laboratory study of the effect of rap conditioning on the mechanical properties of hot mix asphalt...

26
ORIGINAL ARTICLE Laboratory study of the effect of RAP conditioning on the mechanical properties of hot mix asphalt containing RAP Asmaa Basueny Daniel Perraton Alan Carter Received: 10 July 2012 / Accepted: 14 June 2013 Ó RILEM 2013 Abstract This paper evaluates the effect of reclaimed asphalt pavement (RAP) laboratory conditioning on the rheological properties of recycled hot-mix asphalt. Four different conditioning processes were used on a single RAP source before mixing: unheated RAP, RAP heated at 110 °C in a microwave, RAP heated in a covered pan at 110 °C in a draft oven, and RAP heated in a non-covered pan at 110 °C in a draft oven. Dense graded 20 mm HMA was designed using a PG 64-28 binder and mixed with 25 % of the four different conditioned RAPs. Thermal stress restrained specimen test (TSRST) and complex modulus test were used to characterize RAP conditioning effect. Test results showed that the complex modulus of the four mixes has no different rheological behaviour, and did not affect TSRST results as much. Keywords Reclaimed asphalt pavement Recycling Binders Complex modulus TSRST RAP conditioning 1 Introduction Reclaimed asphalt pavement (RAP) has been used in hot-mix asphalt (HMA) since the 1930s. Unlike with crushed Portland cement concrete or recycled aggre- gates, the possibility of using the old asphalt binder in the newly blended mixtures and, therefore, reducing the required new asphalt content (virgin binder) make the use of RAP in HMA economically attractive [9]. The mid-90s saw the start of the implementation of the Superpave mix design method. The original specifications for Superpave did not include guidance on how to integrate RAP into the new mix design system. Interim recommendations were developed through the FHWA Asphalt Mixture Expert Task Group [7] based on experience and on the performance of Marshall’s mixes with RAP. The specifications were changed in 2002 after the results of an NCHRP research project, Incorporation of RAP in the Super- pave System, became available [15]. AASHTO Stan- dards MP2 (now M323), standard specification for Superpave volumetric mix design for hot mix asphalt, describe how to design HMA with RAP. The guiding principle of the AASHTO standard is that mixtures with and without RAP should satisfy the same requirements. The aggregates provided by RAP are included in the determination of the gradation of the mixture and in the consensus properties (except for the sand equivalent value, which is waived because of the inability to test). The bitumen contained in RAP is regarded as part of the total binder content of the mixture. Numerous research studies have been reported in the literature on the laboratory performance, field performance and pavement design of virgin asphalt A. Basueny D. Perraton A. Carter (&) De ´partement du Ge ´nie de la Construction, E ´ cole de Technologie Supe ´rieure, 1100 Notre-Dame Ouest, Montreal, QC H3C 1K3, Canada e-mail: [email protected] Materials and Structures DOI 10.1617/s11527-013-0127-0

Upload: alan-carter

Post on 11-Dec-2016

215 views

Category:

Documents


2 download

TRANSCRIPT

ORIGINAL ARTICLE

Laboratory study of the effect of RAP conditioningon the mechanical properties of hot mix asphalt containingRAP

Asmaa Basueny • Daniel Perraton • Alan Carter

Received: 10 July 2012 / Accepted: 14 June 2013

� RILEM 2013

Abstract This paper evaluates the effect of reclaimed

asphalt pavement (RAP) laboratory conditioning on the

rheological properties of recycled hot-mix asphalt. Four

different conditioning processes were used on a single

RAP source before mixing: unheated RAP, RAP heated

at 110 �C in a microwave, RAP heated in a covered pan at

110 �C in a draft oven, and RAP heated in a non-covered

pan at 110 �C in a draft oven. Dense graded 20 mm HMA

was designed using a PG 64-28 binder and mixed with

25 % of the four different conditioned RAPs. Thermal

stress restrained specimen test (TSRST) and complex

modulus test were used to characterize RAP conditioning

effect. Test results showed that the complex modulus of

the four mixes has no different rheological behaviour, and

did not affect TSRST results as much.

Keywords Reclaimed asphalt pavement �Recycling � Binders � Complex modulus � TSRST �RAP conditioning

1 Introduction

Reclaimed asphalt pavement (RAP) has been used in

hot-mix asphalt (HMA) since the 1930s. Unlike with

crushed Portland cement concrete or recycled aggre-

gates, the possibility of using the old asphalt binder in

the newly blended mixtures and, therefore, reducing

the required new asphalt content (virgin binder) make

the use of RAP in HMA economically attractive [9].

The mid-90s saw the start of the implementation of

the Superpave mix design method. The original

specifications for Superpave did not include guidance

on how to integrate RAP into the new mix design

system. Interim recommendations were developed

through the FHWA Asphalt Mixture Expert Task

Group [7] based on experience and on the performance

of Marshall’s mixes with RAP. The specifications

were changed in 2002 after the results of an NCHRP

research project, Incorporation of RAP in the Super-

pave System, became available [15]. AASHTO Stan-

dards MP2 (now M323), standard specification for

Superpave volumetric mix design for hot mix asphalt,

describe how to design HMA with RAP.

The guiding principle of the AASHTO standard is

that mixtures with and without RAP should satisfy the

same requirements. The aggregates provided by RAP

are included in the determination of the gradation of

the mixture and in the consensus properties (except for

the sand equivalent value, which is waived because of

the inability to test). The bitumen contained in RAP is

regarded as part of the total binder content of the

mixture.

Numerous research studies have been reported in

the literature on the laboratory performance, field

performance and pavement design of virgin asphalt

A. Basueny � D. Perraton � A. Carter (&)

Departement du Genie de la Construction, Ecole de

Technologie Superieure, 1100 Notre-Dame Ouest,

Montreal, QC H3C 1K3, Canada

e-mail: [email protected]

Materials and Structures

DOI 10.1617/s11527-013-0127-0

mix (i.e., mix containing virgin binder and new

aggregates) [15, 21]. However, published studies on

the effect of RAP conditioning on hot mix asphalt are

rather scanty. When mixing materials in the labora-

tory, the goal is to obtain properties that are represen-

tative of the material produced in a batch plant.

However, no clear recommendations are proposed on

how it should be done, at least based on scientific

evaluation.

A review of the Europe standard method (EN

12697-35) on how to prepare specimens of recycled

mixtures in the laboratory for testing provided that the

RAP materials are preheated in an oven at 110 ± 5 �C

for 2.5 ? 0.5 h, however, this specification does not

state whether or not the RAP materials should be

covered during this short-term aging procedure. Since

asphalt binders react with oxygen from the environ-

ment to create oxidation [11], and become more brittle

(age hardening), it is not certain whether having the

mixes covered (or not covered) when inserted an oven

has an influence on the characteristics of the mixes.

Moreover, the Europe standard which is considered

the only existing guideline that specify how to add

RAP in the laboratory, it does not state if the RAP can

be heated also in a microwave oven or added cold as it

done in some plants in the field. Since RAP contains a

certain amount of binder, the RAP conditioning

process must be controlled in order to ensure that the

characteristics in the lab are similar to those observed

in the field. It should be noted that in the field, most

HMA with the addition of RAP are produced by

different hot mix asphalt plants which dominate the

market nowadays such as batch plants with a separate

heating drum (hot addition), batch plants without a

separate heating drum (cold addition), and drum

mixers. RAP is usually added cold and becomes hot

by contact with overheated aggregates.

2 Objective

The main objective of this study is to evaluate, based

on a rheological perspective, the laboratory perfor-

mance of hot mix asphalt containing 25 % RAP mixed

in a laboratory with four different RAP conditioning

processes of the same source of RAP. Thus, to

evaluate whether RAP conditioning does have an

impact on the characteristics of HMA containing RAP,

a decision was made to prepare samples in the

laboratory, with three different conditioning pro-

cesses: (1) add RAP cold (unheated: UH); (2) add

RAP heated in an oven, but covered (heated-covered:

HC), and (3) add RAP heated in an oven and not

covered (heated not covered: HNC). In addition, It has

been demonstrated that using microwaves for heating

asphalt mixtures is fast, deep, and uniform [1].

Therefore, a fourth conditioning process was investi-

gated: RAP heated in Microwave (HM).

Since thermal cracking is considered a common

phenomenon in cold regions [20] and a recognized

problem with pavements in Canada, it was decided

that our mixes would be evaluated for low temperature

cracking resistance. Complex modulus testing was

used to characterize our mixes. Research described

elsewhere has shown that the dynamic modulus of

asphalt mixtures is associated with major distresses

such as fatigue and low temperature cracking [12].

Moreover, findings in the literature have shown that

complex modulus has been widely applied in inves-

tigating the effects of RAP content and RAP type on

the mechanical properties of HMA [8]. Alvarez et al.

[2] showed that the complex modulus varies with the

mode of introduction of RAP into the mix plant. It is

important to verify whether the RAP conditioning

method does not significantly modify the recycled

HMA properties. This verification was accomplished

by the evaluating engineering properties, such as low

temperature cracking resistance with the TSRST, and

linear viscoelastic (LVE) properties with the complex

modulus (stress–strain distribution effect on

pavement).

