C

T=37°C

1.E+03

T=54°C
Fitted |E*|

1.E+02

Measured |E*|
Shifted |E*|
1.E+01
1.E-10

1.E-06

1.E-02

1.E+02

1.E+06

Frequency (Hz)
Figure 29: Illustration of shifting |E*| data at different temperatures to obtain the |E*|
master curve
b
log ( | E* |) = b +
1 1 + exp(-b

3

2
-b

4

log( f

R

))

[17]

where b1, b2, b3 and b4 are the sigmoid coefficients. |E*| increases with the increase in the
loading frequency and decreases with the increases of the testing temperature. |E*| master curve
is useful for estimating the behavior (i.e., rutting and fatigue cracking) of mixtures over a range
of temperatures and rates of loading. Typically, mixtures with relatively low |E*| values at low
temperatures/high frequencies are more flexible (and less brittle), therefore more resistant to
fatigue cracking. On the other hand, mixtures with high |E*| at high temperatures/low
frequencies are stiffer and are more resistant to rutting.

106

In order to better interpret the master curves, they can be plotted in log-log scale (to
better see the difference between the curves at low reduced frequencies) and in linear-log scale
(to better see the difference between the curves at high reduced frequencies). Less cracking is
typically expected when the |E*| is low at high reduced frequencies (high reduced frequency =
high test frequency/low test temperature combination). On the other hand, less rutting is
expected when |E*| is high at low reduced frequencies (low reduced frequency = low test
frequency/high test temperature combination).
Unconfined Dynamic Modulus Test Results and Discussions
The effect of the foamed binder parameters on the rutting and fatigue cracking
performance has not been studied before. Therefore, three foamed binders with different injected
air pressure and water content (i.e., 1% - 10psi, 3%- 15psi, 5% - 20psi) were initially selected to
prepare WMA samples. Two replicates for each WMA mixture set were tested according to
AASHTO TP79, as given from Table 17 to Table 22 in APPENDIX A. Figure 30a shows the
|E*| versus reduced frequency plots in log-log scale. As shown, the curves of three different
mixtures are very similar. Therefore, it is hard to estimate the relative rutting susceptibilities of
these WMA mixtures. Figure 30b shows the |E*| versus reduced frequency plots in linear-log
scale. As shown, the fatigue cracking potential of WMA mixtures decreased with the increase in
the injected water content and air pressure to foam the binder. On the contrary, the fatigue
cracking susceptibility increased with the decrease in the injected water content and air pressure
of the foamed binder. This may be explained by the long half-life and the small bubbles that may
have gotten stuck in the 1% - 10psi mixture, causing fatigue cracking susceptibility. The 2 hours
of curing time may not be adequate for the collapse of all the micro bubbles. Although |E*|
107

master curve is a good ‘indicator’ of performance of asphalt mixtures, the conventional dynamic
modulus test (unconfined) was not sufficient to discuss the rutting and cracking potential of
different WMA mixtures for this study. Additionally, the measured |E*| (not the sigmoid fitted)
at 10 Hz at 10, 21 and 54°C versus the foam quality indicators (i.e., ER, HL, FI, D50 and SAI)
were plotted in Figure 31, in order to investigate if any relation exists in between the measured
|E*| and the foam quality. It was clear that |E*| decreases with the increase in the temperature as
explained in the previous section. However, there was negligible variability in the measured |E*|
of WMAs. Thus, no relation was founded in between the foam quality and |E*| as shown in
Figure 31 and Table 10. Since the dataset is limited with three different mixes, the tau coefficient
and its significance is relatively low, 0.333 and 0.5. In addition, the linear regression fit indicates
the lack of the relation between unconfined |E*| and foamed binder parameters. Therefore, in the
following section, the dynamic modulus tests under confinement were performed using the same
samples, in order to be able to establish relation between the confined |E*| and foam binder
quality.

108

|E*| (Mpa)

1.E+04

1.E+03

Less Rutting

1.E+05

1.E+02

1.E+01
1% - 10psi
1.E+00
1.E-07

1.E-05

3% - 15psi

1.E-03

1.E-01

5% - 20psi

1.E+01

1.E+03

1.E+05

1.E+03

1.E+05

Reduced Frequency (Hz)
3.E+04
1% - 10psi
2.E+04

3% - 15psi

2.E+04
Less Cracking

|E*| (Mpa)

5% - 20psi

1.E+04

5.E+03

0.E+00
1.E-07

1.E-05

1.E-03

1.E-01

1.E+01

Reduced Frequency (Hz)
Figure 30: Unconfined Dynamic Modulus: a) log-log, b) linear –log.

109

a)
10°C
21°C

b)

1.E+04
|E*| (Mpa)

|E*| (Mpa)

1.E+04

1.E+03
54°C

1.E+02

10°C
21°C

1.E+03
54°C
1.E+02

1.E+03
54°C

0
1% -10 psi
1.E+04 d)
|E*| (Mpa)

2
2.5
3
Expansion Ratio (ER)
1% -10 psi 3%- 15 psi 5% -20psi
1.E+04 c)
10°C
21°C
|E*| (Mpa)

1.5

50
100
Half-Life (HL)
3%- 15 psi 5% -20psi
10°C
21°C

1.E+03
54°C

1.E+02

1.E+02

200
300
400
500
Foam Index (FI)
1% -10 psi 3%- 15 psi 5% -20psi
e)
1.E+04
10°C
21°C
|E*| (Mpa)

100

0
1% -10 psi

0.5
D50
3%- 15 psi

1
5% -20psi

1.E+03
54°C
1.E+02
100

300
500
Surface Area Index (SAI)
1% -10 psi
3%- 15 psi
5% -20psi
Figure 31: The comparison of Unconfined Dynamic Modulus at 10 Hz and Foam Binder
Quality Parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index (FI), d)
D50, e) Surface Area Index (SAI)

110

Table 10: Statistical analysis between unconfined dynamic modulus at 10 Hz and foam
binder quality parameters
Correlation

Linear Regression

Kendall's
Tau
Unconfined E* vs.
Expansion Ratio
Unconfined E* vs.
Half-Life
Unconfined E* vs.
Foam Index
Unconfined E* vs.
D50
Unconfined E* vs.
SAI

Kendall's Tau
Significance
(1- pvalue)

R

R

Regression
Significance
(1-pvalue)

0.333

0.500

0.427

0.182

0.281

-0.333

0.500

0.199

0.040

0.128

-0.333

0.500

0.471

0.222

0.312

0.333

0.500

0.374

0.140

0.244

-0.333

0.500

0.0107

0.0001

0.007

2

Confined Dynamic Modulus Test Results and Discussions
The confined dynamic modulus is generally not performed due to the complexity of the
testing procedure. Zeiada et al. (2011) studied the different level of confining stresses (10, 20, 30
and 40 psi) on the moduli with respect to the unconfined stress. It was observed that the
confining stress higher than 20 psi did not increase the moduli of the mixtures. In addition, it was
concluded that the mixtures were less affected form the confinement at lower temperatures due
to the stiffness of the mixtures at lower temperatures. On the other hand, the confining pressure
at higher temperatures significantly affects the moduli due to loss of stiffness at high
temperatures. Therefore, 20 psi confined stress was determined to be used in this study. The
measured and predicted |E*| and phase angle data is available for each test from Table 23 to
Table 28 in APPENDIX A.

111

The |E*| versus reduced frequency curves were plotted in log-log scale in Figure 32a and
in linear-log scale in Figure 32b. The confined dynamic modulus tests indicated that there was
slight difference in the rutting potential of WMA mixtures prepared with different foamed
binders. It was observed that |E*| decreased (i.e., the rutting potential increased) as the injected
water content and air pressure increased in the foamed binder (see 5% - 20psi versus 1% -10psi
and 3% - 15psi). However, this may be misleading as the difference in the |E*| was limited and
varied in a small range. Therefore, the difference may be caused by the sample to sample
variability. Figure 32b shows the |E*| master curve in the linear-log scale, where a clear trend is
not visible. WMA prepared with 3% - 15psi has the highest |E*| as compared to the WMAs with
1% - 10psi and 5% - 20psi. In order to further investigate the relations in between the measured
|E*| and the foam quality, the measured |E*| (not the sigmoid fitted) at 10 Hz at 10, 21 and 54°C
versus the foam quality indicators (i.e., ER, HL, FI, D50 and SAI) were plotted in Figure 33. It
was clearly observed that there is no linear trend in between the |E*| and foam quality. The
statistical analysis is also provided in Table 11. Kendall’s correlation and linear regression are
both relatively low due to nonlinear ranking. In addition, it was not possible to fit a polynomial
regression since the dataset is less than four mixtures. It was concluded that the WMAs prepared
with 3% -15psi was the optimum mix design due to confined dynamic modulus. However, It
should be noted that |E*| is only an “indicator” of performance. Actual performance tests are
Flow Number (for rutting) and Push-Pull fatigue (for fatigue cracking).

112

1.E+05
a) Log-log scale

1.E+03
1.E+02
1.E+01

Less Rutting

|E*| (Mpa)

1.E+04

1.E+00
1.E-07

1% - 10psi
3% - 15psi
5% - 20psi
1.E-05

1.E-03
1.E-01
1.E+01
Reduced Frequency (Hz)

1.E+03

1.E+05

1.E+03

1.E+05

3.E+04
b) Linear-log scale
3.E+04
1% - 10psi
3% - 15psi

2.E+04

5% - 20psi
Less
Cracking

|E*| (Mpa)

2.E+04

1.E+04
5.E+03
0.E+00
1.E-07

1.E-05

1.E-03
1.E-01
1.E+01
Reduced Frequency (Hz)

Figure 32: Confined Dynamic Modulus: a) log-log, b) linear –log.

113

1.E+05

b)

|E*| (Mpa)

a)

|E*| (Mpa)

1.E+05

1.E+04

1.E+04

1.E+03

1.E+03

1.E+02

1.E+02
1.5
1% -10 psi
1.E+05

2
2.5
Expansion Ratio (ER)
3%- 15 psi 5% -20psi

0

3

1% -10 psi
1.E+05

|E*| (Mpa)

|E*| (Mpa)

c)

1.E+04

1.E+03

50
100
Half-Life (HL)
3%- 15 psi 5% -20psi
d)

1.E+04

1.E+03

1.E+02

1.E+02
100

|E*| (Mpa)

1% -10 psi

0

200
300
400
Foam Index (FI)
3%- 15 psi 5% -20psi
1.E+05
e)

1% -10 psi

0.5
D50
3%- 15 psi 5% -20psi

1.E+04

1.E+03

1.E+02
100

200
300
400
500
Surface Area Index (SAI)
1% -10 psi 3%- 15 psi 5% -20psi
Figure 33: The comparison of Confined Dynamic Modulus at 10 Hz and Foam Binder
Quality Parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index (FI), d)
D50, e) Surface Area Index (SAI)

114

1

Table 11: Statistical analysis between confined dynamic modulus at 10 Hz and foam binder
quality parameters
Correlation

Linear Regression

Kendall's
Tau

Kendall's
Tau
Significance
(1- pvalue)

R

R

Regression
Significance
(1-pvalue)

Confined E* vs ER

-0.333

0.500

0.873

0.762

0.676

Confined E* vs HL

0.333

0.500

0.407

0.166

0.267

Confined E* vs FI

0.333

0.500

0.896

0.803

0.707

Confined E* vs D50

-0.333

0.500

0.843

0.711

0.639

Confined E* vs SAI

0.333

0.500

0.590

0.348

0.402

2

Flow Number (FN)
Rutting, depression of pavement surface along the wheel path, is one of the major
pavement distress types. One of the ways to determine rutting potential on pavement surface is
the Flow Number Test, where the flow number is associated to the resistance of the pavement to
permanent deformation.
Flow Number Test Procedure and Analysis
The tests were conducted according to AASHTO TP79: Determining Dynamic Modulus
and Flow Number for Hot Mix Asphalt (HMA) by using the Asphalt Mixture Performance
Tester (AMPT) shown in Figure 28. The loose mixtures were subjected to the short-term (2
hours) aging before compacting in a Superpave gyratory compactor at a diameter of 150mm to a
height of 180mm. Then, the compacted samples were cored to a diameter of 100mm and the ends
115

were cut to obtain a final height of 150mm. The final samples were accepted for testing if
percent of the air void is in the range of 7±0.5%.
Flow Number Test is a repeated load test conducted at relatively high temperatures (3554oC). The test procedure (i.e., test temperature, axial deviator stress, confining stress, contact
stress, duration of stress pulse, duration of rest period and the shape of stress pulse) is not
determined in the AASHTO TP79. However, the deviator stress is generally applied in haversine
shape, 0.1 seconds of loading (stress) followed by 0.9 seconds of unloading (recovery) periods,
in which each loading and unloading periods create one cycle. These repeated stress pulses and
recovery periods through the Flow Number test result a continuously growing permanent strain
in the asphalt mix. These strain values are plotted with respect to each cycle and analyzed.
Since WMA mixture testing experience is very limited, the test parameters were selected
based on the recommendations of NCHRP reports 691 and 714. The flow number test
temperature was 45oC, which is equal to the 50% reliability of 7 day maximum pavement
temperature in Michigan. This was computed at a depth of 20 mm (suggested depth for surface
course mixes) using LTPPBind Version 3.1beta. The tests were conducted under 87 psi (600
kPa) axial deviator stress and 4.4psi (30 kPa) contact stress at 0 psi confining stress (unconfined
mode).
Flow Number Test Results and Discussions
Initially, three foamed binders with different injected air pressures and water contents
(i.e., 1% - 10psi, 3%- 15psi, 5% - 20psi) were selected to prepare WMA samples. Since there
was no significant difference determined from both the confined and unconfined dynamic
116

modulus tests, the other samples (2%-12.5psi and 4-17.5psi) in the test matrix were not prepared
initially. As shown in Figure 34a, the initial test set was identified as “1”. For the simplicity, the
average of two to four replicates was plotted in Figure 34. The data for each replicate is available
in Figure 59 in the APPENDIX B. It was observed that the rutting resistance of the mixes
increased with the increase of the injected air pressure and water content of the foamed binder in
the preparation of WMA mixtures. This may have caused by higher aggregate-to-aggregate
friction, possibly because of poor coating of the aggregates. In addition, relatively quick failure
of the WMA mixtures with low water content and air pressure injected samples may also be the
micro bubbles captured in the samples because of the longer half-life.
In order to further investigate these hypotheses, two more WMA samples with different
foamed binders (i.e., 2%-12.5psi and 4%-17.5psi) were prepared and tested under the same
conditions shown as “2” in Figure 34. It was expected to be that the fatigue behavior of these two
transition samples will follow the similar curves and lay in between 1%-10psi, 3%-15psi and
5%-20psi. However, the “2” set was failed relatively quickly as compared to the “1” set. The
trend in the relation of the rutting and foam binder parameters was same, i.e., rutting potential
increased with increasing the water content and air pressure. In order to better illustrate the trend,
the permanent microstrains were plotted at 50 cycles, as shown in Figure 34b. The trend lines for
the two data sets (“1” and “2”) were separately plotted to illustrate the overall trend. As clearly
shown in Figure 34b, the trends were same. It had been already known that the laboratory foamer
has been having repeatability problems within each trial on the same day and even worse
between the days. Since the samples “1” and “2” were not prepared on the same day, the
variability of the tests was expected. The results shown in Figure 34 clearly show the influence
117

of the foam binder parameters on the rutting susceptibility. Therefore, the permanent strain at
50th cycle versus the foam quality indicators (i.e., ER, HL, FI, D50 and SAI) was plotted in
Figure 35, in order to investigate these relations individually. The statistical analysis was given
in Table 12. Kendall’s tau coefficients are +1 and -1 and their significances are 100%. In
addition, the linear regression significantly fit the data. As shown in Figure 35a and d, Expansion
Ratio (ER) and D50 increase with the increase in the injected water content and air pressure of
the binder as the permanent microstrain at50

th

cycle decreases. On the contrary, half-life (HL),

foam index (FI), and Surface Area Index (SAI) increase with the decrease in the injected water
content and air pressure of the binder as the permanent microstrain at50

th

cycle increases. The

relation of rutting and foam binder parameters was determined in this section as follows:
(i)

As the expansion ratio of the foamed binder increases, its half-life decreases
(Figure 35a and b). This improves the rutting resistance of WMA pavements.

(ii)

As the surface area index decreases, the higher aggregate-to-aggregate friction as
a result of the poor coating improves the rutting resistance of the WMA
pavements.

(iii)

As the bubble sizes in the foamed binder (see D50 in Figure 35d) increases and
the surface area index (see SAI in Figure 35e) decreases, the coating of aggregates
becomes poor. The poor coating increases the shear strength in between the
aggregates. This improves the rutting resistance of WMA pavements at high
temperatures.

