National Academies Press: OpenBook
« Previous: Chapter 2 - Research Approach
Page 15
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 15
Page 16
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 16
Page 17
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 17
Page 18
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 18
Page 19
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 19
Page 20
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 20
Page 21
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 21
Page 22
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 22
Page 23
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 23
Page 24
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 24
Page 25
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 25
Page 26
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 26
Page 27
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 27
Page 28
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 28
Page 29
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 29
Page 30
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 30
Page 31
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 31
Page 32
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 32
Page 33
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 33
Page 34
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 34
Page 35
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 35
Page 36
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 36
Page 37
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 37
Page 38
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 38
Page 39
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 39
Page 40
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 40
Page 41
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 41
Page 42
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 42
Page 43
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 43
Page 44
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 44
Page 45
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 45
Page 46
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 46
Page 47
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 47
Page 48
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 48
Page 49
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 49
Page 50
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 50
Page 51
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 51
Page 52
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 52
Page 53
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 53
Page 54
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 54
Page 55
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 55
Page 56
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 56
Page 57
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 57
Page 58
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 58
Page 59
Suggested Citation:"Chapter 3 - Findings." National Academies of Sciences, Engineering, and Medicine. 2011. Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders. Washington, DC: The National Academies Press. doi: 10.17226/14613.
×
Page 59

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

15 3.1 Introduction This chapter presents key findings from the five major stud- ies conducted in NCHRP Project 9-36 that were described in Chapter 2: (1) identify candidate methods, (2) selection study, (3) VCS study, (4) SAFT optimization study, and (5) verifica- tion study. Each of these studies is documented in detail in an appendix to this report (see TRB website). Conclusions and proposals based on these findings are presented in Chapter 4. 3.2 Identify Candidate Methods 3.2.1 Post-SHRP Approaches Used for Laboratory Aging of Asphalt Binders The review of the literature and research in progress iden- tified three approaches for laboratory aging of binders that have been pursued by various researchers in the United States and abroad since the completion of the SHRP asphalt research program. The three approaches are • Microwave technology, which increases the energy level of the binder molecules and the bulk temperature of the binder, thereby accelerating the oxidation process; • Various techniques for producing thin films of binder and thereby increasing the availability of oxygen and the oppor- tunity for volatilization; and • Air blowing, which increases the availability of oxygen and the opportunity for volatilization. This section discusses the viability of using each of these approaches in an improved aging procedure considering the state of knowledge of the chemistry of binder aging and the requirements of an ideal aging test developed in this project. 3.2.1.1 Microwave Technology A significant amount of work has been completed by Bishara and his colleagues on developing short- and long-term aging procedures based on microwave technology (9–13). Early work used a standard household microwave oven to age a 10-g sam- ple of binder. Based on the performance grading system high temperature parameter, G*/sin δ, 33 minutes of microwave treatment without pressure was found to be equivalent to Thin Film Oven Test (TFOT) aging, but 63 minutes was required to reproduce RTFOT aging. A total of 158 minutes of microwave treatment without pressure was found to be approximately equivalent to the combination of TFOT plus PAV aging based on the performance grading system intermediate tempera- ture parameter G*sinδ. In later work, a MARS5 scientific microwave produced by CEM Corporation with temperature and pressure control was used for the long-term version of the test. In the latest protocol for long-term aging, 66 g of binder are aged at 135°C in a scientific microwave unit under 3,200 kPa (460 psi) of air pressure for 190 minutes. Comparison of rheo- logical properties between long-term microwave aging and RTFOT plus PAV aging has shown good agreement. For Project 9-36, the principal concern with basing improved binder aging methods on microwave technology is a lack of understanding of the mechanism whereby microwave energy accelerates the aging process. Bishara and McReynolds (9) describe how microwave energy is dissipated as heat in an asphalt sample, but its effect on volatile loss and oxidation reac- tions has not been studied. For long-term aging, high pressures are needed to accelerate the aging process, which is one of the major criticisms of the PAV. Approximate cost of the CEM microwave unit is $20,000. Throughput and the inability to measure volatile loss also are significant concerns for the microwave approach. Because the microwave aging mechanism is not well understood and microwave technology must rely on high pressures to simulate long-term aging, it was not consid- ered a viable approach for further development in Project 9-36. 3.2.1.2 Thin Films Laboratory aging of binders in thin films at elevated tem- perature has been the method of choice of asphalt technologists C H A P T E R 3 Findings

for over 60 years. An asphalt binder’s exposure to air during mixing and in asphalt concrete mixtures is in the form of a thin film. The current RTFOT was developed in California in response to the need for a short-term aging test (14). In an attempt to simulate long-term aging, Griffin and his colleagues (15) performed extensive studies with very thin films and developed a special viscometer to characterize the viscosity of the binder (16–17). More recently, Petersen (18) reported on the Thin Film Aging Test in which thin films of asphalt binder are exposed to the atmosphere at service temperatures. The disadvantage of these tests, which may be considered as ultra- thin film tests, is that they yield a very small quantity of material, making them impractical for specification use. Thin film aging is the approach used in the Thin Film Oven Test, (AASHTO T170), the Rolling Thin Film Oven Test (AASHTO T240), the Pressure Aging Vessel (AASHTO R28), and two European methods (the German Rotating Flask and the Rolling Cylinder Aging Test). Welborn (19) presented an excellent summary of the development of the various U.S. laboratory aging tests based on thin films. The primary difference in all of these methods is the thickness of the film and the method used to obtain the thin film. Methods based on thin films have been used extensively to simulate both short- and long-term aging. Our current knowl- edge of the chemistry of binder aging is founded in the interpre- tation of large amounts of data obtained using this approach (20). Clearly, improved methods based on thin films should be considered as viable approaches for NCHRP Project 9-36. 3.2.1.3 Air Blowing Air blowing was one of the first approaches recommended as a laboratory aging test for asphalt binders. In 1937, Nicholson (21), and Rashig and Doyle (22) proposed short duration, high temperature, high airflow tests for accelerated aging of asphalt binders. In these tests, a 250-g sample of asphalt was aged at 218°C for 15 minutes with an airflow rate of 9 L/min. Later, Skidmore (23) proposed a longer, lower temperature air blowing test that aged a 100-g sample at 177°C for 2 hours using an airflow rate of 1 L/min. Although these tests reason- ably reproduced the ductility and penetration of asphalts recovered from recently constructed pavements, an air blow- ing test for simulating short-term aging was never standard- ized. Little work was done using this approach after 1940 until the development of the Stirred Air Flow Test (SAFT) by Glover et al. (1) in 2001. Laboratory aging using thin films and air blowing are mechanically similar, as long as the air bubbles remain small and well dispersed. Smaller air bubbles increase the ratio of surface area to volume of the asphalt, and the reaction becomes less oxygen-diffusion limited and behaves more like a thin film reaction. The primary concern with air blowing is whether air bubbles can be dispersed in highly viscous modified binders at temperatures considered reasonable for a long-term aging test. There also is concern regarding the partial pressure of the volatiles within the bubbles and whether the thermodynamics of volatile transfer to the atmosphere is duplicated within the closed bubbling system. However, since air blowing produces oxidation reactions that are similar to those of thin films, air blowing was considered a viable approach for Project 9-36. In summary, the review of the literature and research in progress identified significant post-SHRP research on binder aging methods using the three approaches of microwave tech- nology, thin films, and air blowing. Of these three, microwave technology was eliminated because the mechanism whereby microwave energy accelerates the aging process is not well understood, and high pressures are needed for simulation of long-term aging. The use of thin films for laboratory aging is well established and accepted by asphalt technologists. Air blowing is similar to thin film reactions, provided the air bub- bles remain small. Candidate methods for an improved short- term binder aging test based on thin films and air blowing are discussed in the following section. 3.2.2 Methods Based on Viable Approaches 3.2.2.1 German Rotating Flask The German Rotating Flask (GRF) is the common name given to German Standard DIN 52 016, Testing the Thermal Stability of Bitumen in a Rotating Flask. This test was developed in Germany as an inexpensive alternative to the RTFOT. It uses a rotary evaporator similar to that used in ASTM D5404 and AASHTO T319 for recovery of binders after solvent extraction. Figure 2-3 showed a schematic of the GRF. In the test, air is introduced into the rotating flask to age the binder. A 100-g sample is aged in the rotating flask at 165°C for 160 min, 10 min without airflow, followed by an addi- tional 150 min with air at ambient temperature flowing at the rate of 500 mL/min. Three studies to evaluate the GRF as an alternative short- term aging procedure have been conducted in the United States. Sirin et al. (24) and Tia et al. (25) evaluated the test for the Florida Department of Transportation for use with mod- ified asphalts that could not be properly aged in the RTFOT. They concluded that the GRF could be used to simulate short-term aging in hot mix plants and that various degrees of aging could be obtained by varying the temperature, air- flow, duration, and sample weight used in the test. They rec- ommended a cover for the oil bath and the use of a Morton flask to better control temperature and provide uniform mix- ing of modified binders. They also recommended testing con- ditions to approximately reproduce the aging that occurred in the TFOT (AASHTO T170) and the RTFOT. Reported mass changes for the modified test incorporating the cover 16

and Morton flask, and using an airflow rate of 4 L/min, were greater than those for the RTFOT. At the same time as the Florida studies, the Western Research Institute under FHWA’s contract, “Fundamentals of Asphalts and Modified Asphalts,” initiated a study to modify the Ger- man Standard DIN 52 016 to simulate RTFOT aging (2). The study included four phases and ultimately led to the develop- ment of a draft AASHTO standard test method. In each of these phases the properties listed in Table 3-1 were compared for binders short-term aged in various modifications to the GRF and the RTFOT. In the first phase, the method was altered to test 200 grams of binder using a 2-L round-bottom flask at 165°C with a flow rate of 500 mL/min. Eight Materials Reference Library (MRL) asphalts (AAA-1, AAB-1, AAC-1, AAD-1, AAF-1, ABM-1, AAK-1, and AAM-1) were aged in the Modified German Rolling Flask (MGRF) with split sam- ples aged in the RTFOT. Comparisons of rheological proper- ties and mass change revealed that the MGRF consistently aged binders less than the RTFOT. In the second and third phases the airflow was increased to 1 L/min, and 2 L/min, respectively. The results were better, but the degree of aging was still less than in the RTFOT. Based on these results, the round-bottom flask was replaced with a 2-L Morton flask to increase the surface area of the binder during the aging process. All eight MRL asphalts plus three modified binders (Styrelf, Ultrapave, and Novophalt) were aged using the Morton flask at 2 L/min airflow during the fourth phase of testing. The resulting rheological property data are shown in Figure 3-1 for the neat binders. The agreement for the two methods is excellent over a very wide range of temperatures. Mass change also was considered in the study. Figure 3-2 compares mass change data from the MGRF and RTFOT. The MGRF pro- duced less mass loss than the RTFOT, particularly for binders with high mass losses. Table 3-2 summarizes the operating conditions for the MGRF. Ramaiah and D’Angelo have conducted a study for FHWA to further evaluate the MGRF (26). This project included stud- ies to assess the effect of variations in the operating parameters to establish initial tolerances, and a study to compare short- term aging from the MGRF with the RTFOT for five polymer modified binders, one air blown binder, and three neat binders. The tolerances established from this study are included in 17 Property Conditions Value G*/sinδ Short-term aged Temperature at 2.2 kPa G*sin δ PAV aged after short-term aging Temperature at 5,000 kPa S PAV aged after short-term aging Temperature at 200 MPa m-value PAV aged after short-term aging Temperature at 0.300 Mass Loss Short-term aged Report Table 3-1. Properties used in the MGRF optimization. -25 -15 -5 5 15 25 35 45 55 65 75 -25 -15 -5 5 15 25 35 45 55 65 75 GRADING TEMPERATURE FOR RTFOT RESIDUE, °C G RA DI NG T EM PE RA TU RE F O R M G RF R ES ID UE , °° C S = 300 MPa m = 0.300 Equality G*/sinδ = 2.2 kPa G*sinδ = 5000 kPa Figure 3-1. Comparison of continuous grade temperatures from the MGRF and RTFOT for eight MRL asphalts.