3 Properties of hot mix asphalt with RAP

Daniel and Lachance [4] found that, as expected,

adding 15 % RAP increased the stiffness of the

mixture, and that mixtures containing 25 and 40 %

RAP did not follow the expected trends. Instead, the

norm of the complex modulus curves was similar to

that of the control mixture. On the other hand, other

researchers have found a contrasting tendency for

complex modulus test results. For example, Shah et al.

[23] observed no increases in stiffness with the

addition of 15 % RAP, compared with the control

mix, while the addition of 25 and 40 % RAP resulted

in an increase in the stiffness of the mix.

Materials and Structures

Huang et al. [9] conducted a study on the laboratory

fatigue characteristics of asphalt surface mixtures

containing screened RAP. They found, especially

when the RAP content represented less than 30 % of

the total mix, that adding RAP generally improves the

fatigue performance of asphalt mixtures. This conclu-

sion appears to contradict the common belief—the

more RAP, the more brittle the mixture, thus lower the

fatigue resistance. However, Similar results were

reported by Sargious and Mushule [22], and results

were even supported from the NCHRP 9–12 study by

some of the results of fatigue tests [15].

Other studies have shown the lack of consistency of

results seen in past research. Tam et al. [25] found that

mixtures with RAP are less resistant to low-tempera-

ture cracking than are non-recycled mixtures, while

Kandhal et al. [12] found no significant difference in

cracking performance in both cases. Sargious and

Mushule [22] found that recycled mixes perform better

than virgin mixtures with respect to thermal cracking.

In general, most studies on laboratory-produced

mixtures conclude that recycled mixes exhibit greater

resistance to rutting than virgin mixes [12, 13]. From

field studies, the rutting performance of recycled

mixes has been found to be better than for virgin mixes

[24], while other studies show no significant differ-

ences between the rutting behaviour of recycled and

virgin mixes [15].

It can be noted from the above discussions that the

performance of recycled mixes in terms of stiffness,

fatigue, thermal cracking or rutting could be better,

worse, or similar to that of the corresponding virgin

mix. For this reason, the authors of the present paper

believe that the RAP conditioning method could lead

to variations in results. A testing program was

conducted to evaluate the impact of RAP heating

conditioning process on the rheological behaviour of

recycling asphalt mixtures.

4 Test materials and methods

In this experimental program, a dense graded 20 mm

HMA commonly used as a base course in Quebec

(GB20) was designed with a PG 64-28 binder and

25 % of conditioned RAP. The design asphalt binder

grade (PG 64–28) was used in this study with no

decrease by one increment on both the high- and low-

temperature grades as it is suggested by Kandhal and

Foo (2008) when using RAP percentage between 15

and 25. Experimental results of a study conducting by

Shah et al. [23] indicated that the binder grade PG

64-22 can be used for RAP contents up to 40 %.

Additional research concluded that a significant

increase in the stiffness of the mix was observed at

high, intermediate and low temperatures when no

change to the virgin binder was made for only the

higher RAP contents ([40 %) [14, 15]. Also, TSRST

is affected by the type of binder. It well known that the

low temperature of the bitumen grade (PG-L) is

related to the fracture temperature of asphalt mixtures.

Hence, our mixtures with binder grade PG 64-28 will

be evaluated for thermal cracking.

The choice of the amount of RAP to add is based on

the fact that 25 % RAP is routinely added in HMA in

Canada. A single source of RAP materials was used.

The RAP was tested for specific gravity and asphalt

content as well as gradation (LC 21-040 standard). The

RAP aggregates were recovered using two methods:

solvent extraction and the ignition oven method. The

RAP source had a nominal maximum aggregate size of

10 mm and an asphalt content of 3.9 % [percent by the

total mass of mixture (m/m)].

The selected virgin binder (PG 64-28) is a medium

grade asphalt binder that can be used in warm

climates. The mixing temperature for the virgin binder

is 155 ? 2 �C, and the corresponding compaction

temperature is 145 ? 2 �C (LC 26-003). It is to be

noted that the LC method is a standard method used by

MTQ, Canada. Five different classes of virgin aggre-

gates were selected to produce the GB20 asphalt

mixtures. Gradations and other properties of the virgin

aggregates are shown in Table 1. The particular

aggregates were selected based on past experience.

The main purpose of this study was to show

whether the initial RAP temperature and virgin

aggregate temperature (with the same final tempera-

ture of the mixture after production at 160 �C) have an

influence on the rheological properties of the final mix.

Four conditioning were carried out with adding RAP

material at two different temperatures: 25 and 110 �C.

The first temperature corresponds to the usual cold

introduction of RAP in the plant mixer. The second

temperature corresponds to a temperature specified in

the standard EN 12697-35. Table 2 presented initial

temperatures of the RAP, virgin aggregate and virgin

binder during the production of HMA in this labora-

tory investigation.

Materials and Structures

For the microwave heating process, a 700 W

microwave oven was used. The following steps were

done to heat our sample from room temperature to

110 �C in the microwave oven:

• Place the sample in the container.

• Place the container in the microwave oven and

heat for a specific time. For our case, we use 5 min

at the beginning.

• When the microwave oven stops, remove RAP

from the oven and stir it for few seconds. After

that, measure the temperature.

• Return the container to the oven and reheat for five

more minutes.

• Repeat the process until to reach to 110 �C.

It was found that the required time to heat our

design mass is 20 min.

When cold RAP is added, the new aggregate is

superheated up to 300 �C in the oven prior to mixing

with the RAP (Table 2). The cold RAP materials is

added to the superheated aggregates and mixed before

adding the virgin binder, and finally, the components

are blended until the aggregate is thoroughly coated by

the binder. When hot RAPs are incorporated (110 �C),

the virgin aggregates are heated to 25 �C above the

virgin binder mixing temperature prior to the mixing

with RAP and virgin binder.

4.1 Characterization of the RAP material used

4.1.1 RAP material preparation

For this research project 200 kg of RAP was sampled

at an asphalt plant near Montreal. The sampling was

done in accordance with the testing method LC

21-010: Sample rate of The Ministere des Transport

du Quebec (MTQ). Afterwards, the RAP was stored in

30 kg buckets in the laboratory.

Table 1 Properties of

virgin aggregates (reported

by supplier)

d/D 10/20 5/10 0/5 2.5/5 Filler

Mineralogy Granitic Granitic Limestone Granitic Limestone

Sieve size Percent passing

28 mm 100 100 100 100 100

20 mm 96 100 100 100 100

14 mm 49 100 100 100 100

10 mm 12 92 100 100 100

5 mm 2.0 9.0 94 90 100

2.5 mm 1.0 3.0 58 2.0 100

1.25 mm 0.0 2.0 29 1.0 100

630 lm 0.0 2.0 14 1.0 100

315 lm 0.0 2.0 10 1.0 100

160 lm 0.0 1.0 8.0 1.0 100

80 lm 0.8 0.7 7.0 0.4 98

% Absorption (m/m) 1.0 1.4 0.6 1.4 1.0

Bulk specific gravity 2.867 2.842 2.766 2.762 2.700

Table 2 Initial temperatures of RAP, virgin materials, and final mixture during HMA production in laboratory

Unheated

(UH)

Microwave

(HM)

Heated cover

(HC)

Heated non-cover

(HNC)

T virgin aggregates (�C) 300 180 180 180

T RAP (�C) 25a (24 h) 110a (20 min) 110a (3 h) 110a (3 h)

T virgin binder (�C) 155 155 155 155

T in a mixer (�C) 160 160 160 160

a The mass of RAP required for preparing a slab is about 6,300 g

Materials and Structures

In the laboratory, two bucket of 30 kg each was

homogenized by mixing with the aid of a concrete

mixer (Fig. 1). The material was then deposited on a

clean, non-absorbent surface. The material was

homogenized manually with a square head shovel

before being divided by quartering in sample mass

reduced (see Fig. 1). The separated samples were

placed in sealed plastic bucket until when they are

needed for different laboratory tests.

4.1.2 Extraction of RAP binder and aggregates

The asphalt binder content of the RAP was determined

using two methods: ignition oven method (LC 26-006)

and solvent extraction (LC 26-100). Table 3 illustrates

the asphalt contents obtained from each of the

extraction methods. The mean values for the asphalt

binder content resulting from both the ignition and

solvent extraction methods were close to the technical

data provided by the supplier. Since the supplier’s

binder content falls in between our lab results, their

binder content was used in the mix design.

The as-received gradation (before extraction) of the

RAP material was determined using the LC 21-040

method. Homogenized RAP material was split to

obtain eight 1,000-g samples, and each sample was

oven-dried at 50 �C before the sieve analysis. The

RAP material gradation is shown in Table 4. It can be

seen that the RAP contains 43 % coarse aggregates,

retained on the 5 mm sieve, less than 1 % of particles

passing on the 80 lm sieve before extraction and a

nominal maximum size aggregate (NMSA) of 10 mm.

The gradation of the aggregate part of the RAP was

also measured after bitumen extraction, ignition and

solvent, and the results are presented in Table 4, and

graphically in Fig. 2a. The results support two

tendencies in the gradation: (a) the intermediate sieve

sizes (5, 2.5, 1.25, 0.630 mm) have more variability

than other sieves, and (b) there is a higher amount of

fines particles after extraction than before extraction.

