EFFECTS OF PHYSICAL HARDENING ON THERMAL CONTRACTION OF ASPHALT BINDERS. A thesis presented to. the faculty of

Transcription

1 EFFECTS OF PHYSICAL HARDENING ON THERMAL CONTRACTION OF ASPHALT BINDERS A thesis presented to the faculty of Russ College of Engineering and Technology of Ohio University In partial fulfillment of the requirements for the degree Master of Science Viswanath Dokka March 2006

2 This thesis entitled EFFECTS OF PHYSICAL HARDENING ON THERMAL CONTRACTION OF ASPHALT BINDERS by VISWANATH DOKKA has been approved for the Department of Civil Engineering and the Russ College of Engineering and Technology by Sang Soo Kim Associate Professor of Civil Engineering Dennis Irwin Dean, Russ College of Engineering and Technology

3 DOKKA, VISWANATH. M.S. March Civil Engineering EFFECTS OF PHYSICAL HARDENING ON THERMAL CONTRACTION OF ASPHALT BINDERS (138 pp.) Director of Thesis: Sang Soo Kim Determining the cracking temperatures of paving and roofing asphalts at cold environment by current test methods is a costly and tedious process. In cooperation with the National Cooperative Highway Research Program (NCHRP) the Ohio University Civil Engineering Department has been conducting advanced studies to determine the cracking temperature of the asphalt binders and physical hardening effects. Approved: Sang Soo Kim Associate Professor of Civil Engineering

4 Acknowledgments My sincere thanks to Dr. Sang Soo Kim, for mentoring me all along this research, to Dr. Lloyd Herman, Dr. Nichole Pavel for their advice, guidance and accepting to be in my thesis review committee; to Sam Khoury for his support and assistance; to my friends Reddy and Pavan for their invaluable help; and to NCHRP for funding this project; to many patient scientists and friends who were just a phone call away; to the readers of this thesis, who I hope will carry on the good work with care and integrity valued in this group. Finally, I wish to express my gratitude and indebtedness to my parents for all the sacrifices they have made and all the troubles they have endured in supporting me to achieve my master s degree.

5 5 Table of Contents Page Abstract... 3 Acknowledgments... 4 List of Tables... 8 List of Figures INTRODUCTION LITERATURE REVIEW Low Temperature Cracking of Flexible Pavements Factors Affecting Low Temperature Cracking of HMA Materials Additives Environment Pavement Structure Determination of Low Temperature Cracking of Asphalt Binders Measuring Low Temperature Cracking of the Asphalt Binders before SHRP Limiting Stiffness Incremental Stress/Strain Calculation a) Hills and Brien Model b) Shahin-McCullough Model TC c) Program COLD... 39

6 6 d) Ruth Model Statistical Models Fracture Mechanics Models Development of the Lytton Model Measuring Low Temperature Cracking after SHRP Bending Beam Rheometer Direct Tension Test Current Specifications to Determine the Low Temperature Cracking Asphalt Binder Cracking Device (ABCD) Determination of Fracture Strength from ABCD Test Repeatability Test with ABCD FHWA Binders Physical Hardening Glass Transition (T g ) and the Physical Hardening OBJECTIVE MATERIALS AND PROCEDURE Materials Sample Preparation Testing Procedure RESULTS Cooling Rate ABCD Conditioning... 90

7 7 5.3 Strain Deflection during Conditioning Graphs of ABCD Specimens Comparison of All Binders Statistical Analysis CONCLUSIONS AND RECOMMENDATIONS REFERENCES

8 8 List of Tables Page TABLE 2.1 Repeatability with ABCD TABLE 2.2 Cracking temperatures of FHWA Binders determined by ABCD and other methods TABLE 2.3 Stiffness data for Highway 118 binder [4] TABLE 2.4 m-values for Highway 118 binders [4] TABLE 5.1 Cracking temperatures at different Cooling Rates TABLE 5.2 Comparison of cooling rate TABLE 5.3 Two way ANOVA with replicates for 1 C/hr and 10 C/hr TABLE 5.4 Comparison of conditioning times (10 C/hr cooling afterward) TABLE 5.5 Two way ANOVA with replication results for 0.5 hr, 24 hr and 72 hr TABLE 5.6 Comparison of binders

9 9 List of Figures Page FIGURE 2.1 Typical viscoelastic response of asphalt cement under creep loading FIGURE 2.2 Nomograph for determining modulus of stiffness of asphalt cement (after McLeod) [38] FIGURE 2.3 Relationship between penetration, PVN, and base temperature for asphalt cements (after McLeod) [38] FIGURE 2.4 Temperature susceptibility of penetration graded asphalt cements FIGURE 2.5 Nomograph for predicting cracking temperatures in C from asphalt penetration (after Gaw) FIGURE 2.6 Schematic of bending beam rheometer [64] FIGURE 2.7 Schematic of direct tension test [64] FIGURE 2.8 Aluminum ring placed inside the mold FIGURE 2.9 Aluminum ring after pouring of the asphalt binder FIGURE 2.10 Superposition of creep curves of asphalt AAM-1 measured at isothermal ages of 2 hour and 16 days [74] FIGURE 2.11 Isothermal volume measurement for three asphalts at -15 C over a period of 24 hours [69] FIGURE 2.12 Correlation of volume change and hardening shift at equi-isothermal ages [69] FIGURE 2.13 Correlation between hardening shift factors and the estimated deviation from equilibrium volume line [69]... 72

10 10 FIGURE 2.14 Correlation between total endothermic enthalpy and hardening potential for 8 core SHRP asphalts [69] FIGURE 2.15 Specification temperatures for Highway 118 binders. (S 1 and m 1 are 60 sec limiting stiffness and m-value temperatures -10 after one hour of isothermal conditioning at the grading temperatures whereas S 3 and m 3 are the same after three days of conditioning. Similarly, MP 1a and strain at fracture specification temperatures are also given for one-hour and threeday conditioning, respectively. Numbers on columns are rounded to the nearest degree.) [4] FIGURE 4.1 Pouring sample from steel cup with aluminum foil spout FIGURE 4.2 Prepared samples with thermocouples FIGURE 4.3 LabVIEW real-time strain plot FIGURE 4.4 Tested samples from silicone molds FIGURE 5.1 Comparison of chamber and specimen temperatures FIGURE 5.2 Effects of cooling rate on average ABCD cracking temperature FIGURE 5.3 Temperature comparison during 24 hr conditioning FIGURE 5.4 Temperature comparison during 72 hr conditioning FIGURE 5.5.a ABCD strain reading during conditioning of coat 209 at -20C for 24 hours (Run#1) FIGURE 5.5.b ABCD strain reading during conditioning of coat 209 at -20C for 24 hours (Run#2) FIGURE 5.6 ABCD strain reading during conditioning of coat 130 at -20C for 24

11 11 hours FIGURE 5.7 ABCD strain reading during conditioning of coat 061 at -20C for 24 hours FIGURE 5.8 ABCD strain reading during conditioning of coat 129 at -20C 24 hours FIGURE 5.9.a ABCD strain reading during conditioning of flux 130 at -20C for 24 hours (Run#1) FIGURE 5.9.b ABCD strain reading during conditioning of flux 130 at -20C for 24 hours (Run#2) FIGURE 5.10 ABCD strain reading during conditioning of PAV at -20C for 24 hours FIGURE 5.11.a ABCD strain reading during conditioning of coat 061 at -20C for 72 hours (Run#1) FIGURE 5.11.b ABCD strain reading during conditioning of coat 061 at -20C for 72 hours (Run#2) FIGURE 5.11.c ABCD strain reading during conditioning of coat 061 at -20C for 72 hours (Run#3) FIGURE 5.11.d ABCD strain reading during conditioning of coat 061 at -20C for 72 hours which cracked during conditioning (Run#4) FIGURE 5.11.e ABCD strain reading during conditioning of coat 061 at -20C for 72 hours (Run#5) FIGURE 5.12.a ABCD strain reading during conditioning of coat 129 at -20C for 72

12 12 hours which cracked during conditioning (Run#1) FIGURE 5.12.b ABCD strain reading during conditioning of coat 129 at -20C for 72 hours (Run#2) FIGURE 5.12.c ABCD strain reading during conditioning of coat 129 at -20C for 72 hours (Run#3) FIGURE 5.13.a ABCD strain reading during conditioning of coat 209 at -20C for 72 hours (Run#1) FIGURE 5.13.b ABCD strain reading during conditioning of coat 209 at -20C for 72 hours (Run#2) FIGURE 5.13.c ABCD strain reading during conditioning of coat 209 at -20C for 72 hours (Run#3) FIGURE 5.14.a ABCD strain reading during conditioning of coat 130 at -20C for 72 hours (Run#1) FIGURE 5.14.b ABCD strain reading during conditioning of coat 130 at -20C for 72 hours (Run#2) FIGURE 5.15.a ABCD strain reading during conditioning of flux 130 at -20C for 72 hours (Run#1) FIGURE 5.15.b ABCD strain reading during conditioning of flux 130 at -20C for 72 hours (Run#2) FIGURE 5.16.a ABCD strain reading during conditioning of PAV at -20C for 72 hours (Run#1) FIGURE 5.16.b ABCD strain reading during conditioning of PAV at -20C for