Table 3-2. Figures 3-3 and 3-4 present rheological and mass change data for the binders used in the FHWA study. The rheological data in Figure 3-3 shows excellent agreement with the RTFOT similar to that from the Western Research Insti- tute study. The mass change data, on the other hand, is sig- nificantly different than that from the Western Research Institute study. The conclusion from the FHWA study is that the mass changes in the MGRF are greater than those in the RTFOT. This conclusion is in general agreement with that from the Florida studies reported by Sirin et al. (24) and Tia et al. (25). Research was conducted in Germany to develop a long- term aging procedure based on the GRF (27). This test, referred to as the Long-Term Rotating Flask (LTRF), uses the same equipment as the GRF to long-term-age binders in an oxygen atmosphere at lower temperatures. Three steel balls are intro- duced into the round-bottom flask from German Standard DIN 52 016 to promote mixing of binders at the lower aging temperatures. The long-term aging is performed at 95°C or 103°C depending on the grade of the binder. During the aging, oxygen is introduced into the flask at the rate of 7 L/h, and the duration of the test is 47 hours. Table 3-3 summarizes oper- ating conditions for the LTRF. The degree of aging obtained with the LTRF was compared to that obtained with the Rotating Cylinder Aging Test (described in the next section of this report) using nine unmodified and nine modified binders. Sörensen (27) reported only moderate correlation for various properties measured on binders aged in the two devices. Linear correlation coefficients ranged from 0.75 to 0.97. Properties considered included softening point, penetration, penetration index, Frass break- ing point, ductility, complex shear modulus, creep stiffness, and m-value. 3.2.2.2 Rotating Cylinder Aging Test The Rotating Cylinder Aging Test (RCAT) was developed in Belgium to accurately simulate long-term aging (28). The device is shown in Figure 3-5. The main components are an oven for temperature control, and a cylindrical vessel and rotating mechanism for aging the binder. The vessel includes a unique rotating shaft mechanism to keep the binder evenly dispersed in the vessel during the test. Although the device was developed to simulate long-term aging, it can be used for short-term aging using temperature and airflow rates that are the same as those for the RTFOT. 18 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 RTFOT MASS CHANGE, % M G RF M AS S CH AN G E, % AAA-1 AAB-1 AAC-1 AAD-1 AAF-1 ABM-1 AAK-1 AAM-1 Figure 3-2. Comparison of mass change data from the MGRF and RTFOT. Table 3-2. Recommended operating parameters for the MGRF. Parameter Condition (2) Tolerance (26) Sample Size 200 g ± 1 g Temperature 165°C ± 1.5°C Airflow 2 L/min ± 0.04 L/min Rotational Speed 20 rpm ± 5 rpm Heat-Up Time 10 min with no air flow ± 1 min Aging Time 200 min under air flow ± 1 min

19 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 GRADING TEMPERATURE FOR RTFOT RESIDUE, °C G RA DI NG T EM PE RA TU RE F O R M G RF R ES ID UE , °° C m = 0.300 or S = 300 MPa Equality G*/sinδ = 2.2 kPa Figure 3-3. Comparison of continuous grade temperatures from the MGRF and RTFOT from the FHWA study. -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 RTFOT MASS CHANGE, % M G RF M AS S CH AN G E, % ALF Control PG 70-22 PG 58-22 Air Blown PG 70-28 PG 58-22 EVA Grafted SBS Radial Grafted Figure 3-4. Comparison of mass change data from the MGRF and RTFOT from the FHWA study.

Table 3-4 summarizes the operating conditions for the RCAT. Using the commercially produced version of the device, both short- and long-term aging can be performed with the same equipment. First, a 500-g sample of binder is short-term aged. Upon completion of the short-term aging, a portion of the sample is removed for physical property measurements. The remainder of the sample is then long- term aged. A unique concept included in the long-term aging procedure is the removal of samples at various times to allow characterization of aging kinetics. Using the operating condi- tions listed in Table 3-4, the RCAT reasonably reproduces the short-term aging that occurs in the RTFOT and the long-term aging that occurs in the PAV (29, 30). 3.2.2.3 Stirred Air Flow Test Glover, et al. (1) proposed an air blowing test called the Stirred Air Flow Test (SAFT) for short-term aging of asphalt binders. The test was shown schematically in Figure 2-2. In 20 Table 3-4. Operating parameters for the RCAT. Parameter Condition Test Sample Size 500 g Temperature 163°C Speed 1 rpm Airflow Air at 4 L/min Short Term Aging Time 235 min Temperature 90°C Speed 1 rpm Airflow Oxygen at 4.5 L/hr Long Term Aging Time 140 hr OVEN ROTATING VESSEL SHAFT AIR SUPPLY Figure 3-5. Rotating cylinder aging test. Table 3-3. Operating parameters for the LTRF. Parameter Condition Sample Size 100 g Mixing 3, 30-mm diameter steel balls Temperature 95 or 103°C depending on binder grade Rotational Speed 4 rpm Atmosphere Oxygen at 7 L/h Aging Time 47 h

the SAFT, air from a nozzle submerged in the binder is dis- persed in the binder by an impeller mounted to an external motor. Through trial and error, operating parameters for the device were determined that approximate the level of aging that occurs in the RTFOT. This work included consid- eration of the effects of sample size, impeller type, nozzle type, airflow rate, and impeller speed on 60°C viscosity measure- ments and carbonyl growth. Impeller speed had the greatest effect on the viscosity and carbonyl growth measurements. Increasing the speed of the impeller significantly increased aging as measured by viscosity and carbonyl growth. The other operating conditions considered had only a minor effect on the aging process. After appropriate operating parameters were selected, changes in 60°C complex viscos- ity and carbonyl growth were used to compare the device and the RTFOT for several unmodified and modified binders. Figure 3-6 presents a typical comparison reported for the SAFT. The proposed operating parameters are sum- marized in Table 3-5. A draft ASTM standard test method for the SAFT was developed. The SAFT also includes a volatile compound collection system that consists of a simple air-cooled condenser through which the exhaust air from the vessel passes before exiting to the atmosphere. Volatile compounds produced during the short-term aging process become trapped in the condenser. They are removed by washing the condenser with solvent, and then they are weighed after the solvent evaporates. Fig- ure 3-7 presents a comparison of the mass change from the RTFOT and the mass of volatile compounds collected with the SAFT as reported by Glover, et al. (1). As shown in Figure 3-7, the mass of volatile compounds collected with the SAFT is approximately an order of magnitude less than the RTFOT mass loss. It is less sensitive to binder source than in the RTFOT mass change and the order of ranking varies. Glover, et al. (1) also reported high variability for the condensed volatile compounds measurements. Coefficients of variation for multiple tests on the same asphalt were high, averaging approximately 35 percent. Glover, et al. (1) dismissed the poor agreement between the RTFOT mass change and the mass of condensed volatile compounds collected in the SAFT as being the result of errors in the RTFOT mass change meas- urements. They hypothesized that the high variability in the SAFT measurements may be the result of condensation of volatile compounds on the lid caused by temperature gradi- ents over the height of the vessel. They recommended using a heating mantel that heats the device over its entire height in an effort to reduce the potential for volatile compounds to condense on the lid. Since its development, additional evaluation of the SAFT has been completed by the Texas Department of Transporta- tion and FHWA. The Texas effort was directed at an inter- laboratory study to develop precision statistics for the SAFT and to compare them to the RTFOT (31). Six laboratories and six materials were used in the study. Precision statistics 21 Table 3-5. Proposed operating parameters for the SAFT. Parameter Condition Sample Size 250 g Temperature 163°C Airflow 2000 mL/Min Stirrer Speed 700 rpm Heat-Up Time 15 Min under Nitrogen Aging Time 30 Min under Air Figure 3-6. Comparison of viscosity for asphalts aged in the SAFT and RTFOT (1).

were developed for G*/sinδ measured at the specified high- temperature grade of the material tested, and the mass of volatile compounds collected. For the degree of aging as meas- ured by G*/sinδ, the authors concluded that 1. The degree of aging in the SAFT at the current operating conditions is somewhat less than that in the RTFOT. The aging time should be increased from 30 to 35 min to provide better agreement between the two methods. 2. The precision statistics for the SAFT were slightly poorer than those for the RTFOT, but should improve as operators become more familiar with the device. The authors also concluded that the volatile loss procedure in the SAFT worked very well and provided a significant improvement of the RTFOT mass change determinations. This conclusion for the volatile loss procedure is suspect because three of the laboratories could not provide valid volatile loss data, and the corresponding data for the RTFOT were not collected. The FHWA evaluation of the SAFT included a comparison of high- and low-temperature rheological property measure- ments for four binders aged in the SAFT and the RTFOT (26). The aging time in the SAFT test was 35 min as recommended by Glover, et al. (31). Figure 3-8 compares the continuous per- formance grade temperature for binders that are short-term aged in the two tests. As shown, rheological properties of short- term-aged binders from the SAFT show a similar degree of agreement with the RTFOT as short-term-aged binders from the MGRF. 3.2.3 Candidate Short-Term Binder Aging Procedures Based on the review of the literature and research in progress, three candidate procedures were identified for further consid- eration in Project 9-36. They are the (1) MGRF, (2) RCAT, and (3) SAFT. This section discusses the advantages of adapting pos- itive elements of the procedures in a hybrid procedure, and presents the procedures that were recommended for further consideration in NCHRP Project 9-36. 3.2.3.1 Comparison of Candidate Tests Table 3-6 compares pertinent details of the three candi- date tests. This table is arranged to quickly compare the three tests. The upper section presents general information about the test, such as the cost and complexity of the equip- ment, the quantity of material that can be aged, how easy it is to use and clean, and whether NCHRP Project 9-36 could take advantage of further development that was ongoing. The second section presents specific information relative to application of the test to short-term aging, and the last sec- tion presents specific information relative to long-term aging. 22 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 MASS CHANGE IN RTFOT, % BY WEIGHT M A SS O F CO ND EN SE D VO LA TI LE S IN S AF T, % B Y W EI G HT AC-10AAA AAF AAM AAS AAD ABM AAG AC-20(2) AC-20(3) Figure 3-7. Comparison of mass of condensed volatile compounds collected in the SAFT and the mass change from the RTFOT.