The purpose of the comparison shown in Table 4

and Fig. 2a is to identify how the asphalt cement

Fig. 1 RAP material preparation

Table 3 RAP binder content by ignition and solvent method

From ignition

(repetition: n = 3)

From solvent

(repetition: n = 2)

% Asphalt binder

(m/m)

SD % Asphalt binder

(m/m)

SD

Asphalt binder content

3.90 0.006 3.66 N.A.

Materials and Structures

present in the RAP material could affect the gradation

of recycled asphalt mixtures. It is assumed that as RAP

is added to a hot mix asphalt mixture, the asphalt

binder of the RAP will tend to separate and disperse in

the mix because of heat. Consequently, the added

virgin aggregate will receive some coating, and

basically, the added RAP reduces the virgin asphalt

needed for the mixtures. As the extracted asphalt

binder leaves the RAP aggregates, the gradation curve

tends to shift to a smaller size (to the left). The shape of

the gradation curve remains the same since the asphalt

binder itself has no gradation, and the components of

the RAP which could constitute a gradation are left

behind after extraction. The only difference is that the

Table 4 Comparison of

gradation before and after

extraction with ignition

oven and solvent method

Sieve size RAP aggregates

RAP-material ‘‘As Received’’

(repetition: n = 8)

Recovered from ignition

(repetition: n = 4)

Recovered from solvent

(repetition: n = 3)

% passing SD % passing SD % passing SD

14 mm 100 0.00 100 0.00 100 0.00

10 mm 98 0.93 98 0.50 99 0.58

5 mm 57 5.21 71 2.89 69 4.00

2.5 mm 34 4.60 52 2.87 51 4.51

1.25 mm 20 2.70 41 1.73 40 3.51

630 lm 11 1.64 32 0.82 31 2.08

315 lm 4.1 0.82 23 0.96 23 1.53

160 lm 1.3 0.16 16 0.58 15 0.58

80 lm 0.4 0.05 10 0.21 10 0.61

100 mm

500 mm 180 mm

(D)

Microwave Non-Covered Covered Cold

(C)

(A) (B)

Fig. 2 Representation of our mix design results and its

production, a gradation of average samples of RAP material

and RAP-aggregate, b 20 mm RAP mixture gradation,

c compacted slabs with different RAP conditioning before

coring, d cylindrical cored sample

Materials and Structures

particles are no longer coated, and are thus smaller by

approximately one sieve size [19]. In addition, it can

be stated that this before and after extraction curves

represent the extremes that will occur when the RAP

material is added to a hot mix asphalt. For this study,

the after extraction gradation was used for the mix

design. That is because, as explained previously, the

RAP particles are separated to smaller sizes and

produce more fine particles because of heating, and

that should be considered in the gradation curve of the

final mix. The produced fine aggregates will dramat-

ically affect the volumetric characteristics of the

resulting mix such as increasing the air void content in

the case of it did not take into account in the mix

design. Problem of compacting using gyratory com-

pacted will be appeared, the required air void at the

number of gyrations 10, 120, and 200 will never be

achieved to be in the limit of the specification. Briefly,

it is improbable that using the gradation of RAP before

extraction would be able to meet gradation and

volumetric requirements of the mix design.

4.1.3 RAP aggregate specific gravity

It is important to obtain the bulk specific gravity

(Gsb) of the combined aggregates because it is one

of the inputs for calculating voids in mineral

aggregate (VMA). The Gsb of the combined aggre-

gates is determined from specific gravities tests

conducted on samples from each component in the

mixture. The Gsb of the RAP aggregate was

estimated using the recommended methodology in

NCHRP Report 452 [16], Recommended Use of

Reclaimed Asphalt Pavement in Superpave Mix

Design Method: Technical manual. The estimated

Gsb values were calculated from a maximum

specific gravity (Gmm) tests on the RAP samples

(the Gmm method). In the Gmm method, the RAP

Gmm and the asphalt content of the RAP were used

to estimate Gsb. The effective specific gravity of the

RAP aggregate (Gse) can be calculated based on the

Gmm and asphalt content values as follow:

Gse ¼100� Pb

100Gmm� Pb

Gb

ð1Þ

where, Pb is the RAP binder content, percent by total

mass of mixture; and Gb is the specific gravity of RAP

binder (assumed to be 1,020 in this study)

This Gse is used to calculate Gsb as follows:

Gsb ¼Gse

Pba�Gse

100�Gb

� �þ 1

ð2Þ

where, Pba is the absorbed binder, percent by Gsb

weight of aggregate (assumed to be 1.4 % in this study

this percent presents about 63 % of the typical water

absorption value of the aggregate).

The RAP materials maximum specific gravity

(Gmm) was measured by performing the Gmm test

(LC 26-045) and it was found to be 2.602. Then, the

estimated Gsb of the RAP aggregate can be calculated

to be 2.679.

4.1.4 Properties of recovered asphalt binder

The binder was extracted from ‘‘as-received’’ RAP

using the procedure recommended in Quebec’s stan-

dard LC 25-001. 200 grams was recovered using the

binder recovery process. The properties of the RAP

binder were evaluated using selected Superpave

binder test procedures including the penetration test

(ASTM test methods D5-06e1), the viscosity test

(ASTM test methods D4402-06), the dynamic shear

rheometer (DSR) test (AASHTO TP5-98), and the

bending beam rheometer (BBR) test (AASHTO test

method T 313).

The penetration test was conducted to measure the

consistency of RAP binder. The viscosity test results

were performed to establish the mixing and compac-

tion temperatures and for grading of hot mix asphalt

mixtures. The DSR test was used to characterize

asphalt binder properties at high and intermediate

service temperatures, while he bending beam rheom-

eter (BBR) test was used to characterize it at low

service temperatures. In this study, the DSR test was

performed at 76 and 82 �C, in an attempt to estimate

the SUPERPAVE performance grade of the RAP

binder. The actual strain and torque were measured

and used to calculate various LVE parameters,

including the norm of the complex modulus, G�j j,and phase angle, d. These test values are commonly

used to calculate two measures: the rutting factor

G�j j= sin d, and the fatigue cracking factor, G�j j sin d.

Test temperatures used for the bending beam

rheometer (BBR) test were -18 �C, and -24 �C.

Superpave binder specification includes a maximum

of 300 MPa for creep stiffness, and the decrease in

Materials and Structures

stiffness leads to smaller tensile stresses in the asphalt

binder and less chance for low temperature cracking

[26].

The results of the penetration test show that the

penetration value of the RAP binder is 21 (0.1 mm).

The rational viscometer test results presented in

Tables 5 shows that the RAP binder has a viscosity

of 2.586 and 0.471 Pa.s at 135 and 165 �C,

respectively.

The results of the DSR test are shown in Table 6.

The G*/sin d value measured using DSR at 76 �C was

found to be 4.0 kPa. The RAP binder would have an

estimated high temperature grade of PG82 as the

binder meet the requirements of G*/sin d[ 2.2 kPa at

the temperature 76 �C. Table 7 shows the results of

the BBR test. From this table, it can be seen that the

stiffness value for the RAP binder is 141 MPa. The

measured creep stiffness value meet the 300 MPa

maximum requirement of MP1 for binders to satisfy a

low temperature grade of -18 �C. Also, the measured

m values are lower than the 0.300 minimum estab-

lished to fulfill the same grade requirement.

5 Recycled asphalt concrete mixes

RAP material, virgin asphalt, and virgin aggregate

were proportioned to produce dense graded 20 mm

HMA (GB20) mix design. In the present study, the

procedures described in LC 26-045 (Determination of

the maximum density) and LC 26-003 (Determination

of the ability of compaction of hot mix asphalt by

means of the gyratory compactor) regarding the

preparation of HMA specimens were followed.

Table 8 gives all procedures used to introduce the

RAP material in the recycled HMA mixes. The

properties of the produced mixtures are summarized

in Table 9. The voids in mineral aggregates (VMA)

and air voids of mixes calculated from the Superpave

gyratory compactor results at 200 gyrations ranged

from 12.3 to 14.5 % and from 2.2 to 4.8 % (v/v),

respectively, and the gradations were consistent. The

gradations of all the 20 mm recycled asphalt mixtures

are illustrated in Fig. 2b; as can be seen in the figure,

the design gradation for the produced recycling

asphalt mixtures did not violate the Superpave control

points recommended for use in the LC method. All

mixes meet LC requirements for the mix design.

6 Sample preparation

Four parallelepiped slabs, 500 mm wide by 180 mm

long by 100 mm in height were prepared for each mix,

and compacted in the laboratory using an MLPC slab

compactor with 5 % target air voids (Fig. 2c). These

slabs were then cut and cored to prepare specimens for

TSRST and complex modulus testing. For the TSRST,

the specimens were 60 mm in diameter, and were saw-

cut to a final height of 250 mm according to AASHTO

designation TP10. For complex modulus tests, the

specimens were 75 mm in diameter and saw-cut to a

final height of 120 mm (ASSHTO TP 62-03). Two

specimens having a diameter of 75 mm and two of

60 mm were extracted from each slab, as shown in

Fig. 2d. The cored specimens were glued to two

aluminum plates for mechanical connection before the

TSRST and complex modulus tests.