13 13 72 hours (Run#2) FIGURE 5.16.c ABCD strain reading during conditioning of PAV at -20C for 72 hours (Run#3) FIGURE 5.17 ABCD strain deflection of the empty invar rings during 72hr conditioning FIGURE 5.18 ABCD strain deflection of the empty invar rings during 24hr conditioning FIGURE 5.19 ABCD strain reading of the coat 129 cracked during conditioning FIGURE 5.20 ABCD strain reading of coat 129 cracked during conditioning FIGURE 5.21 ABCD strain of coat 130 after 72 hrs conditioning FIGURE 5.22 ABCD strain of coat 209 after 72 hrs conditioning FIGURE 5.23.a ABCD strain of coat 061 after 72 hrs conditioning (Run#1) FIGURE 5.23.b ABCD strain of coat 061 after 72 hrs conditioning (Run#2) FIGURE 5.23.c ABCD strain of coat 061 after 72 hrs conditioning (Run#3) FIGURE 5.24 ABCD strain of coat 209 after 72 hrs conditioning FIGURE 5.25.a ABCD strain of flux 130 after 72 hrs conditioning (Run#1) FIGURE 5.25.b ABCD strain of flux 130 after 72 hrs conditioning (Run#2) FIGURE 5.26.a ABCD strain of PAV after 72 hrs conditioning (Run#1) FIGURE 5.26.b ABCD strain of PAV after 72 hrs conditioning (Run#2) FIGURE 5.27 ABCD strain of PAV after 24 hrs conditioning FIGURE 5.28.a ABCD strain of flux 130 after 24 hrs conditioning (Run#1) FIGURE 5.28.b ABCD strain of flux 130 after 24 hrs conditioning (Run#2)

14 14 FIGURE 5.29 ABCD strain of coat 209 after 24 hrs conditioning FIGURE 5.30 ABCD strain of coat 129 after 24 hrs conditioning FIGURE 5.31.a ABCD strain of coat 061 after 24 hrs conditioning (Run#1) FIGURE 5.31.b ABCD strain of coat 061 after 24 hrs conditioning (Run#2) FIGURE 5.32 ABCD strain of coat 129 after 24 hrs conditioning FIGURE 5.33 Comparisons of binders with different cooling rates & conditioning times FIGURE 5.34 Cracking temperature of binders at different conditioning time

15 15 CHAPTER 1 INTRODUCTION Asphalt cement is a dark brown to black cementitious material that is either naturally occurring or is produced by petroleum distillation. Asphalt cement is the oldest engineering material known to man. Its adhesive and water proofing properties were known at the dawn of civilization and it was used by the shipbuilding industry as a water proofing agent. An ancient civilization in the Indus Valley (northwestern India) used asphalt cement in the construction of large public baths or tanks around 3000 B.C. [1]. Commercial types of asphalt can broadly classified into two categories: natural asphalts and petroleum asphalts. Natural asphalts are found in geologic strata whereas, the petroleum asphalts are colloidally dispersed hydrocarbons residue from crude petroleum distillation. Today asphalt from petroleum crude is mostly used for paving roads. Pavements are divided into two types - rigid pavements and flexible pavements. Rigid pavements are constructed using Portland cement concrete while, flexible pavements are constructed using asphalt. Flexible pavement failure can be attributed to causes such as changes in the crude source and refining processes; improper mix design; increased truck traffic volume; tire pressure and axle loading; deficiency in specifications and improper use of additives [2].

16 16 There are four major types of distress which cause asphalt pavement failures [2]. Rutting or permanent deformation caused by progressive movement of materials under repeated application of loads at high temperature. Fatigue cracking caused by repeated application of loading which exceeds the structural design criteria. Low temperature thermal cracking caused by the development of thermal stress that exceeds the fracture strength at low temperatures. Moisture damage or stripping of asphalt pavements caused by the lack of an adhesive bond between the asphalt and aggregate due to the presence of moisture. Stripping may contribute to rutting and fatigue cracking. In the worst case this can lead to disintegration of the pavement. Low-temperature cracking distress is the focus of this project and the cracking mechanism is further discussed below. As the ambient temperature drops, asphalt mixtures shrink due to thermal contraction. The pavement also stiffens and becomes brittle. This is because asphalt is thermoplastic material; i.e., the stiffness increases as the temperature decreases. Thermal stress is induced in the asphalt mixture since the friction between the pavement and underlying pavement structure restricts the asphalt from contraction. When the thermal stress exceeds the tensile strength of the asphalt pavement, a transverse crack will develop at the surface to relieve the stress. This non-load associated crack can occur from a single critically low-temperature or from a thermal cycle that fluctuates just above the

17 17 critical cracking temperature [1]. Jung [3] reported that at colder temperatures or repeated temperature cycles, a crack will penetrate the full depth and width of the asphalt concrete layer. The crack initiates at the surface because this area is cooled first. Thermal stresses are generally equal throughout the length of a road with a constant air temperature thus the cracks tend to be evenly spaced. The transverse cracks caused by low temperatures allow fine aggregates and water into and out of the pavement, weakening the subgrade. Pavement roughness from low-temperature cracks can be especially severe where subgrade soils contain expansive clays. Moisture entering the cracks causes localized swelling, which results in upheaval of the pavement surface and adjacent to each crack. Also, as fines are washed from the subgrade, the pavement structure is left unsupported. Further cracking then occurs upon loading, leaving depressions in the road [2]. Asphalt concrete failures caused by low-temperature cracking are disastrous to pavement performance and service life. A poor riding surface necessitates an increase in maintenance and eventual early replacement of the road. When overlaid, these cracks reflect through the new pavement. This costs taxpayers more money and time waiting for road construction. Therefore, it is imperative to know the critical cracking temperature of the asphalt cement and concrete in order to prevent the aforementioned failures. Stiffness of binder at low temperature is the key factor affecting thermal cracking potential of asphalt binder. Current specifications are based on it. Physical hardening significantly alters stiffness. The phenomenon called physical hardening was observed to cause significant isothermal changes in the creep compliance. Physical hardening is

18 18 caused by the gradual density change that occurs over time when bitumen is held at low temperatures. The cracking temperatures of the binders, predicted by AASHTO MP 1a procedure, increased by 4 C to 14 C when they are tested after three days of physical hardening [4]. The increase in the cracking temperature is more than two PG grades. It can be inferred from the increased cracking temperature that the pavement failure may happen much before the specified PG low temperature is reached when physical hardening effect is considered. According to Shenoy [5], the physical hardening phenomenon observed in asphalt binders, under isothermal conditions at low temperatures is found to be absent when the asphalt binder exists in combination with the aggregates in the aggregate-asphalt mixes as they are restrained from contraction. When Asphalt Binder Cracking Device (ABCD) is used the asphalt binder is restrained from contraction by a metal ring. More discussion is carried out in future aspects. So, the reason to conduct research on physical hardening is very important. The effect of physical hardening on the cracking temperatures of the asphalt binders has not been experimentally measured. The stiffness values after one day and three days of binder physical hardening were determined by some researchers [6]. But the exact cracking temperatures were not experimentally determined. The low temperature cracking of the asphalt binder after considering the physical hardening effect can be determined by Asphalt Binder Cracking Device (ABCD). The operating principle of ABCD is based on the differential thermal contraction between the metal ABCD ring and an asphalt binder placed outside of the ring. As the temperature is lowered, the test binder shrinks more rapidly than the ABCD ring placed inside which results in development of

19 19 thermal stresses. When the developed thermal stress exceeds the strength of the binder, the binder specimen cracks. Strain gauges installed inside of the ABCD ring detect the fracture, and the temperature is recorded as the ABCD cracking temperature [7]. The objective of this research is to determine the effect of physical hardening on the low temperature cracking of the asphalt binder using an ABCD. The method will induce thermal stress similar to field conditions by restraining asphalt from contraction until a low-temperature crack is observed. Asphalt contraction is carried out by varying cooling rates and isothermal conditioning time in order to determine the physical hardening effect. The change in the cracking temperatures is determined. Thus, the change in the cracking temperature when physical hardening effect is considered can be determined. The results of this study show that the cracking temperatures of the binders vary with cooling rate and also with change in the conditioning time. With an increase in the conditioning times the cracking temperatures of some binders decrease and cracking temperatures for some binders increase and cracking temperatures for some remain the same regardless of conditioning time. The change in the cracking temperature when tested with ABCD for physical hardening effect is around 2 C to 5 C.

20 20 CHAPTER 2 LITERATURE REVIEW 2.1 Low Temperature Cracking of Flexible Pavements In cold regions, one of the most prevalent distresses of the asphalt pavement is thermal cracking. It was realized that low temperature cracking characteristics of pavements were not a result of temperature alone, but also were influenced by variations in mixes and climate [8]. During winter, de-icing materials can infiltrate through pavements and thaw base materials, causing depressions to form. Fine materials mixed with water can pump through cracks, creating voids below the pavement, which also causes depressions to form. These problems may reduce rideability and service life of a pavement [3,9,10]. The ingress of water through the cracks also tends to cause loss of bonding, increasing the possibility of stripping and resulting in a depression at the crack brought about by raveling of the lip of the crack and pumping of the fine fraction of base material. Which result in rough riding qualities and often secondary cracks are produced, that parallel to the major cracks. Pavement roughness at low temperature contraction cracks can especially be severe when subgrade soils are expansive clays. Moisture entering the cracks causes localized swelling of subgrade soil, which results in upheaval of the pavement surface at and adjacent to each crack. Low temperature cracking occurs more often after the binder is aged because asphalt becomes more brittle as it ages [11].

21 Factors Affecting Low Temperature Cracking of HMA The low temperature cracking can occur from a single course of low temperatures [12]. It can also be caused by a continuous chain of fluctuations in the temperature [3,10,12]. Once the low temperature cracking occurs it spreads from the top of the pavement to the bottom layers. Oxidation of the asphalt cement may result from the presence of high percentage of air voids after construction of the asphalt pavement, which leads to the stiffening of asphalt mix. It was noted that the pavements with high modulus of stiffness value at low temperatures is more prone towards cracking because of less flexibility [2,8]. Use of soft and age-resistant asphalt binders can reduce failure due to low temperature cracking [2,13]. A soft asphalt binder would have adequate ability to relax stresses and can generate improved resistance to thermal cracking [14]. Furthermore, to tackle the issue of oxidation, the in-place air voids should be kept low. Jung [3] stated that influence of low-temperature cracking in asphalt concrete pavements could be broadly categorized as follows: Material Environmental Pavement structure Materials The most crucial material factor is rheological properties of asphalt cement. Rheology is defined as the study of the deformation and flow of matter, more specifically, a material s stress-strain-time-temperature response. Stiffness of asphalt is

22 22 one of the most important rheological properties to thermal cracking. According to Roberts et al. [2] hot mix asphalt that has high stiffness at low temperatures, is very prone to cracking due to low flexibility as stated above. Mix stiffness at low temperatures is primarily dependent on the stiffness of the asphalt cement. Stiffness (S) as defined by Van der Poel [15] in Equation 2.1 has a modulus that relates stress (σ ) and strain (ε ) as a function of time (t) and temperature (T). ( tt ) S, = ε σ Equation 2.1 This time-temperature dependency signifies asphalt as a viscoelastic material. It behaves like a viscous liquid under slow loading at high temperatures, but like an elastic solid under fast loading at low temperatures. Furthermore, a viscoelastic material combines both elastic behavior, in which the material stores work applied and recovers its initial state after removing applied loads, and viscous behavior in which the material deforms permanently under applied loads and dissipates the stress into mainly permanent deformation [16]. In Figure 2.1, there is a constant static creep load applied and for a viscoelastic response in the bottom of the figure, strain will increase elastically, followed by a nonlinear delayed elastic region until there is a viscous response, which is known as creep. Once the load is removed, there is creep recovery until a constant value is reached. The permanent deformation is due to viscous flow. Viscous flow is the result of stress relaxation, where a binder flows to relieve stress.