23 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 GRADING TEMPERATURE FOR RTFOT RESIDUE, °C G RA DI NG T EM PE RA TU RE F O R SA FT R ES ID UE , °° C m = 0.300 or S = 300 MPa Equality G*/sinδ = 2.2 kPa Figure 3-8. Comparison of continuous performance grade temperatures for binders aged in the SAFT and RTFOT from the FHWA study. Table 3-6. Comparison of candidate methods. Considerations Rotating Cylinder Aging Test Modified German Rotating Flask Stirred Air Flow Test Neat and Modified Binders Yes Yes Yes Amount of Material 500 g 200 g 250 g Equipment Cost $18,000 $3,500 $5,000 estimated Equipment Complexity Moderately complex Simple Moderately complex Availability Low High Low Test Complexity Reasonable Simple Reasonable Binder Recovery Easy Easy Easy Clean Up Solvent Ignition Oven Solvent General Active Development Yes Yes Yes Published Procedure Yes Yes Yes Further Refinement Needed No No Yes Temperature 163°C 165°C 163°C Duration 235 min 210 min 45 min Atmosphere Air at 4 L/min Air at 2 L/min Air at 2 L/min Measure Volatility None—configuration of vessel makes volatile recovery difficult Mass change—adaptable to volatile recovery Volatile recovery system Mimics Aged Binder Chemistry Yes Yes Yes Short- Term Aging Repeatability/Reproducibility Limited data Limited data Some data Adaptable to Long Term Yes—work completed Yes—steel balls to enhance mixing Probably with new impeller Published Procedure Yes Yes No Temperature 90°C 95°C neat 103°C modified Duration 140 hr 47 hr Atmosphere Oxygen at 4.5 L/hr Oxygen at 7 L/hr NA Aging Kinetics Possible Yes Yes Yes Mimics Aged Binder Chemistry Yes Yes Likely Long- Term Aging Repeatability/Reproducibility Limited Limited None

24 As shown, all three tests generally meet the requirements established for an improved short-term binder aging test. Each has been shown to be relatively simple, applicable to both neat and modified binders, and to reasonably reproduce the level of aging that occurs in the RTFOT. Procedures for extend- ing the RCAT and the MGRF to long-term aging at reason- able temperatures and without pressure have been developed, but these will require modification and further development to adapt them to specification testing in the United States. Although a long-term aging procedure has not been devel- oped for the SAFT, it appears that the device is extendable to long-term aging, provided sufficient dispersion of air can be accomplished in stiff modified binders at temperatures around 90°C to 100°C. This may require redesign of the stir- ring mechanism to adequately disperse air in highly viscous media. The flexibility to design this portion of the device is a distinct advantage for extending the test to long-term aging. 3.2.3.2 Promising Tests The RCAT and MGRF are conceptually similar and also very similar to the RTFOT. Both involve exposing a relatively thin film of binder to air or oxygen. The binder film is renewed by rotating the vessel containing the binder. The RCAT vessel and the MGRF both include elements to mix the binder using gravity. The RCAT includes an internal shaft mechanism that keeps the binder evenly dispersed in the vessel. In the MGRF, the indentations in the Morton flask mix the binder during the short-term test. In the long-term test proposed in Germany, steel balls are added to a smooth flask to provide mixing. The MGRF is, by far, the more appealing of these two devices from the perspective of cost, equipment availability, and testing time for long-term aging. The test can be assembled from off- the-shelf components for a reasonable cost. The RCAT is more expensive (approximately $18,000) and available from only a single supplier. Because the long-term RCAT requires 1 week to complete, and the equipment for the MGRF is more readily available at a lower cost, the RCAT was not considered for further development in NCHRP Project 9-36. The SAFT uses a different approach in which air is dispersed in the binder using a stirring mechanism. It is essentially a small, laboratory-scale air blowing still. A thin film is not pro- duced in this device. Instead, air from a nozzle submerged in the binder is dispersed in the binder by an impeller attached to an external motor. It appears that for short-term aging, this approach is much more efficient at mixing air with the binder because the duration of the test is approximately one-half to one-quarter of that for the tests involving thin films with sim- ilar temperatures and airflow rates. It is not known whether the more rapid aging observed in the SAFT is the result of more rapid volatile removal or more rapid oxidation. Of con- cern is the possibility that more vigorous oxidation in the SAFT may be overwhelming volatile loss, especially given the low amounts of volatiles that have been collected by previous researchers using this test. Whether adequate mixing of air and binder will occur at the lower temperatures needed for extend- ing the test to long-term aging is an issue that was addressed experimentally in NCHRP Project 9-36. Another important consideration for an improved short- term aging procedure is the ability to quantify the amount of volatile material lost from the binder. Only the SAFT offers a method for directly measuring the volatile compounds lost during the aging process. It includes a condenser for collect- ing the volatile compounds from the exhaust air from the vessel. The volatile compounds trapped by the condenser can be weighed to determine the amount of material lost and can be analyzed to determine their composition. Although this approach requires additional development to reduce the level of variability reported in the initial development studies, it is better than determining mass change by weighing the filled vessels before and after aging as is done with the RTFOT and MGRF. The change in mass after short-term aging is the net result of a loss in mass due to the loss of volatile compounds and an increase in mass due to oxidation. It appeared that the MGRF could be modified to collect volatile compounds using an approach similar to that used in the SAFT. 3.3 Selection Study 3.3.1 Introduction The primary finding from the review of the literature and research in progress was that the MGRF and SAFT are both promising approaches for an improved short-term aging pro- cedure to be used in the United States with AASHTO M320. The equipment required for both tests is relatively inexpensive, and they are easy to perform, applicable to both neat and mod- ified binders, and reasonably reproduce the level of aging that occurs in the RTFOT. The major issue unresolved through the literature review and review of research in progress was which of these two tests was best suited for extension to long-term aging. Only limited data was available on a long-term version of the MGRF, and no long-term aging data was available for the SAFT. The selection study was designed to investigate the feasi- bility of extending these tests to long-term aging. Since cost, complexity, and ability to simulate the RTFOT were judged to be similar for the two tests, the extendibility to long-term aging became an important factor in selecting the short-term test method for further development in Project 9-36. The selection study was conducted in two parts. The first part of the study was an assessment of various modifications that could easily be made to the MGRF and SAFT to produce prototype long-term versions of these tests. The goal in this effort was to obtain approximately the same degree of aging

that occurs in the PAV. The second part of the selection study was a formal experiment designed to address whether the degree of aging in the prototype long-term versions of the tests is affected by the large differences in viscosities for neat and modified binders at the selected long-term aging temper- ature of 100°C. This section presents key findings from the selection study. The complete selection study report is included as Appendix B. 3.3.2 Development of Initial Prototype Long-Term Aging Tests 3.3.2.1 MGRF Attempts to develop a prototype long-term version of the MGRF that approximates the aging produced in the PAV focused on methods to enhance mixing and to create a film that is continuously renewed within the flask. This was accom- plished by adding various mixers and scrapers and varying the rotational speed of the flask. For all of this testing, a tempera- ture of 100°C, an airflow rate of 36 L/h, and an aging time of 48 hours were used. Table 3-7 summarizes the chronological order of the various configurations that were attempted. Fig- ure 3-9 compares the degree of aging achieved with each con- figuration relative to the aging obtained with the PAV. Schematic diagrams of selected configurations are shown in Figures 3-10 through 3-13. The measure of the degree of aging shown in Figure 3-9 is defined by Equation 1. The relative aging according to this equation is simply the change in vis- cosity above RTFOT aging caused by the prototype long-term test divided by the increase in viscosity that occurs during PAV aging. For all equipment configurations and binders, relative aging is reported based on the dynamic viscosity measured at 60°C and 0.1 rad/s. where RA = relative long-term aging ⏐η∗⏐i = dynamic viscosity for configuration i, measured at 60°C, 0.1 rad/s ⏐η∗⏐RTFOT = dynamic viscosity for RTFOT aged, measured at 60°C, 0.1 rad/s ⏐η∗⏐PAV = dynamic viscosity for PAV aged, measured at 60°C, 0.1 rad/s The investigation of various alternatives for a long-term version of the MGRF procedure started with the MGRF, which uses a 2,000-mL Morton flask. The alternatives that were investigated included a 2,000-mL round flask with the addition of steel balls and rollers to enhance mixing and for- mation of a film and the use of scrapers to create and renew the film. As shown in Figure 3-9, the Morton flask was mar- ginally successful for long-term aging for the PG 58-28 binder, but was not successful for aging the PG 82-22 binder. The use of a round flask with steel balls, as used in Germany, increased the aging of the PG 82-22 slightly while the use of rollers that conformed to the shape of the flask did not. The simple scrap- ers designed to fit in a round flask appear to remove much of the viscosity effect, resulting in similar aging of the PG 58-28 and PG 82-22 binders, but the degree of aging after 48 hours is only one-third of that obtained in the PAV. 3.3.2.2 SAFT Attempts to develop a prototype long-term version of the SAFT that approximates the aging produced in the PAV RA i RTFOT PAV RTFOT = − − ⎛ ⎝⎜ ⎞ ⎠⎟ × η η η η     100 1% ( ) 25 Table 3-7. Summary of long-term rotating flask configurations tested. Number Flask Mixer Speed Figure Observations 1 Morton None 4 rpm Not shown 1. Adequate film for PG 58-22 binder. 2. Does not produce a moving film for PG 82-22 binder. 3. Low relative aging for both binders. 2 Morton 3 Steel Balls 1 rpm Figure 3-10 1. Not used with PG 58-28 binder. 2. Does not produce a moving film for PG 82-22 binder. 3. Low relative aging for PG 82-22 binder 3 Round 1 Football- Shaped Roller 1 rpm Not shown 1. Not used with PG 58-28 binder. 2. Does not produce a moving film for PG 82-22 binder. 3. Low relative aging for PG 82-22 4 Round 2 Football- Shaped Rollers 1 rpm Figure 3-11 1. Not used with PG 58-28 binder. 2. Does not produce a moving film for PG 82-22 binder. 3. Low relative aging for PG 82-22. 5 Round Single Scraper 1 rpm Figure 3-12 1. Does not produce a film for PG 58-28 binder. 2. Generated a renewed film for PG 82-22, but film thickness increased with aging time. 3. Low relative aging for both binders. 6 Round Double Scraper 1 rpm Figure 3-13 1. Not used with PG 58-28 binder. 2. Generated a renewed film for PG 82-22. Film thickness relatively constant with aging time. 3. Low relative aging for PG 82-22 binders.