Table 5 Viscosity of recovery bitumen binder of RAP as it is

(UH)

Test

temperature

(�C)

Speed

(RPM)

Torque

(%)

Viscosity

(mPa.s)

Average

viscosity

(mPa.s)

135 12 64.3 2,679 2,586.0

135 12 61.8 2,579

135 12 60.0 2,500

140 12 45.8 1,908 1,838.7

140 12 43.4 1,808

140 12 43.2 1,800

145 12 35.0 1,475 1,382.0

145 12 32.3 1,345

145 12 31.8 1,325

150 12 24.0 1,000 986.0

150 12 23.5 979

150 12 23.5 979

155 12 18.0 750 750.0

155 12 18.0 750

155 12 18.0 750

160 12 14.1 588 588.0

160 12 14.1 588

160 12 14.1 588

165 12 11.3 471 471.0

165 12 11.3 471

165 12 11.3 471

Materials and Structures

For this experimental program, two other repli-

cates for only the mix of heated microwave (HM)

were produced and labeled as mix No. 2 and mix

No. 3, respectively. Samples from the second mix

(No. 2) and the third mix (No. 3) have very high air

content as compared to samples from the first mix

(No. 1). The difference in air voids is being

probably due to the variability of the RAP material,

as discussed in this paper. Two samples of each mix

were prepared and also exhibited higher air void

ratios [5.7 % (v/v) B air voids B 7.3 % (v/v)]. Each

sample was subjected to the TSRST in order to

investigate the effect of air voids on the TSRST

results.

Table 6 DSR test results of recovery bitumen binder of RAP as it is (UH)

Replicate number Temperature (�C) G*/sin d (kPa) Phase angle, d, (�) Min allow G*/sin d Test status

1 76 4.0915 71.3 2.2 Passed

2 82 2.1087 73.8 2.2 Failed

Table 7 BBR test results of recovery bitumen binder of RAP as it is (UH)

Replicate

number

Temperaturea

(�C)

Creep stiffness,

S (60)

m value

m (60)

Average

S (60) (SD)

Average

m (60) (SD)

Min

allow

m (60) test

results

1 -18 133 0.320 141 0.326 0.300 Passed

2 -18 149 0.331 0.300 Passed

1 -24 313 0.292 259.5 (30.892) 0.290 (0.003) 0.300 Failed

2 -24 259 0.291 0.300 Passed

3 -24 260 0.287 0.300 Passed

a Fluid bath temperature at 60 s; target test temperature = -18 �C and -24 �C

Table 8 Recycled asphalt mixtures [GB20 HMA; PG 64-28; 25 % of RAP; asphalt content of 4.5 % (m/m); virgin binder of 3.5 %

(m/m)]

RAP conditioning Designation Mixing Methoda

Unheated UH Cold RAP ? superheated virgin aggregates

300 �C ? asphalt binder, 155 �C

Cold RAP was mixed with virgin aggregates

superheated to 300 �C. The asphalt binder,

heated at 155 �C, was added when the

blended temperature of RAP material and

virgin aggregates reach the mixing

temperature of 180 �C

Heated in

microwave

HM Heated RAP in a microwave 110 �C ? heated

virgin aggregates, 180 �C ? asphalt binder,

155 �C

Heated RAP in a microwave up to 110 �C was

mixed with virgin aggregates heated to

180 �C. After mixing, the asphalt binder

heated at 155 �C was added and then mixed

Heated in a

covered pan in a

forced draft oven

HC Heated RAP in a non-covered pan in a forced

draft oven, 110 �C ? heated virgin

aggregates, 180 �C ? asphalt binder, 155 �C

Heated RAP to 110 �C in a covered pan in the

oven and heated virgin aggregates to 180 �C

were mixed before mixing with the heated

asphalt binder at 155 �C

Heated in a non-

covered pan in a

forced draft oven

HNC Heated RAP in a covered pan in a forced draft

oven, 110 �C ? heated virgin aggregates,

180 �C ? asphalt binder, 155 �C

Heated RAP to 110 �C in a non-covered pan in

the oven and heated virgin aggregates to

180 �C were mixed and then the asphalt

binder heated at 155 �C was added and then

mixed

a The mixture of RAP, virgin aggregates, and virgin binder were mixed for 2 min. Finally the mixture was heated in a pan non-

covered for 2 h in order to maintain the target temperature for compacting

Materials and Structures

7 Test equipment and procedure

7.1 Tensile stress specimen test (TSRST)

In this project, the thermal stress restrained spec-

imen test (TSRST) was used to evaluate the low-

temperature-cracking resistance of asphalt mixtures.

The principle underlying the test system is to keep

the length of the specimen constant during cooling.

A cylindrical specimen is mounted in the load

frame enclosed by the cooling cabinet. As the

temperature decreases, the thermal stress inside the

sample increases, until the specimen ultimately

breaks.

Table 9 Mixture properties

Mix No.

1 (UH)

Mix No.

1 (HM)

Mix No.

1 (HC)

Mix No.

1 (HNC)

LC

standard

GB20

Grading–percent passing

28 mm 100 100 100 100 100

20 mm 99 99 99 99 95–100

14 mm 82 82 82 82 67–90

10 mm 66 66 66 66 52–75

5 mm 40 40 40 40 35–50

2.5 mm 25 25 25 25 –

1.25 mm 18 18 18 18 –

630 lm 13 13 13 13 –

315 lm 11 11 11 11 –

160 lm 8.3 8.3 8.3 8.3 –

80 lm 6.8 6.8 6.8 6.8 4.0–8.0

RAP conditioning Unheated Heated in

microwave,

110 �C

Heated in a covered pan

in the oven, 110 �C

Heated in a non-covered

pan in the oven, 110 �C

N.A.

% Total binder content (m/m) 4.5 4.5 4.5 4.5 4.5b

% Virgin binder (m/m) 3.5 3.5 3.5 3.5

Binder grade PG 64-28 PG 64-28 PG 64-28 PG 64-28 N.A.

% RAP binder (m/m) 1.0 1.0 1.0 1.0

% RAP binder content (m/m) 3.9 3.9 3.9 3.9

Maximum specific gravity

(Gmm)

2.622 2.617 2.612 2.613

Bulk specific gravity (Gmb) 2.517 2.492 2.521 2.550

Effective bitumen content:

Vbe (%)a10.2 10.2 10.2 10.2 10.2

Voids in mineral aggregate:

VMA (% of bulk volume)

13.9 14.5 13.4 12.3

% Voids filled with asphalt:

VFA (% of VMA)

70.5 66.9 73.6 80.7

Voids at Superpave gyratory compactor (% of total volume)

10 gyrations 16.7 16.7 14.9 14.4 C11

120 gyrations 5.6 6.4 5.0 4.0 4–7

200 gyrations 4.0 4.8 3.5 2.2 C2

a Expressed by part of volume of the HMA without voidb The main target for the MTQ regulation is to reach the prober Vbe

Materials and Structures

7.2 TSRST results

TSRST test results include the fracture temperature,

fracture stress, slope of the thermally induced stress,

and the transition temperature as shown in Fig. 3a. At

the beginning of the test, there is a relatively slow

increase in thermal stress due to a relaxation of the

asphalt mixture. However, beyond a given tempera-

ture, known as the transition temperature, the rela-

tionship between the thermally induced stresses and

the temperature is approximately linear. The transition

temperature is defined as the temperature where the

material changes from elastic to visco-elastic behav-

iour, or vice versa [10]. The transition temperature and

the slope of the stress–temperature curve below the

transition temperature (slope 2) may play a major role

in characterising the rheological behaviour of the

asphalt mixtures at low temperatures. Only the

fracture stress, the fracture temperature and the slope

are discussed below.

The tensile stress and the temperature at the

breaking point represent the fracture stress and the

fracture temperature, respectively. Stress–temperature

curves for all samples tested under TSRST are

presented in Fig. 3b, c, and a summary of the results

are reported in Table 10. In general, the fracture

temperature (Tf) observed for the mixture tested

exhibited minor variability between replicates. Gen-

erally, the following results are observed for the first

mix (No. 1): (1) the average fracture strength and

fracture temperature for UH specimens are 5.0 MPa

and -32.8 �C, respectively; (2) for the HM specimen,

the fracture strength and the fracture temperature are

5.9 MPa and -35.3 �C, respectively; (3) for the HC

mix, the fracture strength and the fracture temperature

are 5.0 MPa and -32.3 �C, respectively; (4) for the

HNC mix, the fracture strength and the fracture

temperature are 5.5 MPa and -35.3 �C, respectively.

The fracture temperature is a good indicator of the low

temperature performance of laboratory-produced

mixes. No significant variation was found in the

results, indicating that RAP conditioning has a limited

effect on low temperature behaviour.

To compare samples of different air voids, the

parameter r0, which is the strength at zero porosity,

was used. The r0 will give an indication if the ranking

between mixes will change or not. To adjust the

strength according to the porosity (air voids) of the

mixes, the following equation was used [27]:

ru ¼ r0 1� Pð Þ3 ð3Þ

where ru and P represent strength and porosity.

The r0 calculated for each replicate is illustrated in

Fig. 4a and reported in Table 10. In Fig. 4, the values

between parentheses indicate the air voids content

(v/v) in each sample. As shown in Fig. 4a, by

considering the calculated (r0) values, we found that

there is no difference between conditioning processes.

