Influence of moisture conditioning on healing of asphalt binders

Construction and Building Materials 146 (2017) 360–369

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Influence of moisture conditioning on healing of asphalt binders Umme Amina Mannan ⇑, Mohiuddin Ahmad, Rafiqul A. Tarefder Civil Engineering Department, University of New Mexico, MSC01 1070, Albuquerque, NM 87131, USA

h i g h l i g h t s
Healing ability of asphalt binder is greatly influenced by moisture conditioning.
A healing model is developed to separate the total healing into instantaneous and long-term healings.
Instantaneous healing decreases as moisture conditioning decreases the cohesion or energy of separation of binder.
Moisture conditioning reduces the long-term healing rate of asphalt binder.

a r t i c l e

i n f o

a b s t r a c t

Article history: Received 11 January 2017 Received in revised form 22 March 2017 Accepted 11 April 2017

The self-healing property of asphalt binder in the asphalt concrete has been reported in recent years. Healing is an important property of asphalt binder as it helps to recover the fatigue damage during rest period. However, the effect of moisture conditioning on asphalt binder healing has not been studied yet. In this study, Moisture Induced Sensitivity Test (MIST) is used to moisture-condition the binders and then tested using a Fourier Transform Infrared (FTIR) and Dynamic Shear Rheometer (DSR) to evaluate the chemical and healing properties respectively. Also, cohesion properties of the binder are calculated from the tack test using DSR. FTIR results show that water is absorbed in the asphalt binder due to the moisture conditioning. Additionally, results show that the healing rate of asphalt binder decreases due to moisture conditioning. A healing model is developed to separate the total healing into instantaneous and longterm healings. The instantaneous healing is the instant healing that occurs just after the loading is removed. Moisture conditioning decreases the amount of instant healing by reducing the cohesion or energy of separation of the binder. The long-term healing occurs only if there is a very long rest period and depends on the activation energy (or diffusion rate). Results show that moisture conditioning reduces the long-term healing rate by increasing the required activation energy for diffusion. Therefore, the overall healing of fatigue damage reduces due to moisture conditioning of the asphalt binder. Ó 2017 Elsevier Ltd. All rights reserved.

Keywords: Healing Moisture DSR FTIR Asphalt binder MIST

1. Introduction Several studies in recent years showed the evidence of healing in asphalt concrete and the effect of healing on the fatigue life of the asphalt pavement [1–5]. Healing is the ability of asphalt to recover microdamage caused due to loading [6]. Healing of asphalt concrete depends on the chemical and physical properties of the asphalt binder and several external factors such as temperature, aging, moisture, etc. This study focuses on asphalt binder healing or cohesive healing in asphalt mixture. Healing of asphalt binder is highly sensitive to aging and moisture conditioning, as aging and moisture conditioning changes the physical and chemical properties of asphalt binder [7–9]. Some studies have evaluated ⇑ Corresponding author. E-mail addresses: [email protected] (U.A. (M. Ahmad), [email protected] (R.A. Tarefder).

Mannan),

http://dx.doi.org/10.1016/j.conbuildmat.2017.04.087 0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.

[email protected]

the effect of aging on the healing of the asphalt binder [10], However, there is no work to the authors’ knowledge which demonstrates the effect of moisture on the healing of asphalt binder. It is already known that moisture plays an important role on the damage of asphalt pavement. As water enters the asphalt concrete it diffuses with the asphalt binder at the molecular level, resulting in binder softening. This leads to cohesive failure within the asphalt binder film of the asphalt concrete, which is called moisture damage [11–15]. However, the focus of this study is the healing of fatigue damage not the moisture damage. Another study by Islam and Tarefder showed that the increase of moisture in the asphalt concrete does not always increase the damage or reduces the strength of the mix [16]. From the above studies, it can be said that moisture reduces the binder’s cohesive properties and the asphalt binder’s chemical property might change. These changes in the binder due to moisture conditioning can also effect the

U.A. Mannan et al. / Construction and Building Materials 146 (2017) 360–369

Solenid Valve Sealed Pressure Vessel

Cyclic Pressure Load

361

changes and their effects on asphalt binder healing due to moisture conditioning, may lead to a better understanding of the effect of moisture on asphalt healing.

3. Laboratory testing

Sample Container

3.1. Moisture conditioning

Asphalt Binder

Diaphragm Deflated State Diaphragm Air Supply Fig. 1. MIST moisture conditioning setup.

healing. However, the effect of chemical changes due to moisture in healing of binder has not been studied yet. Therefore, the effect of moisture conditioning on asphalt binder healing is an important mechanism that should be evaluated. This will provide a better understanding about how moisture affects the healing of asphalt, which eventually affects cracking and long-term performance of the pavement. 2. Objective The main objective of this study is to evaluate the influence of moisture conditioning on the healing of asphalt binders. To achieve this objective, two different asphalt binders are moisture conditioned in three different levels. The chemical changes that might have occurred due to the moisture conditioning are then evaluated using the Fourier Transform Infrared (FTIR) Spectroscopy. The healing and cohesive properties of the binders are measured using the Dynamic Shear Rheometer (DSR). Healing model for all the moisture conditioned binders are developed using the Wool and O’Connor healing model [17]. The study of chemical and mechanical