26 12% 15 % 7% 32% 37% 55% 27% 0% 20% 40% 60% 80% 100% 120% 140% Morton Morton + 3 Balls Round + 2 Rollers Round + Scraper Round + Double Scraper 60 °C DYNAMIC VISCOSITY INCREASE, PERCENT OF PAV PG 58-2 8 PG 82-2 2 More Aged Than PAV Less Aged Than PAV Figure 3-9. Relative aging from Equation 1 for various long-term rotating flask configurations. Figure 3-10. Schematic of 2,000-mL Morton flask with three steel balls. Figure 3-11. Schematic of 2,000-mL round flask with two football-shaped rollers.

focused on the design of an impeller that could efficiently mix air with the binders over a wide range of viscosities. The design proceeded from the impeller used in the short-term version of the test, which is very efficient at mixing air with low viscosity binders, to a helix impeller which is efficient at mixing highly viscous fluids, and finally to a helix/turbine impeller which combines the benefits of both. For all of this testing, a temper- ature of 100°C and an airflow rate of 36 L/h were used. Table 3-8 summarizes the chronological order of the various configurations that were attempted. Figure 3-14 compares the degree of aging achieved with each configuration relative to the aging achieved with the PAV. Schematic diagrams of selected configurations are shown in Figures 3-15 through 3-17. The measure of the degree of aging shown in Figure 3-14 is the same used in Figure 3-9 and defined in Equation 1. The relative aging according to this equation is simply the change in viscos- ity above RTFOT aging caused by the prototype long-term test divided by the increase in viscosity that occurs during PAV aging. The dynamic viscosity was measured at 60°C, 0.1 rad/s. The original impeller worked well with the PG 58-28 binder, but it did not provide adequate mixing of the PG 82-22 binder. When this impeller is used with extremely viscous materials, the entire mass of material spins with the impeller. The helix impeller, which is frequently used to mix very viscous and particulate-filled fluids, worked well with the PG 82-22 binder, but apparently did not disperse air as efficiently in the less viscous PG 58-28 binder. The helix/turbine impeller, which includes a helix to move the binder vertically in the 27 Clamp to hold scraper stationary Figure 3-13. Schematic of 2,000-mL round flask with double scraper. Clamp to hold scraper stationary Figure 3-12. Schematic of 2,000-mL round flask with single scraper.

vessel, and a turbine to mix air with the binder, resulted in the best performance over the range of binders investigated. At 48 hours, the degree of aging obtained in the PG 58-28 and PG 82-22 binders exceeded that obtained in the PAV. The last iteration of the impeller design was a helix/turbine impeller with eight turbine blades. With this impeller, PAV aging conditions were reached after approximately 40 hours. Figure 3-18 shows the degree of aging obtained with this con- figuration for the three binders included in the selection study. The measure of the degree of aging shown in Figure 3-18 is the same used in Figures 3-9 and 3-14 and defined in Equation 1. The relative aging according to this equation is simply the change in viscosity above RTFOT aging caused by the proto- type long-term test divided by the increase in viscosity that occurs during PAV aging. The dynamic viscosity was measured at 60°C, 0.1 rad/s. The degree of aging appears to increase with increasing binder stiffness, which is counterintuitive. The testing described earlier found that it is possible to extend the SAFT to a long-term aging test. The following sec- tion discusses the experiment on viscosity effects that was conducted to quantify the significance of the differences between the aging of the PG 58-28 and the PG 82-22 binder shown in Figure 3-18. 3.3.3 Viscosity Effects Experiment Only the final iteration (a helix/turbine impeller with eight turbine blades) of the long-term version of the SAFT was sub- jected to the experiment on viscosity effects. Table 3-9 sum- marizes the testing conditions for the long-term SAFT. 28 Table 3-8. Summary of long-term SAFT configurations tested. Number Impeller Speed Schematic Observations 1 Original 700 rpm Figure 3-15 1. Good mixing of PG 58-28 binder. 2. Could not stir PG 82-22 binder. 3. High relative aging for PG 58-28. 2 Helix 220 rpm Figure 3-16 1. Good mixing of both PG 58-28 and PG 82-22 binders. 2. Better mixing of PG 82-22 binder occurs at lower speeds. 3. High relative aging of PG 82-22 binder. 4. Moderate relative aging of PG 58-28 binder. 3 Helix/ 4-Bladed Turbine 350 rpm Figure 3-17 1. Good mixing of PG 58-28, PG 82-22, and PG 76-22 binders. 2. Aging at 48 hrs exceeds PAV conditions for the PG 58-28 and PG 82-22. 3. Less difference in aging between PG 58-28 and PG 82-22 than observed with helix. 4 Helix/ 8-Bladed Turbine 350 rpm Not shown 1. Good mixing of PG 58-28, PG 82-22, and PG 76-22 binders. 2. PAV aging obtained at approximately 40 hours. 95% 40% 136% 71% 85% 65% 21% 125% 63% 57% 0% 20% 40% 60% 80% 100% 120% 140% Original 48 hrs Helix 48 hrs Helix 24 hrs Helix /T urbine 48 hrs Helix /T urbine 24 hrs 60 °C DYNAMIC VISCOSITY INCREASE, PERCENT OF PAV PG 76-22 LDP E PG 58-28 Neat PG 82-22 SB S Mo re A ged T han PAV Less Aged T han PAV Figure 3-14. Relative aging from Equation 1 for various long-term SAFT configurations.

The viscosity effects experiment was designed to investi- gate the effect of viscosity on the degree of aging that occurs in the long-term SAFT. This was accomplished by aging split samples of RTFOT-aged PG 58-28 and PG 82-22 binders in the PAV and the long-term SAFT, and comparing rheological properties at high, intermediate, and low pavement tempera- tures. The following properties were measured for the RTFOT, PAV, and long-term SAFT: • Shear modulus and phase angle from a DSR frequency sweep at 60°C using frequencies from 0.1 to 100.0 Hz. • Shear modulus and phase angle from a DSR frequency sweep at 25°C using frequencies from 0.1 to 100.0 Hz. • Creep stiffness and m-value at 60 sec from BBR tests con- ducted at −12°C. Three independent samples were aged in the long-term SAFT and PAV and tested as outlined above. Regression analysis was used to compare the rheological properties between the PAV and the long-term SAFT. Details of the statistical analysis are presented in Appendix B on the project webpage. The sta- tistical analysis found that there was a statistically significant difference in aging between the PAV and the long-term SAFT and that the difference was binder dependent. Table 3-10 summarizes the key differences and compares them to the single operator coefficient of variation for the DSR and BBR 29 600ml/min. Air 700rpm Heating Mantle Probe to Temp. Control Temp. Probe Figure 3-15. Schematic of long-term SAFT with original impeller. 350 rpm 600ml/min. Air Bath Wax Clamp View A 3-D A Figure 3-17. Schematic of long-term SAFT with helix/turbine impeller. 220 rpm 600ml/min. Air Heating Mantle Probe to Temp. Control Figure 3-16. Schematic of long-term SAFT with helix impeller.

tests. The bias of the long-term SAFT relative to the PAV is approximately twice the AASHTO single-operator precision; therefore, these biases have engineering significance as well statistical significance. Probably more important than the finding that the aging was different between the PAV and the long-term SAFT was the fact that the two binders aged differently. The PG 82-22 binder aged more in the long- term SAFT than in the PAV, while the PG 58-28 binder aged more in the PAV than in the long-term SAFT. This differ- ence was most evident in the high pavement temperature tests, but also occurred in the intermediate and low pave- ment temperature tests that are used in AASHTO M320. Differences between the aging produced by the long-term SAFT and the PAV appear to be temperature dependent. The differences were greater at the upper grading tempera- ture than at the lower grading temperature. This implies that the two aging procedures produce materials that are different rheologically. There are two possible explanations for the binder effect. First, the helix/turbine impeller and its rotational speed may not be properly optimized for lower viscosity binders. Second, the air dispersion mechanism in the long-term SAFT may age polymers more than, or in a different way than, the high- pressure aging occurring in the PAV. Additional testing of neat and modified binders, both having a wide range of con- sistency, is needed to determine the cause of this effect and to further improve the long-term SAFT. This additional testing was beyond the scope of NCHRP Project 9-36. 30 1.15 0.80 0% 20% 40% 60% 80% 100% 120% 140% 82-22 SBS 76-22 LDPE 58-28 NEAT 60 °C DYNAMIC VISCOSITY INCREASE, PERCENT OF PAV More Aged Than PAVLess Aged Than PAV 0.93 Figure 3-18. Relative aging from Equation 1 for final iteration of the long-term SAFT (helix/eight-bladed turbine, 200 rpm, 36 L/h airflow, 100°C, 40 hours). Table 3-9. Testing conditions for the long-term SAFT. Condition Value Sample Size 250 g Aging Temperature 100°C Impeller Type Helix + 8-Bladed Turbine Impeller Speed 350 rpm Airflow Rate 36 L/h Aging Time 40 hours Table 3-10. Comparison of long-term SAFT bias with DSR and BBR precision. Long-Term SAFT Bias Relative to PAV, % Property PG 58-28 PG 82-22 AASHTO Single Operator Coefficient of Variation, % G* at 60 °C -10 +14 7.9 G* at 25 °C -6 +13 7.9 S at -12 °C +3 +7 3.2 m at -12 °C -3 +3 1.4

Based on the findings of the selection study, the SAFT was selected for further development in NCHRP Project 9-36. The SAFT was selected because the selection study found that with additional development it may be possible to extend this test to long-term aging. The MGRF, on the other hand, could not be extended to long-term aging because of inadequate mixing of air and binder in this test at the lower aging tem- peratures that should be used to simulate in-service conditions. 3.4 Volatile Collection System Study The objective of the volatile collection system (VCS) study was to design and evaluate an improved VCS for the SAFT. The prototype SAFT included a VCS, which consisted of a copper coil condenser operated at ambient temperature. As shown previously in Figure 3-4, the mass of volatiles collected using this system was a factor of 10 lower than the mass change in the RTFOT and showed little difference between binders. The VCS study, which is described in detail in Appendix C (see the project webpage on the TRB website), included an evaluation of the air-cooled condenser used in the prototype SAFT, the design of an improved VCS employing reusable adsorbents that are commonly used for chromatographic analyses, and evaluation of the improved VCS. This section presents key findings of the VCS study. 3.4.1 Evaluation of Prototype SAFT VCS A mass change experiment was conducted with the SAFT to identify the cause of the unexpectedly low volatile mass collected in the prototype VCS. The neat PG 58-28 binder that was included in the selection study was aged under the following conditions: • Condition 1—RTFOT using standard conditions; • Condition 2—SAFT with the prototype VCS in place, airflow rate 2,000 mL/min; • Condition 3—SAFT without the lid or prototype VCS, airflow rate 2,000 mL/min; and • Condition 4—SAFT without the lid or prototype VCS, airflow rate 4,000 mL/min. The mass of the volatiles was determined for Condition 2 by flushing the volatiles from the VCS with solvent and evapo- rating the solvent. In Conditions 2, 3, and 4 mass change was determined by weighing the vessel and its components before and after conditioning. Table 3-11 summarizes the results of the testing. This experiment provided important insight into the behavior of the prototype VCS. First, the SAFT without the lid and prototype VCS produced mass losses that are less than the RTFOT, but a factor of 10 higher than the mass of the volatiles collected in the prototype VCS. The mass of the volatiles collected in this experiment by the prototype VCS is of the same size as those reported by the original developers of the SAFT, which ranged from 0.013 to 0.051 percent by weight (1). Thus, it appeared that only a small percentage of the volatiles that are produced by the SAFT was being col- lected by the VCS. At a flow rate of 2,000 mL/min, the mass of the volatiles trapped in the collection system was only 0.013 percent by mass compared to a mass loss of 0.11 per- cent when the lid and collection system were removed. Based on this finding, the design of an improved VCS that would capture a greater amount of volatiles was initiated. 3.4.2 Improved VCS Design The design of an improved VCS was an incremental process. In the first iteration, called VCS-I, silica gel and activated car- bon filters were added after the condenser from the prototype VCS to collect moisture and hydrocarbon material passing through the condenser. Silica gel and activated carbon filters also were added before the SAFT vessel to remove moisture and hydrocarbon material from the incoming air. Figure 3-19 is a schematic of VCS-I. This version was used in a study to investigate the effect of pressure inside the SAFT on the amount of volatiles produced. All testing was performed with 31 Table 3-11. Summary of SAFT mass change measurements. Test and Test Parameters Airflow, mL/Min Mass Change, wt %(a) Average Mass Change, wt % Mass of Collected Volatiles, wt % Average Mass of Collected Volatiles, wt % RTFOT 4,000 -0.357 -0.349 -0.35 NA NA SAFT with lid and collection system 2,000 NA NA 0.016 0.011 0.014 SAFT without lid and collection system 2,000 -0.108 -0.112 -0.11 NA NA SAFT without lid and collection system 4,000 -0.184 -0.176 -0.18 NA NA Note: (a) Negative values in this table indicate a loss in mass.