For analyzing the results statistically, STAT-

GRAPHICS Centurion XVI program was used. The

two sample comparing procedure test was used to

determine whether or not there is a significant

difference between the fracture temperatures (t test).

The p value is the probability of obtaining a test static

(or data point) that is significantly different from the

null hypothesis. When the computed p value is less

than 0.05, we can reject the null hypothesis because

the differences observed in the mean values are true

errors or differences and not due to random sampling

errors.

Results of the two sample t-tests on the fracture

temperatures of the RAP mixes samples showed that

no significant differences between the four RAP

conditioning mixes: UH, HM, HC, HNC, as shown

in Table 11. The fracture temperatures of the RAP

mixes produced with different conditioning appeared

to be similar.

7.2.1 Slopes of stress–temperature curves

In order to determine, as precisely as possible, the

value of the slope 2, we will truncate the experimental

values by keeping only the data better representing the

linearity of the stress–temperature curve at the end of

the test associated to the failure point. As a first step,

we subtract the experimental data associated with the

beginning of the test in the range of temperatures from

5 to -10 �C. Subsequently, we determine the evolu-

tion of the coefficient of regression (R2) by reducing

progressively the data from -10 �C, until the calcu-

lated value of R2 reach a maximum. At the end of the

process, we calculate the value of the slope 2 and

report it in Table 10. In the same manner, to fix the

value of the slope 1, we subtract the experimental data

associated with the end of the test and keep only the

data that are representative of the linearity at the

beginning of the test (in the range of temperatures

from 5 to -20 �C). Then, we still determine the R2 of

Materials and Structures

the linear approximation after reducing progressively

the filtered sampling data from -20 �C, until the

calculated R2 reaches a to the maximum. Then, the

value of the slope 1 is calculated and reported in

Table 10.

Figure 4b, c shows slope 1 and slope 2 for all

samples tested in this study, respectively. Slope 1

values range from -0.011 to -0.015 (MPa/�C), which

reveals no clear trend between fracture stress and

fracture temperature regarding RAP conditioning

method. Nevertheless, slope 1 is an indicator of the

stress relaxation potential of the mix: lower is their

absolute value, higher is their relaxation potential.

Curiously, an asphalt mixture prepared with RAP

Fig. 3 TSRST test results,

a typical results, b TSRST

thermal stresses for all tested

samples of the first mix (No.

1), c TSRST thermal stresses

for tested samples of HM

mix produced by mix No. 1,

mix No. 2, and mix No. 3

Materials and Structures

Ta

ble

10

Su

mm

ary

of

all

con

dit

ion

ing

pro

cess

RA

Pm

ixtu

res

TS

RS

Tre

sult

s

Mix

ture

var

iab

leID

UH

HM

HC

HN

C

Mix

No

.1

Mix

No

.1

aM

ixN

o.

2M

ixN

o.

3M

ixN

o.

1M

ixN

o.

1

Sla

bai

rv

oid

s(%

)3

.93

.46

.86

.33

.83

.5

Sam

ple

nu

mb

er1

21

12

12

12

12

Sam

ple

air

vo

ids

(%)

1.9

1.9

1.1

6.6

7.3

5.7

6.1

1.4

2.4

0.9

1.7

Fra

ctu

rest

ress

(MP

a)5

.34

.85

.93

.83

.44

.03

.45

.24

.95

.65

.4

Mea

nfr

actu

rest

ress

(MP

a)5

.0–

3.6

3.7

5.0

5.5

Fra

ctu

rete

mp

erat

ure

(�C

)-

33

.8-

31

.9-

35

.3-

33

.2-

32

.1-

32

.6-

32

.4-

33

.0-

31

.5-

36

.2-

34

.4

Mea

nfr

actu

rete

mp

erat

ure

(�C

)-

32

.8-

35

.3-

32

.6-

32

.5-

32

.3-

35

.3

Fra

ctu

rep

osi

tio

n(m

m)

19

0fr

om

top

30

fro

m

top

50

fro

m

top

20

fro

m

edg

e

17

5fr

om

top

60

fro

m

edg

e

50

fro

m

edg

e

12

5fr

om

top

12

0fr

om

top

Isn

ot

clea

r

25

fro

m

top

Sam

ple

P(%

)1

.91

.91

.16

.67

.35

.76

.11

.42

.40

.91

.7

r0

(MP

a)5

.65

.16

.14

.74

.34

.84

.15

.45

.35

.85

.7

Av

erag

er

0(M

Pa)

5.3

5–

4.5

4.4

55

.35

5.7

5

Slo

pe

1(M

Pa/

�C)

-0

.01

3-

0.0

15

-0

.01

2-

0.0

11

-0

.01

2-

0.0

12

-0

.01

3-

0.0

14

-0

.01

4-

0.0

13

-0

.01

4

Slo

pe

2(M

Pa/

�C)

-0

.29

5-

0.3

02

-0

.31

7-

0.2

36

-0

.21

1-

0.2

41

-0

.21

2-

0.3

04

-0

.32

6-

0.3

13

-0

.32

4

Tra

nsi

tio

nte

mp

erat

ure

(�C

)-

24

.0-

25

.0-

25

.8-

25

.4-

24

.4-

25

.1-

24

.3-

24

.2-

26

.1-

27

.7-

27

.6

Cro

sste

mp

erat

ure

(�C

)-

16

.1-

17

.2-

17

.7-

17

.9-

17

.2-

17

.1-

17

.4-

17

.1-

17

.6-

19

.3-

18

.6

Pp

oro

sity

,r 0

stre

ng

that

zero

po

rosi

tya

Sam

ple

No

.2

(mix

No

.1

)w

asb

rok

end

uri

ng

test

Materials and Structures

Fig. 4 Other TSRST test

results a fracture strength at

zero porosity, b slope 1,

c slope 2, and d TSRST

average fracture

temperature for all

conditioning RAP mixtures

Materials and Structures

conditioned in the microwave (HM) showed a

decrease in its ability to relax stresses, which implies

that the properties of the bitumen were much more

affected. Figure 4c also shows that slope 2 has a

relatively wide range (from -0.211 to -0.326 MPa/

�C). As expected, the decrease in slope 2 is more

related to air void content than to the RAP condition-

ing process.

7.2.2 Fracture temperatures

The relationship between the fracture temperature and

RAP conditioning is presented in Fig. 4d. Preliminary

TSRST results indicate that there is no significant

difference in the fracture temperatures of the four

short-term aged conditions. The maximum difference

in fracture temperature was observed between the

unheated mixture (UH) and the heated non-covered

(HNC) mixture, and came in at about 2.5 �C. The

difference between the unheated and HM (mix No. 1)

was also about 2.5 �C. The mixture prepared with a

RAP material heated in the oven with no cover (HNC)

reached a lower fracture temperature than other

asphalt mixture tested. The higher fracture tempera-

ture recorded in the testing program was associated

with the asphalt mixture prepared using the covered

RAP material process (HC). Lower fracture temper-

atures values mean better low temperature behaviour,

as shown for the HNC and microwave (HM) mixes.

Based on our results, the RAP conditioning could

slightly affect the low temperature behaviour of

asphalt mixture.

Finally, as reported in Table 10, by increasing the

air void content, the fracture stress decreased, but it

still in the magnitude of the fracture stresses resulting

from other samples of low air void content. Moreover,

the effect of fracture temperature thought to be a much

lesser extent.

7.3 Complex modulus test

The complex modulus test is performed to determine

the LVE behaviour of asphalt mixtures at various

temperatures and different loading speeds. For visco-

elastic materials, such as asphalt mixes, the stress–

strain relationship under a continuous sinusoidal

loading can be defined by the complex modulus E�.The complex modulus is defined as the ratio of the

amplitude of the sinusoidal stress of pulsation xapplied to the material r ¼ r0 sin xtð Þ and the ampli-

tude of the sinusoidal strain eðtÞ ¼ e0 sin xt � /ð Þ that

results in a steady state [6, 18]:

E� ¼ re¼ r0eixt

e0ei xt�/ð Þ ¼ E1 þ iE2 ð4Þ

where, E1 is the storage modulus and E2 is the loss

modulus.

The modulus (length of vector) of the complex

number is defined as the dynamic modulus E�j j, the

norm of the complex modulus, where r0 is the

maximum stress amplitude and e0 is the peak recov-

erable strain amplitude:

E�j j ¼ r0

e0

ð5Þ

The complex modulus was evaluated using un-

confined uniaxial tension–compression tests, with a

servo-hydraulic testing system (MTS 810, TestStar II)

used, The specimens were subjected to sinusoidal

oscillating axial loading in both tension and compres-

sion at constant amplitude (50 9 10-6 m/m), Three

extensometers placed 120� apart were used during E�

testing to improve the accuracy of the results, The tests

were performed at different temperatures (-35, -25,

-15, -5, 5, 15, 25, 35 �C), and at each temperature, a

frequency sweep of eight frequencies (20, 10, 3, 1, 0.3,

0.1, 0.03, 0.01 Hz) was done.

In our testing program, the norm of complex

modulus ( E�j j) of the produced RAP mixes was

determined on two specimens of each mix, with air

voids ranging from 1.5 to 3.5 % (2.5 ± 1.0 %),

Authors believe that only 1.0 % variation could have

a minor effect on the results and it was not far from the

current 0.5 % accepted air void variation as mention

by [3].