This study evaluates two unmodified performance grade (PG) asphalt binders from Holly Asphalt Co.: PG 70-22 and PG 58-22. Several methods are developed to moisture condition the samples over the time: Nottingham asphalt test equipment, AASHTO T 283, Moisture Induced Sensitivity Test (MIST), etc. [18]. Among these, only MIST is used in this study for moisture conditioning the binders. MIST conditioning is chosen over the AASHTO T 283 for this study, because MIST does not apply freeze-thaw conditioning. MIST uses a cyclic pressure loading and high water temperatures to simulate a harsh moisture condition for asphalt pavement, which affects the overall material strength [19]. In MIST, high temperature (60 °C) and pressurized (40 psi) water is forced into the asphalt binder and this accelerates the moisture conditioning of the binder. The schematic of the MIST conditioning is presented in Fig. 1. The binders were placed in a small container inside the chamber and fully submerged in water. A thin layer (5 mm) of binder is conditioned to ensure the proper conditioning of the binder from all the sides. Then the MIST lid was closed and a diaphragm at the bottom of the chamber inflated and deflated by hydraulic pump and piston mechanism. Due to this the water pressure inside the chamber increased and decreased. When the pressure increases, water is pushed into the binder sample (blue arrows in Fig. 1) and during depressurizing some of the water is pushed off (green arrows in Fig. 1). This leads to the absorption and diffusion of water in the asphalt binder. The recommended test conditions for MIST consist of 3500 cycles under 40 psi pressure at 60 °C, which is considered as the equivalent of one cycle under AASHTO T 283 conditioning [20,21]. Samples were conditioned by three repeated cycles in the MIST chamber to moisture conditioned the sample into three different degrees. These conditioned samples are labeled as MIST-cycle 1 (3,500 cycles), MIST-cycle 2 (7,000 cycles) and MIST-cycle 3 (10,500 cycles).

Asphalt sample on the ATR

FTIR with UATR

(a) FTIR Test Setup (b) FTIR with ATR mechanism Fig. 2. FTIR setup.

362

U.A. Mannan et al. / Construction and Building Materials 146 (2017) 360–369

3.5

Shear Modulus, G* (MPa)

3

G*i G*r

2.5 2

Rest Period G*L

1.5 1 0.5 0 0

5,000

10,000

15,000

20,000

25,000

Time (s)

(a) DSR test setup

(b) Healing test configuration Fig. 3. Healing test setup.

Force

Debonding

Compression Asphalt binder Time

Contact Fig. 4. Schematic of the tack test by DSR [23].

3.2. FTIR testing Unconditioned (control) and moisture conditioned binders were tested using the FTIR to examine if there is a chemical change in the asphalt binder due to moisture conditioning. The PerkinElmer FTIR with the Universal Attenuated Total Reflectance (UATR) accessory was used for the analysis (Fig. 2(a)). The IR spectra were recorded with a horizontal ATR accessory and a DiComp crystal (composed of a diamond ATR with a zinc selenide focusing element). A small amount ( 1 g) of asphalt binder was placed directly on the ATR plate and a fixed load was applied to the sample to ensure full contact with the crystal. The mechanism of FTIR device with ATR is showed in Fig. 2(b). IR light is shot from the source through a beam splitter. Then the splitter splits the light into two beams: One goes to a stationary mirror and another goes to a moving mirror. Both beams come back to splitter and recombined. The recombined beam passes through the ATR and the sample. The sample absorbed some of the beam with different wavelengths depending on the vibrations of existing different functional groups, and the remined beam is received by a detector. The detector then reports intensity of light versus time for all wavelengths simultaneously. Finally, by using Fourier transform

with help of software an absorbance versus wavenumber IR spectra is reported. For each sample, six scans were collected at a resolution of 4 cm1. For all binder samples, spectra were measured within the range of 4000 to 400 cm1 wavenumber.

3.3. Healing test Time sweep test with rest period (loading–rest–loading) was conducted on dry and moisture conditioned binders using DSR to investigate healing of asphalt binder. Cylindrical samples of 8 mm diameter and 2 mm thick were used and tested under strain control cyclic loading mode. During the cyclic loading a strain of 5% amplitude was applied, and during the rest period the temperature-gap control setting was used. This strain level at the loading period was chosen based on the linear viscoelastic (LVE) limit from the strain sweep test. For all the binders, the LVE range was from 2.5% to 5% strain. To prompt the damage during loading period in the sample, the upper LVE limit (5%) strain was chosen. These tests were conducted at three different temperatures (10, 15 and 20 °C) and at 10 Hz loading frequency. Three replicate samples were tested at each moisture conditions. All samples were

363

U.A. Mannan et al. / Construction and Building Materials 146 (2017) 360–369

loaded till the sample modulus (G ) was reduced to 45% of the initial G value (Gi ), then the loading was stopped and the samples were allowed to heal over different length of times (rest periods). After the rest period, the samples were again loaded until the failure. A schematic of the healing test configuration is shown in th Fig. 3(a). Thus, the healing was estimated as a function of the rest period. The healing index (H) is calculated as follows:

HðtÞ ¼

Gr  GL Gi  GL

ð1Þ

where, Gr is the G after the rest period, GL is the G after loading (Fig. 3). HðtÞ is calculated for different rest periods and temperatures, then used in the healing equation discussed in the following section.

where, R0 is the instant healing occurring due to the wetting and the second part of the equation is the long-term healing occurring due to the inter-molecular diffusion of crack surface. In Eq. (4) K is the _ temperature dependent diffusion parameter, wðtÞ is the diffusion

_ ðtÞ is the wetting rate. The healing initiation function and u processes in asphalt binder can be divided into two parts: instantaneous strength gained due to interfacial cohesion between the crack and long-term strength gained due to diffusion and randomization of molecules. Therefore, during the healing of asphalt binder instant wetting can be considered, which means instantaneous disappearance of crack interface and complete molecular contact. For instant _ ðtÞ ¼ dðtÞ where dðtÞ is the wetting Wool and O’Connor assumed u Dirac delta function. And if the end molecular chain at the microcrack do not need any surface rearrangement then the diffusion _ parameter can be assumed as wðtÞ ¼ dðtÞ. Then the Eq. (4) can be

3.4. Tack test

rewritten as:

DSR was used to conduct the tack test to measure the change in asphalt binder cohesion due to moisture conditioning [22]. For this test, a sample with 25 mm diameter was used and three replicate samples were tested for all moisture conditioning levels. DSR parallel plate geometry is used for the tack test. A schematic of the tack test is shown in Fig. 4 [23]. It can be seen in Fig. 4, that the tack test consists of three intervals. Initially, the sample was compressed into 1.5 mm thickness and the normal force was kept to zero for 200 s. Then an additional compressive normal force of 10 N was applied to the sample to ensure the DSR plate-binder contact. The upper plate of the DSR was moved upward at a speed of 0.1 mm/s. This upward tension causes the debonding of the binder. DSR can measure the applied force and gap increase (vertical displacement) over time. The area under the force-time curve is referred to as the tack factor (Tc), which represents the stickiness of the binder [24]. This can also be represented by the energy of separation (W s ). W s is calculated using the following equation:

RðtÞ ¼ R0 þ Kt 1=4

Ws ¼

1 A

Z

F  v  dt ¼ d

Z

r  de

ð2Þ

where, F is the applied force, v is the speed of separation, A is the contact area and d is the specimen thickness. W s is an important parameter as it measures the energy requires to separate the asphalt binder-binder surface [24]. As all the samples tested in this study showed cohesive failure (failure in the binder), thus W s can be used as a measure of cohesive strength of the binder. 3.5. Healing analysis background Healing is the reverse process of damage; during loading, microcrack damage develops and during the rest period the healing of this microcrack damage occurs. Several models have been proposed to describe healing of polymer, among them the most well-known is the crack healing model by Wool and O’Connor [17]. This model describes five stages of the healing: surface rearrangement, surface approach, wetting, diffusion and randomization. In this model, wetting and diffusion is controlled by the mechanical properties of the material. Wool and O’Connor defined the overall healing (RðtÞ) as a convolution of intrinsic healing (Rh ðtÞ)and wetting distribution (uðtÞ) function. Eq. (3) shows the proposed healing equation for the dimensionless healing ratio RðtÞ .

Z

duðsÞ RðtÞ ¼ Rh ðt  sÞ ds ds 1 t

ð3Þ

The general healing equation derived from Eq. (3) is as follows:

h i _ _ ðtÞ RðtÞ ¼ R0 þ ðKt 1=4  wðtÞÞ u

ð4Þ

ð5Þ

where, K represents the strength gain due to the inter-molecular diffusion between the crack surfaces at certain temperature. K is also called the diffusion rate as a function of temperature. It can be expressed using the Arrhenius diffusion law as K ¼ K 0 eEa =RT . In which, K 0 , Ea , R and T are fitting constant, activation energy, the universal gas constant (8.314 J/mol/K) and temperature in Kelvin respectively. Substituting for K in Eq. (5), the self-healing model for asphalt binder can be written as follows:

  RðtÞ ¼ R0 þ K 0 eEa =RT ðt 1=4 Þ  HðtÞ

ð6Þ

Eq. (6) describes the overall healing process of asphalt binder, where the total healing RðtÞ  HðtÞ can be calculated from healing test data as a function of rest period using Eq. (1). R0 represents the instantaneous healing due to the crack surface wetting and depends on the cohesive property of material. The second part is the long-term healing, which is dependent on the molecular diffusion at the crack surfaces. From the Eq. (6), the diffusion rate is actually dependent on the activation energy (Ea ). Ea is defined as the energy required for time-dependent strength gain during the long-term healing. The activation energy varies with binder type. If Ea is less than the energy at surface due to damage, then the time dependent healing will not occur. Therefore, Ea is an important parameter for understanding the time-dependent healing in the asphalt binder. In this study, the effect of moisture in the asphalt healing ability is evaluated by measuring these healing parameters for different moisture conditioned binders. 4. Results and discussion 4.1. FTIR test results FTIR analysis is conducted to observe the chemical changes in the binder due to moisture conditioning. The functional group in the material is assigned according to the FTIR spectra main bands.