a PG 58-28 binder with an average RTFOT mass change of –0.343 percent. Triplicate 250-g samples were aged for 45 min at 163°C using an airflow rate of 4,000 mL/min. Three differ- ent conditions were used to produce the flow: slight positive pressure, vacuum with 90-kPa absolute pressure, and vacuum with 70-kPa absolute pressure. The subatmospheric pressures (70 kPa and 90 kPa absolute) were produced by applying a vacuum downstream, essentially sucking the air through the SAFT vessel. Figure 3-20 presents the results of this testing. As shown, a significant mass is collected in each component of the VCS. Approximately 73 percent of the total is collected in the hydro- carbon trap, 17 percent in the moisture trap, and 10 percent in the condenser. The amount collected in the hydrocarbon trap exceeds the RTFOT mass change for this binder, which is reasonable considering the mass change measurement includes mass loss due to volatilization and mass gain due to oxidation. VCS-I also was used to collect volatiles from successive runs of the same binder. In this study, dedicated condensers, silica gel filters, and activated hydrocarbon filters were used, and the components of the VCS were not cleaned or purged 32 Silica Gel Moisture Trap Agilent Technologies MT 120-2-S Hydrocarbon Trap Koby Junior Collector 10 ft 1/4 in Copper Tubing Silica Gel Moisture Trap Cole Parmer C-02908-6 2 Hydrocarbon Trap Koby Junior SAFT VESSEL BINDER Figure 3-19. Schematic of VCS-I. 0.00 0.10 0.20 0.30 0.40 0.50 0.60 70 90 101 Absolute Pressure, kPa Qu an tit y C ol le ct ed , w t % Condenser Hydrocarbon Trap Moisture Trap SAFT Mass Loss RTFOT Mass Loss Figure 3-20. Mass of material collected in the improved VCS, VCS-I.

between successive runs. This allowed the volatiles from mul- tiple runs to be combined to collect sufficient materials for analysis. The mass collected in each component for each run is shown in Figure 3-21. The mass collected in the silica gel decreases significantly from the first to the third run. Based on this observation and the observation that the silica gel acquired a brownish color, it appeared that the silica gel adsorbs hydrocarbons as well as water. Further, the results show that the efficiency of the silica gel decreases with succes- sive runs, most likely caused by interference from absorbed polar compounds. These observations prompted the need for a revised VCS design. The second iteration of the improved VCS, called VCS-II, was designed to use filters that are commonly used in chro- matographic analyses. The resulting system is shown in Fig- ure 3-22. Although not shown in Figure 3-22, the inlet silica gel and activated carbon filters from the VCS-I were retained to condition the incoming nitrogen and air. VCS-II used two absorbent polymer resin filters, Tenax TA and HayeSep Q to remove, respectively, the larger molecular weight polar materials (aromatics) and the remaining hydro- carbons, while the molecular sieve was selected to remove water. The activated carbon filter was included only to ensure the efficiency of the resin beads and molecular sieve. Nothing should be collected in the activated carbon filter if the system is properly designed. The system shown in Figure 3-22 was not intended for routine use but instead was selected to provide an understanding of the nature of the volatiles that are being released and to form a basis for selecting a simplified system. Each of the absorbents was contained in a 7-in. (175-mm) long, 0.5-in (12.5-mm) diameter stainless steel tube and were connected with rubber tubing. A photograph of an assembled filter is shown in Figure 3-23. The filter beds can be reused by purging them of collected volatiles by passing high-temperature nitrogen through them. Estimated life of the absorbents, as predicted by the manufacturer’s technical support staff, is in excess of 50 runs. Data on the magnitude and repeatability of the mass changes for each of the collectors were obtained for duplicate runs of binders: Citgo PG 58-28, ABM-2, AAM-1, and AAD-2. The 33 0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 1 2 3 Run Number W ei gh t G ai n, % Collector, wt % Water Filter, wt % Hydrocarbon Filter, wt % Total, wt % Figure 3-21. Average amount of volatiles captured in successive runs, VCS-I. SAFT Air N2 5 4 2 3 60-80 Tenax TA 60-80 Hayesep Q 60-80 Mol Sieve 5A 1 1 2 3 4 5 1. Purge filter system O C C C C 2. Preheat binder C C O C O 3. Age binder C C C O O 4. Flush H2O at end of run O C C C C Process Step Flow Path 100°C 250°C Hydrocarbon Trap* Koby Junior Note: Flow diagram for descriptive purposes only and not to be used to lay out system. Air is pre-dried with a commercial drier and further cleaned by passing the air through a silica gel and charcoal filter before it is introduced into the SAFT. *Hydrocarbon trap may be redundant for this system. Figure 3-22. VCS based on chromatographic filters, VCS-II.

mass of the volatiles collected during replicate runs for the four binders is shown in Table 3-12 for the SAFT with the VCS-II and the RTFOT. Surprisingly, the majority of the volatiles col- lected in the VCS-II were collected in the molecular sieve and not on the Tenax TA and HayeSep Q filter beds. For example, for the 58-28 binder, 80 percent of the volatiles were collected on the molecular sieve while only 20 percent were collected on the Tenax TA and HayeSep Q, with similar results for the other three binders. When the Tenax TA was challenged (purged) at the end of the run, considerable smoking was observed as the volatiles were released. The odor was acrid in nature, unlike the smell of asphalt binder. When the HayeSep Q was purged, no smoke was observed but an asphalt-like odor was observed. The volatiles driven from the molecular sieve have a hydrocarbon odor somewhat similar to asphalt cement. Material collected during the last run for binder AAD-2 was sent to Heritage Research Group to determine the com- position of the volatiles collected in each filter. The procedure used by Heritage was to first sequentially elute each of the three tubes with hexane, methylene chloride, and methanol. Each eluent was analyzed by gas chromatography with flame ionization detection (GC/FID) for quantification of the organic chromatographic material. Use of the three different solvents allowed the removal of compounds of varying polari- ties. Water present in each of the eluents was determined by Karl Fisher testing using a Mettler Toledo DL-38 unit. This analysis showed that essentially all of the water was collected in the molecular sieve. The Tenax TA filter collected 91.2 percent of the organic material, the HayeSep Q filter collected 8.2 per- cent, and the molecular sieve collected the remaining 0.6 per- cent. The mass of organic material collected on each of the filters varied somewhat from the mass changes presented in Table 3-12; however, given the tendency to lose highly volatile components during handling and the nature of the GC mea- surements, the results were considered compatible. The chro- matograms were, as expected, showing compounds similar to those obtained previously in asphalt fume studies conducted by Heritage Research Group. The results of the mass change and chemical analyses indi- cated that further refinements of the VCS were warranted. The compositional analysis of the material collected in each filter indicated that some organic material was collected on the molecular sieve. This indicated that the length of the HayeSep Q was too short. As part of the VCS-III design study, an experiment was conducted to establish the length of HayeSep Q collector required to minimize breakthrough. The length of the HayeSep Q collector was varied from 1.5 in. (37 mm) to 4.5 in. (111) mm by adding one or two 1.5-in. (37-mm) lengths to the original 1.5-in. (37-mm) length and the mass collected in each length was then weighed. This experiment found very little difference in the total amount captured when the HayeSep Q bed varies from 1.5 in. (37 mm) to 4.5 in. (111 mm). A conservative HayeSep Q filter bed length of 3.9 in. (100 mm) was selected for the VCS-III based on these results. Internal discussions as well as input from others outside the study suggested that the Tenax TA filter could be removed from the system. Removing the Tenax TA filter bed would simplify and reduce the cost of the VCS. Replicate SAFT runs were conducted with various lengths of HayeSep Q with and without the Tenax TA. This experiment found very little dif- ference in the weights captured on the HayeSep Q plus Tenax TA filters and on the HayeSep Q filter only. The final configuration for the VCS-III is shown schemat- ically in Figure 3-24. It consists of silica gel and activated carbon filters in the inlet gas stream. The outlet stream passes through a (3.9-in.) 100-mm long HayeSep Q collector to col- lect hydrocarbons and a 5-angstrom molecular sieve to collect water. The HayeSep Q collector and the molecular sieve are challenged prior to testing by passing nitrogen gas at 2 L/min and 250°C inlet temperature for 15 minutes. 3.4.3 Evaluation of Improved VCS The commercial SAFT fitted with the VCS-III was used to condition the 12 binders in the validation study. Three repli- cate samples of each binder were conditioned in the commer- cial SAFT and the RTFOT. The amount of material collected on the HayeSep Q and the molecular sieve for each of the asphalt binders is shown in Table 3-13 and Figure 3-25. Also shown in Table 3-13 is the negative value of the RTFOT mass 34 Table 3-12. Mass change for RTFOT and SAFT with VCS-II. Aging Device Measured Value Filter Media CITGO 58-28 ABM-2 AAM-1 AAD-2 Tenax TA 0.025 0.020 0.007 0.034 HayeSep 0.026 0.012 0.008 0.011 Mol. Sieve 0.204 0.179 0.128 0.166 Tenax + HayeSep 0.051 0.032 0.015 0.045 Mass Change, Percent of Initial Binder Mass Total 0.255 0.211 0.143 0.211 Tenax + HayeSep 20 15 10 21 SAFT Mass Change, Percent of Total Volatiles Collected Mol. Sieve 80 85 90 79 RTFOT Mass Change None 0.345 0.348 -0.122 1.058