23 23 Figure 2.1 Typical viscoelastic response of asphalt cement under creep loading The stiffness or consistency (i.e., viscosity or penetration) and the temperature susceptibility at low temperatures have historically been the most important considerations in thermal crack prevention [17]. Ideally, a high stiffness at higher temperatures is preferred to resist rutting and a low stiffness at lower temperatures to resist thermal cracking. A lower viscosity (or penetration) grade of asphalt cement will produce a lower stiffness with decreasing temperature and reduces the potential for lowtemperature cracking. According to Roberts et al. [2], temperature susceptibility is the rate at which the consistency of asphalt cement changes with a change in temperature. Cement with high susceptibility should be avoided because its viscosity and stiffness are

24 24 high at low service temperatures. Crystalline wax, which is naturally present in asphalt, can be responsible for increasing temperature susceptibility by raising the glass transition temperature. The wax content mainly depends on the source of crude oil from which the asphalt binder is manufactured. Temperature susceptibility has been measured by the Penetration Index or Pen-Vis Number and the wax content with Differential Scanning Calorimeter. Age hardening is another asphalt property that can affect the low temperature performance of binder [18]. Asphalt cement hardens when it is heated and mixed with aggregates at temperatures ranging from 135 C to 165 C. During the short mixing and transporting period at high temperatures, the asphalt cement is oxidized and loses volatile components. This short term aging causes an increase in viscosity and becomes stiffer. Long-term aging at ambient service temperature also contributes to hardening because asphalt cement hardens with age, but at a much slower rate once the asphalt concrete is paved. Short-term aging is simulated by the Rolling Thin Film Oven (RTFO) and long term aging is simulated by Pressure Aging Vessel (PAV) in the current specifications. When the pavement reaches its maximum density under traffic loading, age hardening decelerates further. Age hardening can be increased if the pavement has a high air void content allowing air, water, and light to penetrate. Higher asphalt content can deter hardening because a thicker cement coating will harden at a slower rate. Asphalt being an organic compound will react with oxygen from the environment. Oxidation changes the structure and composition of the asphalt molecules. Oxidation causes the asphalt to become more brittle, leading to the term oxidative, or age, or hardening. Oxidation occurs

25 25 more rapidly at higher temperatures. A considerable amount of hardening occurs during hot mix asphalt (HMA) production, when the asphalt cement is heated to facilitate mixing and compaction. That is why; oxidation is more of a concern when the asphalt cement is used in a hot, desert climate. It is observed that as a pavement ages, low-temperature crack spacing will decrease Additives The problems faced by asphalt pavements leading to premature failure can be attributed to the binder becoming less stiff at high temperatures causing permanent deformation of the pavement. Similarly the mixture may be unable to withstand breakage in bonding between aggregate and binder due to the presence of moisture causing stripping of the asphalt mix. The binder may become too brittle to resist fatigue cracking under repeated loading. The pavement may also undergo low temperature cracking due to oxidative aging and high stiffness at low temperatures [2,19]. In order to overcome these failures the practice of modifying the asphalt binder became common and polymers in particular have received wide spread attention as performance improvers of the asphalt binders [20,21]. The main advantages of using polymers are that they can increase the stiffness of asphalt at high temperatures, which prevents permanent deformation and increases the strain tolerance, thus improving the fatigue resistance at ambient temperatures [20,21,22,23,24,25]. It was found that the polymers showed a dramatic change in properties at high temperatures. Superpave system incorporates modifiers added to the asphalt binders and provides a means to evaluate the performance of the

26 26 modifiers through various tests [21]. Even though conventional asphalt cements have been characterized for a long time, carrying out the same processes for the modified asphalts has proven to lead to unreliable results as these procedures do not accurately describe the mechanical behavior of modified asphalt cements [21,25]. The performance of the modified asphalts cannot be described based on material oriented procedures such as force ductility, elastic recovery, toughness and tenacity which give information regarding the amount and type of asphalt binders [21,22]. But rather must be defined in terms of performance related physical properties, such as resistance to rutting or cracking and related to pavement life [21,25]. The current Superpave binder specification is based on that all simple binders which exhibit the same time temperature and load dependency factors and modified binders are considered to be simple in rheological behavior [2,21]. Also it is assumed that oxidation and volatilization are the only mechanisms that effect binder properties. But these assumptions are not necessarily valid for all types of modified binders. The current protocols cannot measure the special important characteristics shown by polymer modified asphalts [26]. Thus there is a need to evaluate performance characteristics of the modified asphalt mix and its applicability to the Superpave mix design [27,28,29]. The results of modified asphalts with additives are highly dependent upon the concentration, the molecular weight, the chemical composition, the particle size and the molecular orientation of the additive, as well as the crude source, the refining process and the grade of the base asphalt used [21]. There are many reasons to modify an asphalt binder. One of them is it softens the binder at low temperature which improves relaxation

27 27 properties and strain tolerance by minimizing the non load associated thermal cracking [21]. The benefit of polymer modified asphalt binder is reduced vulnerability to temperature [30]. The properties of the modified asphalt depend on the two things after polymer addition. They are polymer system and compatibility. Polymers are classified into two categories. They are Elastomers and Plastomers. Elastomers can be stretched and elastically recover their shape when released but they add only less strength to the asphalt until they are expanded but they get stronger as they are pulled out of shape, and they recover when released, much like a rubber band. Plastomers form a tough, rigid, three dimensional networks and have high early strength to resist heavy loads, but may crack at lower strains [21]. The smaller molecules are called monomers. Two or more different monomers can be called as co-polymers. The polymers improve the thermal cracking resistance at low temperatures by restraining the crack propagation area [21]. The place where there is a temperature extremity, the polymer modified binders are used [31]. It has been observed that binders can be softened and the temperature susceptibility reduced by modifying asphalt with polymer crumb rubber, which is produced from grinding recycled tires. When blended with hot asphalt, the rubber particles swell from three to five times their original volume and soften by absorption of aromatic components from the asphalt. This reaction also increases the high temperature viscosity of the binder improving rutresistance [21,25].

28 Environment There are two main environmental factors that impact thermal cracking; ambient air temperature and cooling rate. The ambient temperature is the most important factor because it correlates to pavement surface temperature, which is used as the design temperature for selecting PG asphalt cement. Strategic Highway Research Program (SHRP) developed this performance based grading system to designate the lowest temperature a binder can endure without cracking. The minimum pavement temperature is the temperature at the surface of the layer [2]. Anderson et al. [8] stated that the cooling rate, which can be considered as strain rate, has a significant effect on the strength of asphalt. Strain rate influences the stress relaxation capabilities of the asphalt mixture because the faster deformation is applied, the less time asphalt has to react and mitigate the stress. Hence, faster cooling would result in a greater incidence of low-temperature cracking. Jung and Vinson [3] showed that the fracture temperatures of asphalt mixtures become warmer as the cooling rate increases from 1 to 5 C/hr. Fabb (1974) [32] stated that the rate of cooling has little or no effect on the failure temperature when it is greater than 5 C/hr. More realistic cooling rates that are seen in the field are 0.5 to 2 C/hr (Olard et al., 2004 [33]). It is ideal for thermal cracking tests to operate at cooling rates that represent field conditions, but test times become lengthy and expensive.

29 Pavement Structure Low-Temperature cracking is also affected by various pavement structural elements. Crack frequency as a function of pavement width has been supported by field documentation, which concludes that wider pavements have longer crack spacing. According to Jung and Vinson (1994) [3], crack spacing for secondary roads 7 m in width is approximately 30 m, where as for general airports, with pavements 15 to 30 m wide, the initial spacing can be greater than 45 m. Jung goes on to say that thicker asphalt concrete layers have a lower incidence of thermal cracking. At the St.Anne Test Road, increasing the asphalt concrete layer from 10 to 25 cm resulted in one half the crack frequencies when all other variable (asphalt mixture and environment) were the same. Therefore, a greater cross-sectional area would withstand a larger thermal stress. Another structural parameter is the coefficient of friction between the sub-grade and the pavement. Since an aggregate base has a lower thermal coefficient than that of asphalt, using a priming tack coat to adhere it to the asphalt concrete would decrease the overall contraction of the bonded layers. This treatment will facilitate the prevention of low-temperature cracking (Jung and Vinson [3]). 2.3 Determination of Low Temperature Cracking of Asphalt Binders The low temperature cracking of the asphalt binders has been determined in different ways as described below.