Table 1 Characteristic bands of functional groups in asphalt binder [25]. Functional groups

Wavenumber (cm1)

vOAH bond vCAH Aliphatic vC@O Carbonyls vC@C Aromatic dCAH Aliphatic index (A(CH2)n) dC-H Aliphatic branched (ACH3) vS@O Sulfoxide Stretching vibration of Benzene

3400–3700 2924, 2853 1700 1600 1460 1376 1030 866, 812

364

U.A. Mannan et al. / Construction and Building Materials 146 (2017) 360–369

Control

1.2

Control

1.2

MIST 1st Cycle

MIST 1st Cycle 1

1

MIST 2nd Cycle

% Absorbance

% Absorbance

MIST 2nd Cycle 0.8 MIST 3rd Cycle 0.6

0.8 MIST 3rd Cycle 0.6

0.4

0.4

0.2

0.2

0

0 600

800

1000

1200

1400

1600

600

1800

800

1000

Wave numbers (cm-1)

1200

1400

(a) PG 58-22 0.08 Control

Control 0.07

0.07 MIST 1st Cycle

MIST 1st Cycle 0.06

0.06 MIST 2nd Cycle

MIST 2nd Cycle

0.05

% Absorbance

% Absorbance

1800

(b) PG 70-22

0.08

MIST 3rd Cycle 0.04 0.03

0.05 MIST 3rd Cycle 0.04 0.03

0.02

0.02

0.01

0.01

0 3100

1600

Wave numbers (cm-1)

3200

3300

3400

3500

3600

3700

0 3100

3200

3300

Wave numbers (cm-1)

3400

3500

3600

3700

Wave numbers (cm-1)

(c) PG 58-22

(d) PG 70-22 Fig. 5. FTIR spectrum for different binders.

The main bands of the functional groups in the asphalt binders and their wave numbers are shown in Table 1 [25] and the FTIR spectra of all asphalt binders are shown in Fig. 5. As the spectrum between the wavenumber 500 to 2000 cm1 is considered as finger printing region, the spectrum pattern at this region is completely different for different materials [26]. Fig. 5(a) and (b) shows the finger print region spectrum for PG 58-22 and PG 70-22 respectively. It shows that due to moisture conditioning there is no change except at the 1030 and 1700 cm1 wavenumbers. However, the water absorption due to moisture conditioning is seen at the wavenumber 3100 to 4000 cm1, which are shown in Fig. 5 (c) and (d) for PG 58-22 and PG 70-22 respectively. A study by Vasconcelos et al.

[27] shows that if asphalt binder is submerged under water for several days then the absorbed water in binder is visible in the FTIR spectrum at the 3100–3700 cm1 wavenumber region. In this study the moisture conditioned samples showed a peak at 3400 cm1, which represents the liquid water region. PG 70-22 has lower water absorbance than the PG 58-22 binder, which means water diffuses into softer binder faster than harder binder. As the hard binder has higher molecular weight than the soft binder, therefore the molecules in hard binder form a packed structure [28]. Due to this packed structure hard binders absorbed less water. This figure shows that the water absorbance percentage is very low compared to the overall spectrum, however, the

Table 2 Functional group indices from FTIR spectrum. Condition

Aliphatic Ial

Aromatic Iar

Benzene Ibz

Carbonyl IC¼O

Sulfoxide IS¼O

Hydroxyl IOH

PG58-22

Control MIST 1st Cycle MIST 2nd Cycle MIST 3rd Cycle

0.685726 0.734866 0.743933 0.725684

0.091195 0.107715 0.107488 0.102994

0.085307 0.092433 0.090814 0.095975

0.095512 0.019288 0.003424 0.022346

0.015046 0.014688 0.022927 0.02249

0.023289 0.025519 0.057764 0.127346

PG70-22

Control MIST 1st Cycle MIST 2nd Cycle MIST 3rd Cycle

0.698141 0.652707 0.662029 0.695088

0.120595 0.091335 0.092389 0.098982

0.088922 0.073702 0.071379 0.083641

0.036283 0.143821 0.135105 0.07951

0.01948 0.006337 0.006679 0.008408

0.05948 0.082105 0.015723 0.080395

365

U.A. Mannan et al. / Construction and Building Materials 146 (2017) 360–369

25

25

Modulus after Rest, G*r (MPa)

Modulus after Rest, G*r (MPa)

Control

20

15 Control 10 MIST 1st Cycle MIST 2nd Cycle

5

MIST 1st Cycle

20

MIST 2nd Cycle 15 MIST 3rd Cycle

10

5

MIST 3rd Cycle 0

0 0

20

40

60

80

100

120

0

140

20

40

60

80

100

120

140

Rest Period, t (mins)

Rest Period, t (mins)

(a) G r* for PG 58-22

(b) G r* for PG 70-22 -1

1 10 °C 0.9

y = 0.2117x + 0.1579 R² = 0.9423

15 °C

-1.4

20 °C

0.7

Linear (10 °C)

0.6

Linear (15 °C)

0.5

Linear (20 °C)

0.4

y = -8836.9x + 28.638 R² = 0.9788

-1.8

y = 0.1438x + 0.0397 R² = 0.9816

ln (K)