Table 11 Two sample t test comparisons of fracture temper-

atures of 25 % RAP mixtures with different RAP conditioning

Pairs tested p value Conclusion

UH vs. HM 0.8132 NSD

UH vs. HC 0.6692 NSD

UH vs. HNC 0.2021 NSD

HM vs. HC 0.4406 NSD

HM vs. HNC 0.0977 NSD

HC vs. HNC 0.1213 NSD

NSD No significant difference

Materials and Structures

7.3.1 Data acquisition and treatment of the signal

At each data acquisition, data is collected for two

consecutive cycles. The time interval is adapted to have

100 points per cycle. These experimental points are not

exactly on a sine curve due to experimental variations

and non-linear behaviour of bituminous mixtures.

Experimental data related to the force and the displace-

ment measured by the load cell and the three extens-

ometers, respectively are then treated as approximate

sinusoidal curves defined by the following equation:

y ¼ y0 þ yA � sinðxt þ /Þ ð6ÞThus, the part of frequency is constant, and the

other parameters of the equation are calculated by the

least squares method.

In the treatment file, the signal of the stress is

obtained from the recorded force by dividing this

value by the cross section area of the sample. This is

regarding to the fact that the test is conducted in

homogenous conditions. Moreover, the deformation

signal is considered as the average value of the three

values obtained from the three extensometers. The

data acquisition is stored in blocks of rows in sequence

for each pair of cycles of solicitation and in column

dedicated to different variables: time, stress, strain,

and temperature in an excel file. Using a macro

calculation developed with Microsoft excel, the data

will be processed by the least squares method to

calculate the approximate sinusoidal signals of stress

and deformation as the average of the three values

obtained from the three extensometers. Then, each

treatment gives a point of the result in the treatment

file that defines the parameters of approximate signals

and the mechanical characteristics of the material, for

each pair of cycles analyzed.

7.4 Complex modulus results

The results can be plotted on a master curve. To that

end, each isothermal curve can be shifted in frequency

in order to obtain a single master curve at a reference

temperature, Tref . The master curve was plotted as a

function of the equivalent frequency fe based on the

assumption that the asphalt mixtures exhibit a ther-

morheologically simple behaviour, which means that

the Time–Temperature Superposition Principle

(TTSP) is applicable. The expression of fe is given

by the following equation [5]:

fe ¼ aTðT; TrefÞ � fr ð7Þ

where aT is the shift factor at temperature T , Tref is the

reference temperature: aTðTref ; TrefÞ ¼ 1, fr is the real

frequency of solicitation.

For example, the construction of the master curve

of the mix UH by the shifting procedure at Tref ¼5:8 �C is illustrated in Fig. 6a.

The shift factor at a temperature T, named aTðTÞ,used for the construction of the master curve can be

determined by means of Eq. (8). For a given reference

temperature Tref , the Williams, Landel and Ferry

(WLF) Eq. (8) gives log aT as a function of three

constants C1, C2, T , and Tref�Cð Þ [17]:

logðaTÞ ¼�C1 T � Trefð ÞC2 þ T � Trefð Þ ð8Þ

7.5 Modulus in the Cole–Cole plane, in the black

space and master curve

The complex modulus and other parameters, deter-

mined at eight temperatures and eight frequencies, can

be used to determine nine parameters included in the

2S2P1D rheological model that characterize the

asphalt concrete response in the linear visco-elastic

domain. The nine parameters are Einf , Ezero, d, k, h, b,

s0, C1, and C2, and Table 12 shows the nine of them

obtained using the 2S2P1D model for the tests

performed on two samples from the four hot recycling

mixtures prepared with different RAP processes. The

model’s parameters were determined by obtaining the

best fit curve for the measured complex modulus

values plotted in the Cole–Cole and Black diagrams.

Table 12 Parameter of different mixtures (mean value of two

samples tested per mix)

Conditioning UH HM HC HNC

Parameters 2S2P1d

Einfini (MPa) 35,950 36,300 34,500 34,800

Ezero (MPa) 115 82 120 105

d 1.80 1.85 1.70 1.83

k 0.177 0.177 0.177 0.177

h 0.500 0.544 0.520 0.530

b 500 500 500 500

s0 (s) 0.20 0.20 0.20 0.155

C1 27.88 26.24 21.97 24.16

C2 185.48 175.81 152.53 164.08

Materials and Structures

The fitting of laboratory data according to the Cole–

Cole and Black diagrams are shown in Figs. 5b and

Fig. 6, respectively.

A detailed 2S2P1D model and its parameters are

not discussed in this paper, as they have already been

presented in the literature by Olard and Di Benedetto

[17].

Statistical parameters such as, R2, the standard error

of predicted values (se) divided by the standard

deviation of measured values (sy) were used in this

study to perform the goodness-of-fit statistics for the

2S2P1D predicted model in arithmetic scale. The R2

parameter is a measure of correlation between the

measured and the predicted values, and its function is

to determine the level of accuracy of the fitting model,

and thus a higher value of R2 is indicative of higher

accuracy. Table 13 shows the goodness-of-fit statistics

(R2) for the different RAP conditioning mixes. It was

Fig. 5 Representation of some complex modulus test results, a construction of the master curve of E�j j at Tref ¼ 5:8 �C with the

shifting procedures for the mix UH, b Cole–Cole plane for all four mixes, experimental results and 2S2P1D

Materials and Structures

Fig. 6 Black space diagram

for all four mixes,

experimental results and

2S2P1D model

Materials and Structures

observed R2 = 0.9871–0.9989, and therefore the

2S2P1D model showed a good correlation with the

values of E�j j of the mixes, This finding indicates that

this linear visco-elastic rheological model can predict

the bituminous mixes precisely.

The test results presented according to the 2S2P1D

rheological model effectively discriminate mixes

according to their rheological properties. When the

asphalt mixtures are tested within their linear visco-

elastic domain of behaviour, the results will fit on a

single curve in the Cole–Cole plane as well as in the

Black space. As can be seen in Figs. 5b and Fig. 7, our

results fit well on a single curve, which means the tests

were properly performed.

However, when the results are shown in a Black

space (Fig. 6), a scattering of the results can be seen

for high phase angle values, which represent results at

high temperatures or low frequencies. This can be

attributed to the bitumen type. The PG 64-28 used in

this study contains polymer (SBS); it was shown by

Olard and Di Benedetto [17] that mixes made with

polymer-modified binder sometimes exhibit a behav-

iour that does not conform to time–temperature

superposition principle (TTSP). Even if all results

cannot be presented on a single line in a Black space

domain, the results are still considered good since they

fit on a single curve on a Cole–Cole plane (Fig. 5b) as

well as on a single master curve plotted at a reference

temperature using a shifting procedure. In this case,

this property is called the ‘‘Partial Time–Temperature

Superposition Principle’’ (PTTSP), as the shifting

procedure gives a unique and continuous master curve

only for the norm of the modulus.

The master curves for the complex modulus for the

four mixes UH, HM, HC, and HNC are shown in

Fig. 8a. A visual inspection of the curves indicates that

for most frequencies (or temperatures), the dynamic

modulus of the mixtures containing the same percent-

age of RAP (25 %), and prepared with different

conditioning RAP addition processes are similar and

do not vary greatly. The highest modulus values are

observed for the mixture prepared with the addition of

heated-microwave RAP and with the cold process.

This trend is for equivalent frequencies higher than

0.1 Hz, and the HC mix is very slight lower in this

range. The non-covered RAP condition mixture shows

the slight lowest modulus.

7.6 Data analysis for complex modulus

By comparing two samples procedure test on the

modulus of the RAP conditioning samples, at 3 Hz,

and at different three temperatures: 15, 25, and 35 �C,

results from these t tests showed that the differences in

the mean E�j j were found to be statistically not

significant at 3 Hz and at all test temperatures.

Figure 9 shows the E�j j values at three test temper-

atures conducting at 3 Hz. The p values for these two

sample t tests are shown in Table 14. Similar statis-

tical analysis were conducted at 20 Hz and at the same

three temperatures yielded to almost comparable

conclusions, except the mixes (mix HM vs. mix

HNC, and mix HC vs. mix HNC) at 25 �C, were found

to be statistically significant as shown in the low p

values in the comparison between them. The p values

for the other two sample t tests examined at 20 Hz are

shown in Table 14. The average E�j j and coefficients

of variation (CV) of the replicate specimens for each

mixture at each test temperature and at 3 and 20 Hz are

shown in Table 15. As shown in this table the CV of

the complex modulus data from different RAP con-

ditioning range from 2.2 to 19.5 % and from 1.6 to

26.7 % at 3 and 20 Hz, respectively.