30 Measuring Low Temperature Cracking of the Asphalt Binders before SHRP Engineers have categorized cracking into two different groups. They are load associated cracking and non load associated cracking. The load associated cracking is described as the fatigue cracking where as the non load associated cracking is described as the low temperature thermal shrinkage cracking. A number of researchers have studied low temperature shrinkage cracking and have developed analytical and predictive models. Thermal fatigue cracking has received less attention from researchers. However, several models have been developed that account for both types of thermal cracking. Two distinct mechanisms for thermal cracking are now recognized. The first mechanism that was recognized by researchers is typified by a single low temperature excursion that causes the pavement to shrink to the extent that the thermal shrinkage stresses exceed the tensile strength of the asphalt concrete. The second mechanism is called thermal fatigue and results from the accumulated damage caused by stresses associated with repeated thermal cycling. In reality, most thermal cracking is probably caused by a combination of the two mechanisms. Before Strategic Highway Research Program (SHRP) the temperature at which asphalt binder fails was determined in different ways. The asphalt stiffness, penetration, temperature susceptibility are the properties which influence low temperature cracking [2]. If the stiffness of the asphalt cement is high, then it is more susceptible for the low temperature cracking. For this reason, the concept of the limiting stiffness was introduced [34,35,36,37]. By using the limiting stiffness value and McLeod s nomograph method [38] the minimum allowable penetration viscosity numbers are determined for different

31 31 values of penetration (Figure 2.2 shows McLeod s nomograph). Minimum kinematic viscosities were then determined from the corresponding penetration and penetration viscosity number (PVN) values as shown in Figure 2.3. By specifying the minimum kinematic viscosity thus determined for each penetration value, it was ensured that the PVN was not lower than the permissible value. At higher penetration values, the temperature viscosity lines are shifted relatively lower at 77 F, thus the asphalt cement does not exceed the limiting stiffness value in spite of steeper slopes which can be observed in the Figure 2.4. The research on this was based on the St. Anne Test Road in Canada [34]. The pavements cracked at an asphalt binder stiffness of 1x 10 9 N/m 2 at 0.5 hr loading time. As recommended by Gaw [34] the low temperature asphalt cement specifications are based on the Figure 2.5 at a performance level based on the pavement surface temperature. The asphalt cement specification limits are represented by a single line drawn on a logarithm plot of asphalt cement penetration at 25 C and 5 C as shown in the Figure 2.5. Asphalt cements for which the penetration at 25 C and 5 C appear on, or to the right of the specification line are considered acceptable. Both unaged and aged asphalt cement penetrations yield similar nomographic cracking temperature.

32 32 Fig 2.2 Nomograph for determining modulus of stiffness of asphalt cement (after McLeod) [38].

33 33 Fig 2.3 Relationship between penetration, PVN, and base temperature for asphalt cements (after McLeod) [38].

34 34 PENETRATION (0.1 MM) C B A 77 TEMPERATURE (F) A- Low Temperature Susceptibility Asphalt B- Medium Temperature Susceptibility Asphalt C- High Temperature Susceptibility Asphalt Figure 2.4 Temperature susceptibility of penetration graded asphalt cements

35 35 Figure 2.5 Nomograph for predicting cracking temperatures in C from asphalt penetration (after Gaw)

36 Limiting Stiffness The most commonly used approach for predicting thermal cracking in bituminous pavements and among the first to be used by researchers is limiting stiffness. This approach is based on relationship between mixture or asphalt cement stiffness at the minimum service temperature and the incidence of low temperature cracking. Gaw and others suggested a limiting mixture stiffness of 2.6 x 10 6 psi (18GPa) at 30 minutes loading time [39]. Suggested values for the limiting stiffness of asphalt cement range from 20,000 lb/inch 2 (140MPa) at a loading time of 2.8 hours to 145,000 psi (1 GPa) at 30 minutes [35,40]. The later value is suggested by the Asphalt Institute in a comprehensive report on designing asphalt concrete pavements to resist low temperature cracking [40] Incremental Stress/Strain Calculation Various computer models use the calculation of incremental stress or strain to predict thermal cracking in asphalt concrete pavements. These include the Hills and Brien Model, Shahin-McCullough model (Program TC-1), program COLD and the RUTH model. [41,42,43]. a) Hills and Brien Model Hills and Brien proposed that low temperature cracking can be predicted through the incremental calculation of thermal shrinkage stresses in the pavement [44]. The pavement is assumed to be free of thermal stresses above 32 F. The stiffness of the asphalt

37 37 cement at this temperature is estimated from Van der Poel s nomograph and a cooling rate of 18 F/hr is generally assumed. The thermal strain in the asphalt cement is calculated for each temperature increment by multiplying an assumed linear coefficient of thermal expansion of 1.1 x 10-4 / F by the change in temperature which is generally 9 F. The stress in the asphalt cement at this temperature level can then be calculated by multiplying the asphalt stiffness by the thermal strain. This stress is compared with typical tensile strength and stiffness data for asphalt cement such as that presented by Heukelom [45]. The stresses are summed for each temperature increment and the calculation proceeds to successively low temperatures until the stress calculated exceeds the tensile strength. According to this method, failure by thermal cracking is probable at this temperature. This method of calculation assumes that asphalt concrete behaves elastically a major shortcoming of this type of analysis because although at very low temperatures this assumption may be reasonable, at intermediate temperatures, asphalt cement and asphalt concrete behave viscoelastically & the stress relaxes. In the Hills and Brien procedure neglects the stress relaxation which occurs during the cooling cycle, and assumes that the behavior of the asphalt cement is linear elastic, and also assumes that the strain in the asphalt cement is the same as in the hot mix asphalt concrete.

38 38 b) Shahin-McCullough Model TC-1 The Shahin-McCullough model TC-1 is based on a probabilistic empirical/mechanistic computer program that computes the extent of thermal cracking as a function of time [41]. The main program contains four sub models that are described briefly below: The pavement temperature model, an improved version of the model developed by Barber [46], is used to predict an hourly pavement temperature as a function of air temperature, wind velocity, solar radiation, asphalt concrete thermal properties, and depth below the pavement surface. The thermal stress model is used to calculate the thermal stresses and strains in the asphalt mixtures as a function of its stiffness and changes in pavement temperature. This model in turn consists of four interacting sub models for predicting aging of asphalt, asphalt stiffness, asphalt concrete stiffness, and thermal stresses and strains. The low temperature cracking model is used to predict the percentage of the pavement surface that is cracked at any time, t. Probabilistic methods are used to predict whether the thermal stresses exceed the mixture strength. Both the thermal stress and mixture strength are assumed to be random variables and are defined in terms of their means and standard deviations. Thermal fatigue model adds the effect of thermal fatigue caused by daily temperature cycling to the effect of low temperature cracking. The model grew out of the realization that thermal cracking of asphalt concrete pavements occurs in the milder climatic zones of the United States as well as in the northern zones having

39 39 much lower temperatures. For calculating thermal stresses and strains, the bitumen and asphalt concrete mixture stiffnesses are determined using the relationship developed by Van der Poel and later modified by Heukelom and Klomp [47,48]. The model assumes that the surface layer is fully restrained, the surface behaves as an infinite beam, the thermal stresses, at the end of each daily temperature cycle, are negligible and the maximum thermal stress occurs at the minimum daily temperature. Thermal stresses calculated over the range of temperatures used as input to the program, are then compared with a tensile strength vs. temperature relationship that is an input to the program. However, no test procedure for measuring this tensile strength is recommended in the TC-1 documentation. To account for the variability of properties of the hot mix asphalt concrete, it is assumed that both the stress and tensile strength are normally distributed random variables. c) Program COLD Program COLD is a sophisticated model that estimates the temperature and the resulting thermal stresses in the pavement and the time at which low temperature cracking is likely to occur [42]. The first part of the program is used to calculate temperatures in the pavement at 1/8 th hr increments for each day. The model is based on thermodynamic principles that account for the rate of heat transfer through a solid medium. The second part of the model calculates the thermal stresses caused by the temperature differentials calculated from the first part of the program. The thermally induced stresses are calculated using a pseudo-elastic beam analysis similar to the one used in the Shahin-McCullough

40 40 TC-1 program. A strength/temperature relationship for the hot-mix asphalt is a required input to the program and is compared with the thermally induced stresses at 2 hr intervals. The program predicts the expected time at which low-temperature cracking will occur. The materials inputs for the COLD program include the absorptivity, emissivity, and convection coefficient of the surface. Thermal conductivity of the asphalt mixtures (both unfrozen and frozen), dry density and moisture content of the asphalt mixtures, thicknesses of the layers, and both creep modulus and tensile strength of the asphalt mixture as functions of temperature. The creep modulus versus temperature relationship can be estimated using Van der Poel s nomograph as modified by Heukelom and Klomp [47]. Although any time of loading can be used with the nomograph, 2 hr was recommended [42]. The tensile strength can obtained with either the diametral indirect tension or the uniaxial tension test. The loading rate suggested in the program documentation is 0.01 in/min (4.23µm/s) [42]. Unlike the Shahin-McCullough TC-1 model, the thermal coefficient of expansion used in COLD is assumed independent of temperature or stiffness. The results predicted by the COLD program were compared with field observations made in Canada, including those from the Saint Anne Test Road and a test road in Edmonton. Efforts in a previous study to verify the COLD subsystems were unsuccessful because of a lack of materials and environmental information for the test sections. The same lack of information was a problem when attempts were made to verify the TC-1 program [49].

41 41 d) Ruth Model Based on work conducted in Florida, Ruth and associates developed procedures for predicting thermally and load-induced pavement cracking [50]. Low temperature asphalt viscosity measurements are used in Ruth s model to estimate various parameters for the asphalt concrete mixtures. These parameters are then used to compute thermal stresses and strains, which are then used to predict relative cracking potential. Either the temperature/viscosity relationship developed from asphalt viscosity data for three or more temperatures or the actual temperature/viscosity measurements can be used as input to the program. A pavement cooling rate curve and a minimum pavement temperature typical of the temperatures at the site are calculated. Incremental strains are then computed at 15 to 30 min increments using the stress and viscosity of the mix. The absolute viscosity of the asphalt cement is measured with the Cannon constant stress rheometer (Schweyer rheometer). The incremental strain values are accumulated to obtain the total creep strain. The computer program output lists all of the computed parameters including stresses and strains, according to time and temperature. Failure criteria are applied to the appropriate output parameters to identify the failure (critical) temperature or the maximum stress & strain attained. The stiffness values at a loading time of 20,000 s obtained for the materials at the Saint Anne Test Road were used by Ruth in the development of the model [50].