Healing Index, H

0.8

y = 0.0731x + 0.073 R² = 0.9442

-2.2

0.3 -2.6

0.2 0.1 0 0.00

0.50

1.00

1.50

Rest Period,

2.00

t1/4

2.50

3.00

3.50

-3 0.0034

0.00344

0.00348

0.00352

0.00356

1/T (1/Kelvin)

(mins)

(d) Calculation of E a from K

(c) H vs.t1/4 for PG 70-22 (control)

Fig. 6. Results and analysis of healing test data from DSR.

absorbance increases with moisture conditioning. For PG 58-22 (soft binder) with the increase in MIST cycle the water absorption increases. The 3rd MIST cycle shows the highest water absorption peak. Whereas for PG 70-22 (hard binder), with the increase in moisture conditioning the water absorption peak increases initially, but it starts to decrease at 3rd MIST cycle. Therefore, due to the MIST conditioning, asphalt binder absorbs water and it can be detected using the FTIR spectrum.It can be seen from Fig. 5, that the % absorbance of some functional groups change due to the moisture conditioning. To quantify these changes, different indices are calculated for all the functional groups in Table 1. The indices are calculated based on Eqs. (7)–(12) as aliphatic index (Ial ), aromatic index (Iar ), benzene index (Ibz ), carbonyl index (IC¼O ), sulfoxide index (IS¼O ) and hydroxyl index (IOH ). All the calculated index values are shown in Table 2.

Ial ¼

A1460 þ A1376 P A

A1600 Iar ¼ P A Ibz ¼

A866 þ A812 P A

ð7Þ ð8Þ ð9Þ

A1700 IC¼O ¼ P A

ð10Þ

A1030 IS¼O ¼ P A

ð11Þ

A3400 IOH ¼ P A

ð12Þ

P where, A ¼ Area under 2000 and 600 cm1 ¼ A1700 þ A1600 þ A1460 þ A1376 þA1030 þ A866 þ A812 þ A723 Table 2 shows that both control binders have the same aliphatic index. However, they change after moisture conditioning. The aromatic index is higher for PG 70-22 control binder; thus, it can be concluded that the aromatic groups in the binder are main reason for the increase in stiffness as PG 70-22 has higher stiffness than the PG 58-22. Also, Table 2 shows that the values Ial and Iar do not change with moisture conditioning. However, Ibz shows an increase with moisture conditioning. Asphalt samples conditioned by MIST are subject to high pressure and relatively high temperature (60 °C). Therefore, the conditioning might be associated with some aging concerns, especially under longer cycles such as MIST 3rd cycle. Hence the IS¼O and IC¼O have been studied to see whether there is any aging of binder caused by the moisture conditioning.

366

U.A. Mannan et al. / Construction and Building Materials 146 (2017) 360–369

Table 3 Different parameters of healing equation. Binder type

Conditioning

Temperature (°C)

PG 70-22

Control

10 15 20

MIST 1st Cycle

10 15 20

MIST 2nd Cycle

10 15

MIST 3rd Cycle

10 15

PG 58-22

Control

10 15 20

MIST 1st Cycle

10 20

MIST 2nd Cycle

10 20

MIST 3rd Cycle

10 15 20

Healing equation parameters HðtÞ  R0 þ Kt 1=4 R0 (95% CI bounds)

K (95% CI bounds)

0.073034 (0.0588, 0.0873) 0.039698 (0.0239, 0.0555) 0.15791 (0.1159, 0.1999) 0.010942 (0.0031, 0.0188) 0.092599 (0.0669, 0.1183) 0.18675 (0.1405, 0.2330) 0.063822 (0.0470, 0.0806) 0.16806 (0.1245, 0.2116) 0.020438 (0.0110, 0.0298) 0.15686 (0.1184, 0.1953)

0.073115 (0.0669, 0.0793) 0.14381 (0.1370, 0.1507) 0.21168 (0.1159, 0.1999) 0.085688 (0.0823, 0.0891) 0.16503 (0.1538, 0.1762) 0.26136 (0.2412, 0.2815) 0.1033 (0.0960, 0.1106) 0.21713 (0.1982, 0.2360) 0.099973 (0.0959, 0.1041) 0.21376 (0.1971, 0.2305)

0.32041 (0.2596, 0.41395 (0.3248, 0.26775 (0.2093, 0.31147 (0.2531, 0.18675 (0.1405, 0.31927 (0.2593, 0.23336 (0.1790, 0.29485 (0.2397, 0.49268 (0.4020, 0.12955 (0.0901,

0.21159 (0.1851, 0.35977 (0.3210, 0.26636 (0.2410, 0.20578 (0.1804, 0.26136 (0.2412, 0.20519 (0.1791, 0.29259 (0.2690, 0.19863 (0.1746, 0.2386 (0.1992, 0.29505 (0.2779,