7.7 Conditioning effect coefficient C�CEC

In this research, in order to objectively compare the

results of complex modulus of mixes with RAP in

numbers, a condition effect coefficient, C�CECwas

calculated. The calculation of the conditioning effect

coefficient (C�CEC) was developed by Di Benedetto [5],

and is defined as the ratio between the complex

Table 13 Goodness-of-fit statistical analysis for the four

recycled mixtures

RAP conditioning mix se/sy R2

UH 0.0791 0.9937

HM 0.1134 0.9871

HC 0.0328 0.9989

HNC 0.0820 0.9933

Materials and Structures

modulus of a specific mix, in our case the recycled

mixture made with a specific condition, E�S:C: (i.e.,

HM, or HC, or HNC) at the equivalent frequency fe

[defined by Eq. (7)], and the complex modulus of a

reference mix, herein the unheated process (UH) at the

same frequency fe as written in Eq. (9):

Fig. 7 Master curves for

the four mixes, experimental

results and 2S2P1D

Materials and Structures

C�CEC feð Þ ¼E�S:C:E�UH

i:e:; E�S:C: can be E�HM or E�HC or E�HNC

� � ð9Þ

C�CEC is calculated at an equivalent frequency fe. It

means that the complex modulus of the mixture

prepared with a specific condition at frequency fe�SC

and complex modulus of the UH at fe�U:H: must be

considered at the same frequency fe ¼ fe�S:C: ¼fe�U:H: for the calculation.

C�CEC is a complex number, as shown in Eq. (10). Its

norm, calculated by Eq. (11), is the ratio of the norms

of the complex modulus of the recycled mixture made

with a specific condition to the E�j j of the mix

Fig. 8 Representation of

other complex modulus test

results for the recycled

mixtures prepared by four

different RAP processes,

a Master curves of the

2S2P1D model

(Tref ¼ 5:0 �C), b C�CEC

�� ��versus equivalent frequency

(Tref = 5.0 �C), c phase

angle versus equivalent

frequency (Tref ¼ 5:0 �C)

Fig. 9 Average complex dynamic modulus of RAP mixes,

f = 3 Hz

Materials and Structures

prepared with the unheated RAP (UH) material. Its

phase angle, determined by Eq. (12), is the difference

between the phase angle of the recycled mixture made

with a specific condition and the phase angle of the UH

material.

C�CEC ¼ C�CEC

�� ��eiuCEC ð10Þ

C�CEC

�� �� ¼ E�S:C:E�UH

�������� ð11Þ

/CEC ¼ /E�S:C:� /E�

UHð12Þ

7.8 Relationship between C�CEC and equivalent

frequency (fe ¼ aT � fr)

The norm of the complex conditioning effect coeffi-

cient C�CEC is plotted for all mixes in Fig. 8b in

accordance to the 2S2P1D model. This figure shows

that the norm of C�CEC for UH (reference mix, so

C�CEC

�� �� ¼ 1), HM, HC, and HNC mixtures are very

close at high frequencies and/or low temperatures. It

should be noted that a C�CEC

�� �� value of one mean that

no difference exists between the conditioning pro-

cesses under comparison. These results are in accor-

dance with the TSRST results presented in the

previous section, which indicate no significant differ-

ence between the four investigated RAP conditioning

situations at low temperatures. Nevertheless, at low

frequency, from 1E-4 to 1 Hz, corresponding to high

temperature condition, the RAP conditioning could no

longer be considered as negligible. Results presented

in Fig. 8 seem to show a different tendency of what we

found for high frequencies. Noted that, results analysis

of Fig. 8 are really tricky by the fact that the modelling

process used to get that curves is very sensitive to

errors due to the calibration procedures, especially for

very low frequencies and high temperatures.

Table 14 Two sample t test results for the norm of the com-

plex modulus

Pairs

tested

T (�C) f = 3 Hz f = 20 Hz

p value Conclusion p value Conclusion

UH vs.

HM

15 0.6526 NSD 0.6864 NSD

25 0.5883 NSD 0.5675 NSD

35 0.2836 NSD 0.3160 NSD

UH vs.

HC

15 0.7867 NSD 0.8148 NSD

25 0.3569 NSD 0.3435 NSD

35 0.8358 NSD 0.2244 NSD

UH vs.

HNC

15 0.3904 NSD 0.4112 NSD

25 0.1095 NSD 0.1329 NSD

35 0.5138 NSD 0.4114 NSD

HM

vs.

HC

15 0.7631 NSD 0.8448 NSD

25 0.4114 NSD 0.2250 NSD

35 0.1294 NSD 0.7079 NSD

HM

vs.

HNC

15 0.3075 NSD 0.3030 NSD

25 0.1862 NSD 0.0268 SD

35 0.4191 NSD 0.5003 NSD

HC vs.

HNC

15 0.3050 NSD 0.3382 NSD

25 0.1826 NSD 0.0212 SD

35 0.1766 NSD 0.1856 NSD

Table 15 Average

complex dynamic modulus

of mixtures

Mix T (�C) f = 3 Hz f = 20 Hz

E�j j (MPa) CV for the mean (%) E�j j (MPa) CV for the mean (%)

UH 15 5,925 6.2 9,833 5.1

25 2,307 4.8 4,769 3.4

35 743 17.0 1,715 12.5

HM 15 6,173 9.1 10,070 5.1

25 2,535 19.5 5,058 12

35 574 18.5 1,327 26.7

HC 15 6,023 4.3 9,962 4.6

25 2,165 2.9 4,547 1.9

35 764 2.2 1,793 7.6

HNC 15 5,361 11.9 9,137 8.9

25 1,759 14.7 4,041 1.6

35 664 10.1 1,587 3.3

Materials and Structures

As we noted previously, our mixes did not follow

the concept of TTSP at high temperatures (low

frequencies) and that part could be a sensitive part of

our modelling. Therefore, in order to check the

accuracy of our modelling results for the three

mixtures (HM, HC, HNC), we compute the ‘‘true’’

values of C�CEC

�� �� according to the experimental data

obtained for the three high testing temperatures (15,

25, and 35 �C) and for all tested frequencies. To do

this, we still use the UH mix data, as a baseline. The

results are shown in Fig. 10. For HM and HC mixes,

‘‘true’’ values of C�CEC

�� ��shows that values slightly

balance around the equality line ( C�CEC

�� �� = 1.0) like

what we obtained from modelling (Fig. 8b). For HNC

mix, C�CEC

�� �� values results from experimental data

(‘‘true’’ values) show that the tendency confirms the

results obtained from modelling calculations, that the

HNC gives always a lower stiffness values than the

reference mix (UH). Nevertheless, we found a little

disperse tendency between the variation of C�CEC

�� ��from ‘‘true’’ calculated values (Fig. 10) to that from

modelling (Fig. 8b).

Figure 10 shows that the C�CEC

�� �� values, obtained

from experimental data and modelling, for the HM at

low frequencies (less than 1 Hz) and high temperature

(35 �C) seem to be more effective and give us

indication that the condition might be significant.

Nevertheless, we need to keep in mind that for these

condition the stiffness of mixes are very low, less than

300 MPa. At this level, the measurement capacity of

the apparatus (load cell precision) and setting of the

hydraulic press are not optimized.

In overall, our analysis using the C�CEC

�� �� resulting

from experimental and modelling data showed that the

difference modulus between mixes is low even at low

frequencies (high temperatures).

8 Conclusion

This paper investigates the influence of warming RAP

materials with different conditions on the mechanical

properties of the manufacture recycled asphalt mix-

tures containing 25 % RAP. Four asphalt mixtures

including one RAP source, one RAP content (25 %),

one type of binder (PG 64-28), four RAP conditioning

processes were investigated in this study.

The TSRST has been carried out on all mixtures

including two additional mixes with high air voids,

and the complex modulus, the key parameter in the

mechanistic-empirical design guide, was also con-

ducted on the four different mixes at different

temperature and frequencies.

Based on the analysis of TSRST laboratory data,

the following conclusions are made:

• The low-temperature behaviour of the recycled

asphalt mixtures is not significantly affected by

RAP conditioning processes. A maximum

Fig. 10 The true C�CEC

�� �� values versus frequency at different

high temperatures: 15, 25, and 35 �C

Materials and Structures

difference of about 2.5 �C in fracture temperature

was observed between the HNC and the UH mixes.

• Fracture strength is highly influenced by air void

content. Fracture strength was greater for mixes

with lower air voids, as compared with those with

higher air voids. Also, the air void affects the

fracture temperature, albeit to a much smaller

extent.

• The binder grade PG 64-28 can be used for 25 RAP

content percent since the fracture temperatures for

the four mixed were found to be ranged from

-31.5 to -36.2 �C, and these values are lower

than the low temperature grade of the used virgin

binder (-28 �C).

• An analysis of the influence of RAP conditioning

on the complex modulus of the recycled mixtures

is carried out with the help of the two coefficients,

C�CEC. The main results are:

• The modulus differences between the four inves-

tigated recycled mixtures are very low over all the

frequencies higher than 1.0 Hz whatever the RAP

condition process, and were found to be not

statistically significant. At low frequency (lower

than 1.0 Hz) and/or high temperature, the influ-

ence of heated RAP in a pan covered can be

considered as low from the results of the C�CEC

�� ��calculated either from the model (13 %) or from

the experimental data (5 %).