change and the combined mass of the material collected on the HayeSep Q and the molecular sieve. The negative value of the RTFOT mass change is used so that a positive value for the absorbents and the RTFOT indicate material lost dur- ing the conditioning procedure. From Figure 3-25 it appears that the mass of hydrocarbon volatiles collected on the HayeSep Q filter is similar for all of the binders tested rang- ing from 0.00 to 0.07 percent of the original mass of binder. This range is similar to the range reported for the SAFT pro- totype VCS. The mass of water collected on the molecular sieve is much lower for the five polymer-modified binders compared to the neat binders and the air blown binder. There is no apparent correlation between the mass lost during RTFOT conditioning and the mass collected on the individual or combined absorbents. This is illustrated graph- ically in Figure 3-26 where the mass collected on the absorbents is plotted versus the mass lost during the RTFOT condition- ing. This is not surprising given that the RTFOT represents the balance between the oxygen consumed by oxidation, loss of hydrocarbon volatiles, and water resulting from oxidation or other chemical reactions. 3.5 SAFT Optimization Study The objective of the SAFT optimization study was to enhance the efficiency of the operating parameters for the commercial SAFT so that it more nearly reproduced the degree of aging that occurs in the RTFOT for neat binders. During the VCS study, it was observed that the commercial version of the SAFT, when operated using the parameters rec- ommended for the prototype, produced significantly less aging than the RTFOT. Recall, the developers of the SAFT documented good agreement between the prototype SAFT and the RTFOT. The likely cause of the lower aging in the commercial SAFT is the improved temperature control pro- vided by this device. In the prototype, temperature control was provided by a heating mantle that was in direct contact with the SAFT vessel. In the commercial SAFT, the vessel is heated in an oven. The process controller limits the maximum temperature in the oven to 176°C. The heating mantle, on the other hand, can reach very high temperatures, and because it is in direct contact with the steel SAFT vessel, the binder at the wall of the vessel can also reach temperatures well above the test temperature of 163°C. More rapid aging of the binder occurs in areas of the SAFT vessel with the highest temperature. The SAFT optimization study included two experiments. The first was an experiment to verify that the heat-up phase of the test does not result in significant aging of the binder. During the heat-up phase, the temperature of the binder is increased from approximately 100°C to the testing tempera- ture of 163°C while nitrogen flows through the SAFT vessel. Since the starting temperature (temperature of the binder after charging the SAFT vessel) cannot be accurately controlled, it is critical that the heat-up phase not contribute significantly 35 Table 3-13. Mass collected on VCS-III filters and RTFOT mass change. Mass Collected on Filters, % of Original Mass Binder HayeSep Q Molecular Sieve Combined HayeSep Q and Molecular Sieve Negative of RTFOT Mass Loss, % of Original Mass AAC-1 0.072 0.141 0.214 0.058 AAD-2 0.041 0.134 0.175 1.058 AAF-1 0.043 0.114 0.157 0.008 ABM-2 0.024 0.113 0.137 0.349 ABL-1 0.049 0.129 0.178 0.654 AAM-1 0.014 0.085 0.100 -0.122 Citgoflex 0.020 0.052 0.072 0.196 ALF 64-40 0.046 0.007 0.053 0.207 Airblown 0.020 0.061 0.081 -0.031 Elvaloy 0.000 0.003 0.003 0.173 EVA 0.020 0.005 0.026 0.132 Novophalt 0.025 0.006 0.031 0.132 Indicating Silica Gel Moisture Trap Bottled Nitrogen SAFT VESSEL BINDER Activated Charcoal Hydrocarbon Trap Kolby Junior Laboratory Air Supelco HaySepQ® Supelco Molecular Sieve 5A Ambient 100 mm Figure 3-24. Final version of VCS-III.

to the degree of aging that occurs in the test. The second experiment was an experiment to determine the effects of impeller speed, airflow rate, and test duration on the degree of aging measured by the rheology of the binder at high in-service pavement temperatures. Operating parameters for the com- mercial SAFT were selected from this experiment. The SAFT optimization study is described in detail in Appen- dix D (see the project webpage on the TRB website). Its key findings are discussed in the following sections. 3.5.1 Heat-Up Effects Experiment During the heat-up portion of the SAFT test, the tempera- ture of the binder is increased from approximately 100°C to the testing temperature of 163°C while nitrogen flows through the SAFT vessel. Since the starting temperature (temperature of the binder after charging the SAFT vessel) cannot be accu- rately controlled, it is critical that the heat-up phase not con- tribute significantly to the degree of aging that occurs in the 36 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 Pe rc en t o f O rig in al M as s, % Hayesep Q Molecular Sieve Figure 3-25. Mass collected on filters. 0.00 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 0.05 0.10 0.15 0.20 0.25 M as s Co lle ct ed o n Ab so rb en ts , % o f O rig in al B in de r M as s Negative of the RTFOT Mass Change, % Hayesep Q Molecular Sieve Combined Hayesep Q & Molecular Sieve Figure 3-26. Material collected on molecular sieve and HayeSep Q versus RTFOT mass change.

test. To assess the change in the properties of the asphalt binder during the heat-up period DSR, measurements at 10 rad/s were made at the high pavement temperature for three binders (PG 58-28, PG 76-22, and PG 82-22). DSR measurements were made for each binder in the tank condi- tion and on material removed from the SAFT after comple- tion of the heat-up phase. Initial testing showed that the SAFT trapped air bubbles in the PG 82-22 binder; therefore, for this experiment and the operating parameters experiment discussed in the next section, both the tank and the materials sampled at the end of the heat up-up period were exposed to the PAV vacuum degassing procedure to remove the trapped air before they were tested in the DSR. The DSR data from this experiment are summarized in Table 3-14. For a given binder, all of the DSR data are within the AASHTO T315 single-operator precision of 9.5 percent for test- ing of original binder. Since the complex modulus before and after the heat-up phase varies by less than the single- operator precision, this experiment found that the binder does not significantly stiffen during the SAFT heat-up phase. 3.5.2 Operating Parameters Experiment The purpose of this experiment was to determine the effect of three operating parameters—impeller speed, airflow rate, and conditioning time—on the properties of asphalt binder aged in the commercial SAFT. Three different unmodified binders, listed in Table 3-15, were chosen for this experiment. The binders, which have been studied extensively elsewhere, represent a range in chemical composition, mass change, and sensitivity to short-term aging. No modified binders were included because the purpose of the experiment was to deter- mine the SAFT operating parameters that best mimic the aging that occurs in the RTFOT for neat binders. The experiment design for the operating parameters experi- ment is presented in Table 3-16. It employed a Plackett-Burman design to simultaneously assess the effects of changes in impeller speed, airflow rate, and conditioning time on the aging that occurs in the commercial SAFT. Plackett-Burman designs are often used in ruggedness testing to assess the effect of changes in multiple test parameters. These are extremely efficient designs that allow the main effects to be determined with a limited amount of testing. For the three variables included in this experiment only four test results are needed to assess the main effects. ASTM E1169-02 (32) presents detailed informa- tion on the design and analysis of the type of experiment used. The rheological data from the operating parameters exper- iment are summarized in Table 3-17. Two statistical analyses were performed on the DSR data shown in Table 3-17. The first analysis was performed to determine if the PAV degassing procedure affects the DSR data obtained after short-term aging of neat binders in the SAFT. Recall that previous work with modified binders during the heat-up effects experiment showed that degassing was a necessary step for modified 37 Table 3-14. High-temperature DSR data from heat-up effects experiment. Tank After Heat Up Binder Temp, °C Rep G*, kPa δ, deg G*/sinδ, kPa Average G*/sinδ, kPa G*, kPa δ, deg G*/sinδ, kPa Average G*/sinδ, kPa 1 1.18 87.1 1.18 1.20 87.0 1.20 PG 58-28 58 2 1.30 86.9 1.30 1.24 1.29 87.0 1.29 1.25 1 1.45 83.6 1.46 1.44 84.1 1.45 PG 76-22 76 2 1.50 83.3 1.51 1.48 1.33 83.9 1.34 1.39 1 1.54 66.3 1.68 1.54 65.9 1.69 PG 82-22 82 2 1.59 66.1 1.74 1.71 1.58 65.6 1.73 1.71 Table 3-15. Binders for the optimization study. Binder/Source Comments PGGrade RTFOT Mass Change, % Aging Index, 135°C Viscosity California Coastal (AAD-2) SHRP core binder except AAD-1 used during SHRP 52-28 -1.058 2.86 West Texas Intermediate (AAM-1) SHRP core asphalt 64-16 +0.122 1.98 California Valley (ABM-2) Replacement for SHRP core binder AAG-2, except AAG-1 used during SHRP 58-16 -0.348 1.62

binders aged in the SAFT. For simplicity, the study team used the vacuum degassing procedure currently specified in AASHTO R28-06, Standard Practice for Accelerated Aging of Asphalt Binder Using a Pressurized Aging Vessel (PAV). This procedure exposes the binder to 170°C for 40±2 minutes. For the last 30±1 minutes, the binder is exposed to vacuum with a residual pressure of 15±2.5 kPa absolute. The effect of the degassing process was evaluated by performing DSR tests on three split samples of the SAFT-aged material. One sample was immediately cooled at ambient temperature, a second sample was exposed to 170°C for 40 minutes, and the third sample was exposed to the PAV degassing procedure. A paired difference analysis was used to assess the effect of degassing. Details of the analysis are presented in Appendix D (see the project webpage on the TRB website). The following three differences were considered: 1. Difference between degassed and air-cooled binder, 2. Difference between air-cooled binder and binder heated per the degassing procedure but without the application of vacuum (oven treatment), and 3. Difference between degassed binder (degassed) and binder heated per the degassing procedure but without the application of vacuum (oven treatment). The results of this analysis are summarized in Table 3-18 and include the mean difference ( – d ), its standard deviation (Sd), the calculated t statistic (T), and its percentage value (p). Low p values (bold and underlined in Table 3-18) indicate significant differences. This analysis found that the degassing procedure significantly affects the rheology of neat SAFT- aged binders, and that it is probably the additional exposure to high temperature that causes this additional aging. Since degassing is needed to remove entrapped air from stiff mod- ified binders, this experiment shows that it must be included as part of the procedure for all binders, and that the degassing must be standardized. The second analysis that was performed investigated the effects of impeller speed, airflow rate, and conditioning time on the degree of aging that occurs in the SAFT to determine the sensitivity of the SAFT aging to these parameters. Tem- perature was not included in the experiment because the 38 Table 3-16. Operating parameters experiment design. Binder Source Impeller Speed (rpm) Air Flow (L/Min) Aging Time (Min) Testing Plan 45 Yes 2 60 -- -- 45 -- -- 700 4 60 Yes 45 -- -- 2 60 Yes 45 Yes AAD-2 1400 4 60 -- -- 45 Yes 2 60 -- -- 45 -- -- 700 4 60 Yes 45 -- -- 2 60 Yes 45 Yes AAM-1 1400 4 60 -- -- 45 Yes 2 60 -- -- 45 -- -- 700 4 60 Yes 45 -- -- 2 60 Yes 45 Yes ABM-2 1400 4 60 -- -- Table 3-17. DSR data from the operating parameters experiment. DSR at 58°C Operating Parameters Degassed Oven Treatment Air Cooled RTFOT, AVG Binder Order Speed Flow Time G*, δ, G*/sinδ, G*, δ, G*/sinδ, G*, δ, G*/sinδ, G*, δ, G*/sinδ, rpm L/Min Min kPa deg kPa kPa Deg kPa kPa Deg kPa kPa deg kPa AAD-2 1 700 4 60 1.72 81.5 1.74 1.69 81.6 1.71 1.68 81.5 1.70 AAD-2 2 1,400 4 45 4.16 74.6 4.31 4.20 74.6 4.36 4.03 75.0 4.17 AAD-2 3 1,400 2 60 4.73 73.2 4.94 4.71 73.4 4.91 4.37 73.6 4.56 AAD-2 4 700 2 45 1.30 83.4 1.31 1.36 83 1.37 1.32 83.0 1.33 2.60 79.0 2.65 AAM-1 1 700 4 60 5.43 82.4 5.48 5.31 82.3 5.36 5.04 82.5 5.08 AAM-1 2 1,400 2 60 8.99 78.7 9.17 9.22 78.6 9.41 8.52 78.7 8.69 AAM-1 3 1,400 4 45 7.37 79.9 7.49 7.57 79.9 7.69 7.03 80.0 7.14 AAM-1 4 700 2 45 4.13 83.8 4.15 4.37 83.5 4.40 3.77 83.2 3.80 5.94 81.9 6.00 ABM-2 1 1,400 4 45 4.39 88.8 4.39 4.27 88.8 4.27 4.44 88.8 4.44 ABM-2 2 1,400 2 60 5.24 88.5 5.24 5.04 88.5 5.04 5.39 88.5 5.39 ABM-2 3 700 2 45 2.45 89.4 2.45 2.50 89.5 2.50 2.54 89.5 2.54 ABM-2 4 700 4 60 2.85 89.4 2.85 2.83 89.3 2.83 2.97 89.3 2.97 3.14 89.5 3.14