42 42 This model is unique in that the viscosity and flow measured by the Schweyer rheometer are used to determine the stiffness of the asphalt mix. Application of the model has been confined to its development, and its validity has not bee examined by others Statistical Models Statistical Models also have been developed as predictors of the occurrence and extent of thermal cracking. Both studies discussed below were completed in the early 1970 s using observations of pavements in Ontario, Canada, as a database. The Hajek and Haas statistical model requires values of asphalt stiffness (predicted from a nomograph), pavement thickness, pavement age, subgrade type, and a winter design temperature [51,52]. Thirty-two observations were used by Hajek and Haas to develop the regression model, including seven observations used to verify the model. The reported coefficient of determination, R 2, for the regression equation was 0.82 in the following prediction of the cracking index, I [53]. I = D M + ( t A) log S B (0) S B (0) log D Equation 2.2 Where, I = cracking index, number of transverse cracks per 500 ft (152 m) of roadway. D = indicator variable for subgrade type, unitless: clay, D=2; loam, D=3; sand, D=5.

43 43 M = winter design temperature, C, defined as the temperature below which only 1 percent of the hourly temperatures occur during the coldest January for a 10- year period. A = age of pavement, years t = thickness of asphalt concrete surface layer, S B (0) = stiffness of the original bitumen at 20,000 s loading time and at the winter design temperature, kg/m 2. The bitumen stiffness used in this model is calculated using McLeod s suggested modification of Heukelom and Klomp s nomograph. [47,54] To use the nomograph, the penetration at 77 F (25 C) and kinematic viscosity at 275 F (135 C) (or penetration at 77 F (25 C) and ring and ball softening point temperature) must be measured for each asphalt cement. The Hajek-Haas model was used to predict the cracking index for 32 sections of pavement in Michigan [49]. Linear regression of the predicted versus the observed cracking gave an R 2 of This agreement between the observed and predicted values is poor, demonstrating the same problem shown with the TC-1 model the models are inaccurate when they are used for conditions other than those for which they were developed. Data collected from six sites in Texas were also used with the Hajek-Haas model. Once again, the predicted values showed poor agreement with the observed values, explainable, in part, because the primary mode of thermal cracking in West Texas is thermal fatigue, whereas

44 44 the primary mode of cracking in the northern United States and Canada is the result of shrinkage stresses associated with a single low temperature excursion. Fromm and Phang also developed a statistical model to predict low temperature cracking using 33 observations of Ontario roads as a database [55]. They used a stepwise regression procedure to develop several different models applicable to different climatic regions in Ontario. Thirty two parameters were used as possible predictors in this stepwise procedure, with the final general model resulting in nine predictor variables, the most significant of which were freezing index, viscosity ratio, critical temperature, and air voids in the mix. The freezing index is the cumulative number of degree-days below 32 F (0 C). The viscosity ratio is the viscosity at 140 F (60 C) divided by the viscosity at 275 F (135 C). The critical temperature was defined at which the viscous flow observed in the asphalt concrete under a 44-lb/sq.inch (3l0-KPa) tensile stress equals the shrinkage during a 10 F (5.5 C) drop in temperature. Viscous flow was measured by loading a 1.5 in X 1.5 in X 8 in (38mm X 38mm X 220mm) asphalt concrete beam in direct tension under a 100-lb (440-N) load. The coefficient of thermal expansion was also measured experimentally on 2in X 2in X ll.3in (51mm X 5lmm X 287mm) asphalt concrete beams Fracture Mechanics Models None of the models discussed previously in this chapter provided an acceptable prediction of low-temperature shrinkage cracking or thermal cracking. Either these models must be improved or alternative models must be developed. One of the more

45 45 promising, from a theoretical standpoint, is the fracture mechanics model. A short review of fracture mechanics is warranted before this model is presented. Fracture mechanics theory provides a rational explanation for the strength of a wide range of materials. In this theory, the strength of a material is related to the existence of flaws and the stress concentration at the tips of' these flaws. As the load applied to a material containing flaws increases, the stresses around the flaws reach a limiting value, and fracture becomes possible. The fracture mechanics approach to failure analysis leads to the determination of fracture toughness, a fundamental, or characteristic, material property. Linear elastic fracture mechanics (LEFM) is based on the Griffith failure hypothesis and is applicable to brittle materials. According to Griffith, brittle materials have microscopic flaws (or cracks) that are distributed randomly throughout their volume. Griffith's theory further states that the most critical crack (the flaw with the greatest stress concentration at its tip) will grow only when the elastic energy that is released during crack growth exceeds the energy of the newly created surface area. Therefore, crack growth will occur when [56]: σ c 1/ 2 a c = ( 2Eγ e/ π ) Equation 2.3 Where, σ c = remote critical stress, lb/in 2 (Pa) a c = a critical crack length, in (mm) E = elastic modulus, lb/in2 (Pa)

46 46 γ e = elastic surface energy per unit area, lb-in/in 2 (J/m 2 ) Equation indicates that a crack extension in brittle materials occurs when the product on either side of equation attains a critical value. This equation was modified by Irwin for a material displaying plastic deformation [57]: σc a = [2E( γ + γ p )/π ] c e 1/ 2 Equation 2.4 Where, σ c = remote critical stress, lb/in2 (Pa) a c = critical crack length, in (mm) E = elastic modulus, 1b/in2 (pa) γ e = elastic surface energy per unit area, lb-in/in 2 (J/m 2 ) γ = plastic surface energy per unit area, lb-in/in 2 (J/m 2 ) p According to fracture mechanics theory, the variable that governs fracture is the critical stress intensity factor, K c, and is given by: K c = σc πac x ao f ( /d) Equation 2.5 Where, K c = critical stress intensity factor for plane stress, lb/in 2 -in 1/2 σ c = remote critical stress, lb/in 2

47 47 a c = critical crack length, in (mm) f ( a o /d) = dimensionless variable that depends on the geometry of the specimen and crack length a = initial crack length, in (mm) o d = depth of beam, in (mm) K c is the maximum allowable stress intensity factor; it cannot be exceeded, and when it is equaled, catastrophic failure will occur. The remote stress, σ c is the stress in the body at a distance removed from the crack equal to ten times the size of the crack. The maximum constraint of the crack occurs with plane strain. For the special conditions where plane strain exists and the opening mode of crack extension is present, Kc becomes, by definition, K lc. The value of K lc at a particular temperature depends on specimen thickness and constraint. With increasing specimen thickness, the value of K c approaches K lc. Therefore, K lc is a fundamental material property characterizing crack resistance and is therefore called the plane strain fracture toughness. Thus, in principle, the same value of K lc should be found by testing specimens with different geometries and with different critical combinations of crack size and shape. The fracture toughness for a material that is brittle in nature and has very little crack-tip plasticity is obtained by using linear elastic fracture mechanics (LEFM) theory and a test procedure such as the compliance method, crack opening displacement (COD) method, or R-curve method [58,59,60].

48 48 During the last two decades, attempts were made to characterize the fracture properties of asphalt systems by applying fracture mechanics principles. Some of the research studies on this topic were conducted by Moavenzadeh [61] and Majidzadeh et al. [62]. Lytton applied fracture mechanics concepts to develop a computer model for predicting thermal crack initiation [63] Development of the Lytton Model Lytton's model, called THERM, predicts the amount of cracking, as well as the time of crack initiation [63]. Non-load-associated pavement cracking is assumed to result from the thermal fatigue mode of failure, rather than low-temperature tensile strength failure. A total of 576 runs of the computer model for four sites in Michigan and four sites in northern Texas provided data. A pavement design procedure was also developed on the basis of the model. Thermal fatigue is defined in the THERM program as fatigue caused by thermal cycling occurring below 75 F (25 C), which was assumed by Lytton to be a general lower bound for the stress-free temperature (a temperature at which the residual stresses resulting from thermal shrinkage are minimal or nonexistent). Shahin and McCullough's revisions of Barber's equations were used to compute pavement temperatures using the air temperature, wind speed, and solar radiation [41,46]. The temperatures thus determined were used in a fracture mechanics-based model for predicting the thermal fatigue cracking frequency.

49 Measuring Low Temperature Cracking after SHRP Tests to determine the low temperature cracking of the asphalt binders in the current specifications developed through the Strategic Highway Research Program (SHRP) are as follows. Bending Beam Rheometer [64] Direct Tension Test [64] Bending Beam Rheometer The Bending Beam Rheometer (BBR) measures the mid point deflection of a simply supported beam of asphalt binder at low temperature subjected to a constant load applied to the mid point of the beam. The device operates only in the loading mode, recovery measurements are not obtained. A test beam is placed in the controlled temperature fluid bath and loaded with a constant load for 240 seconds. The test load (980mN) and the mid point of deflection of the beam are monitored versus time using a computerized data acquisition system. Specimens are made by pouring heated asphalt into aluminum molds measuring 6.35 mm thick by mm wide by 127 mm long. The maximum bending stress at the midpoint of the beam is calculated from the dimensions of the beam, the span length, and the load applied to the beam for loading times of 8,15,30,60,120, and 240 seconds. The maximum bending strain in the beam is calculated for the same loading times from the dimensions of the beam and the deflection of the beam. The load and deflection at 0.0 and at 0.5 second are reported to verify that the full testing load (980mN) during the test is applied within the first 0.5 second. They are not

50 50 used in the calculation of the stiffness and m-value and should not be considered to represent material properties. The rise time of the load (time to apply full load) can be affected by improper operation of the pressure regulators, improper air bearing pressure, malfunctioning air bearing, and other factors. The schematic diagram of the Bending Beam Rheometer is shown in Figure 2.6. Figure 2.6 Schematic of bending beam rheometer [64] The test temperature for this test is related to the temperature experienced by the pavement in the geographical area for which the asphalt binder is intended. The flexural creep stiffness or flexural creep compliance, determined from this test, describes the low temperature stress-strain-time response of asphalt binder at the test temperature within