Results showed that IC¼O does not follow any trend (increasing or decreasing) with the increase of moisture conditioning and the change in IC¼O is not substantial. However, IS¼O increases with the increase in moisture conditioning. A study by Ouyang et al. [29] showed that S@O increases due to the short-term aging and then decreases with the long-term aging in the asphalt binder. It implies that moisture conditioning only causes short-term aging in the binder; it does not affect the long-term aging. IOH increases due to the moisture conditioning. One explanation may that be due to the moisture conditioning, water is absorbed into the binder showing a higher area under the OAH peak of the FTIR spectrum. This absorption of water into the binder might result in deterioration of asphalt binder cohesion which can affect the binder healing property. Also, absorption of water into the binder can reduce the rate of diffusion at the microcrack surface. Therefore, binder cohesion and healing activation energy is studied in the later sections to understand the effect of moisture conditioning on the healing. 4.2. Healing test results Fig. 6(a) and (b) show the Gr values over various rest periods for different conditioned binders at 10 °C. It can be seen that PG 58-22 has faster and higher G recovery at rest period than that of PG 7022. The shear modulus of the control binder recovers faster over the rest period, thus having higher values than that of the moisture

0.3813) 0.5031) 0.3262) 0.3698) 0.2330) 0.3792) 0.2877) 0.3500) 0.5833) 0.1690)

R2

Ea (kJ/mole)

0.944

73.45

0.982 0.942 0.987

77.03

0.964 0.953 0.96

100.78

0.941 0.986

103.10

0.952 0.886

16.22

0.2380) 0.913 0.3985) 0.95 0.2917) 0.889

16.49

0.2311) 0.953 0.2815) 0.883

24.48

0.2312) 0.949 0.3162) 0.893

27.29

0.2226) 0.817 0.2780) 0.973 0.3122)

conditioned binders for both PG 70-22 and PG 58-22. However, moisture conditioning has more influence on the softer binder (PG 58-22) than the harder binder (PG 70-22). Due to moisture conditioning, the G recovery after the rest period is lower than the control binder for PG 58-22. Whereas, for PG 70-22, the effect of moisture conditioning on the G recovery after the rest period is not noticeable at 10 °C. However, at 15 °C and 20 °C, both binders show that moisture conditioning has a substantial effect on G recovery after the rest period. Also, with increase in temperature, the healing after the rest period increases for both binders. It is observed from Fig. 6(a) and (b) that there is a rapid recovery of the G in approximately first 20 s, then the recovery rate gradually decreases. Therefore, the overall healing in the binders can be divided into two parts: instantaneous and long-term healing. This result verifies the assumption in the healing analysis background section.Fig. 6(c) presents the healing index (H) calculated from Eq. (1) versus the fourth root of the rest period for the control PG 70-22 binder. At low temperature, H shows better linear relationship with the rest period than at higher temperature. Also, the healing increases with the increase in temperature, which can be due to the increase in the molecular movement at high temperature. Table 3 shows the healing equation for all moisture conditioned binders at three different temperatures. For all the binders, H shows a strong correlation with the rest period and the correlation coefficient (R2) for all fitted curves are above 0.81.

367

U.A. Mannan et al. / Construction and Building Materials 146 (2017) 360–369

moisture conditioning, although the control binder shows the highest R0 value than the moisture conditioned binders. Therefore, it can be concluded that moisture conditioning decreases the instant healing of the binders.The fitted parameters R0 andKare calculated by linear regression (Table 3). Then from the lnðKÞ vs. 1 plot (Fig. 6 (d)) the activation energy, Ea is calculated using the T following linear equation:

0.14 0.1273

Control 0.12

MIST 1st Cycle MIST 2nd Cycle

0.1

MIST 3rd Cycle 0.08 0.0578

0.06

lnðKÞ ¼ lnðK 0 Þ  0.04 0.0245 0.0273

0.0233 0.0255

0.0162 0.0165

0.02 0

Ea (MJ/mole)

IO-H

Fig. 7. Activation energy from healing test vs. hydroxyl index (IOH ) from FTIR test.

PG 70-22 has smaller R0 values than PG 58-22. Hence, the harder binder has less instant healing than the softer binder. Also, the R0 value initially decreases and then increases for a higher degree of

ð13Þ

Table 3 shows Ea for all moisture conditioned binders. Ea is the energy required for the inter-molecular diffusion between the crack surfaces. Hence, if a binder has a higher activation energy it results in a slower time-dependent healing. Results show that moisture conditioned binders have higher activation energy, which means moisture conditioning reduces the rate of time-dependent healing. This change in the activation energy can be due to the water diffusion into the binder. Fig. 7 shows the comparison of the trends of Ea values with that of the IOH values for PG 58-22. As expected the value of Ea increases as the value of IOH increases. Therefore, as moisture is absorbed in the binder the diffusion of microcrack surfaces requires higher energy, thus the long-term healing of the binder diminishes.

5

5

0

0 0

0.5

1

1.5

2

2.5

3

0

-5

0.5

1

-10 -15 -20

Control

2

2.5

-25

-10 -15 -20

Control

MIST 1st Cycle

-25

MIST 1st Cycle

-30

MIST 2nd Cycle

-30

-35

MIST 3rd Cycle

-35

-40

MIST 2nd Cycle MIST 3rd Cycle

-40 Displacement (mm)

Displacement (mm)

(a) PG 58-22

(b) PG 70-22

9000

1200

Control

Control

MIST 1st Cycle

8000

MIST 2nd Cycle 7000

1000

MIST 1st Cycle

MIST 3rd Cycle MIST 2nd Cycle

6000

Tack Factor, Tc (Ns)

Energy of Separation, w (N/m)

1.5

-5

Normal Force (N)

Normal Force (N)

  Ea 1 R T

5000 4000 3000

800 MIST 3rd Cycle 600

400

2000 200 1000 0 PG 58-22

PG 70-22

0 PG 58-22

Binder Type

PG 70-22 Binder Type

(c) Energy of separation (Ws )

(d) Tack factor (Tc) Fig. 8. Tack test results.