118

70000
a)

1

Microstrain

60000
2

50000
40000
30000
20000

1

10000

2

1% - 10psi
3% - 15psi
5% - 20psi
2% - 12.5psi
4% - 17.5psi

0
0

50

100
150
Number of Cycles

200

250

Permanent Microstrain @ 50 cycles

31000
29000

b)
2

27000
25000
23000
1

21000
19000
17000
15000
13000
1% - 10 psi

2% - 12.5psi

3% - 15psi

4% - 17.5psi

5% - 20psi

Figure 34: a) Permanent (plastic) strain with cycles obtained from unconfined FN tests, b)
Permanent strain at 50 cycles

119

14000
12000
10000

24000

14000
12000
10000
0

20000
18000
14000

5 -20

5 - 20

16000

12000

100

d)

22000

3 - 15

3 - 15

18000

50
Half-Life (HL)

1 - 10

20000

14000

16000

24000

c)

16000

18000

3

1 - 10

22000

2
2.5
Expansion Ratio (ER)

20000

Permanent Microstrain
@ 50 cycles

1.5

Permanent Microstrain
@ 50 cycles

5 - 20

16000

3 - 15

18000

b)

22000

3 - 15
5 - 20

Permanent Microstrain
@ 50 cycles

20000

24000

1 - 10

22000

a)
1 - 10

Permanent Microstrain
@ 50 cycles

24000

12000

10000

10000

300
Foam Index (FI)
24000

0

e)

20000

14000

3 - 15

18000
16000

0.5
D50
1 - 10

22000

500

5 - 20

Permanent Microstrain
@ 50 cycles

100

12000
10000
200

300
400
500
Surface Area Index (SAI)

Figure 35: The comparison of permeate strain at 50 cycles and foam binder quality
parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index (FI), d) D50, e)
Surface Area Index (SAI)
120

1

Table 12: Statistical analysis between the permeate strain at 50 cycles and foam binder
quality parameters
Correlation
Kendall's
Kendall's
Tau
Tau
Significance
(1- pvalue)
Permanent Microstrain @
50 cycles vs. Water
Content/Air Pressure
Permanent Microstrain @
50 cycles vs. ER
Permanent Microstrain @
50 cycles vs. HL
Permanent Microstrain @
50 cycles vs. FI
Permanent Microstrain @
50 cycles vs. D50
Permanent Microstrain @
50 cycles vs. SAI

Linear Regression

R

R

Regression
Significance
(1-pvalue)

2

-1.000

1.000

0.999

0.998

0.974

-1.000

1.000

0.971

0.943

0.847

1.000

1.000

0.921

0.847

0.745

1.000

1.000

0.958

0.918

0.815

-1.000

1.000

0.983

0.967

0.884

1.000

1.000

0.983

0.964

0.879

Push-Pull (Compression-Tension) Fatigue Test
Fatigue cracking is typically caused by many repetitions of the heavy traffic load, inadequate
support in between the pavement layers and subgrade, very stiff binder in the surface layer and

poor drainage. Laboratory cyclic push-pull (tension-compression) test is one of the common
methods to predict field fatigue life of the pavements. In addition, this test was preferred in lieu
of Four Point Bending Beam (FPBB) test (AASHTO T 321: Standard Method of Test for
Determining the Fatigue Life of Compacted Hot Mix Asphalt (HMA) Subjected to Repeated
Flexural Bending) since this test is much easier to conduct as well as the use of less material,
energy and time, and it provides more information than the FPBB test. The fatigue performance
121

of the pavements is quantified by the number of cycles to failure (Nf) in Push-Pull test at a given
strain level, temperature and loading frequency.
Push-Pull Test Procedure
Laboratory samples were prepared in the Superpave gyratory compactor to a height of
approximately 180 mm and then cut and cored to a cylindrical sample, 76 mm in diameter and
150 mm tall. The samples were accepted for testing if air void was in acceptable range (7±0.5%).
Initially, LVDT tabs were mounted with two components high strength epoxy (120° apart from each
other). LVDT gauge lengths were about 70 mm and the top and bottom tabs were about 37.5 mm

away from the top and bottom edges of the samples. Then, the specimens were glued with steel
epoxy to aluminum top and bottom plates using a custom made gluing jig to provide perfectly

parallel specimen ends, as shown in Figure 36. Finally, the tests were conducted using Asphalt
Mixture Performance Tester (AMPT). Samples were carefully mounted to AMPT to eliminate
eccentricity during the test, which may cause non-uniform stress distribution and thus localized
failure (in general close to one of the end platens), as shown in Figure 36c.
Push-pull test can be conducted in both stress controlled and strain controlled loading mode.
However, the push-pull tests in this study were performed only at strain controlled mode at 10 Hz
frequency at both 10°C and 20°C. For both temperatures, two replicates of each sample were
prepared and tested. However, some tests were removed from the analysis due to localized failures

at the top and bottom of the samples during testing, as shown in Figure 36c. The samples with
the mid-failure as shown in Figure 36d were used for the analysis. The analysis were performed

122

using viscoelastic continuum damage concept (VECD) and using non-commercial software
developed by Dr. M. Emin Kutay, called PP-VECD v0.1.
Brief summary of Viscoelastic Continuum Damage (VECD) concept
Viscoelastic Continuum Damage (VECD) theory is based on Schapery’s elasticviscoelastic correspondence principle and the work potential theory. Schapery (1999) divided the
total strain (εtot) as viscoelastic (εve) and viscoplastic (εvp) strains, as follows:
ε

tot

= ε

+ε

ve

[18]

vp

The viscoelastic strain is the sum of both the linear viscoelastic strains and the strains due
to the microcracks. The time dependent viscoelastic behavior can be simplified into a linear
elastic solution through the pseudostrain using convolution integral, as follows:

ε

R

1

=

E

where ε

R

is the pseudostrain, E

R

R

∫E ( t - τ )

∂
ε
∂
τ

dτ

[19]

is the reference modulus, E(t) is the linear viscoelastic
R

relaxation modulus, t is time and τ is the time variable of integration. When E is taken as unity,
εR is equal to the linear viscoelastic stress (σ) as follows:

σ = ε

R

123

E

R

[20]

a)

b)

d) ACCEPTED
c) REMOVED

Figure 36: Push-Pull Test: a) Custom made gluing jig, b) AMPT fixture, c) Non-accepted
tests (end failure), d) Accepted tests (mid failure)
However, this simplified equation does not consider the nonlinear behavior due to the
continuum damage of micro cracks. Therefore, the stress-strain behavior of the viscoelastic
material is determined based on the time dependent damage growth. The following equations
utilize the pseudostrain energy density function and the damage parameter (S):

σ = IC ( S ) ε

R

=

∂
W
∂
ε

124

R

R

[21]

W

dS
dt

R

=

I
2

C (S ) ε

= (- ∂
W

R

R

2

∂)
S

[22]

α

[23]

R

where W is the pseudostrain energy density function, C(S) is the pseudostiffness, I is the initial
stiffness parameter (0.9-1.1), which eliminates the sample to sample variability, α is the material
constant related to the damage growth.
Independent of loading history, testing mode (stress or strain controlled), magnitude or
rate of the loading, and testing temperature, all C versus S curves for each PP test for the same
mixture should collapse on a single curve. Once C versus S from different tests (on the same type
of WMA) seems to collapse on a single curve, an exponential best fit line is as follows:

C = exp( aS

b

)

[24]

where a and b are the constants defining the best fit. However, this C versus S curves are not
sufficient to determine the fatigue performance. The number of cycles to failure (Nf) was
determined based on the 50 % reduction in pseudostiffness (i.e., C=0.5).
Push-Pull Test Result and Discussions
As it was stated before, the push-pull tests were performed at strain controlled mode at 10
Hz frequency at both 10°C and 20°C. Although the best fitted C versus S curves were given in

Figure 37, the individual C versus S curves for each test were plotted from Figure 60 to Figure 64 in
APPENDIX C.
125

As shown in Figure 38, the fatigue cracking performance of the tested mixtures under
constant strain (300 microstrain) were compared at different temperatures (5, 10, 15, 20, 25 and
30°C) using the VECD theory. As the temperature increased, the fatigue resistance of the asphalt
mixtures increased (as expected). In addition the fatigue performance of the WMA mixtures at
20°C calculated under different traffic loads (strain levels) (100, 200, 300, 400 and 500
microstrain). As shown in Figure 38 and Figure 39, the best fatigue performance was observed at
3% - 15psi and 4% -17.5psi mixtures. The worst performance was in 5% - 20psi, which is
consistent with coating analysis. It was also mentioned during FN data analysis that 5% - 20psi
might have “exposed” uncoated aggregates causing more friction between the aggregates. These
“exposed” aggregates can lead to increased fatigue cracking potential. On the other hand, the 1%
- 10psi and 2% - 12.5psi samples also showed worse performance as compared to 3% - 15 psi
and 4% - 17.5psi mixes. This could be due to trapped moisture bubbles in 1% - 10psi and 2% 12.5psi since these samples had small foam bubbles. Therefore, it can be concluded that the
“optimum” water content and air pressure is around 3% -15psi and 4% -17.5 psi. The number of
cycles to failure at 300 microstrain, 10 Hz, and 20°C versus the foam quality indicators (i.e., ER,
HL, FI, D50 and SAI) was plotted in Figure 40 and the statistical analysis were studied
individually. Although the expansion ratio (ER) increases as the water content and air pressure
increases, the there is no trend in between ER and fatigue resistance. As shown in Figure 40a,
3% - 15 psi showed the best performance as compared to the other mixes. In addition, the other
graphs (i.e., HL, FI, D50, SAI) were analyzed from Figure 40b to Figure 40e, in which the best
fatigue performance was observed at 3% - 15psi and 4% -17.5psi mixtures. The statistical

126

analysis, in which the correlation and linear regressions studied in Table 13, indicated that the
lack of linearity. Therefore, polynomial regression is fitted to the data as given in Table 13. The
2

coefficient of determination (R ) and significance of the polynomial fit are relatively high for
ER, FI and D50. However, the significance of the relation with SAI and HL is relatively low. As
a result, it is crucial to consider foam quality in the WMAs design for the optimum performance.
1

1%-10psi
2%-12.5psi
3%-15psi
4%-17.5psi
5%-20psi

0.9
0.8
0.7
C

0.6
0.5
0.4
0.3
0.2
0.1
0
0.E+00

2.E+05

4.E+05

6.E+05
S

Figure 37: C versus S curves of different WMAs

127

8.E+05

1.E+06

8.00E+04

6.00E+04

4.00E+04

1.00E+05

2.00E+04

5
10
Temperature (°C)
15

Figure 38: Number of cycles to failure at 300 microstrain.
128
20
25

1.65E+05
1.62E+05

1.36E+05
1.24E+05

1%-10psi
2%-12.5psi
3%-15psi
4%-17.5psi
5%-20psi

1.20E+05

9.23E+04

6.10E+04
5.59E+04
7.43E+04
7.29E+04
4.15E+04

1.40E+05

3.46E+04
3.18E+04
4.22E+04
4.14E+04
2.35E+04

1.60E+05

2.67E+04
2.45E+04
3.26E+04
3.19E+04
1.82E+04

1.80E+05

2.93E+04
2.69E+04
3.57E+04
3.50E+04
1.99E+04

4.62E+04
4.24E+04
5.63E+04
5.52E+04
3.14E+04

Nf
2.00E+05

0.00E+00

30

1.00E+04

100
200
300

Strain

Figure 39: Number of cycles to failure at 20°C
129
400

3.48E+03
3.19E+03
4.24E+03
4.15E+03
2.36E+03

1.00E+05

9.49E+03
8.70E+03
1.16E+04
1.13E+04
6.45E+03

1.00E+06

3.46E+04
3.18E+04
4.22E+04
4.14E+04
2.35E+04

4.86E+06
4.45E+06
5.92E+06
5.80E+06
3.30E+06

1.00E+07

2.15E+05
1.97E+05
2.62E+05
2.56E+05
1.46E+05

Nf
1.00E+08
1%-10psi

2%-12.5psi

3%-15psi

4%-17.5psi

5%-20psi

1.00E+03
500

5.E+04
a)

Nf

4 - 17.5
5 - 20

2.E+04

2 - 12.5

1 - 10

3.E+04

3 - 15

4.E+04

1.E+04
0.E+00
2

2.2

2.4
2.6
Expansion Ratio (ER)

2.8

3

5.E+04
3 - 15

b)

4 - 17.5

5 - 20

2.E+04
1.E+04

1 - 10

3.E+04

2 - 12.5

Nf

4.E+04

0.E+00
0
5.E+04

20

60
Half-Life (HL)

80

100

c)
1 - 10

2.E+04

2 - 12.5

3 - 15

3.E+04

4 - 17.5

5 - 20

4.E+04
Nf

40

1.E+04
0.E+00
100

150

200

250
300
350
400
450
500
Foam Index (FI)
Figure 40: The comparison of Number of Cycles to failure at 300 microstrain, 10 Hz, 20°C
and Foam Binder Quality Parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam
Index (FI), d) D50, e) Surface Area Index (SAI)

130

Figure 40 (cont’d)
5.E+04

Nf

4 - 17.5

1.E+04

3 - 15

2.E+04

2 - 12.5

1 - 10

3.E+04

5 - 20

d)
4.E+04

0.E+00
0

0.2

0.4

D50

0.6

0.8

1

5.E+04

Nf

1 - 15

2.E+04

2 - 12.5

3.E+04

3 - 15

4 - 17.5

5 - 20

e)
4.E+04

1.E+04

0.E+00
150

200

250

300
350
400
450
500
Surface Area Index (SAI)
Table 13: Statistical analysis between Number of Cycles to failure at 300 microstrain, 10
Hz, 20°C and Foam Binder Quality Parameters
Correlation

Linear Regression

Polynomial
Regression

Nf vs. ER

-0.200

Kendall's
Tau
Significance
(1- pvalue)
0.592

Nf vs. HL

0.200

0.592

0.264

0.070

0.170

0.335

0.112

Nf vs. FI

0.200

0.592

0.523

0.273

0.350

0.526

0.277

Nf vs. D50

-0.200

0.592

0.478

0.228

0.317

0.688

0.473

Nf vs. SAI

0.000

0.5

0.387

0.150

0.253

0.0837

0.007

Kendall
's Tau

R

Regression
Significance
(1-pvalue)

R

R

0.602

0.363

0.412

0.881

0.776

2

R

131

2

Tensile Strength Ratio (TSR)
Moisture susceptibility is another major concern related to the WMA pavements.
Moisture damage can occur due to the loss of cohesion within the binder film, the failure of
adhesive bond between the binder and the aggregate and the degradation of aggregates. There are
various test methods for evaluation of moisture susceptibility of mixtures. These include
Tunnicliff & Root (ASTM D4867), Lottman (ASTM D4867), Immersion Compression, Boiling
water test (ASTM D3625), Texas freeze-thaw pedestal test, Marshall Stability index,
Environmental Conditioning System (ECS), Standard Method of Test for Resistance of
Compacted Hot Mix Asphalt (HMA) to Moisture-Induced Damage (AASHTO T283).
In this study, AASHTO T283, which is a typical method for determining the moisture
susceptibility of HMA pavements, was performed. It was also suggested methodology for WMA
mixtures in NCHRP Reports 691 and 714. The test measures the change of the tensile strength
of the mixtures resulting from the freeze-thaw cycle of the conditioned mixes, relative to the
unconditioned mixes.
TSR Test Procedure
The sample preparation for the AASHTO T283 is significantly different than the previous
performance tests discussed. For each mixture, six to eight samples are prepared to achieve the
target air void range (7.0% ± 0.5%). As soon as the loose mixture is prepared, it is left in the
room temperature for 2±0.5 hours in order to cool down. Then, the loose mixture is placed into a
conventional oven at 60±3°C for 16 hours. Finally, the sample is transferred to another oven at
CT±3ºC for 2±10 minutes hours before the compaction. The mixture is compacted to at a size of
150 mm diameter to 95±5 mm height (see Figure 41). After overnight cooling, the maximum
132

bulk specific gravity of the sample should be checked and grouped into two subsets (three
samples each), which has approximately same average air voids.

Force at
Failure, Pt
σt=Pt/(пDt)

σxx
σxx

Vertical Def.
At Failure, yt

σyy

Vertical
Deformation

Tensile
Strength
σyy

Tensile Strain
At Failure
Figure 41: Indirect tensile strength test specimen and stress distribution
The first subset, called as unconditioned set, is left for drying and stored at room
temperature. This set should be kept at 25 ± 0.5ºC for 2 h ± 10 min before testing. The second
subset, called as conditioning set, saturated in the range of 70% to 80% in a vacuum container.
The saturated samples are covered with a plastic film and placed into separate sealed bags with
10 ± 0.5 ml of extra water. These bags are kept in an environmental chamber at –18 ± 3°C for a
minimum of 16 hours. After the freezing cycle, the samples are rapidly transferred to a water
bath at 60± 1°C for 24±1 h without the seal bags and plastic films. Following the thaw cycle,
the samples are transferred to a water bath at 25 ± 0.5°C for 2 h ± 10 min. The tests are
133

conducted using the Material Testing System (MTS) and loaded at a constant rate (50 mm/min).
The maximum load achieved at failure is used to compute the tensile strength, as follows:

S

t

2P

=

π tD

[25]

where St is tensile strength in psi and P is the maximum load in lbf. D and t are the sample
diameter and thickness in inches, respectively. Tensile Strength Ratio (TSR) is defined as the
ratio of conditioned to unconditioned indirect tensile strength of mixes, as follows:
S
TSR =

S

1

× 100 %

[26]

2

where S1 and S2 are the average conditioned and unconditioned tensile strength, respectively. If
this ratio is greater than 80 percent, the mixtures are not considered to be susceptible to moisture
damage. If low TSR values for WMA mixes are obtained, antistripping agents such as hydrated
lime and liquid additives can be used.
TSR Test Result and Discussions
The samples for AASHTO T283 were prepared for each injected water content and air
pressure combinations. The detailed data analysis for each mixture is available from Table 29
and Table 33 in APPENDIX D.
The conditioned and unconditioned strength of the mixtures were given in Figure 42. The
change in the conditioned strength did not follow any trend with the change of the mixture. On
the other hand, as shown in Figure 42b, the unconditioned strength of the mixtures decreased as
134

the water content and air pressure increases. The highest strength in the conditioned set was
achieved in 3%-15psi mixture. As stated before, the lower water content and air pressure
injected foamed binder was anticipated to be composed of relatively small bubbles and the
collapse of the foamed binder takes longer. Therefore, it was observed that these mixtures had
better aggregate coating, which also increased the strength between the aggregate and binder
bonds. Thus, the tensile strength of these mixes was relatively high. On the contrary, the higher
water content and air pressure injected foamed binder was anticipated to be composed of
relatively large bubbles and foamed binder collapse more rapidly. Therefore, the bubbles
evaporate quickly from the mixture, as well as resulting poor coating during mixture production.
The poor bond between the aggregate and binder resulted relatively lower tensile strength. The
statistical analysis between the sample strength and the injected water content and air pressure
relation is given in Table 14. The Kendall’s tau coefficient is 0.8 and its significance is equal to
0.958 for the unconditioned strength. This indicated that there is strong correlation between the
unconditioned strength of the mixtures and foamed binder. In addition, the linear regression
fitted well for this relation. Therefore, the coefficient of determination and significance of the
regression are relatively high. However, both Kendall’s tau and linear regression were poor for
the conditioned strength.
TSR value is a function of both the conditioned and unconditioned tensile strength of
mixtures. As given in Figure 43, the trend in the TSR increased with the increase in the water
content and air pressure of the binder, mixture prepared. The statistical analysis between the TSR
and foamed binder is given in Table 14. The Kendall’s tau correlation and significance was
relatively low, the coefficient of determination for the linear regression is 0.78 and its
135

significance is 69%. However, the mixture prepared with 3%-15psi foamed binder had the
highest moisture resistance as well as the highest conditioned and unconditioned strengths. The
outcomes of the moisture susceptibility tests were very similar to the fatigue cracking. The
optimum mix design was observed to be 3%-15psi.
The unconditioned and conditioned tensile strength versus the foam quality indicators
(i.e., ER, HL, FI, D50 and SAI) was plotted in Figure 44 due to high variability in the strength
only in order to briefly investigate these relations individually. As shown in Figure 44a and d,
Expansion Ratio (ER) and D50 increase with the increase in the injected water content and air
pressure of the binder as the unconditioned strength decreases. On the contrary, half-life (HL),
foam index (FI), and Surface Area Index (SAI) increase with the decrease in the injected water
content and air pressure of the binder as the unconditioned strength increases as shown in Figure
44b, Figure 44c and Figure 44e. However, the conditioned strength has a different trend as stated
earlier, where the highest strength in the conditioned set was achieved in 3%-15psi mixture with
respect to the other WMAs. In addition, tensile strength ratio (TSR) versus the foam quality
indicators (i.e., ER, HL, FI, D50 and SAI) was plotted in Figure 45, in which the optimum mix
was determined to be 3%-15psi mix. It can be also concluded that similar to the performance
tests discussed previously, the moisture resistance of the WMA mixtures can also be controlled
with foamed binder quality.