51 51 the linear viscoelastic response range. The low temperature performance of paving mixtures is related to the creep stiffness and the slope of the logarithm of the creep stiffness versus the logarithm of the time curve of the asphalt binder contained in the mix. The creep stiffness and the slope of the logarithm of the stiffness versus the logarithm of the time curve are used as performance based specification criteria for asphalt binders in accordance with AASHTO M320. Deflection of an elastic beam using the elementary bending theory, the mid span deflection of an elastic prismatic beam of constant cross-section loaded in three points loading can be obtained by applying following equations. δ = PL 3 /48EI Equation 2.6 Where, δ = deflection of the beam at mid-span, m P = Load applied, N L = span length, m E = modulus of elasticity, MPa I = moment of inertia, m 4 (bh 3 /12 for rectangle) The maximum bending stress in the beam occurs at the mid-span at the top and bottom of the beam which is calculated as follows: Where, σ = 3PL/2bh 2 Equation 2.7

52 52 σ = maximum bending stress in the beam, MPa P = constant load, N L = span length, mm b = width of beam, mm h = thickness of beam, mm The maximum bending strain in the beam occurs at the mid-span, at the top and bottom of the beam which is calculated as follows: ε = 6 δh/l 2 mm/mm Equation 2.8 Where, ε = maximum bending strain in beam, mm/mm δ = deflection of beam, mm h = thickness of beam, mm L = span length, mm Linear Viscoelastic Stiffness Modulus: According to the elastic-viscoelastic correspondence principle, it can be assumed that if a linear viscoelastic beam is subjected to a constant load applied at t = 0 and held constant, the stress distribution is the same as that in a linear elastic beam under the same load. Further, the strains and displacements depend on time and are derived from those of the elastic case by replacing E with 1/D (t).

53 53 Since 1/D (t) is equivalent to S (t), rearranging the elastic solution results in the following relationship for the stiffness: S (t) = 3 PL 3 4bh ( t) Equation 2.9 Where, S (t) = time dependent flexural creep stiffness, MPa P = constant load, N L = span length, mm b = width of beam, mm h = thickness of beam, mm δ (t) = deflection of beam, mm δ (t) and S (t) indicate that deflection and stiffness respectively are functions of time. The limitations for the Bending Beam Rheometer are It is applicable to material having flexural stiffness value from 20 MPa to 1 GPa. Both aged and unaged binders can be tested. Can test the sample between -36 C to 22 C Direct Tension Test The Direct Tension Test (DTT) measures the stress at failure and strain at failure in an asphalt binder test specimen pulled at a constant rate of elongation. Test specimens are prepared by pouring hot asphalt binder into an aluminum mold that is throated at the

54 54 ends to form an 18 mm long gage length. Two G10 phenolic end tabs are used to bond the asphalt binder during the test and to transfer the tensile load from the test machine to the asphalt binder. The test method was developed for asphalt binders at temperature where they exhibit brittle or brittle-ductile failure. A brittle or brittle-ductile failure will result in a fracture of the test specimen as opposed to a ductile failure in which the specimen simply stretches without fracturing. The test is not applicable at temperatures where failure is by ductile flow. A displacement transducer is used to measure the elongation of the test specimen as it is pulled in tension at a constant rate of 1 mm/min. The load developed during the test is monitored and the tensile strain and stress in the test specimen when the load reaches a maximum are reported as the failure strain and failure stress, respectively. The schematic diagram of the Direct Tension Test is shown in Figure 2.7. Figure 2.7 Schematic of direct tension test [64]

55 55 Stress at failure is used in a mechanistic pavement cracking model to compute critical cracking temperature. The procedure to compute critical cracking temperature is described in AASHTO PP42 or AASHTO MP1. The critical cracking temperature is then used in specifying the low temperature grade of asphalt binder in accordance with AASHTO M320. The test is designed to measure the strength of the asphalt binder at the critical cracking temperature. The asphalt binder has limited ability to resist stress without cracking. In the asphalt binder specification, failure stress is used to determine the critical cracking temperature. For evaluating an asphalt binder for conformance to AASHTO M320, the elongation rate of the gage section is 1.0mm/min and the test temperature is selected according to the grade of the asphalt binder. Other rates of elongation and test temperatures may be used to test asphalt binders. The failure stress can be calculated by dividing the failure load by the original area of the test specimen cross-section (6 by 6mm) as shown below: σ f = P f /A Equation 2.10 Where, σ f = failure stress, MPa P f = failure load, N A = original area of cross section, m 2 (36 X 10-6 m 2 ) The failure strain can be calculated by dividing the elongation at failure by the original gage length, as shown below. ε f = δ f /L c Equation 2.11

56 56 Where, ε f = failure strain, mm/mm δ f = elongation at failure, mm L c = effective gage length (33.8mm) The limitations for the Direct Tension Test are as follows. Can be used for both aged and unaged binders. The test is limited to asphalt binders containing particulate material having dimensions less than 250µm. The test is designed within the temperature range from 6 C to -36 C. 2.4 Current Specifications to determine the Low Temperature Cracking The Association of American State Highway and Transportation Officials (AASHTO) M320 binder specification is used currently [65] where, asphalt binder is graded based on its performance characteristics (Performance Graded or PG grading system). For instance, PG asphalt binder has a low chance of rutting up to 70 C pavement temperatures and will not crack at pavement temperatures above -22 C. In order to comply with the AASHTO M320 low temperature specification limit, the asphalt binder should have a stiffness less than 300 MPa and an m-value (slope of log [time]-log [BBR stiffness] plot) larger then at 60 second loading time and at a specification temperature plus 10 C (for instance, for PG X-22 grading, test is performed at 12 C). The

57 57 m-value is linked to the relaxation time of the asphalt binder. Asphalt binder with a high m-value will reduce the thermal stress quickly avoiding rapid accumulation of thermal stress. However if the creep stiffness is greater than 300 MPa but is less than 600 MPa, the asphalt binder is acceptable if the failure strain is more than 1.0 percent when tested by the DTT. The m-value requirement must be satisfied in both cases and the DTT is not required if the creep stiffness is less than 300 MPa. Failure strain is a decisive factor for a ductile-brittle transition. Asphalt binder with high failure strain is ductile and has an enhanced resistance to thermal cracking. Despite the fact that AASHTO M320 performs quite well with most original asphalt binders, it doesn t fit for some types of asphalt binders including physically and chemically modified asphalt binders. Consequently, an alternative asphalt binder specification (AASHTO MP1a) was lately adopted, which requires that the critical temperature for the thermal cracking of an asphalt binder be found using both BBR and DTT results. The stiffness at two temperatures are used to calculate the thermal stress in the binders, and DTT at multiple temperatures is used to find the tensile strength as a function of temperature. The temperature at which the calculated induced thermal stress equals the tensile strength of the binder is called the critical temperature. The AASHTO MP1a standard uses a thermo-viscoelastic model to calculate thermal stresses using binder relaxation modulus following procedures described in AASHTO PP42. The stress curve is compared with binder strength data obtained from the DTT to arrive at a critical temperature (T cr ) below which transverse cracking occurs

58 58 in the pavement [66]. Thermal stress calculations are based on one of the fundamental equations in linear viscoelasticity: σ ξ ' dε ξ ' dξ ( ) ( ) ( ' ) ' ξ E ξ ξ dξ = Equation 2.12 Where, ξ = Reduced time ( ξ ' ) σ = stress ε ( ξ ') = strain E ( ξ ') = Relaxation Modulus The coefficients of thermal expansion/contraction (α) of asphalt binders are presumed to be all equal (170 x 10-6 / C). Subsequently, the thermal stress is expressed as a convolution integral [66] as follows: ξ σ = ( ) ( th ε ε ) E ξ ξ ( ξ ) ' dξ ' Equation 2.13 o ( ξ ) Where, σ ( ξ ) = thermal stress at reduced time ξ, MPa ξ = reduced time, s ξ ' = time, dummy integration variable, s E = relaxation modulus of asphalt binder, MPa ε = mechanical strain th ε = α. T, thermal strain

59 59 In the method proposed by Bouldin et al. [66], BBR creep data is converted to relaxation modulus based on another fundamental equation in linear viscoelasticity: Where, ξ = Reduced time t E ( ξ ) = Relaxation Modulus ( ) Dt ( ξ ) D ( ξ ) = BBR creep compliance = 1/S ( ξ ) E ξ dξ = t Equation 2.14 The creep compliance is the inverse of the stiffness reported as part of the BBR test method. The previous equation can be solved numerically by the Hopkins and Hamming algorithm (Hopkins & Hamming, 1957) [66] to get E (ξ ) : ( t ) E( ti+ 1/ 2 ) [ f ( tn+ 1 ti ) f ( t1 t 1 )] ( t t ) t E Equation 2.15 n+ 1 n+ 1 / 2 = i+ f n+ 1 n This relaxation modulus is used in previous equation to calculate thermal stress. Bouldin et al. [66] proposed an empirical Pavement Constant (PC) to convert binder stress to mix stress and compare the thermal stress plot with DTT strength to arrive at T cr. Bahia et al. [67] suggested that the tensile strengths of asphalt binders be computed by combinations of five temperatures and four strain rates for accurate cracking temperature estimation.