3

368

U.A. Mannan et al. / Construction and Building Materials 146 (2017) 360–369

40

the findings from the test results, the following conclusions can be drawn:

Control MIST 1st Cycle

35 32.04

31.15

31.93 MIST 2nd Cycle

29.49

30

MIST 3rd Cycle

25 20 15 11.00

10.53

11.79

10.82

10 5 0

R0 (%)

Ws (N/m)

Fig. 9. R0 from healing test vs. energy of separation (W s ) from tack test.

4.3. Tack test results Fig. 8 shows the results from the tack test for all the binders. Fig. 8(a) and (b) shows the load-displacement curves for PG 5822 and PG 70-22 binders respectively. As PG 58-22 is a softer binder, it takes less force to pull out the binder compared to PG 70-22 binder. Also, the control binder requires the highest force to pulloff. Fig. 8(c) shows the energy of separation (W s ) for all the binders. This figure shows that for PG 58-22, W s decreases with the increase in moisture conditioning, except for MIST 2nd cycle sample. PG 7022 shows a decreasing trend with the increase in moisture conditioning. Similar patterns are found for the tack factor (Tc) as well (Fig. 8(d)). Fig. 8(c) and (d) shows that the stickiness decreases with the moisture conditioning, which basically leads to the decrease in the cohesion. Therefore, the cohesion or the tackiness of asphalt binder decreases with the moisture conditioning. R0 in the healing equation (Eq. (5)) corresponds to instantaneous healing of wetted surface, which can be related to the binder work of cohesion. Many researchers have related R0 with the surface energy measurement, as the work of cohesion is a function of surface free energy [6,30]. While the energy of separation has been previously used as the measure of cohesion property of the binder [22]. In this study, the W s from the tack test has been correlated with R0 from the healing test. Fig. 9 shows the comparison of the trends of R0 values with that of W s from the tack test for PG 58-22 binder. As expected, the values of R0 show the same trend as W s . With moisture conditioning, the energy of separation or cohesion decreases for MIST 1st cycle then it increases for the 2nd cycle and again decrease for 3rd cycle and so does the R0 from the healing equation. Thus, by using Ws from the tack test method the effect of moisture on the instantaneous healing can be explained. So, it can be concluded that moisture conditioning decreases the instantaneous healing by decreasing the cohesion or Ws.

5. Conclusions The effect of moisture conditioning on the healing of asphalt binder is discussed in this study. Asphalt binders are moisture conditioned in three different levels. Healing models and all the healing parameters are estimated from the healing test for moisture conditioned binders. In addition, the effect of moisture conditioning on the chemical and cohesive properties of binder are determined; and their effect on asphalt healing is examined. Based on