136

500

Conditioned
Unconditioned

Tensile Strength(psi)

400
300

200
100
0
1% - 10psi

2% - 12.5psi

3% - 15psi

4% - 17.5psi

5% - 20 psi

Figure 42: Unconditioned and conditioned tensile strength

Tensile Strength
Ratio (%)

120
110
100
90
80
70
60
50
1% - 10psi

2% - 12.5psi

3% - 15psi

4% - 17.5psi

5% - 20psi

Figure 43: Tensile Strength Ratio (%)
Table 14: Statistical analysis of unconditioned strength, conditioned strength and TSR with
respect to water content and air pressure

Correlation

Linear Regression

Kendall's
Tau
Uncondotioned Strength vs.
Water Content/Air Pressure
Condotioned Strength vs.
Water Content/Air Pressure
TSR vs. Water Content/Air
Pressure

Kendall's
Tau
Significance
(1- pvalue)

R

R

Regression
Significance
(1-pvalue)

-0.8

0.958

0.912

0.832

0.731

0.2

0.592

0.879

0.772

0.683

0.2

0.592

0.884

0.782

0.690

137

2

a)

400
300
200
150

4 - 17.5

2 - 12.5

250

5 - 20

3- 15

350

1 - 10

Tensile Strength (psi)

450

Conditioned

100

2.8

3

3 - 15

1- 10

400

2.4
2.6
Expansion Ratio (ER)

2- 12.5

450
b)
400
350
300 5- 20
250
200
150
100
0
450

Conditioned

20

Unconditioned

40
60
Half-Life (HL)

80

100

c)

300

3- 15

350

4 - 17.5

200
150

Conditioned

100
50

150

1- 10

250

2 - 12.5

5- 20

Tensile Strength (psi)

2.2

4- 17.5

Tensile Strength (psi)

2

Unconditioned

Unconditioned

250
350
450
Foam Index (FI)
Figure 44: The comparison of Unconditioned/Conditioned Tensile Strength and Foam
Binder Quality Parameters: a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index
(FI), d) D50, e) Surface Area Index (SAI)

138

Figure 44 (cont’d)
e)

400
300
200
150

Conditioned

100
0

4- 17.5

2- 12.5

250

5- 20

3- 15

350

1- 10

Tensile Strength (psi)

450

Unconditioned

0.2

0.4

0.6

0.8

1

D50
400

200
150

Conditioned

100
150

200

250

300
350
400
Surface Area Index (SAI)

139

1 - 10

250

2 - 12.5

5 - 20

300

3 - 15

350

4 – 17.5

Tensile Strength (psi)

450

Unconditioned
450

500

120
a)

20

4- 17.5

40

2 - 12.5

60

1 - 10

TSR

80

5 - 20

3 - 15

100

0
2
120

2.2
b)

2.8

3

80

100

3 - 15

100
5 - 20

80

20

1 - 10

40

2 - 12.5

60
4 - 17.5

TSR

2.4
2.6
Expansion Ratio (ER)

0
0

20

40
60
Half-Life (HL)

120
c)
5 - 20

80

20

1 - 10

40

2 - 12.5

60

4 - 17.5

TSR

3 - 15

100

0
50

150

250
Foam Index (FI)

350

450

Figure 45: The comparison of Tensile Strength Ratio (TSR) and Foam Binder Quality
Parameters a) Expansion Ratio (ER), b) Half-Life (HL), c) Foam Index (FI), d) D50, e)
Surface Area Index (SAI)

140

Figure 45(cont’d)
120
d)

40

4 - 17.5

2 - 12.5

60
1 - 10

TSR

80

5 - 20

3 - 15

100

20
0
0

0.2

0.4

0.6

0.8

1

D50
120
e)
5 - 20

80

40
20

1 - 10

2 - 12.5

60
4 - 17.5

TSR

3 - 15

100

0
150

200

250

300
350
400
Surface Area Index (SAI)

141

450

500

CHAPTER 6

INVESTIGATION OF FOAM DISIPATION USING SYNHROTRONBASED X-RAY MICROTOMOGRAPHY

The quality of the foamed binder depends various factors such as the binder type, grade
and modification, the foaming technology used, amount of water, temperature etc. AFCT
analysis, which was verified with X-ray Microtomography (XRM) imaging technique, proved
the foamed binder parameters depends on the injected water content and air pressure in
CHAPTER 4. In order to further investigate the effects of some of the other characteristics such
as the binder type and foaming technology on the generation and evolution of the foam, XRM
imaging technique was also utilized to frozen foamed binder samples. Change in overall volume
of the moisture bubbles as well as the size distribution of the bubbles was computed using 3D
image processing methods. Variation of volume and size distributions of the bubbles in different
types of asphalt binders as well as a mastic specimen was discussed in this chapter.
MATERIALS AND METHODS
Four different types of binders and a mastic sample were investigated by XRM as given
in Table 15. Two of the selected binders were unmodified binders with PG grades of 58-28 (the
binder source was different than the binder used in CHAPTER 4) and 64-22. One of the other
binders was a polymer (Elvaloy) modified binder and the last one was a Crumb Rubber (CR)
modified binder. The CR modified binder had 15 % CR (by weight of binder), which was

142

prepared using a technique called Wet Process. The binder was mixed with CR at 190 oC using a
mixer at a rate of 2000±100 rpm (revolutions per minute) for 60 ± 5 minutes.
Table 15: Description of the specimens utilized in this study.
Identification
PG58-28F
PG70-22F
PG64-22F
PG70-22CRMF
PG58-28A

Binder Modification
Unmodified
Elvaloy
Unmodified
Crumb Rubber Wet Process
Unmodified

Warm Mix Method
Laboratory Foaming
Laboratory Foaming
Laboratory Foaming
Laboratory Foaming
Advera (Synthetic Zeolite)

PG70-22A

Elvaloy

Advera (Synthetic Zeolite)

PG70-22CRMA
PG58-28SANDF

Crumb Rubber Wet Process
PG 58-28+ fine aggregate

Advera (Synthetic Zeolite)
Laboratory Foaming

Each binder was foamed using two methods: (i) a laboratory foaming device (Wirtgen
WLB 10) and (ii) Advera, a synthetic Zeolite additive (details were given in CHAPTER 2). The
nozzle-based foamed binders were prepared by injecting air (5%), water (1.5%) and asphalt
binder (93.5%) at 160°C.The zeolite based foamed binders were prepared with Advera at 120°C.
4.5% Advera by weight of the binder was added to the asphalt binder, where the crystallized
water in the Advera transforms into moisture bubbles and foams the binder. The same binder
foaming procedure was followed independent of the independent of the binder type.
In order to observe the effect of foaming on the microstructure when aggregates are
present, mastic specimens were prepared by mixing the foamed PG58-28 binder with fine
aggregates retained on #200 sieve (passing #100 sieve). Binder content for the mastic specimens
were selected to be 10% by weight of the mix. The mixing temperature was approximately 150°C
because of rapid cooling of the foamed binder after the foaming process.
143

3D IMAGING USING SYNCHROTRON BASED X-RAY MICROTOMOGRAPHY
The 3D image acquisition of the specimens was done at the 5-BM-C Microtomography
beam line at the Advanced Photon Source (APS) facility in Argon National Laboratory (ANL),
similar to the analysis in CHAPTER 4. In this part of the research, 20keV parallel beam was
utilized, which provided a volume scan of 7 mm diameter, 7 mm tall cylinders. This size was the
maximum size sample that can be scanned with the default camera and optical system (i.e., the
X-ray detector). The final image size was 1299 by 1299 by 1299 pixels. This corresponded to 7
mm/ 1299 pixels = 0.0054 mm/pixel (5.4 micron) image resolution. Same procedure explained in
CHAPTER 4 used in the analysis of 3D XRM images in this chapter.
The 7 mm diameter and 7 mm tall specimens were appropriate for this study because the
maximum sizes of the bubbles were much smaller (~0.4mm) than the overall sample size, as
shown in Figure 50 and Figure 51. The maximum size of the bubbles was larger in binders made
with Zeolite (Figure 52); however, it was still smaller than the overall size of the specimen. As
shown in Figure 56 and Figure 57, the bubbles in mastic specimens were also very small
(maximum size ≈ 0.2mm), therefore the 7 mm diameter tubes were appropriate in this study.
BINDERS PREPARED USING DIRECT FOAMING
Figure 46 shows the 3D visualization of the labeled (with different colors) bubbles for
two of binders frozen at different times. These views were generated from XRM images after
thresholding and labeling operations described in CHAPTER 4. Figure 46qualitatively shows the
reduction in bubble size and density in about 10 minutes for both of the binder specimens PG5828F and PG70-22F. However, the rate of reduction in the bubble density and size seems to be
slower in PG70-22F as compared to PG58-28F specimen. This phenomenon might be due to (i)
144

smaller diffusion coefficient of PG 70-22 binder as compared to PG 58-28 and PG 64-22 binders,
and/or (ii) PG70-22 is a stiffer binder than others, which perhaps slows down the movement of
the bubbles within the binder. The reduction in size of the bubbles is quantified in the following
section.

6

Z (mm)
Z (mm)

6
55
44
33
22
1
1
0

Z (mm)
Z (mm)

7
7
6
6
5
5
4
4
3
3
2
2
1
1
0
0
0
0 2
2

0
0
0

6

4
4

6

4

6
6

X (mm)
X (mm)

4
2 Y (mm)
Y (mm)

2

2
0

0

6

6

4

4
4

X (mm)
X (mm)

(a) t= 0.4 min (PG58-28F unmodified binder)
a) t=0.4min (PG58-28F unmodifiedbinder)

6
6

2
2

0

0

4

YY (mm)
(mm)

(b)10 min (PG58-28F unmodified binder)
b) 10min (PG58-28F unmodified binder)
7

7
7
6
6
5
5
4
4
3
3
2
2
1
1

Z Z (mm)
(mm)

7
66
55
44
33
22
11

Z Z (mm)
(mm)

0
0

2

0
0

2

2

4

4

4
6
X X (mm) 6
(mm)

00

2
2

6
6

0

4

6

2

2

4

4

Y (mm)
Y (mm)

4

X (mm)
X (mm)

6

6

2
0

0

2

6

4

Y (mm)
Y (mm)

(d) t=13min
(PG70-22F
d) t=13 min modified binder) Elvaloy
(PG70-22F Elvaloy
polymer polymer
modified
binder)
Figure 46: 3D temporal view of bubbles in two asphalt binder specimens: (a) & (b) PG5828 unmodified binder and (c) & (d) PG70-22 Elvaloy polymer modified binder.
(c) t=1 (PG70-22F Elvaloy modified
c) t= 1minmin (PG70-22F Elvaloy modified
binder)
binder)

145

Temporal change in the total volume of the bubbles
Figure 47 shows the reduction in the volumetric proportion of the bubbles with time. The
volumetric proportions of the bubbles were calculated from the 3D XRM binary images using
the Matlab® algorithms, explained in CHAPTER 4. The volumetric percentage of the bubbles is
the ratio of the total volume of the bubbles to the total scanned volume. As shown in Figure 47a,
rate of change of volumetric percentage of bubbles varies in different binders. Figure 47b shows
that curves follow approximately linear trend when plotted in log-log scale. In order to quantify
the change in the total bubble volume, a parameter called moisture dissipation index (MDI) is
defined to quantify the speed of dissipation of moisture within the binder. The MDI is anticipated
to correlate well with the amount of time that takes for moisture to dissipate in the field after
foam mix asphalt is placed. The MDI is defined as the slope of the linear best-fit line equation
fitted to logarithm of volumetric bubble percentage versus time graph (Figure 47b) as follows:
log ( V ) = - ( c log( t ) + c )
1
2

[27]

MDI = c1
where V = volumetric percentage of bubbles (%) at any time t (min), c1 and c2= are the fit
coefficients as shown in Figure 47. Figure 48 shows the MDI values for different binders
investigated in this study. As seen from the Figure 48, MDI values are higher for unmodified
binders as compared to the polymer and crumb rubber modified binders. As it was hypothesized
before, this phenomenon can be attributed either to the (relatively high) stiffness of the polymer
modified binders or (possibly to) their low diffusion coefficient.

146

On the other hand, the volume of the bubbles in the crumb rubber modified binder
(PG70-22CRMF) seemed to have increased with time. While this may be because of a
measurement or sampling error, it may also be an artifact of the existence of crumb rubber (CR)
particles in the binder as shown in Figure 49. It is hypothesized that, while mixing within the
foaming nozzle, the CR particles caused the bubbles to be even smaller than the resolution (i.e.,
5.4 micron) of the XRM images. The bubbles in CR specimens might have diffused and
disappeared in a much longer period of time.

Volumetric portion of
bubbles (%)

1.0%

PG64-22F
PG 70-22CRMF
PG 70-22F
PG 58-28F

0.8%
0.6%
0.4%
0.2%
0.0%
0

10

20
Time (min)

30

40

50

log(Volumetric portion of
bubbles)

-1.5
-2.0

y = -0.18x - 2.0504

-2.5

y = 0.158x - 3.0802

-3.0
y = -0.60x - 2.2997

PG64-22F
PG 70-22CRMF
PG 70-22F
PG 58-28F

-3.5
-4.0

y = -1.15x - 2.23

-4.5
-0.5

0.0

0.5
1.0
1.5
2.0
log(Time) (min)
Figure 47: Reduction in the volume of bubbles with time; (a) in linear x-y scale and (b)
logarithmic x-y scale.
147

Moisture Dissipation Index (MDI)

1.50

Unmodified Binders

1.30
1.10
0.90
Polymer/CR Modified
Binders

0.70
0.50
0.30
0.10
-0.10

PG58-28F

PG64-22F

PG70-22F

PG70-22CRMF

-0.30

Figure 48: Moisture Dissipation Index (MDI) values for different binders.

Z (mm)
Z (mm)

7mm

66
55
44
33
CR particles
22
11
00
0
66
0 1
1 2
44
2 3
3 4
2
4 5
2 Y (mm)
5
Y (mm)
00
X (mm)
X (mm)
Figure 49: (a) 2D slice XRM image of PG70-22CRMF and (b) 3D visualization of crumb
rubber particles.
Temporal change in the size distribution of the bubbles
In addition to the total volume of the bubble, size distribution of the bubbles was also
analyzed. In order to calculate the size distribution, first, equivalent diameters of each bubble
were computed using the labeled binary XRM images. The equivalent diameter is herein defined
148

as the diameter of the equivalent sphere that has the same volume as a given bubble. Then
cumulative frequency distribution of the bubble sizes was computed. Figure 50 and Figure 51
show the change in bubble size distribution with time for the specimens utilized in this study. In
general, the size of the bubbles reduces with time and they become more uniformly graded.
Figure 50a and Figure 50b, the size distribution in t=30 min is not shown because there were
only few bubbles left in the binder and illustration of size distribution would not have been
realistic. In Figure 51c and Figure 51d, size distribution of the bubbles in PG70-22CRMF seems
to indicate that at 43 min, the sizes are larger (in general) than t=32 min. This may be due to an

0.20

100%
80%

60%
40%

0 min
10 min

20%

0.15

Median Diameter
Mean Diameter

0.10
0.05
0.00

0%
0

0.1 0.2 0.3 0.4
Bubble Size (mm)

PG 58-28F
(0min)

0.5
0.25

100%

Percent Finer (%)

b)
Size (mm)

a)

c)

80%

0.20

Size (mm)

Percent Finer (%)

error in the sampling in PG70-22CRMF.

0.15

60%

1 min
10 min
30 min

40%
20%

d)

PG 58-28F
(10min)
Median Diameter
Mean Diameter

0.10
0.05
0.00

0%
0

0.05 0.1 0.15 0.2
Bubble Size (mm)

PG64-22F PG64-22F PG64-22F
(1min) (10min) (30min)

0.25

Figure 50: The change in size distribution of bubbles over time for: (a) & (b) PG58-28F and
(c)&(d) PG64-22F.

149

Size (mm)

80%
60%

t=3min
t=13min
t=33min

40%
20%
0%
0.05

100%

Percent Finer (%)

0.20

a)

0.25
0.20
t=1min
t=32min
t=43min

20%

0.05
PG 70-22F PG 70-22F PG 70-22F
(3min)
(13min)
(33min)

80%
40%

Median Diameter
Mean Diameter

0.10

0.15
0.25
Bubble Size (mm)

c)

60%

0.15

b)

0.00

Size (mm)

Percent Finer (%)

100%

d)

Median Diameter
Mean Diameter

0.15
0.10
0.05
0.00

0%
0

0.2
0.4
Bubble Size (mm)

PG 70-22
CRMF
(1min)

0.6

PG 70-22
CRMF
(2min)

PG 70-22
CRMF
(32min)

Figure 51: Change in size distribution of the bubbles with time for binder (a) & (b) PG7022 and (c)&(d) PG70-22CRMF.
BINDERS PREPARED USING SYNTHETIC ZEOLITE
It was observed that binders foamed using Zeolite additive (Advera) exhibited quite
different bubble size distribution as shown in Figure 52. Bubbles were larger and more scarce.
This indicates that different foaming methods can exhibit different bubble volume and size
distribution. Figure 53 shows the change in the overall volume of the bubbles in specimens
prepared with synthetic Zeolite. Typically, the rate of dissipation of moisture was larger as
compared to the foamed specimens. Figure 54 shows the comparison between the MDI values of
specimens prepared using direct foaming and Zeolite additive, where MDI values are larger in
binders prepared with Zeolite. This (with the limited data available) indicates that the moisture

150

disspates faster in Zeolite as compared to direct foaming process. In these figures, PG58-28A is
not shown because no air bubble was observed within the XRM images of this specimen. This
was attributed to the insufficient mixing of the Zeolite during sample preparation. An example
image of PG58-28A is shown in Figure 55, where a clear bright spot in the center is visible. This
bright area is probably the Zeolite additive, which is not mixed with the binder.

Z (mm)
Z (mm)

Z (mm)
Z (mm)

7
7
6
6
5
5
4
4
3
3
2
2
1
1
0
00
0 22

66
55
44
33
22
11
00

0
0

6
6
2 3
4
4
34 5
4
2
4
4
4
2
4
2
6 0 2 Y (mm) X (mm) 5 6 0
Y (mm)
6 0
X (mm)
6 0
XX (mm)
(mm)
Y (mm)
Y (mm)
0
t=17 min
t=0 min
t=17 min
t=0 min
Figure 52: 3D XRM image of PG70-22CRMA binder foamed using Zeolite additive.
mimmm
min

151

11 2

Volumetric portion of
bubbles
(%)

4.0%
3.5%
3.0%
2.5%
2.0%
1.5%
1.0%
0.5%
0.0%

PG 70-22CRMA

0

10

20
Time (min)

log(Volumetric portion of
bubbles)

30

40

PG 70-22CRMA
PG 70-22A
Linear (PG 70-22CRMA )
Linear (PG 70-22A )

0.00
-1.00
-2.00
-3.00
-4.00

PG 70-22A

y = -0.484x - 1.9529

y = -1.1837x - 3.6079

-5.00
-6.00
-1.50

-1.00

-0.50
0.00
0.50
1.00
1.50
log(Time) (min)
Figure 53: Reduction in the volume of bubbles with time in specimens prepared with
Zeolite.