review of the literature and research in progress indicated a strong desire to perform the short-term aging test at 163°C, which reasonably simulates hot mix plant operating temper- atures. Figures 3-27 through 3-29 show the effect of the three operational parameters on G*/sinδ measured after SAFT aging and degassing. From these figures it is clear that the aging in the SAFT is affected by impeller speed and conditioning time, but not by airflow rate over the ranges studied. Using the slopes in these figures, the RTFOT G*/sinδ values measured for three binders, and assuming linear relationships, the impeller speed required to reproduce RTFOT aging at the average conditioning time of 52.5 minutes and the conditioning time required to reproduce RTFOT aging at the average impeller speed of 1,050 rpm can be estimated. These are summarized for the three binders in Table 3-19. Based on Table 3-19, it appeared that using an impeller speed of 1,000 rpm, an airflow rate of 2,000 mL/min, and a conditioning time of 45 minutes would result in aging in the commercial SAFT that is approximately equivalent to that which occurs in the RTFOT. The analysis presented in this section assumes linear effects. Developmental work on the VCS-II provided an opportunity to assess the tentative operating parameters. Independent rheological data (average of duplicate runs) collected during the VCS-II testing are presented in Table 3-20. These data show that the operating conditions listed above for the SAFT underestimates the aging that occurs with the RTFOT in three of the four cases. Either the conditioning time or the impeller speed can be increased to increase the degree of aging in the SAFT. The magnitude of these changes was estimated to be a 100-rpm increase in impeller speed or a 5-minute increase in conditioning time to increase G*/sinδ by 7 percent. After careful consideration, the longer conditioning time was selected because of concern that further increases in impeller speed could result in significant quantities of binder splash- ing onto the lid of the SAFT with the potential for asphalt droplets to exit into the VCS. Visual observation from com- pleted tests indicates that at 1,000 rpm no asphalt splashes on the lid, but at 1,400 rpm a significant amount of asphalt splashes onto the lid. Table 3-21 presents rheological data for the Citgo PG 58- 28 binder where the conditioning time was increased from 45 to 50 minutes. As shown, this change increased G*/sinδ by approximately 6.5 percent. With the new operating param- eters, the SAFT ages the Citgo PG 58-28 binder slightly more than the RTFOT, but based on Table 3-20, this increased aging should provide better agreement when a wide range of binders is considered. The final operating parameters selected from the operating parameters experiment for the commercial SAFT for use in the verification study were • 163°C aging temperature, • 2,000 mL/min airflow, 39 Table 3-18. Summary of degassing effects. Paired Differences Parameter Degassed—Air Cooled Oven Treatment— Air Cooled Degassed—Oven Treatment d 0.14 0.17 -0.03 Sd 0.23 0.34 0.14 T 2.11 1.75 -0.64 p 0.03 0.05 0.74 0 1 2 3 4 5 6 7 8 9 500 1,000 1,500 G */s in δ, k Pa Impeller Speed, rpm ABM-2 AAM-1 AAD-2 Figure 3-27. Effect of impeller speed on G*/sin for SAFT-aged material. 0 1 2 3 4 5 6 7 0 2 4 6 G */s in δ, kP a Air Flow, L/min AAD-2 AAM-1 ABM-2 Figure 3-28. Effect of airflow on G*/sin for SAFT-aged material. 0 1 2 3 4 5 6 7 8 40 50 60 70 Duration, min G */s in δ, k PA AAD-2 AAM-1 ABM-2 Figure 3-29. Effect of conditioning time on G*/sin for SAFT-aged material.

• 1,000 rpm impeller speed, • 50-minute aging time, • 250-g sample mass, and • Vacuum degassing per AASHTO R28 after short-term aging in the SAFT. Figure 3-30 illustrates the sequence of operations for the commercial SAFT determined from the SAFT optimization study. This sequence was used in the verification study. 3.6 Verification Study The objectives of the verification study were to (1) confirm that the commercial versions of the SAFT and MGRF repro- duce the degree of aging obtained in the RTFOT for a wide range of neat binders and (2) compare the aging from the SAFT, MGRF, and RTFOT with that from mixture samples aged in a forced-draft oven in accordance with the performance testing protocol contained in AASHTO R30. Initially, the verification study only included the SAFT, but it was expanded at the request of the project panel to include the MGRF. The verification study consisted of two main components— the RTFOT verification experiment and the oven-aged mix- tures experiment. The RTFOT verification experiment, which included DSR and BBR measurements, was designed to pro- vide master curves for the binders in the tank condition and after SAFT, MGRF, and RTFOT conditioning. These binder master curves served the following two purposes: 1. To enable a comparison of the rheological properties of material conditioned in the SAFT, MGRF, and RTFOT, and 2. To allow a comparison of the master curves measured for the binders with master curves back-calculated from mix- ture properties. In the oven-aged mixtures experiment, hot mix asphalt using the binders from the RTFOT verification experiment were prepared and aged in accordance with AASHTO R30. Dynamic modulus master curve tests were performed on the mixture samples. From the mixture modulus master curves, the binder stiffness was estimated using the Hirsch Model (6) and compared to the measured stiffnesses obtained from the SAFT, MGRF and RTFOT. The verification study used six neat MRL binders and six polymer-modified binders in both experiments. A single limestone mixture was used in the oven-aged mixture exper- iment. Detailed information for the binders and the mixture was presented earlier in Chapter 2. The sections that follow present key findings from the veri- fication study. Full details of the study are included in the ver- ification study report in Appendix E (see the project webpage on the TRB website). The verification study was the last study in NCHRP 9-36 and provided the basis for making the final recommendation with respect to a replacement for the RTFOT. 3.6.1 RTFOT Verification Experiment The RTFOT verification experiment included compar- isons of the short-term conditioned specification parameter, G*/sinδ, mass change, master curves, and aging indices for binders conditioned in the RTFOT, SAFT, and MGRF. It was not possible to obtain good quality data for the EVA modified 40 Table 3-19. Estimated SAFT operational parameters to reproduce RTFOT aging. Binder RTFOT G*/sinδ, kPa Estimated Impeller Speed, for 52.5-Min Conditioning, rpm Estimated Conditioning Time for 1,050 rpm Impeller Speed, Min AAD-2 2.65 962 40 AAM-1 6.00 935 47 ABM-2 3.14 857 38 Average 918 42 Table 3-20. Rheological properties from VCS-II development study. Property Aging Condition Citgo 58-28 ABM-2 AAM-1 AAD-2 Original 1.18 1.89 2.95 0.88 After SAFT 2.44 3.17 4.94 2.19 G*, kPa After RTFOT 2.48 3.15 5.41 2.60 Original 87.3 89.7 85.6 85.8 After SAFT 83.7 89.2 82.7 79.7 Phase Angle, deg After RTFOT 84.1 89.5 80.3 79.0 Original 1.18 1.89 2.96 0.88 After SAFT 2.45 3.17 4.98 2.23 G*/sinδ After RTFOT 2.49 3.15 5.49 2.65 Table 3-21. Rheological properties for the Citgo PG 58-28 for SAFT aging times of 45 and 50 minutes compared to RTFOT. Property RTFOT SAFT 45-Min Aging SAFT 50-Min Aging G*, kPa 2.48 2.44 2.65 Phase Angle, deg 84.1 83.7 83.9 G*/sinδ, kPa 2.49 2.45 2.66

binder because the polymer tended to separate during testing; therefore, this binder was not included in the comparisons described below. 3.6.1.1 Specification Parameter G*/sinδ The continuous-grading temperatures for RTFOT-, SAFT-, and MGRF-conditioned binders were calculated by inter- polating between the logarithm of two measurements, one obtained above the specification criterion of 2.20 kPa and the other below the specification criterion. The continuous- grading temperatures are summarized in Table 3-22. This table includes the average and standard deviation from two separate runs for each device. Figure 3-31 compares the average continuous-grading tem- peratures from SAFT and MGRF conditioning to RTFOT conditioning. Figure 3-31 includes trend lines for the SAFT and MGRF data. This plot clearly shows the better agreement for the MGRF compared to the SAFT. The trend line for the MGRF data coincides with the line of equality, while that for the SAFT 41 SAFT oven heated to 176°C Binder Temperature = 163°C Oven maintained at 176°C Process controller adjusts oven temperature as needed to bring binder to 163°C N2 flowing at 2,000 mL/min Vessel placed in oven, Binder Temperature 120°C Air flowing at 2,000 mL/min End of conditioning period Heat-up phase Conditioning period, 50 min Binder Temperature = 160°C < 10 min < 20 min Note: 1,000 rpm impeller speed and 250 g sample Figure 3-30. Sequence of operations used in final SAFT configuration. Table 3-22. Continuous-grade temperatures for RTFOT, SAFT, and MGRF conditioning. Continuous-Grade Temperature, °C RTFOT MGRF SAFT Binder Average Standard Deviation Average Standard Deviation Average Standard Deviation AAC-1 56.3 0.21 58.1 0.35 57.2 0.07 AAD-2 59.7 0.49 59.5 0.14 58.0 0.00 ABM-2 60.7 0.35 61.1 0.21 60.5 0.35 AAF-1 67.0 0.44 68.4 2.90 65.5 0.28 AAM-1 68.0 0.35 68.6 1.27 65.0 0.28 ABL-1 69.5 0.35 68.0 0.21 65.3 0.57 Elvaloy 69.9 1.13 68.9 0.14 61.6 0.35 ALF 75.4 0.00 75.2 0.49 69.2 0.21 Novophalt 78.9 0.57 80.0 0.28 73.9 0.14 Airblown 80.2 0.92 79.9 2.05 76.5 0.57 Citgoflex 85.3 0.21 85.9 0.14 82.2 0.21