• In the same range of frequency (lower than

1.0 Hz), the C�CEC

�� �� difference between the UH

mix and the HNC mix are close to 30 %. Never-

theless, its ‘‘true’’ value shows only less than 15 %

variation between the two mixes at 35 �C but it

reach to 25 % at 25 �C and it is in the same

magnitude of what we get referred to the C�CEC

�� ��from the modelling. It is probably could be

explained by the very sensitive accuracy of the

model. For the materials made with HM RAP

added, it seems to exhibit slightly different with a

maximum modelling C�CEC

�� �� value equal to 11 %

and its ‘‘true’’ C�CEC

�� �� value obtained from exper-

imental data showed that the difference could be

higher or lower this value. But by considering that

the stiffness of all the mixes is very low at low

frequencies (less than 1.0 Hz), less than 300 MPa,

it means that the effect of heating RAP by

microwave is also low.

• The repeatability of complex modulus is quite

good, as the coefficient of variation for the E�j jdata ranged from 1.6 to 27 %.

The main finding of this study is the four proposed

conditions of RAP adding to the virgin aggregates

(cold, heated in a microwave, heated in an oven in a

pan covered, and heated in an oven in a pan non

covered) had little effect on the changes in the 25 %

RAP added mixtures stiffness and its resistance to

thermal cracking. Moreover, this study can answer the

question concerning finding the best laboratory RAP

heating methods by studying the four proposed RAP

conditioning. Based on our results, we can not propose

a specific method from the four methods to be used in

the laboratory since each method has its advantage and

disadvantages from the degree of handling and the

required time saving. For example, for RAP added

cold, we found that it is a terrible way for handling,

since we have to superheat the virgin aggregate.

Heating RAP in a microwave is a fast way, but

sometimes the size of the specimen could be a problem

for some type of microwaves. From our point of view,

covered heating RAP in an oven could be the best.

Nevertheless the specific time (3 h) for heating is

important.

Finally, additional research is needed to validate

these results for other RAP sources and other RAP

contents. Also, it is recommended for RAP source

containing polymer to be studied in detail.

Acknowledgments This work was funded by the Missions

Sectors in Egypt and the Ecole de Technologie Superieure. The

authors would like to thank the companies in Quebec that provided

us with the materials and with all the needed data for the project.

References

1. Al-Ohaly AA (1987) Laboratory evaluation of microwave

heated asphalt pavement materials. Ph.D thesis, University

of Washington, Seattle

2. Alvarez C, Bonneau D, Dupriet S, Le Noan C, Olard F

(2008) Very high rate (50%) in hot mix and warm mix

asphalts for sustainable road construction. In: Proceedings

of the 4th eurasphalt and eurobitume congress, Copenhagen,

Denmark

3. Shah A, McDaniel RS, Huber GA, Gallivan VL (1998).

Investigation of properties of plant-produced reclaimed

asphalt pavement mixtures. Transp Res Rec 1998:103–111

4. Daniel JS, Lachance A (2004) Rheological properties of

asphalt mixtures containing recycled asphalt pavement

Materials and Structures

(RAP). In: Transportation research board annual meeting

proceeding, TRB paper no.: 04-4507

5. Delaporte B, Di Benedetto H, Chaverot P, Gauthier G (2007)

Linear viscoelastic properties of bituminous materials: from

binders to mastics. Assoc Asphalt Paving Technol 76:

488–494

6. Di Benedetto H, Partl MN, Francken L, De la Roche Saint

Andre C (2001) Stiffness testing for bituminous mixtures.

Mater Struct 34(2):66–70. doi:10.1007/BF02481553

7. FHWA Superpave Mixture Expert Task Group (1997)

Guidelines for the design of Superpave mixtures containing

reclaimed asphalt pavement (RAP). http://www.utexas.edu/

research/superpave/articles/rap.html. Accessed 5 Oct 2011

8. Guthrie W, Cooley D, Eggett D (2007) Effects of reclaimed

asphalt pavement on mechanical properties of base mate-

rials. Transp Res Rec 2005:44–52

9. Huang B, Kingery WR, Zhang Z, Zuo G (2004) Laboratory

study of fatigue characteristics of HMA surface mixtures

containing RAP. In: Transportation research board annual

meeting proceeding, TRB paper no.: 04-4088

10. Isacsson U, Zeng H (1998) Low-temperature cracking of

polymer-modified asphalt. Mater Struct 31:58–63

11. Jaffee BI (2001) Implementation of the SUPERPAVE(TM)

level 1 mixture design system in the Cooper Union Asphalt

Technology Laboratory by classifying an asphalt binder and

compacting samples in the gyratory compactor. M.E. thesis,

The Cooper Union for the Advancement of Science and Art,

New York

12. Kandhal PS, Rao SS, Watson DE, Young B (1995) Perfor-

mance of recycled hot mix asphalt mixtures in the state of

Georgia. NCAT report 95-1. National Centre for Asphalt

Technology, Auburn

13. Malpass GA (2003) The use of reclaimed asphalt pavement

in new Superpave asphalt concrete mixtures. Ph.D thesis,

North Carolina State University, Raleigh

14. McDaniel R, Soleymani H, Shah A (2002) Use of reclaimed

asphalt pavement (RAP) under superpave specifications: A

regional pooled fund study. FHWA/IN/JTRP-2002/6

15. McDaniel R, Soleymani H, Anderson RM, Turner P, Pet-

erson R (2000) Recommended use of reclaimed asphalt

pavement in the Superpave mix design method. NCHRP

Web document no. 30. TRB, National Research Council,

Washington DC

16. McDaniel R, Anderson RM (2001) Recommended use of

reclaimed asphalt pavement in the Superpave mix design

method: technician’s manual. NCHRP report 452. Wash-

ington DC

17. Olard F, Di Benedetto H (2003) General ‘‘2S2P1D’’ model

and relation between the linear viscoelastic behaviours of

bituminous binders and mixes. Road Mater Pavement Des

4(2):185–224

18. Pellinen TK, Witczak MW (2002) Stress dependent master

curve construction for dynamic (complex) modulus. Assoc

Asphalt Paving Technol 71:281–309

19. Potyondy AJ (1996) Recycling waste roofing material in hot

mix asphalt pavement. M.Sc. thesis, Technical University of

Nova Scotia, Canada

20. Raad L, Saboundjian S, Sebaaly P, Epps J (1998) Thermal

cracking models for AC and modified mixes in Alaska.

Transportation research record, no. 1545. J Transp Res

Board 1629:117–126

21. Robert FL, Kandhal PS, Brown ER, Lee DY, Kennedy TW

(1996) Hot mix asphalt materials, mixture design, and

construction. National Asphalt Pavement Association

Education Foundation, Lanham

22. Sargious M, Mushule N (1991) Behaviour of recycled

asphalt pavement at low temperatures. Can J Civil Eng 18:

428–435

23. Shah A, McDaniel RS, Gerald AH, Gallivan VL (2007)

Investigation of properties of plant-produced reclaimed

asphalt pavement mixtures. Transportation research record,

no. 1998. J Transp Res Board:103–111. doi:10.3141/

1998-13

24. Sullivan J (1996) Pavement recycling executive summary

and report. Report no. FHWA-SA-95-060. Federal Highway

Administration, Washington, DC

25. Tam KK, Joseph P, Lynch DF (1992) Five-year experience

of low-temperature performance of recycled hot mix.

Transportation research record, no. 1362. J Transp Res

Board:56–65

26. The Asphalt Institute (2003) Performance graded asphalt

binder specification and testing, SP-1. The Asphalt Institute,

Lexington

27. Young JF, Mindess S, Gray RJ, Bentur A (1998) The sci-

ence and technology of civil engineering materials. Pre-

ntice-Hall, Upper Saddle River

Standards

28. AASHTO (1998) Standard test method for determining the

rheological properties of asphalt binder using a dynamic

shear rheometer. Test method TP5-95, American Associa-

tion of State Highway and Transportation Officials

(AASHTO), Washington, DC

29. AASHTTO (2001) Standard test method for thermal stress

restrained specimen tensile strength. Test method TP10.

American Association of State Highway and Transportation

Officials (AASHTO), Washington, DC

30. AASHTTO (2003) Standard method of test for determining

dynamic modulus of hot-mix asphalt concrete mixtures.

AASHTO TP 62-03, American Association of State High-

way and Transportation Officials (AASHTO), Washington,

DC

31. AASHTO (2008) Standard method of test for determining

the flexural creep stiffness of asphalt binder using the

bending beam rheometer (BBR). Test method T 313-08,

American Association of State Highway and Transportation

Officials (AASHTO), Washington DC

32. ASTM (2006a) Standard test method for viscosity deter-

mination of asphalt at elevated temperatures using a rational

viscometer (D4402-06). American Society for Testing and

Materials (ASTM), West Conshohocken

33. ASTM (2006b) Standard test method for penetration of

bituminous materials (D5-06el)

34. EN 12697-35 (2007) Bituminous mixtures: test methods for

hot mix asphalt—part 35: Laboratory mixing

35. LC 21-010, Sample rate

36. LC 21-040, Gradation analysis

37. LC 25-001, Recovery of bitumen solution by Evaporation

rotative

Materials and Structures

38. LC 26-003, Determination of the ability of compaction of

hot mix asphalt by means of the gyratory compactor

39. LC 26-006, Determination of asphalt content by ignition

oven

40. LC 26-100, Determination of content asphalt (by extraction

with trichloroethylene)

41. LC 26-350, Aggregate grading analysis

42. LC 26-045, Determination of the maximum density

Materials and Structures