Moisture Dissipation
Index (MDI)

1.50
Foam
Zeolite

1.00
0.50
0.00
-0.50

PG70-22
PG70-22CRM
Figure 54: Moisture Dissipation Index (MDI) comparison of binders prepared with foam
and Zeolite binders.
152

Figure 55: 2D XRM image slice of PG58-28A, which was prepared with Zeolite.
INVESTIGATION OF FOAMED MASTICS
In addition to the asphalt binders, mastics (asphalt binder + fine aggregate) were also
included in the research program. It is hypothesized that morphology and type of fine aggregate
can play a crucial role in the moisture retention and dissipation in the mixture. Similar to the
procedure for binders, the mastic specimens were poured into the 7 mm diameter tubes, frozen
using liquid nitrogen and scanned using XRM. Figure 56 shows example XRM image slices of
these two specimens as well as 3D visualization of the bubbles. It was observed that large
elongated voids appeared in mastic specimens. This may be due to (i) fracturing during freezing
using liquid nitrogen or (ii) distortion of the bubbles because of the fine aggregate grains. Figure
57 illustrates the change in the size distribution as well as mean and median size of the bubbles
where reduction in bubble size was observed between times 6 min and 34 min. However, as
shown in Figure 58, overall volume of the bubbles actually increased with time. It may be a
sampling problem or non-homogenous mixing. Therefore, further specimen preparation and
testing are needed to better understand the actual behavior.

153

(b) 2D slice of mastic specimen (t = 34min)

(a) 2D slice of mastic specimen (t = 6min)

1
0
0

1
0
0

2

3 4

4
X (mm)

6

5 6

2
0 1
Y (mm)

(c) Voids of the mastic at t=6 min.

1 2

3

4

X (mm)

4

5 6

0 1

5

6

2 3
Y (mm)

(d) Voids of the mastic at t=34 min.

Figure 56: (a) & (b) 2D slices from XRM images and (c) & (d) 3D visualization of pores of
foamed asphalt mastics.

154

a)
80%
Size (mm)

Percent Finer (%)

100%

60%

40%
t=6min
t=34min

20%
0%
0

0.1
0.2
Bubble Size (mm)

0.135
0.130
0.125
0.120
0.115
0.110
0.105
0.100

Median
Diameter
Mean
Diameter

b)

PG 58-28
PG 58-28
SANDF(6min) SANDF(34min)

0.3

-2.05

1.0%
0.8%

PG 58-28SANDF

log(Volumetric portion
of bubbles)

Volumetric portion of
bubbles (%)

Figure 57: Change in (a) size distribution and (b) mean & median size of the bubbles in
PG58-28SANDF

-2.10

0.6%

Linear (PG 5828SANDF)

-2.15

0.4%
0.2%

PG 58-28SANDF

-2.20
y = 0.0619x - 2.1678

-2.25

0.0%
0

20
Time (min)

40

-2

-1
0
1
log(Time) (min)

2

Figure 58: Change in the overall volumetric percentage of the bubbles with time in
specimen PG58-28SANDF.

155

CHAPTER 7

CONCLUSION AND RECOMEDATIONS
SUMMARY
Most of the current knowledge on foamed Warm Mix Asphalt (WMA) is based on
limited empirical studies and there is a significant lack of understanding of the behavior of the
foamed binder used in WMA pavements. Understanding the WMA foamed binder
characteristics, which affect the mechanical behavior of pavements, is crucial to accurately
predict and improve their long-term performance. Therefore, there is a growing need for
understanding the WMA pavements’ behavior, from binder production to mixture performance.
As of today, the importance of foam quality has never been considered in WMA
pavement design. It has been hypothesized that the quality of the foamed binder depends various
factors such as the binder type, grade and modification, the foaming technology used, amount of
water, air pressure and temperature. The quality of the foamed binder plays a crucial role during
mixing, laying and compaction stages of WMA pavement production. However, there is no study
that came up with foamed binder quality indicators for WMA applications. Even though there
are some candidate foamed binder quality indicators that were developed for stabilized base
applications (i.e., Expansion Ratio (ER), half-life (HL), Foam Index (FI)), their relation to WMA
performance have never been investigated. Furthermore, there is no available method to
precisely measure the potential foamed binder quality indicators. Therefore, an accurate and
repeatable procedure is needed for the measurement of reduction in height of foamed asphalt in
order to calculate these indicators. In this research, an automated test device, called as Asphalt
156

Foam Collapse Test (AFCT), was developed to measure the reduction in the height of the
foamed binder over time. Once the reduction in height of the asphalt foam with time is measured,
the foamed binder’s quality parameters can be precisely calculated. Tracking the change of the
foam structure with time helps to understand the workability of the binder as well as the physical
properties of the binder such as the residual water. In addition, in this research, new parameters
(i.e., Bubble Size Distribution (BSD) and Surface Area Index (SAI)) are introduced to assess the
quality of the foamed binder. These indicators are potentially very important parameters, since
they directly relate to the ability of the foamed binder to coat the aggregates, the workability of
the mix and the mixture performance. The AFCT test and associated BSD computation method
was verified using X-Ray Microtomography imaging technique. Moreover, both AFCT and XRay Microtomography imaging were utilized to investigate foaming characteristics of different
kinds of binders prepared at different levels of injected air pressure and water content. As part of
this research, the relationship between various binder quality indicators and the mixture
performance tests were investigated and the effects of air pressure and water content on the
foamed binder and mixture performance were also observed.
CONCLUSIONS
This research presented a novel testing methodology called Asphalt Foam Collapse Test
(AFCT) for determining foamed binder parameters such as expansion ratio, half-life and foam
index. The AFCT is an automated, accurate and repeatable test method for measuring the height
reduction of the foamed binder as it collapses. A new procedure utilizing the AFCT data is also
introduced to calculate the bubble size distribution and bubble surface area of the foamed binder
from AFCT measurements. As a result of this new approach, a new dimensionless parameter
157

called Surface Area Index (SAI), which was directly related to the mixture coating, workability
as well as the performance, was introduced.
Based on the study on the foamed binder characteristics, the following major conclusions
were drawn:


The expansion ratio (ER) is an indicator of overall volume of bubbles, but it can
be a misleading parameter for the size distribution of the bubbles, which affects
the coating and workability.



Water content and air pressure have significant effect on the ER, half-life (HL)
and foam index (FI). ER increases with the increase of water content and air
pressure, whereas HL and FI decreases with the increase of water content and air
pressure.



The size distribution of the bubbles in foamed binder becomes coarser (i.e., the
bubbles become larger and larger) as the water content and air pressure increase.



As the injected water content and air pressure in the foamed binder increased, the
maximum diameter of the bubbles increased. This leads to the short half-life and
high expansion ratio.



SAI, which is an indication of total surface area of the bubbles, increased with
decreasing water content and air pressure. Also, better mixture coating was
observed at high SAI values.



AFCT procedure to compute the bubble size distribution was validated using 3D
X-Ray microtomography (XRM) imaging. 3D image-based bubble size

158

distribution was compared against bubble size distribution computed from the
AFCT and a good match (within the sample to sample variability) was observed.
Scope of this study also included evaluation of the performance of the WMA mixtures
prepared at different foaming water contents and air pressures. Testing program included
unconfined and confined dynamic modulus (|E*|), Flow Number (FN), Push-Pull Fatigue (PPF),
and Tensile Strength Ratio (TSR) tests. Based on the laboratory tests, the performance of the
mixtures are ranked as given in Table 16, in which the rank is 1 to 3 or 1 to 5 from better to
relatively worse performance. Additionally, the following conclusions were drawn:
Table 16: Performance ranking of mixtures based on the laboratory tests
Performance Tests

Performance test sample descriptions
1% - 10psi 2% - 12.5psi 3% - 15psi 4% - 17.5psi 5% - 20psi

Unconfined |E*|

No Ranking

Confined |E*|

3

NA

1

NA

2

Flow Number

5

4

3

2

1

Push-Pull Fatigue

4

3

1

2

5

Tensile Strength Ratio

3

3

1

3

2



The unconfined |E*| of different WMA mixtures prepared at different foaming
water content/air pressure combinations were very similar. Therefore, it is hard to
estimate the relative rutting or fatigue cracking susceptibilities (as estimated using
ME-PDG) of these WMA mixtures, as well as their relation to the foamed binder
quality indicators through unconfined |E*| tests for this study.

159



The confined |E*| tests revealed differences in |E*|s of mixtures made at different
foaming water content/air pressure combinations. The confined |E*| is perhaps a
more appropriate test as the pavements in the field is in ‘confined’ state.



FN increased (i.e., better rutting performance was observed) as the SAI decreases.
It is hypothesized that this is because of ‘poorer’ coating at low SAI values,
allowing the aggregate-to-aggregate friction to be relatively large, which is in-turn
helping towards the rutting performance.



The fatigue analysis indicated the importance of the foamed binder characteristics
on the mix performance. As a result of the high-injected water content and air
pressure in the binder, higher percentage of uncoated aggregates increased the
fatigue cracking potential. On the other hand, small foam bubbles trapped in the
mix due to the relatively low water content and air pressure in the foamed binder
also increased the fatigue cracking resistance. Therefore, there is an optimum
range of bubble sizes (and the surface area) and D50 that maximizes the resistance
to fatigue cracking.



The unconditioned strength of the mixtures decreases as the injected water
content and air pressure in the foamed binder increases. However, a similar trend
was not observed on the conditioned strength of the WMA mixtures.



The lower water content and air pressure results better aggregate coating, which
also increase the strength between the aggregate and binder bonds. Thus, the
tensile strength of these mixes is relatively high. On the contrary, the higher water

160

content and air pressure caused poor bond between the aggregate and binder
results relatively lower tensile strength.
As part of the study, further analyses with XRM imaging were performed. Several types
of asphalt binders were prepared using two different foaming methods: (i) direct foaming and (ii)
using synthetic Zeolite. Once the images of these samples were obtained, image processing and
analysis techniques were utilized to quantify the change in the microstructure of the moisture
bubbles with time. The findings of this study are summarized as follows:


High quality 3D images of foamed binder specimens can be captured using the
synchrotron based XRM system. Quantitative information such as the speed of
reduction in the volume of bubbles can be obtained from the XRM images.



Rate of moisture dissipation in foamed asphalt binders depends on the type and PG
of asphalt. It was observed in this study that high PG (stiff) asphalt binders
dissipated moisture slower than low PG (soft) binders. This is meaningful because
the stiffer binders possibly have lower diffusion coefficient. Therefore, high PG
binders might be more susceptible to moisture damage in foamed asphalt
pavements.



Foaming process (i.e., direct foaming versus use of foaming agents such as
Synthetic Zeolite) influences the moisture retention and dissipation. It was observed
that Zeolite dissipated moisture faster than the direct foaming method. Therefore,
mixtures prepared with Zeolite will probably be less susceptible to moisture
damage. However, quick dissipation of bubbles may also lead to reduced
workability during construction.
161



Moisture dissipation in mastic (binder + fine aggregate) is different from the
dissipation of moisture from the binder. Size distribution and overall volume of the
moisture bubbles in binders decrease with time. However, limited data obtained in
this study suggests that even though sizes of the bubbles decrease, overall volume
of the bubbles initially increase in mastics. It is hypothesized that this phenomenon
is because of coalescence of the micro-bubbles and these coalesced bubbles
dissipate in a longer period of time.

RECOMMENDATIONS
The AFCT test and the described methodologies for calculating the foam quality
indicators are valuable. The absence of an accurate and repeatable testing method in measuring
the foam quality in the current practice can be filled with AFCT, which is also practical and
affordable test that can be used by the practitioners. In addition, the findings of the performance
tests indicate the importance of foam quality in the performance of the pavement.
The majority of this research was limited with one type of binder and one aggregate
gradation. In the further studies, it is recommended to change the binder type and gradation to
increase the data range. In addition, this research was limited with the laboratory study. It is
recommended to study the relation in between the foam binder quality and field performance.
The crumb rubber (CR) usage in the asphalt pavements is very widespread and
beneficially reuses the scrap tires. There are numerous laboratory and field studies that showed
superior performance of CR modified asphalt pavements over traditional HMA. However, the
initial cost of the CR modified pavements is comparably high than HMA, partially because of the
high production temperatures. Constructing CR modified WMA can compensate for the cost
162

associated with high temperatures (~190oC) used during mixing CR with asphalt binder.
However, effect of moisture content/air pressures, in the presence of CR particles can be
different and should be investigated.

163

APPENDICES

164

APPENDIX A

165

Table 17: Unconfined Dynamic Modulus for the first replicate of 1% - 10psi WMA mixture
Shift Factor
Coefficients:
0.00082
a1=
-0.15022
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
0.12275
b1=
4.47288
b2=
0.90776
b3=
0.40460
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

10
10
10
10
10
10
21
21
21
21
21
21

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

9809
8460
7237
5040
4305
2778
5217
4140
3420
2065
1658
875.4

Phase
angle
15.10
16.93
18.60
23.02
24.53
28.65
23.53
25.99
27.56
31.15
31.73
32.53

Gaussian Model
(phase angle fit):
33.35002
a=
-2.42217
b=
5.07082
c=

Reduced
Frequency

log(aT)

fR
(Hz)

Log fR
(Hz)

1.373
1.373
1.373
1.373
1.373
1.373
0.000
0.000
0.000
0.000
0.000
0.000

590.260
236.104
118.052
23.610
11.805
2.361
25.000
10.000
5.000
1.000
0.500
0.100

2.771
2.373
2.072
1.373
1.072
0.373
1.398
1.000
0.699
0.000
-0.301
-1.000

166

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
11909.128
19.74
9936.447
21.33
8535.660
22.52
5687.247
25.20
4662.058
26.30
2777.269
28.65
5777.513
25.11
4435.721
26.56
3570.583
27.59
2041.093
29.75
1568.405
30.56
813.318
32.06

Total
Error:
Ʃ
Error1
Ʃ
Error2

0.5403
0.2724
0.2679

Error
Error1
(|E*|)
0.021
0.018
0.019
0.014
0.010
0.000
0.012
0.008
0.005
0.002
0.007
0.011

Error2
(Phase
Angle)
0.031
0.026
0.021
0.009
0.007
0.000
0.007
0.002
0.000
0.004
0.004
0.001

Table 17 (cont’d)
Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

37
37
37
37
37
54
54
54
54
54
54

10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

1163
880.1
402.5
282.9
150.9
495
316.9
211.5
101.3
82.8
53.7

Phase
angle
36.22
36.05
34.83
33.67
30.08
38.92
35.06
33.74
29.33
26.85
22.99

Reduced
Frequency

log(aT)

fR
(Hz)

Log fR
(Hz)

-1.643
-1.643
-1.643
-1.643
-1.643
-2.930
-2.930
-2.930
-2.930
-2.930
-2.930

0.227
0.114
0.023
0.011
0.002
0.029
0.012
0.006
0.001
0.001
0.000

-0.643
-0.944
-1.643
-1.944
-2.643
-1.532
-1.930
-2.231
-2.930
-3.231
-3.930

167

Predicted Data
Sigmoid
Fit, |E*|
Mpa
1145.193
858.592
425.987
312.150
150.900
477.343
316.815
231.634
112.344
82.837
42.115

Predicted
Phase
Angle
31.36
31.96
32.96
33.20
33.32
32.84
33.19
33.33
33.18
32.93
31.91

Error
Error1
(|E*|)
0.002
0.004
0.009
0.017
0.000
0.006
0.000
0.017
0.022
0.000
0.061

Error2
(Phase
Angle)
0.013
0.011
0.005
0.001
0.011
0.016
0.005
0.001
0.013
0.023
0.039

Table 18: Unconfined Dynamic Modulus for the second replicate of 1% - 10psi WMA mixture
Shift Factor
Coefficients:
0.000465
a1=
-0.12891
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
0.475519
b1=
3.908692
b2=
0.951278
b3=
0.475207
b4=

Shift
Factor

Measured Data
T (C)
-10
-10
-10
-10
-10
-10
10
10
10
10
10
10

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
18797
17651
16639
14235
13013
9770
9745
8320
7235
5126
4422
2765

Phase
angle
6.16
6.91
7.71
9.78
10.98
15.13
15.27
17.46
19.27
23.85
25.58
29.86

log(aT)

Gaussian Model
(phase angle fit):
34.71557
a=
-1.330173
b=
3.257148
c=

Reduced Frequency
fR
(Hz)

3.838 172064.979
3.838 68825.992
3.838 34412.996
3.838
6882.599
3.838
3441.300
3.838
688.260
1.260
454.459
1.260
181.784
1.260
90.892
1.260
18.178
1.260
9.089
1.260
1.818
168

Log fR
(Hz)
5.236
4.838
4.537
3.838
3.537
2.838
2.657
2.260
1.959
1.260
0.959
0.260

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
18310.230
4.551
17311.789
5.779
16476.661
6.855
14282.714
9.860
13240.334
11.368
10665.334
15.309
9982.793
16.408
8481.256
18.913
7372.819
20.852
5009.052
25.308
4122.902
27.121
2451.035
30.817

Total
Error:
Ʃ
Error1
Ʃ
Error2

0.4187
0.2573
0.1613

Error
Error1
(|E*|)
0.003
0.002
0.001
0.000
0.002
0.010
0.003
0.002
0.002
0.003
0.008
0.015

Error2
(Phase
Angle)
0.026
0.016
0.011
0.001
0.004
0.001
0.007
0.008
0.008
0.006
0.006
0.003

Table 18 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
5418
4197
3401
1974
1593
817
1715
1141
837.1
377.9
262.9
127.2
388.5
241.7
160
75.4
63.3
44.7

Phase
angle
24.45
26.91
28.65
32.46
33.17
34.15
35.8
36.07
36.23
35.4
34.58
31.61
38.26
35.81
34.85
30.27
27.14
22.31

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.631
-1.631
-1.631
-1.631
-1.631
-1.631
-3.104
-3.104
-3.104
-3.104
-3.104
-3.104

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.584
0.234
0.117
0.023
0.012
0.002
0.020
0.008
0.004
0.001
0.000
0.000

169

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.233
-0.631
-0.932
-1.631
-1.932
-2.631
-1.706
-2.104
-2.405
-3.104
-3.405
-4.104

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
5446.218
24.445
4239.406
26.877
3436.818
28.592
1973.061
31.938
1511.068
33.025
769.975
34.538
1606.587
32.802
1108.359
33.926
824.239
34.458
399.444
34.567
289.808
34.127
137.943
32.053
368.974
34.485
241.237
33.750
175.073
32.876
85.008
29.931
63.300
28.341
33.692
24.157

Error
Error1
(|E*|)
0.001
0.001
0.001
0.000
0.007
0.009
0.009
0.004
0.002
0.009
0.017
0.017
0.009
0.000
0.018
0.028
0.000
0.074

Error2
(Phase
Angle)
0.000
0.000
0.000
0.002
0.000
0.001
0.008
0.006
0.005
0.002
0.001
0.001
0.010
0.006
0.006
0.001
0.004
0.008