60 60 The current AASHTO M320 and MP1a specifications do not clearly measure cracking temperatures of asphalt binder. They rely on overview of binder rheology and/or an unfairly large number of tests. Due to poor repeatability of DTT, six similar specimens are tested and the average of the highest four values is used as the strength value at each temperature and strain rate, which suggests that many DT tests are needed to describe the tensile strength envelop over the critical temperature range. Besides, strain rates used in DT test are orders of magnitude faster than field thermal strain rates, so the DTT derived tensile strengths may not be representative of field strength. The current AASHTO M320 and MP1a specifications do not consider the effect of physical hardening. The effects of the stress accumulated in the asphalt binder due to the low temperature contraction cannot be determined by BBR & DTT. 2.5 Asphalt Binder Cracking Device (ABCD) The Asphalt Binder Cracking Device is shortly called as ABCD [7]. It consists of a metal ring of uniform thickness and a strain gauge glued to the inner side of the metal ring. The radius of the metal ring is one inch. The metal ring is arranged on a steel mold as shown in Figure 2.8. The operating principle of ABCD is based on the differential thermal contraction between the metal ABCD ring and an asphalt binder placed outside of the ring which is shown in the Figure 2.9. As the temperature is lowered, the test binder shrinks more rapidly than the ABCD ring placed inside which results in development of thermal stresses. When the developed thermal stress exceeds the strength of the binder, the binder specimen cracks. Strain gauges installed inside of the ABCD

61 61 ring detect the fracture, and the temperature is recorded as the ABCD cracking temperature. Using a number of modified and unmodified binders, ABCD repeatability was studied. Based on this study, the ABCD and its test procedures were modified to produce repeatable results. From laboratory tests, ABCD showed the better correlation with Thermal Stress Restrained Specimen Test (TSRST) results than with the AASHTO M 320 and MP1a procedure. When the strength data determined by ABCD were used in the MP1a procedure, its correlation with TSRST data was good. Figure 2.8 Aluminum ring placed inside the mold

62 62 Figure 2.9 Aluminum ring after pouring of the asphalt binder Determination of Fracture Strength from ABCD Test: To determine ABCD fracture stress, calibration data were obtained by running the ABCD test with empty rings. Corrected strain for each temperature was obtained by subtracting the calibration strain from the uncorrected strain gauge reading. It was assumed that at the conditioning temperature of -20 C the ABCD ring is not subjected to any thermal stress. Both the calibration and uncorrected strain curves starting off from zero strain at -20 C but, upon fracture, strain of the ABCD ring jumped above the calibration curve. This indicates the asphalt binder was under measurable thermal stress after 30 minutes of conditioning at -20 C [67].

63 63 From force equilibrium, stress in the binder can be calculated by the following equation; εe A ABCD σ AC = ABCD Equation 2.16 AAC Where, σ = thermal stress in binder AC ε = corrected strain of ABCD ring E ABCD = Young s modulus of ABCD ring (E Al, E Steel, E Invar = 68.9, 200, 141 GPa) A ABCD, A AC = cross-sectional areas of ABCD ring and binder, respectively Repeatability Test with ABCD: The investigation of repeatability of the new ABCD procedure using the silicone mold and invar ring was performed with roofing binders. Twelve unmodified roofing binders were each tested two times with the invar ring and silicone mold ( coating was tested four times). Five of the binders were of a coating grade and seven were a flux grade. The coating and flux binders cover wide range of consistencies. The flux samples only had to be heated to 150 C in order to be sufficiently fluid to pour, but the coating samples were required to be heated to 240 C. For both flux and coating, there was no problem in the preparation and performing the ABCD tests. The results shown in Table 2.1 show good repeatability with an average standard deviation of 0.65 C.

64 64 Table 2.1: Repeatability with ABCD Binders Coating Flux ID ABCD Cracking Temp, C Std. Trial 1 Trial 2 Avg Dev. C (1) (2) FHWA Binders The Federal Highway Administration (FHWA) conducted a study on some binders and the results were published in the FHWA-RD report which contained TSRST results and the critical temperatures of binders based on MP1a, limiting stiffness (S), limiting m-value (m) and are given in Table 2.2. The same binders were tested using ABCD and the cracking temperatures obtained from ABCD are compared with the cracking temperatures obtained from TSRST and are shown in Table 2.2. The fracture stress was higher and the fracture temperature was lower for the binders when tested with ABCD. ABCD fracture strengths of FHWA binders are compared with the strengths measured by DTT and no apparent relationship was found. The fracture strength of a visco elastic material increases as temperature decreases or rate of loading increases.

65 65 Table 2.2 Cracking Temperatures of FHWA Binders determined by ABCD and other methods ID Description FHWA-RD Report TSRST Strength MPa Avg Tcr, C ABCD Avg Stress, MPa MP1a C MP 1 C S C m C TSRST C B6224 flux B6225 Unm base B6226 Unm high B6227 air-blown B6229 SBS L G B6230 SBS L B6231 SBS R G B6232 EVA B6233 EVA G B6243 ESI B6251 CMCRA MRL Physical Hardening The phenomenon called physical hardening is very important issue related to low temperature testing. The Physical hardening is similar to physical aging for many amorphous solids. This is caused by the gradual density change that occurs over time when bitumen s are held at low temperatures. The mechanical stiffness of the asphalt increases markedly in response to this decrease in volume. It is a reversible process that occurs to asphalt binders below room temperatures [69]. Polymers experience an increase in stiffness when stored isothermally below room temperature. This is primarily because temperature drops too fast for the molecules to readjust and for the volume to decrease, resulting in increasing stiffness with increasing conditioning times [70]. The

66 66 experimental measurements prove that the physical hardening of the paving grade asphalt is no different than physical aging of other amorphous solids. The creep compliance is affected by the isothermal storage temperature, isothermal storage time, and asphalt source. The lower the temperature higher the hardening level and rate. The hardening rate is observed to be very rapid initially. Most polymers experience physical hardening only below their glass transition temperatures. Exceptions to this general behavior include asphalt binders, which experience physical hardening even above their glass transition temperature T g [7]. The effect of physical hardening on the creep compliance can be defined by the single parameter called the hardening shift factor, a ti, in which the shape of the relaxation spectra is unchanged by physical hardening as shown in Figure 2.10.

67 67 Figure 2.10: Superposition of creep curves of asphalt AAM-1 measured at isothermal ages of 2 hour and 16 days [74]. Similar to that of the thermo rheologically simple materials, asphalts are believed to change their properties with temperature mainly due to the changes in free volume which decreases when the temperature is decreased [72,73]. The decrease in the free volume results in more closely packed molecular arrangement and reduced molecular stability. During the physical hardening, the time dependent collapse of free volume is hypothesized to result in volumetric creep that continuously increases the degree of packing, thus producing hardening. This hypothesis of free volume was tested by the

68 68 isothermal volume measurements. A specially designed dilatometer equipped with precise capillary tubes was used to measure the volume changes for a period of 24 hours [69]. The dilatometers were kept in a liquid bath controlled to within ± 0.1 C and volume changes were measured to within ml 9the size of the sample was not mentioned in the research paper). Figure 2.11 is an example of the measurements for three of the asphalts at the temperature of -15 C. Curves shown in the figure represent best fit curves for three independent replicates of asphalt. Using the isothermal-isobaric volume measurements, the reduction in volume relative to an initial volume is collected to compare with the hardening shift factors, a ti.. Isothermal ages of 2, 6, and 24 hours were used and the correlations were very high (R 2 =92%) as shown in Figure 2.12 for all asphalts. Each asphalt shows its own relation which, if the free volume hypothesis is true, should have a slope that is equivalent to the thermal coefficient of contraction at sub-t g temperatures (α g ). The slopes have an average slope of 0.45 log(s)/mm 2 /g, which can be converted to a α g (4.03*10-4 / C) using the average temperature shift function of log(s)/ C calculated for these asphalts [69]. The value of α g measured in this study is within the range of values reported by others [72].

69 69 Figure 2.11: Isothermal volume measurement for three asphalts at -15 C over a period of 24 hours [69].

70 70 Figure 2.12: Correlation of volume change and hardening shift at equi-isothernal ages [69] Glass Transition (T g ) and the Physical Hardening Physical aging of amorphous solids is known to be predominant below glass transition (T g ) [69,72]. In fact, the results of isothermal volume measurements suggest that physical hardening is essentially a continuation of the glass transition phenomenon. When the asphalt is cooled from high temperature its volume shrinks, which is mainly a reduction in free volume. When the glass transition region is approached, the transport mobility of molecules is reduced and, at some point, results in non-equilibrium volumes.

71 71 At this point the asphalt is in a meta-stable state causing the material to continuously shrink isothermally. Asphalts, however, have a wide glass transition region that reflects the complexity of their composition and a multiplicity of transition temperatures. To investigate the relation of the glass transition phenomenon to physical hardening, dilatometer glass transition measurements were made for the asphalts using the same dilatometers used in the isothermal measurements but equipped with larger capillary tubes and a well-controlled ramping thermal bath [69]. At the cooling rate of 1 C/min, the volume change in a 10 ml specimen was measured in a cooling and heating mode to an accuracy of ml over a temperature range of -60 C to 40 C. The measurement clearly reflected a wide transition that for some asphalts extend to temperatures well above 0 C. Following the concept of free volume, the deviation of measured volume from the hypothetical thermodynamic equilibrium was used as an indicator of free volume that needs to be recovered (or collapsed) during isothermal storage (1). Figure 2.13 depicts the relation between isothermal hardening shift factors (a ti ) and the estimated deviation from thermodynamic equilibrium volume line. The correlation shown (R 2 = 85%) reinforces the finding that free volume entrapped at the onset of the glass transition region is the cause of the meta-stable state that leads to the time dependent physical hardening. Differential scanning calorimetry has been used by several asphalt researchers to determine glass and melting transition. In the early 1970 s Noel and Corbett stated that the traditional concept of asphalt wax content is of questionable significance [75]. They concluded that asphalts contain some kind of waxes that are neither completely

72 72 crystalline nor completely amorphous. They called the material as crystallized fractions and offered a method for calculating them from DSC thermograms. Recently, this concept was used by other researchers in the US and in Europe and relations between the crystallized fractions and physical properties were reported [76]. They also suggested that the asphalts harden faster when they contain more CF and when their T g is lower i.e. there is more molecular mobility at the conditioning temperature because the solvent phase is further from its glassy state. Figure 2.13: Correlation between hardening shift factors and the estimated deviation from equilibrium volume line [69].