FTIR spectrum captures the changes in the chemical functional groups due to moisture conditioning. The FTIR spectrum shows an increase in the hydroxyl group due to moisture conditioning. Also, from the sulfoxide group index it can be concluded that moisture conditioning results in short-term aging of asphalt binder.
The proposed healing model of asphalt binder is divided into two part: instantaneous healing and time-dependent healing. Results show that time-dependent healing rate is inversely proportional to percent water absorbed in the binder due to moisture conditioning. And the instantaneous healing (R0 ) is proportional to the cohesion of binder, which can be calculated from the work of separation of the tack test.
Moisture conditioning has a negative impact on the healing property of asphalt binder. cohesion decreases and activation energy increases due to moisture conditioning, which results in a decrease in instantaneous and time-dependent healing respectively. However, moisture conditioning has greater effect on the healing of softer binder (PG 58-22) than that of the harder binder (PG 70-22). In summary, the healing property of asphalt binder deteriorate when the binder is moisture conditioned. References [1] Y.R. Kim, H.-J. Lee, D.N. Little, Fatigue characterization of asphalt concrete using viscoelasticity and continuum damage theory (with discussion), J. Assoc. Asphalt Paving Technol. 66 (1997). [2] R.L. Lytton, J. Uzan, E.G. Fernando, R. Roque, D. Hiltunen, S.M. Stoffels, Development and Validation of Performance Prediction Models and Specifications for Asphalt Binders and Paving Mixes, Strategic Highway Research Program, 1993. [3] H.-J. Lee, Y.R. Kim, Viscoelastic continuum damage model of asphalt concrete with healing, J. Eng. Mech. 124 (1998) 1224–1232. [4] J.S. Daniel, Y.R. Kim, Development of a simplified fatigue test and analysis procedure using a viscoelastic, continuum damage model (with discussion), J. Assoc. Asphalt Paving Technol. 71 (2002). [5] M. Mamlouk, M. Souliman, W. Zeiada, Optimum testing conditions to measure HMA fatigue and healing using flexural bending test, in: TRB Annu. Meet, 2012. [6] D.N. Little, R.L. Lytton, D. Williams, C.W. Chen, Microdamage healing in asphalt and asphalt concrete, volume I: microdamage and microdamage healing, project summary report, 2001. [7] U.A. Mannan, Effect of Recycled Asphalt Shingles (RAS) on Physical and Chemical Properties of Asphalt Binders MS Thesis, The University of Akron, 2012. https://etd.ohiolink.edu/pg_10?100142627915166::NO:10:P10_ETD_ SUBID:47903. [8] M. Hossain, A.A.R. Khan, H. Faisal, R. Tarefder, Finite Element and Mechanical Modeling of Fatigue Behavior of Partial Vapor-Conditioned Viscoelastic Material, in: American Society of Mechanical Engineers, 2015. pp. V010T13A010-V010T13A010. [9] U.A. Mannan, M.R. Islam, R.A. Tarefder, Effects of recycled asphalt pavements on the fatigue life of asphalt under different strain levels and loading frequencies, Int. J. Fatigue 78 (2015) 72–80. [10] A. Bhasin, S. Palvadi, D. Little, Influence of aging and temperature on intrinsic healing of asphalt binders, Transp. Res. Rec. J. Transp. Res. Board (2011) 70–78. [11] A. Bhasin, E. Masad, D. Little, R. Lytton, Limits on adhesive bond energy for improved resistance of hot-mix asphalt to moisture damage, Transp. Res. Rec. J. Transp. Res. Board (2006) 3–13. [12] A.R. Copeland, Influence of moisture on bond strength of asphalt-aggregate systems, (2007). [13] R. Moraes, R. Velasquez, H. Bahia, Measuring the effect of moisture on asphaltaggregate bond with the bitumen bond strength test, Transp. Res. Rec. J. Transp. Res. Board (2011) 70–81. [14] Y.Z.M. Htet, An Assessment of Moisture Induced Damage in Asphalt Pavements, 2015. [15] J. Zaniewski, A.G. Viswanathan, Investigation of Moisture Sensitivity of Hot Mix Asphalt Concrete, Asphalt Technology Program, 2006. [16] M.R. Islam, R.A. Tarefder, Tensile strength of asphalt concrete due to moisture conditioning, Int. J. Civ. Archit. Struct. Constr. Eng. 8 (2014) 942–945. [17] R. Wool, K. O’connor, A theory crack healing in polymers, J. Appl. Phys. 52 (1981) 5953–5963. [18] E.D. Shrum, Evaluation of moisture damage in warm mix asphalt containing recycled asphalt pavement, 2010.

U.A. Mannan et al. / Construction and Building Materials 146 (2017) 360–369 [19] InstroTek, Moisture induced sensitivity test (MIST)., Raleigh, NC, 2012. http:// instrotek.com/wordpress/wp-content/uploads/MIST_rev10_03_23_2012.pdf (accessed July 27, 2016). [20] M. Ahmad, U.A. Mannan, M.R. Islam, R.A. Tarefder, Chemical and mechanical changes in asphalt binder due to moisture conditioning, Road Mater. Pavement Des. (2017) 1–14. [21] AASHTO T 283-14, Standard Method of Test for Resistance of Compacted Asphalt Mixtures to Moisture-Induced Damage, 2014. http://www. techstreet.com/standards/aashto-t-283-14?product_id=1886606 (accessed July 27, 2016). [22] H.U. Bahia, D. Hanson, M. Zeng, H. Zhai, M. Khatri, R. Anderson, Characterization of Modified Asphalt Binders in Superpave Mix Design, National Cooperative Highway Research Program, Washington, D.C., 2001. [23] C. Creton, P. Fabre, Chapter 14 - Tack A2 - Chaudhury, M., in: A.V. Pocius (Ed.), Adhes. Sci. Eng., Elsevier Science B.V., Amsterdam, 2002: pp. 535–575. http:// www.sciencedirect.com/science/article/pii/B978044451140950041X (accessed July 27, 2016).

369

[24] K. Kanitpong, H. Bahia, Relating adhesion and cohesion of asphalts to the effect of moisture on laboratory performance of asphalt mixtures, Transp. Res. Rec. J. Transp. Res. Board (2005) 33–43. [25] J. Coates, Interpretation of infrared spectra, a practical approach, Encycl. Anal. Chem. (2000). [26] A.R. Abbas, U.A. Mannan, S. Dessouky, Effect of recycled asphalt shingles on physical and chemical properties of virgin asphalt binders, Constr. Build. Mater. 45 (2013) 162–172, http://dx.doi.org/10.1016/j.conbuildmat.2013.03.073. [27] K.L. Vasconcelos, A. Bhasin, D.N. Little, C. Glover, Influence of the physical state of water in the diffusion process in asphalt binders, Transportes 20 (2012) 12–18. [28] J. Read, D. Whiteoak, The shell bitumen handbook, Thomas Telford, 2003. [29] C. Ouyang, S. Wang, Y. Zhang, Y. Zhang, Improving the aging resistance of asphalt by addition of Zinc dialkyldithiophosphate, Fuel 85 (2006) 1060–1066. [30] D. Cheng, D. Little, R. Lytton, J. Holste, Surface energy measurement of asphalt and its application to predicting fatigue and healing in asphalt mixtures, Transp. Res. Rec. J. Transp. Res. Board (2002) 44–53.