Table 19: Unconfined Dynamic Modulus for the first replicate of 3% - 15psi WMA mixture
Shift Factor
Coefficients:
0.00055
a1=
-0.13572
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
0.621746
b1=
3.752085
b2=
0.909983
b3=
0.468356
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

18416.0
17385.0
16535.0
14343.0
13259.0
10212.0
9239.0
8053.0
7241.0
5054.0
4187.0
2598.0

Phase
angle
5.9
6.7
7.3
9.2
10.3
14.0
15.4
17.5
19.2
23.7
25.2
29.4

log(aT)

Gaussian Model
(phase angle fit):
33.27342
a=
-1.23645
b=
3.275035
c=

Reduced Frequency
fR
(Hz)

4.019 261303.522
4.019 104521.409
4.019 52260.704
4.019 10452.141
4.019
5226.070
4.019
1045.214
1.305
504.409
1.305
201.764
1.305
100.882
1.305
20.176
1.305
10.088
1.305
2.018
170

Log fR
(Hz)
5.417
5.019
4.718
4.019
3.718
3.019
2.703
2.305
2.004
1.305
1.004
0.305

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
18116.011
4.225
17187.807
5.368
16410.465
6.372
14361.502
9.181
13383.229
10.595
10947.846
14.304
9802.605
16.141
8363.851
18.544
7300.530
20.396
5024.480
24.624
4165.845
26.332
2530.358
29.785

Total
Error:
Ʃ
Error1
Ʃ
Error2

0.3378
0.2017
0.1361

Error
Error1
(|E*|)
0.002
0.001
0.001
0.000
0.001
0.008
0.006
0.004
0.001
0.001
0.001
0.003

Error2
(Phase
Angle)
0.029
0.019
0.013
0.000
0.003
0.002
0.005
0.006
0.006
0.004
0.004
0.001

Table 19 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
5250
4095
3304
1947
1575
809
1663
1169
849.9
377.5
264.8
138.3
444.6
289
193.6
92.5
75.1
45.9

Phase
angle
24
26.36
28.01
31.64
32.23
32.97
34.56
34.86
35
34.27
33.38
30.29
35.34
32.24
31.32
27.75
25.82
22.59

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.660
-1.660
-1.660
-1.660
-1.660
-1.660
-3.114
-3.114
-3.114
-3.114
-3.114
-3.114

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.547
0.219
0.109
0.022
0.011
0.002
0.019
0.008
0.004
0.001
0.000
0.000

171

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.262
-0.660
-0.961
-1.660
-1.961
-2.660
-1.716
-2.114
-2.415
-3.114
-3.415
-4.114

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
5306.757
24.077
4155.530
26.353
3388.408
27.942
1981.390
30.985
1533.054
31.944
804.340
33.187
1586.485
31.832
1109.594
32.761
835.294
33.156
418.825
32.997
309.004
32.470
153.395
30.275
395.790
32.919
264.711
32.101
195.579
31.188
99.065
28.232
75.093
26.670
41.548
22.620

Error
Error1
(|E*|)
0.001
0.002
0.003
0.002
0.004
0.001
0.006
0.007
0.003
0.018
0.028
0.021
0.019
0.015
0.002
0.015
0.000
0.026

Error2
(Phase
Angle)
0.000
0.000
0.000
0.002
0.001
0.001
0.008
0.006
0.005
0.004
0.003
0.000
0.007
0.000
0.000
0.002
0.003
0.000

Table 20: Unconfined Dynamic Modulus for the second replicate of 3% - 15psi WMA mixture
Shift Factor
Coefficients:
0.00058
a1=
-0.13696
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
0.415353
b1=
4.155925
b2=
0.796877
b3=
0.417205
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

24226.0
22654.0
21349.0
18120.0
16598.0
12726.0
8896.0
7579.0
6824.0
4830.0
4092.0
2620.0

Phase
angle
24.2
35.7
25.3
32.8
33.8
21.8
15.7
17.8
19.5
24.0
25.6
29.6

log(aT)

Gaussian Model
(phase angle fit):
33.52415
a=
-1.206494
b=
3.205231
c=

Reduced Frequency
fR
(Hz)

4.048 279257.696
4.048 111703.078
4.048 55851.539
4.048 11170.308
4.048
5585.154
4.048
1117.031
1.309
509.121
1.309
203.648
1.309
101.824
1.309
20.365
1.309
10.182
1.309
2.036
172

Log fR
(Hz)
5.446
5.048
4.747
4.048
3.747
3.048
2.707
2.309
2.008
1.309
1.008
0.309

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
24363.112
3.890
22653.609
4.995
21276.864
5.973
17858.283
8.745
16322.264
10.156
12735.514
13.892
11034.439
15.910
9147.617
18.372
7816.359
20.276
5139.194
24.639
4188.485
26.407
2465.313
29.979

Total
Error:
Ʃ
Error1
Ʃ
Error2

0.756343
0.2549
0.5015

Error
Error1
(|E*|)
0.001
0.000
0.000
0.001
0.002
0.000
0.024
0.021
0.015
0.007
0.003
0.008

Error2
(Phase
Angle)
0.084
0.086
0.076
0.073
0.070
0.036
0.002
0.003
0.004
0.003
0.003
0.001

Table 20 (cont’d)
Shift
Factor

Measured Data

T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
5051
3947
3221
1899
1490
788.1
1712
1171
850.9
406.3
297.9
149
412.9
267.8
177.2
86.9
72.1
46.9

Phase
angle
24.08
26.57
28.14
31.64
32.14
32.81
35.02
35.2
35.11
33.86
32.6
29.16
36.94
33.99
33.14
29.48
26.59
21.24

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.653
-1.653
-1.653
-1.653
-1.653
-1.653
-3.085
-3.085
-3.085
-3.085
-3.085
-3.085

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.555
0.222
0.111
0.022
0.011
0.002
0.021
0.008
0.004
0.001
0.000
0.000

173

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.255
-0.653
-0.954
-1.653
-1.954
-2.653
-1.687
-2.085
-2.386
-3.085
-3.386
-4.085

Predicted Data
Sigmoid
Fit, |E*|
Mpa
5444.640
4165.391
3344.949
1905.711
1464.559
764.040
1525.227
1061.407
798.247
402.208
297.900
149.158
388.974
261.535
193.947
98.625
74.648
40.769

Predicte
d
Phase
Angle
24.098
26.452
28.094
31.231
32.213
33.455
32.080
33.029
33.421
33.200
32.624
30.277
33.150
32.288
31.329
28.234
26.605
22.399

Error
Error1
(|E*|)
0.009
0.007
0.005
0.000
0.002
0.005
0.016
0.014
0.009
0.002
0.000
0.000
0.010
0.004
0.017
0.028
0.008
0.036

Error2
(Phase
Angle)
0.000
0.000
0.000
0.001
0.000
0.002
0.008
0.006
0.005
0.002
0.000
0.004
0.010
0.005
0.005
0.004
0.000
0.005

Table 21: Unconfined Dynamic Modulus for the first replicate of 5% - 20psi WMA mixture
Shift Factor
Coefficients:
0.000517
a1=
-0.12781
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
0.361519
b1=
4.012953
b2=
1.004746
b3=
0.458357
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

Phase
angle

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

18303.0
17051.0
16030.0
13492.0
12207.0
9067.0
9229.0
7875.0
7022.0
4981.0
4207.0
2661.0

6.6
7.3
8.1
10.3
11.5
15.6
15.6
17.7
19.4
23.8
25.4
29.4

log(aT)

Gaussian Model
(phase angle fit):
33.52834
a=
-1.232364
b=
3.2651
c=

Reduced Frequency
fR
(Hz)

3.786 152720.884
3.786 61088.354
3.786 30544.177
3.786
6108.835
3.786
3054.418
3.786
610.884
1.230
424.241
1.230
169.696
1.230
84.848
1.230
16.970
1.230
8.485
1.230
1.697

174

Log fR
(Hz)
5.184
4.786
4.485
3.786
3.485
2.786
2.628
2.230
1.929
1.230
0.929
0.230

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
17476.096
4.863
16484.016
6.133
15660.903
7.238
13523.385
10.291
12518.746
11.807
10061.780
15.723
9494.144
16.670
8077.273
19.111
7034.330
20.984
4813.326
25.232
3979.452
26.933
2397.680
30.330

Total
Error:
Ʃ
Error1
Ʃ
Error2

0.355764
0.2035
0.1523

Error
Error1
(|E*|)
0.005
0.003
0.002
0.000
0.003
0.011
0.003
0.003
0.000
0.004
0.007
0.013

Error2
(Phase
Angle)
0.026
0.016
0.011
0.000
0.003
0.001
0.007
0.008
0.008
0.006
0.006
0.003

Table 21 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
5282
4183
3374
1990
1559
825.3
1750
1211
928.1
434.6
311.5
146.4
463.2
306.4
204.7
98.3
78.6
53.5

Phase
angle
24.23
26.59
28.21
31.81
32.42
33.2
34.85
35.02
34.75
33.61
32.47
29.41
37.94
34.26
33.44
29.66
27.28
22.82

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.565
-1.565
-1.565
-1.565
-1.565
-1.565
-2.938
-2.938
-2.938
-2.938
-2.938
-2.938

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.680
0.272
0.136
0.027
0.014
0.003
0.029
0.012
0.006
0.001
0.001
0.000

175

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.167
-0.565
-0.866
-1.565
-1.866
-2.565
-1.540
-1.938
-2.239
-2.938
-3.239
-3.938

Predicted Data
Sigmoid
Fit, |E*|
Mpa

Predicted
Phase
Angle

5315.027
4169.307
3403.171
1990.505
1537.903
799.673
1727.880
1211.208
911.037
451.073
329.237
157.322
462.836
305.151
221.933
106.687
78.600
40.296

24.238
26.540
28.147
31.223
32.192
33.444
31.791
32.836
33.318
33.354
32.902
30.848
33.379
32.754
31.971
29.250
27.756
23.783

Error
Error1
(|E*|)
0.001
0.000
0.001
0.000
0.002
0.005
0.002
0.000
0.003
0.006
0.010
0.014
0.000
0.001
0.015
0.018
0.000
0.071

Error2
(Phase
Angle)
0.000
0.000
0.000
0.002
0.001
0.001
0.009
0.006
0.004
0.001
0.001
0.005
0.012
0.004
0.004
0.001
0.002
0.004

Table 22: Unconfined Dynamic Modulus for the second replicate of 5% - 20psi WMA mixture
Shift Factor
Coefficients:
0.000477
a1=
-0.12936
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
0.457706
b1=
3.904656
b2=
1.028713
b3=
0.462209
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

Phase
angle

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

17766.0
16661.0
15804.0
13743.0
12750.0
10083.0
9909.0
8494.0
7317.0
5087.0
4301.0
2805.0

6.0
6.8
7.5
9.4
10.4
13.8
15.2
17.4
19.1
23.5
25.0
29.1

log(aT)

Gaussian Model
(phase angle fit):
33.47423
a=
-1.393113
b=
3.294088
c=

Reduced Frequency
fR
(Hz)

3.847 175941.202
3.847 70376.481
3.847 35188.240
3.847
7037.648
3.847
3518.824
3.847
703.765
1.260
455.268
1.260
182.107
1.260
91.054
1.260
18.211
1.260
9.105
1.260
1.821

176

Log fR
(Hz)
5.245
4.847
4.546
3.847
3.546
2.847
2.658
2.260
1.959
1.260
0.959
0.260

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
17481.864
4.393
16568.672
5.564
15806.063
6.588
13804.104
9.443
12851.940
10.876
10490.297
14.617
9829.581
15.712
8437.170
18.097
7401.276
19.944
5159.003
24.200
4301.948
25.940
2646.355
29.512

Total
Error:
Ʃ
Error1
Ʃ
Error2

0.332706
0.1810
0.1518

Error
Error1
(|E*|)
0.002
0.001
0.000
0.000
0.001
0.004
0.001
0.001
0.001
0.002
0.000
0.007

Error2
(Phase
Angle)
0.027
0.018
0.012
0.001
0.005
0.006
0.003
0.004
0.005
0.003
0.004
0.001

Table 22 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
5357
4252
3614
2153
1713
923.1
1794
1287
960.4
463.5
340.5
168.9
524.8
331.2
216.8
95.5
75.9
50.7

Phase
angle
24.04
26.43
28.03
31.49
31.93
32.5
34.53
34.67
34.41
33.37
32.01
28.72
34.91
33.23
33.37
30.89
27.78
22.97

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.627
-1.627
-1.627
-1.627
-1.627
-1.627
-3.089
-3.089
-3.089
-3.089
-3.089
-3.089

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.590
0.236
0.118
0.024
0.012
0.002
0.020
0.008
0.004
0.001
0.000
0.000

177

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.229
-0.627
-0.928
-1.627
-1.928
-2.627
-1.691
-2.089
-2.390
-3.089
-3.390
-4.089

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
5575.598
23.379
4413.164
25.710
3627.820
27.360
2158.409
30.611
1680.097
31.684
888.597
33.237
1785.946
31.449
1259.256
32.582
951.625
33.143
476.783
33.390
349.942
33.035
169.573
31.206
446.828
33.338
296.343
32.736
216.785
31.977
106.098
29.321
78.907
27.857
41.476
23.950

Error
Error1
(|E*|)
0.005
0.004
0.000
0.000
0.003
0.006
0.001
0.003
0.001
0.005
0.005
0.001
0.026
0.019
0.000
0.023
0.009
0.051

Error2
(Phase
Angle)
0.003
0.003
0.002
0.003
0.001
0.002
0.009
0.006
0.004
0.000
0.003
0.009
0.005
0.001
0.004
0.005
0.000
0.004

Table 23: Confined Dynamic Modulus for the first replicate of 1% - 10psi WMA mixture
Shift Factor
Coefficients:
1.03E-05
a1=
-0.10815
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
2.678669
b1=
1.610201
b2=
-0.092824
b3=
0.675849
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

Phase
angle

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

17011.0
16000.0
15156.0
13250.0
12342.0
9920.0
10154.0
8948.0
8041.0
5983.0
5210.0
3637.0

5.5
6.1
6.9
8.6
9.6
12.6
12.9
14.6
16.0
19.8
21.3
25.1

log(aT)

Gaussian Model
(phase angle fit):
28.49632
a=
-0.755005
b=
2.650591
c=

Reduced Frequency
fR
(Hz)

3.349 55846.894
3.349 22338.758
3.349 11169.379
3.349 2233.876
3.349 1116.938
3.349
223.388
1.186
383.755
1.186
153.502
1.186
76.751
1.186
15.350
1.186
7.675
1.186
1.535

178

Log fR
(Hz)
4.747
4.349
4.048
3.349
3.048
2.349
2.584
2.186
1.885
1.186
0.885
0.186

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
16614.081
3.305
15868.245
4.463
15202.703
5.518
13303.173
8.594
12339.898
10.180
9859.927
14.354
10721.016
12.888
9255.675
15.397
8142.133
17.352
5724.346
21.794
4819.255
23.532
3140.772
26.756

Total
Error:
Ʃ
Error1
Ʃ
Error2

0.428176
0.2192
0.2090

Error
Error1
(|E*|)
0.002
0.001
0.000
0.000
0.000
0.001
0.006
0.004
0.001
0.005
0.009
0.018

Error2
(Phase
Angle)
0.039
0.027
0.019
0.000
0.006
0.014
0.000
0.005
0.008
0.010
0.010
0.006

Table 23 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
6417
5246
4310
2794
2317
1412
2539
1901
1523
965.9
832.8
607.9
1144
918.2
789.1
634.3
606
550.9

Phase
angle
20.67
22.89
24.4
27.76
28.27
29.56
30.51
30.66
30.14
28.07
26.06
22.51
26.19
22.62
20.26
15.58
13.32
10.26

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.721
-1.721
-1.721
-1.721
-1.721
-1.721
-3.543
-3.543
-3.543
-3.543
-3.543
-3.543

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.475
0.190
0.095
0.019
0.010
0.002
0.007
0.003
0.001
0.000
0.000
0.000

179

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.323
-0.721
-1.022
-1.721
-2.022
-2.721
-2.145
-2.543
-2.844
-3.543
-3.844
-4.543

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
6417.837
20.489
5152.719
22.887
4312.791
24.516
2795.308
27.363
2319.362
28.081
1544.280
28.375
2288.550
28.120
1805.625
28.494
1526.142
28.352
1085.518
26.666
960.097
25.421
762.891
21.645
916.773
24.833
803.113
22.695
737.986
20.886
634.189
16.386
603.836
14.447
554.605
10.261

Error
Error1
(|E*|)
0.000
0.002
0.000
0.000
0.000
0.012
0.013
0.007
0.000
0.017
0.021
0.035
0.031
0.020
0.010
0.000
0.001
0.001

Error2
(Phase
Angle)
0.001
0.000
0.000
0.001
0.001
0.004
0.008
0.007
0.006
0.005
0.002
0.004
0.005
0.000
0.003
0.005
0.008
0.000

Table 24: Confined Dynamic Modulus for the second replicate of 1% - 10psi WMA mixture
Shift Factor
Coefficients:
-0.00020
a1=
-0.10224
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
2.674677
b1=
1.664104
b2=
-0.011619
b3=
0.619542
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

Phase
angle

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

17829.0
16775.0
15912.0
13783.0
12711.0
10063.0
11213.0
10066.0
9025.0
6712.0
5665.0
3837.0

5.8
6.5
7.1
8.9
9.9
12.9
12.6
14.3
15.7
19.4
20.8
24.7

log(aT)

Gaussian Model
(phase angle fit):
28.21331
a=
-0.774841
b=
2.637479
c=

Reduced Frequency
fR
(Hz)

3.237 43096.333
3.237 17238.533
3.237 8619.267
3.237 1723.853
3.237
861.927
3.237
172.385
1.192
388.673
1.192
155.469
1.192
77.735
1.192
15.547
1.192
7.773
1.192
1.555
180

Log fR
(Hz)
4.634
4.237
3.935
3.237
2.935
2.237
2.590
2.192
1.891
1.192
0.891
0.192

Total
Error:
Ʃ
Error1
Ʃ
Error2

Predicted Data
Sigmoid
Fit, |E*|
Mpa
17725.916
16792.655
15984.807
13781.021
12710.015
10061.692
11416.915
9889.066
8740.208
6257.497
5322.773
3558.343

Predicted
Phase
Angle
3.444
4.640
5.726
8.875
10.489
14.702
12.506
14.988
16.931
21.367
23.114
26.381

0.405294
0.2062
0.1991

Error
Error1
(|E*|)
0.001
0.000
0.000
0.000
0.000
0.000
0.002
0.002
0.004
0.008
0.007
0.009

Error2
(Phase
Angle)
0.041
0.028
0.019
0.000
0.006
0.014
0.001
0.005
0.008
0.010
0.011
0.007

Table 24 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
6438
5332
4695
3074
2552
1620
2695
2057
1683
1096
954.4
727.6
1184
948.9
817.4
656.9
617.8
563.2