73 73 As part of another SHRP project, DSC measurements were conducted on eight of the asphalts used in this study [76]. Very distinct endothermic peaks were observed for several of the asphalts in the temperature region of 0.0 C and 90 C, and the enthalpies of these peaks were reported to an accuracy of ± 10%. Asphalts showing the most hardening and the most isothermal volume change also showed the largest endothermic peaks, Figure Although the temperatures at which these endothermic peaks did not correlate well with the melting point temperatures of the waxes extracted from the corresponding asphalt, the peaks were within the same temperature range as the melting points, 30 C to 90 C. In contrast, physical hardening was observed to be 30 C below the melting point temperatures and became more pronounced as the isothermal temperature decreased [74]. Further, the physical hardening is completely destroyed by heating the asphalt.

74 74 Figure 2.14: Correlation between total endothermic enthalpy and hardening potential for 8 core SHRP asphalts [69]. The most recent study about the physical hardening was done by Basu et al. [6]. The stiffness of the asphalt binder was calculated after conditioning the asphalt binder (conditioning times varied). There was an increase in the stiffness values as the conditioning time increased. Basu et al, [6] have investigated the validity of the equivalence of stiffness and m-value after 2 hours and 60 seconds at temperatures 10 C apart for a number of modified and plain asphalt binders. The effects of physical hardening on the time temperature equivalence factors were also analyzed. The

75 75 assumption they made was that the stiffness after 60 seconds loading at T 1 is approximately equal to the stiffness after 2 hours loading at T 1-10 C. Testing was conducted using BBR. Hesp et al [4] investigated the effect, if any, storage time and loading time have on the BBR and DTT limiting temperatures. For this, samples were tested in the BBR for two hours after both one-hour and three-day conditioning at the grading temperatures. Properties determined included creep stiffness (S) and m-value. Samples were also tested in the DTT to determine their failure strain (ε f ) and failure stress in tension (σ f ) also after one-hour and three-day conditioning and at a single strain rate (1 mm/min) as specified by the AASHTO MP1a protocol. The results obtained from the research conducted by Hesp et al. [4] are shown in Tables 2.3 and 2.4. The results in this study show there are significant effects due to physical hardening. For section 118-1, for instance, the limiting m-value grade loses almost 10 C after three days of conditioning, as shown in Figure An inspection of the data in Figure 2.15 shows that for these binders the MP1a critical temperatures are not very different from the AASHTO M320 temperatures. The biggest difference for the one hour conditioning time is 6 C for section However, on average, after one hour conditioning the MP1a temperatures are 2.4 C lower than the M320 temperatures for this set of five binders. Comparing the three-day data, the conclusions are similar, although for these the average difference is only 0.6 C with MP 1a still providing the lower temperatures. From Figure 2.15 it can be inferred that the cracking temperature changes for 1 hour and three day conditioning. The increase in the temperatures getting warmer for

76 76 sections 118 ranges from 4 C to 14 C. In all the test cases, the specimens that were conditioned for three days were expected to crack at warmer temperatures. Table 2.3: Stiffness data for Highway 118 binder [4] S, S, Differences, T1, 60 s T1-10 C, 2hr % Section PG T1, C T1-10 C hour days hour days hour days Table 2.4: m-values for Highway 118 binder [4] T1, 60 s T1-10 Differences, C, 2hr PG % T1, C T1-10 C Section hour days hour days hour days

77 Figure 2.15: Specification temperatures for Highway 118 binders. (S 1 and m 1 are 60 sec limiting stiffness and m-value temperatures -10 after one hour of isothermal conditioning at the grading temperatures whereas S 3 and m 3 are the same after three days of conditioning. Similarly, MP 1a and strain at fracture specification temperatures are also given for one-hour and three-day conditioning, respectively. Numbers on columns are rounded to the nearest degree.) [4] 77

78 78 However, according to Shenoy [5], the physical hardening phenomenon observed in asphalt binders, held under isothermal conditions at low temperatures is found to be absent when the asphalt binder exists in combination with the aggregates in the aggregate-asphalt mixes. The physical hardening phenomenon is absent when asphalt binder is present as a film in an aggregate mix. The asphalt binders which were proven in SHRP studies to physically harden were tested following the new Superpave Direct Tension Test method. Asphalt binders are held under restraint and experiments are conducted at -15 C, change in the failure strength is not identified because, stress relaxation overcomes isothermal physical hardening in asphalt binders held under restraint. It is shown by inspection of the formulated equations that stress relaxation can, indeed perturb and prevent the physical hardening of asphalt binders that are held under restraint, as is the case when the binders are present in the aggregate-asphalt mix. The stress build-up during the physical hardening can be written as: σ σ t 0 2 t / kb β f 0 = e Equation 2.17 And the equation for the stress relaxation can be written as: σ σ t 0 =e t λ Equation 2.18 Where, σ t = stress at time, t = t

79 79 σ 0 = strain at time, t = 0 λ = stress relaxation time of the binder f 0 = free volume at time, t = 0 k = constant B = constant β = coefficient dependent on temperature (constant in isothermal condition) t = conditioning time It has been observed that the time scale of stress relaxation is less than the time scale of physical hardening. The stress relaxation will occur many times faster than the stress build up due to crystal growth of crystallizable fractions. If it is assumed that the physical hardening is due to crystal growth or network formation, it is obvious that undisturbed nuclei would be needed to promote crystallization or the network formation. The crystallization or network formation will basically be a very slow process at the low isothermal temperature. Thus, it will be difficult to alter the free volume to result in an increase in the stiffness, if the process is perturbed by the stresses in the counteractive direction. This perturbation can occur if the temperature changes even slightly away from the isothermal conditions. Romero et al. [77] conducted research to determine if low temperature physical hardening, which has been reported to be exhibited by asphalt binders, also effects the hot mix asphalts. Two asphalt binders, designation AAM-1 and AAM-2 of the SHRP, which are known to show the effects of physical hardening, were used to prepare asphalt

80 80 mixtures with different mineral fillers. The asphalt mixtures were compacted into slabs from which cores were obtained and tested at low temperatures in the thermal stress restrained specimen test. Before testing, the cores were cooled unrestrained to -15 C and held isothermally for 1 hour and 24 hours. After conditioning, the specimens were restrained and held at a constant length while the temperature was dropped at a rate of 15 C/hr until the specimens fractured. The temperature at which the specimens fractured, the stress at the time of fracture, and the slope of the temperature-stress curve were measured and analyzed using several statistical techniques. The results showed that the mixture made with asphalt binder AAM-2 was affected by conditioning time. The mixture made with asphalt binder AAM-1 was not affected by conditioning time. Other factors, such as mineral fillers and air voids of the mixtures, had a greater influence on the results than did physical hardening.

81 81 CHAPTER 3 OBJECTIVE The purpose of the research is to determine the effect of the physical hardening on the low temperature cracking of the asphalt binder. The main objective of the research is as follows: To determine the effect of physical hardening (conditioning time) on the low temperature cracking of the asphalt binder. To determine the change in the low temperature cracking with the change in the rate of cooling for three different binders (Coating Binders, Flux Binders, PAV 70-28).

82 82 CHAPTER 4 MATERIALS AND PROCEDURES 4.1 Materials The PG binder tested with the ABCD were obtained from Imperial Oil, PG (PAV aged) and roofing binders (coating and flux) were obtained from the Asphalt Institute. The PG was polymer modified with Styrene-Butadiene-Styrene. 4.2 Sample Preparation The binder was heated in tins for 1 hour in order to be sufficiently fluid to pour. The AASHTO binder specifications recommend minimum pouring temperature that produces a consistency equivalent to that of SAE 10W30 motor oil (readily pours but not overly fluid) at room temperature. PAV samples were degassed with a vacuum oven for 15 minutes to remove entrapped air prior to pouring. An aluminum foil spout was wrapped around the tins which were placed in a heated steel cup to slow heat dissipation when pouring several samples as shown in Figure The silicone molds were prepared by applying a release agent of talc and glycerol (with a 1:1 mass ratio) to the specimen forming surfaces. The coating was applied with a small paint brush to produce a thin uniform film to prevent bonding with the mold. The outside and bottom of the rings were also coated and then seated in the mold making sure

83 83 it was level with the top. A strain gauge was positioned next to one of the protrusions in order to accurately measure the fracture strain. Figure 4.1 Pouring sample from steel cup with aluminum foil spout The hot asphalt binder was poured into the mold starting at one spot, letting the binder reach the top and then move around mold in a single pass. It was poured in a continuous stream as quickly as possible to avoid a drop in temperature and from entraining air bubbles or gaps. The mold was slightly overfilled because the binder shrinks upon cooling. After allowing the entire assembly to cool at ambient temperature (25 C or lower) for minutes, the excess binder was trimmed flush with the top of the mold using a heated spatula.

84 84 For the silicone molds, the corners were bent to make sure it was separated from the binder, and then the ring was carefully twisted to separate the ring from the sample to release any bonding or friction. The rings would release easily if they were well coated with the agent and no deformation was caused to the sample. The samples were then ready to be placed in the environmental chamber as shown in Figure 4.2. Figure 4.2 Prepared samples with thermocouples 4.3 Testing Procedure The data acquisition software LabVIEW was opened and the gauges were checked to make sure they were working properly before placing the samples in the environmental chamber. Then the samples were conditioned for 30 minutes; 24 hours; 72

85 85 hours at -20 C. Then cooling was started with a rate of 10 C/hr. Some tests were run at 1 C/hr to see the effects of cooling rate on the cracking temperature. The LabVIEW program was begun to record strain and temperature readings on 10 second intervals. Two thermocouples were used to monitor the air and sample temperatures. The sample temperature was simulated with a gauge that was embedded between a 0.25 inch thick piece of asphalt and inch bar of aluminum. A real-time plot of the strain was shown by LabVIEW and the tests were ended when the samples cracked, producing a sudden jump in strain as in Figure 4.3. The cracking temperature could then be determined by opening the data file in Excel and plotting strain versus sample temperature. The critical temperature is taken directly from the plot at the jump in strain. Tested samples are shown in Figure 4.4. Figure 4.3 LabVIEW real-time strain plot

86 Figure 4.4 Tested samples from silicone molds 86