Phase
angle
20.1
22.27
23.7
27.04
27.51
28.81
29.38
29.26
28.65
26.91
24.95
21.58
23.96
20.65
18.3
13.19
11.21
8.5

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.818
-1.818
-1.818
-1.818
-1.818
-1.818
-3.860
-3.860
-3.860
-3.860
-3.860
-3.860

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.380
0.152
0.076
0.015
0.008
0.002
0.003
0.001
0.001
0.000
0.000
0.000

181

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.420
-0.818
-1.119
-1.818
-2.119
-2.818
-2.462
-2.860
-3.161
-3.860
-4.161
-4.860

Predicted Data
Sigmoid
Fit, |E*|
Mpa

Predicted
Phase
Angle

6951.046
5651.048
4780.814
3176.105
2658.118
1788.678
2479.022
1975.727
1678.822
1198.182
1057.414
830.722
931.466
820.498
755.667
649.665
617.798
564.822

20.095
22.497
24.135
27.022
27.762
28.111
27.959
28.210
27.974
26.090
24.776
20.899
22.992
20.640
18.737
14.234
12.374
8.501

Error
Error1
(|E*|)
0.009
0.007
0.002
0.004
0.005
0.013
0.011
0.005
0.000
0.013
0.015
0.020
0.034
0.021
0.012
0.002
0.000
0.000

Error2
(Phase
Angle)
0.000
0.001
0.002
0.000
0.001
0.002
0.005
0.004
0.002
0.003
0.001
0.003
0.004
0.000
0.002
0.008
0.010
0.000

Table 25: Confined Dynamic Modulus for the first replicate of 3% - 15psi WMA mixture
Shift Factor
Coefficients:
-0.00031
a1=
-0.08666
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
2.696276
b1=
1.556474
b2=
0.023775
b3=
0.730306
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

Phase
angle

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

15626.0
14551.0
13706.0
11605.0
10677.0
8396.0
11296.0
9811.0
8635.0
6350.0
5484.0
3738.0

7.2
8.0
8.8
11.0
12.2
15.8
13.1
14.9
16.4
20.2
21.6
25.3

log(aT)

Gaussian Model
(phase angle fit):
28.63675
a=
-0.731871
b=
2.551689
c=

Reduced Frequency
fR
(Hz)

2.791 15459.103
2.791 6183.641
2.791 3091.821
2.791
618.364
2.791
309.182
2.791
61.836
1.058
285.697
1.058
114.279
1.058
57.139
1.058
11.428
1.058
5.714
1.058
1.143

182

Log fR
(Hz)
4.189
3.791
3.490
2.791
2.490
1.791
2.456
2.058
1.757
1.058
0.757
0.058

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
15295.542
4.460
14551.133
5.951
13879.388
7.285
11943.884
11.040
10961.576
12.903
8466.335
17.564
10844.926
13.123
9439.720
15.753
8340.743
17.797
5875.002
22.392
4930.048
24.155
3163.931
27.297

Total
Error:
Ʃ
Error1
Ʃ
Error2

0.385169
0.1938
0.1914

Error
Error1
(|E*|)
0.002
0.000
0.001
0.003
0.003
0.001
0.004
0.004
0.004
0.009
0.012
0.020

Error2
(Phase
Angle)
0.038
0.026
0.017
0.000
0.006
0.012
0.000
0.006
0.009
0.011
0.012
0.008

Table 25 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
6699
5536
4696
3041
2503
1581
2607
1980
1611
1054
923.9
717.9
1138
926.9
808.3
639.3
605.7
553.5

Phase
angle
20.87
23
24.47
27.78
28.27
29.46
30.16
29.88
29.1
27.03
24.98
21.53
25.51
21.76
19.41
15.09
13
10.5

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.671
-1.671
-1.671
-1.671
-1.671
-1.671
-3.620
-3.620
-3.620
-3.620
-3.620
-3.620

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.533
0.213
0.107
0.021
0.011
0.002
0.006
0.002
0.001
0.000
0.000
0.000

183

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.274
-0.671
-0.972
-1.671
-1.972
-2.671
-2.222
-2.620
-2.921
-3.620
-3.921
-4.620

Predicted Data
Sigmoid
Fit, |E*|
Mpa

Predicted
Phase
Angle

7041.839
5685.684
4759.783
3046.238
2503.009
1623.611
2548.095
1977.782
1649.779
1141.358
1000.135
783.507
907.868
795.572
732.326
633.784
605.675
561.097

20.214
22.745
24.471
27.483
28.231
28.479
28.178
28.629
28.510
26.760
25.445
21.452
24.150
21.781
19.823
15.095
13.117
8.972

Error
Error1
(|E*|)
0.006
0.003
0.002
0.000
0.000
0.004
0.003
0.000
0.003
0.011
0.012
0.013
0.032
0.022
0.015
0.001
0.000
0.002

Error2
(Phase
Angle)
0.003
0.001
0.000
0.001
0.000
0.003
0.007
0.004
0.002
0.001
0.002
0.000
0.005
0.000
0.002
0.000
0.001
0.015

Table 26: Confined Dynamic Modulus for the second replicate of 3% - 15psi WMA mixture
Shift Factor
Coefficients:
-9.7E-05
a1=
a2=
0.100664
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
2.591216
b1=
1.943639
b2=
0.119098
b3=
0.612476
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

Phase
angle

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

26657.0
25304.0
24241.0
21247.0
19577.0
15113.0
16097.0
14136.0
12655.0
9407.0
8184.0
5620.0

23.1
34.8
24.5
31.7
32.6
20.1
37.2
6.8
28.1
37.4
38.6
32.8

log(aT)

Gaussian Model
(phase angle fit):
29.37255
a=
-0.357151
b=
2.888489
c=

Reduced Frequency
fR
(Hz)

3.154 35614.754
3.154 14245.902
3.154 7122.951
3.154 1424.590
3.154
712.295
3.154
142.459
1.140
345.416
1.140
138.166
1.140
69.083
1.140
13.817
1.140
6.908
1.140
1.382

184

Log fR
(Hz)
4.552
4.154
3.853
3.154
2.853
2.154
2.538
2.140
1.839
1.140
0.839
0.140

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
27173.509
6.931
25595.388
8.677
24237.897
10.155
20573.521
14.033
18813.189
15.841
14521.569
20.131
16901.021
17.772
14439.598
20.211
12605.170
21.997
8700.770
25.679
7258.180
26.958
4595.622
28.940

Total
Error:
Ʃ
Error1
Ʃ
Error2

1.274143
0.3659
0.9082

Error
Error1
(|E*|)
0.002
0.001
0.000
0.003
0.004
0.004
0.005
0.002
0.000
0.009
0.013
0.023

Error2
(Phase
Angle)
0.070
0.075
0.058
0.056
0.051
0.000
0.052
0.196
0.022
0.031
0.030
0.012

Table 26 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
6855
5617
4590
3002
2496
1594
3644
2774
2122
1367
1188
905.7
1141
918.5
790.9
618.1
578.6
521.1

Phase
angle
20.71
22.75
24.13
27.27
27.62
28.65
49.76
45.41
44.22
32.21
31.17
28.03
26.26
22.52
20.16
15.46
13.26
10.18

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.701
-1.701
-1.701
-1.701
-1.701
-1.701
-3.562
-3.562
-3.562
-3.562
-3.562
-3.562

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.498
0.199
0.100
0.020
0.010
0.002
0.007
0.003
0.001
0.000
0.000
0.000

185

Log fR
(Hz)

Predicted Data
Sigmoid
Fit, |E*|
Mpa

1.398 10060.963
1.000 8006.931
0.699 6644.844
0.000 4176.754
-0.301 3398.830
-1.000 2128.367
-0.303 3394.825
-0.701 2591.617
-1.002 2126.003
-1.701 1393.376
-2.002 1186.066
-2.701
862.480
-2.164 1093.723
-2.562
912.995
-2.863
809.922
-3.562
646.289
-3.863
598.475
-4.562
520.597

Predicted
Phase
Angle
24.421
26.303
27.473
29.149
29.367
28.654
29.367
29.166
28.650
26.361
24.978
21.135
24.152
21.948
20.160
15.871
14.061
10.180

Error
Error1
(|E*|)
0.043
0.041
0.044
0.041
0.039
0.039
0.009
0.009
0.000
0.003
0.000
0.007
0.006
0.001
0.004
0.007
0.005
0.000

Error2
(Phase
Angle)
0.018
0.016
0.014
0.007
0.006
0.000
0.041
0.036
0.035
0.018
0.020
0.025
0.008
0.003
0.000
0.003
0.006
0.000

Table 27: Confined Dynamic Modulus for the first replicate of 5% - 20psi WMA mixture
Shift Factor
Coefficients:
0.00021
a1=
-0.11398
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
2.478079
b1=
1.60413
b2=
-0.273441
b3=
0.713126
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

Phase
angle

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

10624.0
10061.0
9560.0
8356.0
7713.0
6038.0
5794.0
4959.0
4344.0
3201.0
2931.0
2052.0

8.3
9.2
9.2
10.1
10.6
12.7
15.2
17.6
19.8
25.8
30.5
20.7

log(aT)

Gaussian Model
(phase angle fit):
27.09391
a=
-0.246991
b=
2.64152
c=

Reduced Frequency
fR
(Hz)

3.462 72408.157
3.462 28963.263
3.462 14481.631
3.462 2896.326
3.462 1448.163
3.462
289.633
1.182
380.290
1.182
152.116
1.182
76.058
1.182
15.212
1.182
7.606
1.182
1.521

186

Log fR
(Hz)
4.860
4.462
4.161
3.462
3.161
2.462
2.580
2.182
1.881
1.182
0.881
0.182

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
10445.031
4.181
9981.899
5.531
9563.088
6.733
8346.743
10.111
7721.542
11.789
6099.357
16.015
6384.865
15.281
5419.680
17.751
4698.020
19.585
3181.189
23.405
2636.363
24.732
1669.872
26.739

Total
Error:
Ʃ
Error1
Ʃ
Error2

0.550183
0.2576
0.2925

Error
Error1
(|E*|)
0.002
0.001
0.000
0.000
0.000
0.001
0.011
0.010
0.009
0.001
0.013
0.027

Error2
(Phase
Angle)
0.050
0.040
0.027
0.000
0.011
0.026
0.000
0.001
0.001
0.009
0.019
0.029

Table 27 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
3615
2908
2421
1507
1225
763.5
1281
965.7
773.3
535.2
477.9
374.5
708.3
567.2
482.6
391
371.6
339.8

Phase
angle
19.92
22.12
23.71
27.65
28.27
28.7
27.98
27.35
26.55
23.63
21.67
18.97
22.95
20
18.1
14.08
12.05
9.67

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.629
-1.629
-1.629
-1.629
-1.629
-1.629
-3.241
-3.241
-3.241
-3.241
-3.241
-3.241

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.588
0.235
0.118
0.024
0.012
0.002
0.014
0.006
0.003
0.001
0.000
0.000

187

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.231
-0.629
-0.930
-1.629
-1.930
-2.629
-1.844
-2.241
-2.542
-3.241
-3.542
-4.241

Predicted Data
Sigmoid Predicted
Fit, |E*|
Phase
Mpa
Angle
3616.195
22.319
2842.400
24.237
2344.243
25.411
1483.076
26.976
1225.154
27.088
819.871
26.015
1280.058
27.093
1006.038
26.812
850.775
26.204
611.784
23.629
545.255
22.118
442.311
18.044
562.629
22.571
491.981
20.374
452.056
18.573
389.533
14.250
371.589
12.442
342.959
8.636

Error
Error1
(|E*|)
0.000
0.003
0.004
0.002
0.000
0.011
0.000
0.006
0.014
0.021
0.021
0.028
0.035
0.022
0.011
0.001
0.000
0.002

Error2
(Phase
Angle)
0.012
0.010
0.007
0.002
0.004
0.009
0.003
0.002
0.001
0.000
0.002
0.005
0.002
0.002
0.003
0.001
0.003
0.011

Table 28: Confined Dynamic Modulus for the second replicate of 5% - 20psi WMA mixture
Shift Factor
Coefficients:
2.21E-06
a1=
-0.11398
a2=
Reference
Temperature:
21
Tref=

Sigmoid
Coefficients:
1.911767
b1=
2.587999
b2=
0.486123
b3=
0.418533
b4=

Shift
Factor

Measured Data
T (C)

f (Hz)

|E*|
(Mpa)

Phase
angle

-10.0
-10.0
-10.0
-10.0
-10.0
-10.0
10.0
10.0
10.0
10.0
10.0
10.0

25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

20047.0
18922.0
17977.0
15639.0
14370.0
11111.0
11558.0
10134.0
9040.0
6565.0
5529.0
3723.0

5.6
6.2
6.8
8.5
9.4
12.8
13.2
14.9
16.4
20.2
21.6
25.4

log(aT)

Gaussian Model
(phase angle fit):
28.2353
a=
-1.038716
b=
2.990919
c=

Reduced Frequency
fR
(Hz)

3.533 85201.906
3.533 34080.762
3.533 17040.381
3.533 3408.076
3.533 1704.038
3.533
340.808
1.253
447.636
1.253
179.054
1.253
89.527
1.253
17.905
1.253
8.953
1.253
1.791
188

Log fR
(Hz)
4.930
4.533
4.231
3.533
3.231
2.533
2.651
2.253
1.952
1.253
0.952
0.253

Total
Error:
Ʃ
Error1
Ʃ
Error2

Predicted Data
Sigmoid
Fit, |E*|
Mpa
20524.429
19105.683
17971.772
15188.635
13952.450
11096.081
11572.969
9992.013
8848.067
6443.113
5539.356
3784.307

Predicted
Phase
Angle
3.854
4.981
5.978
8.781
10.190
13.842
13.193
15.409
17.127
21.053
22.626
25.721

0.445971
0.2387
0.2073

Error
Error1
(|E*|)
0.002
0.001
0.000
0.003
0.003
0.000
0.000
0.002
0.002
0.002
0.000
0.002

Error2
(Phase
Angle)
0.031
0.019
0.012
0.004
0.008
0.008
0.000
0.003
0.005
0.004
0.005
0.001

Table 28 (cont’d)
Shift
Factor

Measured Data
T (C)
21
21
21
21
21
21
37
37
37
37
37
37
54
54
54
54
54
54

f (Hz)
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1
25
10
5
1
0.5
0.1

|E*|
(Mpa)
6567
5299
4466
2901
2394
1497
2694
2034
1645
1125
993.6
724
789.7
587.2
471.3
342.2
320.2
299

Phase
angle
20.65
22.78
24.22
27.43
27.85
28.79
29.09
28.63
27.8
25.3
23.16
19.53
30.21
26.92
25.19
19.73
16.95
12.88

log(aT)
0.000
0.000
0.000
0.000
0.000
0.000
-1.822
-1.822
-1.822
-1.822
-1.822
-1.822
-3.756
-3.756
-3.756
-3.756
-3.756
-3.756

Reduced Frequency
fR
(Hz)
25.000
10.000
5.000
1.000
0.500
0.100
0.377
0.151
0.075
0.015
0.008
0.002
0.004
0.002
0.001
0.000
0.000
0.000

189

Log fR
(Hz)
1.398
1.000
0.699
0.000
-0.301
-1.000
-0.424
-0.822
-1.123
-1.822
-2.123
-2.822
-2.358
-2.756
-3.057
-3.756
-4.057
-4.756

Predicted Data
Sigmoid
Fit, |E*|
Mpa

Predicted
Phase
Angle

6908.401
5677.771
4847.994
3267.664
2730.776
1776.154
2535.130
1984.539
1645.501
1066.969
889.121
593.587
773.423
615.731
522.227
367.293
320.200
241.113

20.261
22.382
23.850
26.583
27.389
28.233
27.645
28.161
28.224
27.284
26.441
23.639
25.619
23.946
22.487
18.689
16.970
13.044

Error
Error1
(|E*|)
0.006
0.008
0.010
0.015
0.017
0.023
0.008
0.003
0.000
0.008
0.016
0.030
0.003
0.007
0.017
0.012
0.000
0.038

Error2
(Phase
Angle)
0.002
0.002
0.002
0.003
0.002
0.002
0.005
0.002
0.002
0.008
0.014
0.021
0.015
0.011
0.011
0.005
0.000
0.001

APPENDIX B

190

70000

60000

Microstrain

50000

40000

30000

1% wc- 10 psi-1
3% wc- 15 psi-1
5% wc- 20 psi-1
5% wc- 20 psi-3
4% wc- 17.5 psi-1
4% wc- 17.5 psi-3
2% wc- 12.5 psi-1
2% wc- 12.5 psi-3

20000

10000

1% wc- 10 psi-2
3% wc- 15 psi-2
5% wc- 20 psi-2
5% wc- 20 psi-4(6.23)
4% wc- 17.5 psi-2
4% wc- 17.5 psi-4
2% wc- 12.5 psi-2
2% wc- 12.5 psi-4

0
0

50

100

150
Number of Cycles

Figure 59: Raw Flow Number Data

191

200

250

300

APPENDIX C

192

1%-10psi-1-20C
1%-10psi-2-20C
1%-10psi-3-10C
1%-10psi-4-10C
Best Fit: C=exp(-8.326e-006*S^^(0.97645))

1
0.8

C

0.6
0.4
0.2
0
0

0.5

1

1.5
S

2

2.5

3
5
x10

8

10

6

Nf

10

4

10

2

10
100

30
200

20
300

400

10
500

0
Temperature (C)

Microstrain

Figure 60: Push Pull Tests for WMA mixtures prepared with foamed binder - 1% water
content -10 psi air pressure

193

2%-12.5psi-1-10C
2%-12.5psi-2-10C
Best Fit: C=exp(-5.5158e-005*S^^(0.82102))

1

C

0.8

0.6

0.4

0.2
0

2

4

6

8

10

12

14
4
x10

S

8

10

6

Nf

10

4

10

2

10
100

30
200

20
300

400

Microstrain

10
500

0
Temperature (C)

Figure 61: Push Pull Tests for WMA mixtures prepared with foamed binder - 2% water
content -12.5 psi air pressure

194

3%-15psi-1-20C

1

3%-15psi-2-20C
0.8

3%-15psi-3-10C
Best Fit: C=exp(-5.5158e-005*S^^(0.82102))

C

0.6
0.4
0.2
0
0

0.5

1

1.5
S

2

2.5

3
5
x10

8

10

6

Nf

10

4

10

2

10
100

30
200

20
300

400

Microstrain

10
500

0

Temperature (C)

Figure 62: Push Pull Tests for WMA mixtures prepared with foamed binder - 3% water
content -15 psi air pressure

195

4%-17.5psi-1-10C
4%-17.5psi-2-10C
4%-17.5psi-3-20C
4%-17.5psi-4-20C
Best Fit C=exp(-1.1656e-005*S^^(0.94413))

1
0.8

C

0.6
0.4
0.2
0
0

0.5

1

1.5
S

2

2.5

3
5
x10

8

10

6

Nf

10

4

10

2

10

100

30
200

20
300

400

10
500

0
Temperature (C)

Microstrain

Figure 63: Push Pull Tests for WMA mixtures prepared with foamed binder - 4% water
content -17.5 psi air pressure

196

1

5%-20psi-1-20C
5%-20psi-2-20C
5%-20psi-3-10C
5%-20psi-4-10C
Best Fit C=exp(-8.5218e-006*S^^(0.98427))

0.8

C

0.6
0.4
0.2
0
0

0.5

1

1.5
S

2

2.5

3
x10

8

10

6

Nf

10

4

10

30

2

10

200

20
300

400

10
500

0
Temperature (C)

Microstrain

Figure 64: Push Pull Tests for WMA mixtures prepared with foamed binder - 5% water
content -20 psi air pressure

197

5

APPENDIX D

198

Table 29: TSR test of the WMA mixtures prepared with foamed binder- 1% water content -10 psi air pressure
G=(FE)/F*100

A
B
C
D=(B-C)
E=(A/D)
F
Sample Weight in
SSD
Weight in Volume of
#
Air
Weight
Water
Sample
Gmb
Gmm
Air Voids
1_10_1
3374.7
3428.6
1952.2
1476.4
2.286
2.442
6.40
1_10_2
3377.8
3438
1959
1479
2.284
2.442
6.48
1_10_3
3373.7
3434.9
1958.9
1476
2.286
2.442
6.40
1_10_4
3386.5
3439.1
1961
1478.1
2.291
2.442
6.18
1_10_5
3368.3
3427.6
1951
1476.6
2.281
2.442
6.59
1_10_6
3385.4
3450.9
1955.5
1495.4
2.264
2.442
7.29
CONDITIONED SET
J=
L=((KN=2M/(3.
(G/100)*D B
K
B)/J)*100 M
14*5.9*I) S1
Sample Volume of Initial Dry New SSD
Percent
Tensile
Avg.Tensile
#
Air Voids
Weight
Weight
Saturation Load (Lbs) Strength
Strength
1_10_1
94.46
3374.7
3442.5
71.78
9819.84
303.05
1_10_2
95.79
3377.8
3452.3
77.77
8735.71
268.94
1_10_5
97.28
3368.3
3437.8
71.44
7478.37
230.33
267.44
UNCONDITIONED SET
N=2M/(3.
M
14*5.9*I)
S2
(S1/S2)*100
Avg.
Sample
Tensile
Tensile
Tensile Strength Ratio
#
Load (Lbs) Strength
Strength
(TSR)
1_10_3
13978.63
429.28
62.0
1_10_4
13872.60
430.07
1_10_6
14107.05
433.76
431.04
199

H
Height
(mm)
88.85
89.07
89.29
88.45
89.03
89.18

I=H/25.4
Height
(inch)
3.50
3.51
3.52
3.48
3.51
3.51

Table 30: TSR test of the WMA mixtures prepared with foamed binder- 2% water content -12.5 psi air pressure
G=(FE)/F*100

A
B
C
D=(B-C)
E=(A/D)
F
Sample Weight in
SSD
Weight in Volume of
#
Air
Weight
Water
Sample
Gmb
Gmm
Air Voids
2_12_1
3334.8
3397.7
1929.9
1467.8
2.272
2.442
6.96
2_12_2
3333.3
3399.7
1931.7
1468
2.271
2.442
7.02
2_12_3
3346.9
3412.3
1943.9
1468.4
2.279
2.442
6.66
2_12_4
3339.2
3397.7
1927.9
1469.8
2.272
2.442
6.97
2_12_5
3349.6
3417.4
1942.1
1475.3
2.270
2.442
7.02
2_12_6
3341.5
3409.9
1936.6
1473.3
2.268
2.442
7.12
CONDITIONED SET
J=
L=((KN=2M/(3.
(G/100)*D B
K
B)/J)*100 M
14*5.9*I) S1
Sample Volume of Initial Dry New SSD
Percent
Tensile
Avg.Tensile
#
Air Voids
Weight
Weight
Saturation Load (Lbs)
Strength
Strength
2_12_3
97.84
3346.9
3424.7
79.52
9197.136
283.52
2_12_5
103.64
3349.6
3426.8
74.49
8586.38
265.45
2_12_6
104.95
3341.5
3423.9
78.51
8543.08
264.87
271.28
UNCONDITIONED SET
N=2M/(3.
M
14*5.9*I)
S2
(S1/S2)*100
Avg.
Sample
Tensile
Tensile
Tensile Strength Ratio
#
Load (Lbs)
Strength
Strength
(TSR)
2_12_1 14289.235
441.96
63.9
2_12_2 13682.962
421.08
2_12_4 13381.318
409.72
424.25
200

H
Height
(mm)
88.655
89.105
88.95
89.555
88.6975
88.4425

I=H/25.4
Height
(inch)
3.49
3.51
3.50
3.53
3.49
3.48

Table 31: TSR test of the WMA mixtures prepared with foamed binder- 3% water content -15 psi air pressure
G=(FE)/F*100

A
B
C
D=(B-C)
E=(A/D)
F
Sample Weight in
SSD
Weight in Volume of
#
Air
Weight
Water
Sample
Gmb
Gmm
Air Voids
3_15_1
3331.9
3398.7
1936.4
1462.3
2.279
2.442
6.69
3_15_2
3338.0
3396
1933.6
1462.4
2.283
2.442
6.53
3_15_3
3331.2
3395.2
1935.2
1460
2.282
2.442
6.57
3_15_4
3340.8
3395.4
1934.1
1461.3
2.286
2.442
6.38
3_15_5
3336.7
3391.3
1937.9
1453.4
2.296
2.442
5.99
3_15_6
3357.2
3412.8
1945.8
1467
2.288
2.442
6.29
CONDITIONED SET
J=
L=((KN=2M/(3.
(G/100)*D B
K
B)/J)*100 M
14*5.9*I) S1
Sample Volume of Initial Dry New SSD
Percent
Tensile
Avg.Tensile
#
Air Voids
Weight
Weight
Saturation Load (Lbs)
Strength
Strength
3_15_1
97.89
3331.9
3409.5
79.28
11995.6
371.64
3_15_2
95.49
3338
3405.5
70.69
13100.6
409.36
3_15_5
87.02
3336.7
3405.7
79.29
1.34E+04
411.79
397.60
UNCONDITIONED SET
N=2M/(3.
M
14*5.9*I)
S2
(S1/S2)*100
Tensile
Avg.
Sample
Strength
Tensile
Tensile Strength Ratio
#
Load (Lbs)
(TS)
Strength
(TSR)
3_15_3
12059.76
370.66
107.6
3_15_4
11637.16
357.07
3_15_6
12339.01
380.78
369.50
201

H
Height
(mm)
88.50667
87.75333
89.21667
89.36667
89.00667
88.85667

I=H/25.4
Height
(inch)
3.48
3.45
3.51
3.52
3.50
3.50

Table 32: TSR test of the WMA mixtures prepared with foamed binder- 4% water content -17.5 psi air pressure
G=(FE)/F*100

A
B
C
D=(B-C)
E=(A/D)
F
Sample Weight in
SSD
Weight in Volume of
#
Air
Weight
Water
Sample
Gmb
Gmm
Air Voids
4_17_1
3285.4
3352.7
1904.3
1448.4
2.268
2.442
7.11
4_17_2
3326
3397.7
1927.9
1469.8
2.263
2.442
7.33
4_17_3
3333.1
3404.7
1935.3
1469.4
2.268
2.442
7.11
4_17_4
3340.8
3407
1939.5
1467.5
2.277
2.442
6.78
4_17_5
3333
3398.6
1935.2
1463.4
2.278
2.442
6.73
4_17_6
3318.6
3385.9
1924.9
1461
2.271
2.442
6.98
CONDITIONED SET
J=
L=((KN=2M/(3.
(G/100)*D B
K
B)/J)*100 M
14*5.9*I) S1
Sample Volume of Initial Dry New SSD
Percent
Tensile
Avg.Tensile
#
Air Voids
Weight
Weight
Saturation Load (Lbs)
Strength
Strength
4_17_1
103.03
3285.4
3368.1
80.27
6857.16
212.90
4_17_3
104.49
3333.1
3416.8
80.10
8127.94
249.74
4_17_4
99.44
3340.8
3420.4
80.05
8520.68
264.53
242.39
UNCONDITIONED SET
N=2M/(3.
M
14*5.9*I)
S2
(S1/S2)*100
Tensile
Avg.
Sample
Strength
Tensile
Tensile Strength Ratio
#
Load (Lbs)
(TS)
Strength
(TSR)
4_17_2
12797.44
395.63
61.7
4_17_5
14048.82
433.59
4_17_6
11380.32
349.38
392.87
202

H
Height
(mm)
88.3175
89.24333
88.6975
88.8475
88.325
89.3175

I=H/25.4
Height
(inch)
3.48
3.51
3.49
3.50
3.48
3.52

Table 33: TSR test of the WMA mixtures prepared with foamed binder- 5% water content -20 psi air pressure
G=(FE)/F*100

A
B
C
D=(B-C)
E=(A/D)
F
Sample Weight in
SSD
Weight in Volume of
#
Air
Weight
Water
Sample
Gmb
Gmm
Air Voids
5_20_1
3305.4
3368.7
1917.2
1451.5
2.277
2.442
6.75
5_20_2
3350.9
3410.7
1928.8
1481.9
2.261
2.442
7.40
5_20_3
3342.2
3412.7
1933.4
1479.3
2.259
2.442
7.48
5_20_4
3335.9
3401.3
1938
1463.3
2.280
2.442
6.65
5_20_5
3333.3
3405.9
1939.5
1466.4
2.273
2.442
6.92
5_20_6
3323.3
3385.6
1927.4
1458.2
2.279
2.442
6.67
CONDITIONED SET
J=
L=((KN=2M/(3.
(G/100)*D B
K
B)/J)*100 M
14*5.9*I) S1
Sample Volume of Initial Dry New SSD
Percent
Tensile
Avg.Tensile
#
Air Voids
Weight
Weight
Saturation Load (Lbs) Strength
Strength
5_20_2
109.71
3350.9
3427.8
70.10
10403.7
322.36
5_20_4
97.25
3335.9
3409.2
75.37
10738.21
330.49
5_20_5
101.41
3333.3
3414.9
80.46
10214.06
315.14
322.66
UNCONDITIONED SET
N=2M/(3.
M
14*5.9*I)
S2
(S1/S2)*100
Avg.
Sample
Tensile
Tensile
Tensile Strength Ratio
#
Load (Lbs)
Strength
Strength
(TSR)
5_20_3
11009.98
340.43
97.8
5_20_1
9872.10
305.89
5_20_6
11132.43
343.91
330.07
203

H
Height
(mm)
88.49667
89.09667
88.68333
88.15667
88.87333
88.76333

I=H/25.4
Height
(inch)
3.48
3.51
3.49
3.47
3.50
3.49

BIBLIOGRAPHY

204

BIBLIOGRAPHY
AASHTO Standard Specifications for Transportation Materials and Methods of Sampling and
Testing (Part 2 – Tests), Twenty-Sixth Edition, American Association of State Highway
and Transportation Officials, Washington D.C., 2011.
Abel, F., “Foamed asphalt base stabilization”, 6
State University, 1978.

th

Annual Asphalt Paving Seminar, Colorado

Acott, S.M., “Sand stabilization using foamed bitumen”, Proceeding of 3
Asphalt Pavements for South Africa, 155-172, 1979.

rd

Conference on

Asphalt Academy, “The Design and Use of Foamed Bitumen Treated Materials”, Technical
Guideline 2 (TG2), Pretoria, 2002.
Anderson, R. M., Baumgardner, G., May, R., and Reinke, G., "NCHRP 9-47: Interim Report:
Engineering Properties, Emissions, and Field Performance of Warm Mix Asphalt
Technologies", Transportation Research Board: National Highway Research Council,
Washington, D.C., 2008.
ASTM Designation D1173-53 (Reapproved 1986), “Standard Test Method for Foaming
Properties of Surface-Active Agents (a.k.a. Ross and Miles Test)”, American Society for
Testing and Materials: Philadelphia, PA, 1988.
ASTM Designation D1881-86, “Standard Test Method for Foaming Tendencies of Engine
Coolants in Glassware”, American Society for Testing and Materials: Philadelphia, PA,
1988.
ASTM Designation D3427 – 86, “Standard Test Method for Air Release Properties of Petroleum
Oils”, American Society for Testing and Materials: Philadelphia, PA, 1986.
ASTM Designation D3519-88, “Standard Test Method for Foam in Aqueous Media (Blender
Test)”, American Society for Testing and Materials: Philadelphia, PA, 1988.
ASTM Designation D3601-88, “Standard Test Method for Foam in Aqueous Media (Bottle
Test)”, American Society for Testing and Materials: Philadelphia, PA, 1988.
ASTM Designation D892-74 (Reapproved 1984), “Standard Test Method for Foaming
Characteristics of Lubricating Oils”, American Society for Testing and Materials:
Philadelphia, PA, 1988.

205

Ayman, A., Abbas, A., Nazzal, M., Powers, D., “Laboratory Evaluation of Foamed Warm Mix
Asphalt”, International Journal of Pavement Research and Technology, 5(2), 93-101,
2012.
Bahia, H.U., Miller, T.D., “Establishing a Framework for Analyzing Asphalt Pavements
Sustainability”, Enviroad 2009 Conference, Warsaw, Poland, October 15-16, 2009.
Barel, A., Maib, B.P., Paye, M., Maibach, H.I., “Handbook of Cosmetic Science and
Technology”, Taylo r& Francis, 2005.
Barinov, E.N., “Formation and properties of bituminous foams”, Chemistry and Technology of
Fuels and Oils, 26(10), 544-548, 1990.
Bennert, T., “Evaluation of Warm Asphalt Technology –Feasibility Study”, NJDOT Project
2008-1, Center for Advanced Infrastructure and Transportation, Rutgers University,
2008.
Bhusal, S., “A laboratory study of warm mix asphalt for moisture damage potential and
performances issues”, Ms. Thesis , Oklahoma State University, 2008.
Bissada, A.F., “Structural response of foamed-asphalt-sand mixtures in hot environments”,
Transportation Research Record: Journal of the Transportation Research Board, 1115,
134-149, 1987.
Blades, C., Kearney, E., “Asphalt Paving Principles”, Cornell Local Roads Program, 2004.
Brennen, M.T., Altschaeffl, A.G. and Wood, L.E., “Laboratory Investigation of the use of
Foamed Asphalt for Recycled Bituminous Pavements”, Transportation Research Record:
Journal of the Transportation Research Board, 911, 80-87, 1983.
Bonaquist, R., ”Mix Design Practices for Warm Mix Asphalt”, NCHRP Report 691,
Transportation Research Board of the National Academies. Washington, D.C., 2011.
Bowering, R.H., “Properties and Behavior of Foamed Bitumen Mixtures for Road Building”,
th
AARB 5 Conference, 1970.
Bowering, R.H., Martin, C.L., “Foamed Bitumen: Production and Application of Mixture,
Evaluation and Performance of Mixture”, Proceedings of the Association of Asphalt
Paving Technologists, 45, 453-477, 1976.
Castedo-Franco, L.H., Wood, E.L., “Stabilization with foamed asphalt of aggregates commonly
rd
used in low volume roads”, In: Low-volume roads: 3 international conference,

206

Washington, D.C., Transportation Research Record: Journal of the Transportation
Research Board, 898, 297-302, 1983.
Cervarich, M., “Cooling Down the Mix – New “Warm Mix Asphalt”, Technologies Developed
in Europe”; Hot Mix Asphalt Technology, National Asphalt Pavement Association,
Lanham, MD, March/April.13-16, 2003.
Cervarich, M., “Foaming the Asphalt: New Warm-Mix Technique Challenges Conventional
Wisdom”, Hot Mix Asphalt Technology, National Asphalt Pavement Association,
Langham, MD. 12(4), 23-24, 2007.
Clark,M.F., Culley, R.W., “Evaluation of Air-Blown Asphalts to reduce Thermal Cracking of
Asphalt Pavements”, Proceedings of the Association of Asphalt Paving Technologists,
45,530-555, 1976.
Copeland, A., D'Angelo, J., Dongré, R., Belagutti,S., Sholar,G., “Field Evaluation of High
Reclaimed Asphalt Pavement-Warm-Mix Asphalt Project in Florida”, Transportation
Research Record: Journal of the Transportation Research Board, 2179, 93-101, 2010.
Csanyi, L.H.. “Bituminous mixes prepared with foamed asphalt”, Ames, IA: Iowa State
University, Iowa Engineering Experiment Station, Bulletin 189, 1960.
D’Angelo, J., Harm, E., Bartozsek, J., Baumgardner, G., Corrigan, M., Cowsert, J., Harman, T.,
Jamshidi, M., Jones, W., Newcomb, D., Prowell, B., Sines, R. and Yeaton B., ” Warm
Mix Asphalt: European Practice”, FHWA report no: FHWA-PL-08-007, 2008.
Das, B.M., “Principles of Geotechnical Engineering”, Cengage Learning, Stamford, CT, 7
edition, 2009.

th

De Groot, P., Bowen, C., Koenders, B., Stoker, D., Larsen, O., and Johansen, J., “A Comparison
of Emissions from Hot Mixture and Warm Asphalt Mixture Production”, IRF World
Meeting, Paris, 2001.
Diefenderfer, S. and Hearon, A., “Laboratory Evaluation of a Warm Asphalt Technology for Use
in Virginia”, Report No FHWA/VTRC 09-R11, Virginia Transportation Research
Council, Charlottesville, VA., 2008.
Din 53902-Part 1, “Testing of Surface Active Agents; Determination of Foaming Power; Beating
Method using A Perforated Disc”, German Standards, Edition 81, 1992.
Engelbrecht, J.C., Roberts, F.L., Kennedy, T.W., “ Cold Recycled mixtures, with emphasis on
the curing of foamed specimens – a laboratory study”, Annual Transportation

207

Convention, Session- Maintenance of Transport Infrastructure, Volume 350, Paper 7,
1985.
EPA, “Hot Mix Asphalt Plants Emission Assessment Report”, United States Environmental
Protection Agency, EPA-454/R-00-019, 2000.
FHWA, “Reclaimed Asphalt Pavement in Asphalt Mixtures: State of the Practice”, Federal
Highway Administration, FHWA-HRT-11-021, 2011.
Gandhi T, Amirkhanian S., “Laboratory investigation of warm asphalt binder properties-a
preliminary analysis”, The fifth international conference on maintenance and
rehabilitation of pavements and technological control (MAIREPAV5), Park City, Utah,
USA, p. 475–80. 2007.
German, B.J. and McCarthy, M.J., “Stability of Aqueous Foams: Analysis using Magnetic
Resonance Imaging”, Journal of Agricultural Food Chemistry, 37(5), 1321-1324, 1989.
Gibson, N., Qi, X., Shenoy, A., Al-Khateeb, G., Kutay, M.E.,. Andriescu, A., Stuart, K., Youtcheff,
J., Harman,.T., “Full-Scale Accelerated Performance Testing for Superpave and
Structural Validation”, Report FHWA-HRT-11-045. FHWA, U.S. Department of
Transportation, 2012.
Goh, S.W., You Z., “Mechanical Properties of Warm Mix Asphalt Using Aspha-min”
th
Transportation Research Board (TRB), 87 Annual Meeting Compendium of Papers
DVD, Washington, D.C., 2008.
Guillerme, C., Loisel, W., Bertrand, D., Popineau, Y., “Study of Foam Stability by Video Image
Analysis: Relationship with the Quantity of Liquid in the Foams”, Journal of Texture
Studies, V.24-3, 287-302, 1993.
Haffmans,
B.V.
“http://www.ssco.com.tw/Haffmans/PDF_catalog/QC_instruments/ENIBEMT.pdf “, 2012.
Hanz, A.J., Faheem, A., Mahmoud,E., Bahia, H.U., “Measuring Effects of Warm Mix Additives
using a Newly Developed Asphalt Binder Lubricity Test for the Test”, Transportation
Research Record: Journal of the Transportation Research Board, 2180, 85-92, 2010.
Hassan, M.M., “Life-Cycle Assessment of Warm-Mix Asphalt: an Environmental and Economic
th
Perspective”, Transportation Research Board, 88 Annual Meeting Compendium of
Papers DVD, Washington, D.C., 2009.
He, G. and Wong W., “Decay properties of the foamed bitumens”, Construction and Building
Materials, 20, 866–877, 2006.
208

Heitzman, M., “State of the Practice – Design and Construction of Asphalt Paving Materials with
Crumb Rubber Modifier”, FHWA-SA-92-022, Federal Highway Administration,
Washington, D.C., May 1992.
Hossain, Z., Bhudhala, A, Zaman,M., O'Rear,E., Cross, S., and Lewis, S., “Evaluation of the Use
of Warm Mix Asphalt as a Viable Paving Material in the United States”, Report 06-H0044, Federal Highway Administration, McLean, 2009.
Howard, I.L., Cooley, A., Doyle, J.D., “Laboratory Testing and Economic Analysis of High RAP
Warm Mixed Asphalt”, Mississippi Department of Transportation, FHWA/MS-DOTRD-09-200, 2009.
Huang, B., Zhang,Y., Shu, X., Liu, Y., Ye, P. X., “Neutron scattering for moisture detection in
foamed asphalt”, Journal of Materials in Civil Engineering, 2012, in press.
Hurley, G.C. and Prowell, B.D., “Evaluation of Aspha-Min zeolite for use in warm mix
asphalt”, NCAT Report 05-04, 2005.
Hurley, G.C. and Prowell, B.D., ”Evaluation of Sasobit for use in warm mix asphalt”, NCAT
Report 05-06, 2005.
Hurley, G., “Evaluation of New Technologies for Use in Warm Mix Asphalt”, Master’s Thesis,
Auburn University, Auburn, AL, 2006.
Hurley, G. and Prowell, B., “Evaluation of Potential Processes for Use in Warm Asphalt Mixes,”
Journal of the Association of Asphalt Paving Technologists, Vol. 75, 41-85, 2006.
Hurley, G.C. and Prowell, B.D., “Field performance of warm mix asphalt”, Transportation
th
Research Board (TRB), 87 Annual Meeting Compendium of Papers DVD, Washington,
D.C.,2008.
Hutzler, S., Verbist, G., Weaire, D. Van der Steen, J.A., “Measurement of foam density with AC
capacitance”, Europhysics Letters, 31 (8), pp. 497-502, 1995.
Jackson, S., “Application of Warm Mix Asphalt on Non-Typical Paving Projects”, West
Contracting, 2011.
Jenkins, K.J., “Mix Design Considerations for Cold and Half-warm Bituminous Mixes with
emphasis on Foamed Bitumen”, Ph.D. Dissertation, University of Stellenbosch, South
Africa, 2000.

209

Jones, D., Fu, P., Harvey, J., Halles, F., “Full-Depth pavement reclamation with foamed asphalt:
Final report.”, California Department of Transportation Division of Research and
Innovation Office of Roadway Research, UCPRC-RR-2008-07, 2008.
Jones, D., Wu, R., Tsai, B.W., and Harvey, J.T., “Warm-Mix Asphalt Study: First-Level
Analysis of Phase 2 HVS and Laboratory Testing and Forensic Investigation”, Davis and
Berkeley, CA: University of California Pavement Research Center (RR-2009-02), 2009.
Kanitpong, K., Sonthong, S., Nam, K. Martono, W., Bahia, H., “Laboratory Study on Warm Mix
th
Asphalt Additives”, 86 Annual Meeting Compendium of Papers DVD, Washington,
D.C., 2007.
Kavussi, A., Hashemian, L., “Properties of WMA-Foam Mixes based on Major Mechanical
Tests”, Journal of Civil Engineering and Management, 17(2), 207-216, 2011.
Kendall, M. and Gibbons, J., “Rank Correlation Methods”, Fifth Edition, Oxford University
Press, Oxford, England, 1996.
Kim, Y., Lee H., and Heitzman, M., “Validation of New Mix Design Procedure for Cold Inplace Recycling with Foamed Asphalt for Iowa Department of Transportation”,
Transportation Research Board (TRB) Annual Meeting CD-ROM, 2006.
Kim, Y., Lee H., “Development of Mix Design Procedure for Cold In-Place Recycling with
Foamed Asphalt”, Journal of materials in civil engineering, 116-124, 2006.
Kim, H., Lee, S.J., Amirkhanian, S.N., “Rheology of warm mix asphalt binders with aged
binders”, Construction and Building Materials, 25, 183-189, 2011.
Kutay, M.E., Ozturk, H.I., “Investigation of Moisture Dissipation in Foam-based Warm Mix
Asphalt Using Synchrotron-Based X-Ray Microtomography”, ASCE Journal of Materials
in Civil Engineering, Vol. 24, No 6, pp. 674-683, 2012.
Kutay, M.E., Arambula, E., Gibson, N.H. and Youtcheff, J., “Three-Dimensional Image
Processing Methods to Identify and Characterize Aggregates in Compacted Asphalt
Mixtures”, International Journal of Pavement Engineering, Vol. 11, Issue 6, pp. 511-528.
2010.
Kutay, M.E., Ozturk, H., Abbas, A. and Hu, C., “Comparison of 2D and 3D Image-Based
Aggregate Morphological Indices”, International Journal of Pavement Engineering, Vol.
12, Issue 4, pp. 421-431. 2011.

210

Kristjánsdóttir, O., Muench, S.T., Michael, L. and Burke, G., “Assessing the potential for Warm
Mix Technology Adoption”, Transportation Research Record: Journal of the
Transportation Research Board, No. 2040, Washington DC., 2007.
Kvasnak, A., West, R., Moore,J., Nelson,J., Turner,P. and Tran, N., “Case Study of Warm Mix
Asphalt Moisture Susceptibility in Birmingham”, Transportation Research Record: TRB
th
88 Annual Meeting Compendium of Papers DVD, 2009.
Lamb, H., “Hydrodynamics”, Cambridge, England: Cambridge University Press, pp. 597. 1932.
Lange, C.R., Stroup-Gardiner, M.,” Temperature-Dependent Chemical-Specific Emission Rates
of Aromatics and Polyaromatic Hydrocarbons (PAHs) in Bitumen Fume”, Journal of
Occupational Environmental Hygiene, Cincinnati, Ohio, USA, 4(S1):72-6, 2007.
Larsen, O., Moen, Ø., Robertus, C., and Koenders, B., “WAM Foam Asphalt Production at
Lower Operating Temperatures as an Environmental Friendly Alternative to HMA”, In
rd
Proceedings of the 3 Eurasphalt and Eurobitume Conference, Book 1, Foundation
Eurasphalt, Breukelen, The Neterhlands, 641-650, 2004.
Lee, D.Y., “Treating Iowa's Marginal Aggregates by Foamix Process”, Iowa Department of
Transportation, Ames, HR-212, ISU-ERI-AMES-80221, 1980.
Lee D.Y., ”Treating Marginal Aggregates and Soils with Foamed Asphalt”, Association of
Asphalt Paving Technologists, 50, 211-250, 1981.
Leek, C., Jameson G., “Review of Foamed Bitumen Stabilization Mix Design Methods”,
Austroads Technical Report, Austroads Publication No. AP–T178/11, 2011.
Lesueur, D., Clechc, H., Brosseaudc, A., Suchc, C., Cazacliuc, B., Koendersd, B., Cérinoe, P.J.,
Bonvalletf, J., “Foamability and Foam Stability”, Road Materials and Pavement Design,
5(3), 2004.
Maccarone, S., Holleran, G., and Ky, A., “Cold Asphalt System as an Alternative to Hot Mix”,
th
9 AAPA International Asphalt Conference, 1995.
MaineDOT, “Maine’s Experience Utilizing Full Depth Reclamation with Foamed Asphalt”,
presentation at NESMEA 2004, Portsmouth, NH, 2004.
Mallick, R.B., Kandhal P.S., Bradbury R.L., “Using Warm-Mix Asphalt Technology to
Incorporate High Percentage of Reclaimed Asphalt Pavement Material in Asphalt
Mixtures”, Transportation Research Record: Journal of the Transportation Research
Board, V. 2051, 71-79, 2008.
211

Mallick, R. B., Bergendahl, J. and Pakula, M., “A Laboratory Study on CO2 Emission
th
Reductions through the Use of Warm Mix Asphalt”, Transportation Research Board 88
Annual Meeting, Washington, D.C., 2009.
Marquis, B., Bradbury, R.L., Colson, S., Mallick, R.B., Nanagiri, Y.V., Jonathan S. Gould, J.S.,
Marshall, M., “ Design, construction and early performance of foamed asphalt full depth
th
reclaimed (FDR) pavement in Maine”, Transportation Research Board 82 Annual
Meeting, Washington, D.C., 2003.
MDOT, ”Procedures for Aggregate Inspection”, Michigan Department of Transportation,
Construction and Technology Division, 2009.
Middleton, B., Forfylow, R.W., “Evaluation of warm-mix asphalt produced with the double
barrel green process”, Transportation Research Record: Journal of the Transportation
Research Board, V.2126, 19-26, 2009.
Mohammad, L.N., Abu-Farsakh, M.Y., Wu, Z., Abadie, C., “Louisiana experience with foamed
recycled asphalt pavement base materials.” Transportation Research Record: Journal of
the Transportation Research Board, V. 1832, 17-24, 2003.
Munch, B., Trtik, P., Marone, F., Stampanoni, M., “Stripe and ring artifact removal with
combined wavelet-Fourier filtering”, Optical Express, Vol. 17(10), 2009.
Muthen, K.M., “Foamed Asphalt Mixes: Mix Design Procedure”, CSIR TRANSPORTEK
Report No: CR-98/077, South Africa, 1998.
Nadeau, G., “Warm Mix and the ‘Every Day Counts’ Initiative”, Asphalt Pavement, National
Asphalt Pavement Association (NAPA), pp.16-21., 2012.
Namutebi, M., “Some Aspects of Foamed Bitumen Technology”, Licentiate Thesis, Division of
Highway and Railway Engineering, Department of Transport Science School of
Architecture and the Built Environment, Royal Institute of Technology SE-100 44
Stockholm, 2011.
Namutebi,M., Birgisson, B., Bagampadde, U., “Foaming Effects on Binder Chemistry and
Aggregate Coatability using Foamed Bitumen”, Road Materials and Pavement Design,
12:4, 821-847, 2011.
Nataatmadja, A., “Some Characteristics of Foamed Bitumen Mixes”, Transportation Research
Record: Journal of the Transportation Research Board, Volume.1761, 120-125, 2001.
Nazzal, M.D., Qtaish, L.A., “The Use of Atomic Force Microscopy to Evaluate Warm Mix
Asphalt”, FHWA/OH-2012/19, 2013.
212

NCAT Asphalt E-news,” Warm-Mix and RAP: A Winning Combination”, National Center for
Asphalt Technology NCAT at Auburn University, 24(1), 2012.
NCHRP Report 714, “Special Mixture Design Considerations and Methods for Warm Mix
Asphalt: A supplement to NCHRP Report 673: A Manual for Design of Hot Mix Asphalt
with Commentary”,
Transportation Research Board of the National Academies.
Washington, DC, 2012.
NCHRP, Research Results Digest 374, 2012.
Neu, G.E., "Techniques of Foam Measurement”, Journal of Society of Cosmetic Chemists, V.11,
No.7, Pp. 390-414, 1960.
NIOSH Health Hazard Evaluation Report: Crumb –Rubber Modified Asphalt Paving:
Occupational Exposures and Acute Health Effects (HETA #2001-0536-2864), 2001.
Powers, D., “Internal Report on 2008 Foamed Asphalt WMA Projects”, Ohio Department of
Transportation, 2008.
Prasad, R.J.C., Sreenivasu, T., Rao, M.V.G., “Polar Wavelet-Gaussian Filter for ring artifact
suppression in CT image systems”, International Journal of Computer Science &
Communications Networks, Vol. 1(2), 186-195, 2011.
Prowell, B.D., “Warm Mix Asphalt: The international technology scanning program summary
report”, U.S. Department of Transportation, Federal Highway Administration American
Association of State Highway and Transportation Officials, National Cooperative
Highway Research Program, 2007.
Prowell, B., “Warm-Mix Asphalt: Best Practices”, Presentation at 53
2008.

rd

NAPA Annual Meeting,

Prowell, B.D., Hurley, G. and Frank, B., “Warm Mix Asphalt: Best Practices”, National Asphalt
rd
Pavement Association (NAPA) Quality Improvement Publication 125, 3 Edition, 2012.
Raffaelli, D., “Foamed asphalt base stabilization”, University of California Berkeley Institute of
Transportation Studies, Technology Transfer Program, 2004.
Rand, D.A., “TxDOT Perspective of WMA and Where We are Headed”, 2008 TxAPA Annual
Meeting ,updated for 2009 Corpus Christi, TX.
Ramanujam, J.M. and Jones, J.D., “Characterization of Foamed-Bitumen Stabilization”,
International Journal of Pavement Engineering, 8(2), 111-122, 2007.
213

Rubio, M.C., Martinez, G., Baena, L., Moreno, F., “Warm Mix Asphalt: An Overview”, Journal
of Cleaner Production, 24, 76-84, 2012.
Ruckel, P.J., Acott, S.M., Bowering, R.H., “Foamed-Asphalt Paving Mixtures: Preparation of
Design Mixes and Treatment of Test Specimens”, Transportation Research Record:
Journal of the Transportation Research Board, V. 911, USA, 88-95, 1983.
Ruhl, R., Lindemeier, B., “The German Bitumen Forum”, Heinrich Lauck GmbH, 2006.
Sakr, H.A., Manke, P.G., “Innovations in Oklahoma foamix design procedures”, Transportation
Research Record: Journal of the Transportation Research Board, V. 1034, 26-34, 1985.
Saleh, M., “Characterization of Foam Bitumen Quality and the Mechanical Properties of Foam
th
Stabilized Mixes”, 10 International Conference on Asphalt Pavements (ICAP 2006),
Quebec City, Canada, 2006.
Solaimanian, M., Milander, S., Boz, I., Stoffels, S.M., “Development of Guidelines for Usage of
High Percent RAP in Warm-Mix Asphalt Pavements”, Commonwealth of Pennsylvania
Department of Transportation, Contact No. 510602, Project No. PSU 032, 2011.
Sunarjono, S., “Influence of foamed bitumen characteristics on cold mix asphalt properties”,
Thesis submitted to University of Nottingham for the degree of Doctor of Philosophy,
2008.
Schramm, L.L., Wassmuth, F., “In Foams: Fundamentals and Applications in the Petroleum
Industry”, Advances in Chemistry Series, 1994.
Schramm, L.L., “Emulsions, Foams, and Suspensions: Fundamentals and Applications”,
WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim, 2005.
Sebba, F., “Foams and Biliquid Foams – Aphrons”, Wiley: NewYork, 1987.
Shua, X., Huang, B., Shrum, E.D., Jia, X., “Laboratory evaluation of moisture susceptibility of
foamed warm mix asphalt containing high percentages of RAP”, Construction and
Building Materials, Volume 35, 125–130, 2012.
Su, K., Maekawa, R., Hachiya, Y., “Laboratory Evolution of WMA Mixture for use in Airport
Pavement Rehabilitation”, Construction and Building Materials, 23(7), 2709-2714, 2009.
Xiao, F., Zhao, P.E.W., Amirkhanian, S.N., “Fatigue behavior of rubberized asphalt concrete
mixtures containing warm asphalt additives”,Construction and Building Materials, 23
(2009) 3144–3151 ,2009.
214

Xiao, F., Jordan, J., Amirkhanian,S.N., “Laboratory Investigation of Moisture Damage in Warm
Mix Asphalt Containing Moist Aggregate”, Transportation Research Record: Journal of
the Transportation Research Board, 2126, 115-124, 2010.
Xiao, F., Punith, V.S., Putman, B.J., “Effect of Compaction Temperature on Rutting and
Moisture Resistance of Foamed Warm Mix Asphalt Mixtures”, Journal of Materials in
Civil Engineering, 2012.
Wasiuddin, N.M., Selvamohan, S., Zaman, M.M, Guegan, M.L.T.A., “Comparative Laboratory
Study of Sasobit and Aspha-Min Additives in Warm-Mix Asphalt”, Transportation
Research Record: Journal of the Transportation Research Board, 1998 , 82-88, 2007.
West, R., “Field Testing of Warm Mix Asphalt”, Presentation at the Warm Mix Asphalt
&Recycling Symposium, Sacramento, California, USA, 2009.
Wielinski, J., Hand, A., Rausch, D.M., “Laboratory and Field Evaluations of Foamed Warm-Mix
Asphalt Projects”, Transportation Research Record: Journal of the Transportation
Research Board, 2126, 125-131, 2009.
Wirtgen GmbH, “Wirtgen Cold Recycling Manual”, Windhagen, Germany, 2004.
You, Z., Goh, S.W., Dai, Q., “Laboratory Evaluation of Warm Mix Asphalt”, Final Report,
Michigan Technological University, 2011.
nd

Yunus, C., Boles, M., ”Thermodynamics:An Engineering Approach”, 2
McGraw-Hill, Inc., 1994.

Edition, New York:

Zaumanis, M., “Warm Mix Asphalt Investigation”, MS Thesis, Riga Technical University,
Denmark, 2010.
Zeiada,W.A., Kaloush, K.E., Biligiri, K.P., Reed, J.X., Stempihar, J.J., “Significance of
Confined Dynamic Modulus Laboratory Testing for Asphalt Concrete”, Transportation
Research Record: Journal of the Transportation Research Board, 2010, 9-19, 2011.
Zhao, S. Huang, B., Shu, X., Jia, X., Woods, M.,” Laboratory Performance Evaluation of Warm
st
Mix Asphalt containing High Percentages of RAP”, 91 Annual Meeting of
Transportation
Research
Board,
Washington,
D.C.,
2012.

215