indicates less stiffening, particularly for high-stiffness binders. Figure 3-32 shows the difference between the continuous- grading temperatures for the SAFT- and MGRF-conditioned binders compared to the RTFOT-conditioned binders. The difference for the SAFT appears to depend on the stiffness of the binder, and for stiffer binders the difference can reach as much as a full grade. The difference for the MGRF is within ±1.8°C and does not appear to depend on the stiffness of the binder. Paired difference t-testing was used to assess the signifi- cance of the differences shown in Figures 3-31 and 3-32. The analysis was conducted for three groups: (1) neat binders, (2) modified binders, and (3) all binders. The results of this analysis are summarized in Tables 3-23, 3-24, and 3-25 for 42 52 .0 58 .0 64 .0 70 .0 76 .0 82 .0 88 .0 52.0 58.0 64.0 70.0 82.0 88.0 M G R F or S A FT H ig h Te m pe ra tu re G ra de , º C RTFOT High Temperature Grade, ºC MGRF SAFT Line of Equality MGR F SAF T 76.0 Figure 3-31. Comparison of SAFT and MGRF continuous-grading temperatures to RTFOT continuous-grading temperature. -12.0 -9.0 -6.0 -3.0 0.0 3.0 6.0 D iff e re n c e in Hi gh Te m pe ra tu re G ra de (M G RF o r SA FT m in u s R TF O T) , º C RTFOT High Temperature Grade, C MGRF SAFT 52.0 58.0 64.0 70.0 76.0 82.0 88.0 Figure 3-32. Difference in continuous-grading temperature for SAFT and MGRF residue compared to RTFOT residue.

the three groups. Details of the analysis are presented in Appendix E. This analysis shows that the continuous grade is the same for RTFOT and MGRF conditioning. The high- temperature grade is lower for SAFT conditioning. This finding holds for both neat and modified binders. Since the testing included replicate samples of each binder conditioned in the three short-term conditioning devices, an initial evaluation of the variability of the SAFT and MGRF relative to the RTFOT was made. This evalua- tion was done using an F-test on the pooled variance com- puted from the variances of the duplicate tests for the 11 binders. This analysis is summarized in Table 3-26 and shows that the test variability is somewhat greater for the MGRF compared the RTFOT. Variability for the SAFT is similar to the RTFOT. Details of this analysis are presented in Appendix E. The MGRF variability was high for 2 of the 11 binders tested: AAF-1 and Airblown. The variability for the other binders was similar to that from the RTFOT. It is expected that as technicians gain more experience with the MGRF, its variability should become similar to that for the RTFOT. 3.6.1.2 Mass Change The MGRF test includes a mass change measurement that is determined in the same manner as the RTFOT: the change in mass is calculated from the mass of the flask measured before and after aging. As discussed in Section 3.4, the SAFT uses a VCS to collect and then weigh the mass of volatiles exiting the vessel during aging. The performance of the VCS was discussed earlier in Section 3.4.3, so only the compari- son of the mass change in the MGRF and RTFOT will be pre- sented here. Mass change data from the MGRF and RTFOT are sum- marized in Table 3-27 and compared in Figure 3-33. Fig- ure 3-33 shows that there is a good relationship between the mass change measured in the MGRF and that measured in the RTFOT. The MGRF values are approximately 40 percent of the RTFOT values. This is in agreement with the data col- lected in the Western Research Institute study of the MGRF (2) that was discussed earlier in Section 3.2.2.1. 3.6.1.3 Christensen-Anderson Master Curve Parameters Binder master curves were developed from the combined DSR and BBR data using the Christensen-Anderson Model (33). The Christensen-Anderson Model was used because the parameters in the model are useful in interpreting changes in rheology that occur during laboratory conditioning or in- service aging. Equation 2 presents the Christensen-Anderson 43 Table 3-23. Summary of paired t-test for neat binders. Continuous-Grading Temperature, °C Differences (MGRF or SAFT minus RTFOT), °C Binder RTFOT MGRF SAFT MGRF SAFT AAC-1 56.3 58.1 57.2 1.8 0.9 AAD-2 59.7 59.5 58.0 -0.2 -1.7 ABM-2 60.7 61.1 60.5 0.4 -0.2 AAF-1 67.0 68.4 65.5 1.4 -1.5 AAM-1 68.0 68.6 65.0 0.6 -3.0 ABL-1 69.5 68.0 65.3 -1.5 -4.2 Average Difference, °C 0.42 -1.62 Standard Deviation of Differences, °C 1.18 1.84 Calculated t 1.034 -2.151 tcritical (0.05, 5 degrees of freedom) 2.015 2.015 Conclusion No Difference SAFT Lower Table 3-24. Summary of paired t-test for modified binders. Continuous-Grading Temperature, °C Differences (MGRF or SAFT minus RTFOT), °C Binder RTFOT MGRF SAFT MGRF SAFT AAC-1 56.3 58.1 57.2 1.8 0.9 AAD-2 59.7 59.5 58.0 -0.2 -1.7 ABM-2 60.7 61.1 60.5 0.4 -0.2 AAF-1 67.0 68.4 65.5 1.4 -1.5 AAM-1 68.0 68.6 65.0 0.6 -3.0 ABL-1 69.5 68.0 65.3 -1.5 -4.2 Average Difference, °C 0.42 -1.62 Standard Deviation of Differences, °C 1.18 1.84 Calculated t 1.034 -2.151 tcritical (0.05, 5 degrees of freedom) 2.015 2.015 Conclusion No Difference SAFT Lower Table 3-25. Summary of paired t-test for all binders. Continuous-Grading Temperature, °C Differences (MGRF or SAFT minus RTFOT), °C Binder RTFOT MGRF SAFT MGRF SAFT AAC-1 56.3 58.1 57.2 1.8 0.9 AAD-2 59.7 59.5 58.0 -0.2 -1.7 ABM-2 60.7 61.1 60.5 0.4 -0.2 AAF-1 67.0 68.4 65.5 1.4 -1.5 AAM-1 68.0 68.6 65.0 0.6 -3.0 ABL-1 69.5 68.0 65.3 -1.5 -4.2 Elvaloy 69.9 68.9 61.6 -1.0 -8.3 ALF 75.4 75.2 69.2 -0.2 -6.2 Novophalt 78.9 80.0 73.9 1.1 -5.0 Airblown 80.2 79.9 76.5 -0.3 -3.7 Citgoflex 85.3 85.9 82.2 0.6 -3.1 Average Difference, °C 0.27 -3.24 Standard Deviation of Differences, °C 1.00 2.65 Calculated t 0.882 -4.054 tcritical (0.05, 10 degrees of freedom) 1.812 1.812 Conclusion No Difference SAFT Lower

Model for the frequency dependency of the binder shear modulus. where G(ω) = complex shear modulus Gg = glass modulus, approximately equal to 1GPa ωr = reduced frequency at the reference temperature, rad/s ωc = cross over frequency at the reference temperature, rad/s R = rheological index The shift factors relative to the defining temperature are given by Equations 3 and 4 for temperatures above and below the defining temperature, respectively (33). G Gg c r R R  ω ω ω ( ) = + ⎛⎝⎜ ⎞⎠⎟ ⎡ ⎣ ⎢⎢ ⎤ ⎦ ⎥⎥ − 1 2 2 2log log ( ) where a(T) = shift factor T = temperature, °K Td = defining temperature, °K Above Td the Williams-Landel-Ferry (WLF) equation is valid, but below Td it is no longer valid and the Arrhenius equation must be used. This is because the asphalt binder is not in an equilibrium condition as a result of physical hardening. The four unknown parameters: Gg, ωc, R, and Td, were obtained through non-linear least squares fitting of Equations 2, 3, and 4 using the data from the testing program. The parameter, ωc, changes with temperature and therefore is always given at the reference temperature, selected as 22°C for direct comparison with the data back-calculated from the mixture master curves. To construct the complete master curve, the bending beam rheometer creep stiffness was converted to shear modulus using the following approximate interconversions. where G(ω) = shear modulus S(t) = creep stiffness ω( ) ≈ 1 6 t ( ) G S t  ω( ) ≈ ( ) 3 5( ) log . ( )a T T Td ( ) = −⎛⎝⎜ ⎞⎠⎟13016 07 1 1 4 log ( )a T T T T T d d ( ) = − −( ) + − 19 92 3 44 Table 3-26. Analysis of variability of MGRF and SAFT relative to RTFOT. Standard Deviation, °C Variance, (°C)2Binder RTFOT MGRF SAFT RTFOT MGRF SAFT AAC-1 0.21 0.35 0.07 0.043 0.125 0.005 AAD-2 0.49 0.14 0.00 0.243 0.020 0.000 ABM-2 0.35 0.21 0.35 0.120 0.045 0.125 AAF-1 0.44 2.90 0.28 0.190 8.405 0.080 AAM-1 0.35 1.27 0.28 0.123 1.620 0.080 ABL-1 0.35 0.21 0.57 0.120 0.045 0.320 Elvaloy 1.13 0.14 0.35 1.280 0.020 0.125 ALF 0.00 0.49 0.21 0.000 0.245 0.045 Novophalt 0.57 0.28 0.14 0.320 0.080 0.020 Airblown 0.92 2.05 0.57 0.845 4.205 0.320 Citgoflex 0.21 0.14 0.21 0.045 0.020 0.045 Number of Replicates 2 2 2 Pooled Variance 0.3027 1.3482 0.1059 Computed F NA 4.45 2.86 Critical F (0.05, 10, 10) NA 2.98 2.98 Conclusion NA MGRF Higher No Difference Table 3-27. Mass change data for RTFOT and MGRF. Mass Change, % Binder RTFOT MGRF AAC-1 -0.058 -0.232 ABL-1 -0.654 -0.345 AAM-1 0.122 0.113 Citgoflex -0.196 -0.103 Airblown 0.031 0.033 Novophalt -0.132 -0.045 AAF-1 -0.008 -0.063 ABM-2 -0.349 -0.100 AAD-2 -1.058 -0.362 Elvaloy -0.173 -0.060 ALF 64-40 -0.207 -0.073

ω = frequency in rad/s t = time in seconds Figure 3-34 presents an example of the fitted master curve and the nomenclature used with the Christensen-Anderson Model. A major advantage of the Christensen-Anderson Model is that the model parameters have physical significance. The glassy modulus is the limiting maximum modulus and is approximately equal to 1 GPa, reflective of the stiffness of carbon–carbon bonds that predominate in asphalt binders. The viscous asymptote is the 45° line that the master curve approaches at low frequencies and is an indicator of the steady-state viscosity of the binder. The crossover frequency is the frequency where the phase angle is 45 degrees and is typically close to the point where the viscous asymptote inter- sects the glassy modulus. The crossover frequency, ωc, is an indicator of the hardness of the binder. Finally, the rheologi- cal index, R, is the difference between the log of the glassy modulus and the log of the dynamic modulus at the crossover frequency. It is an indicator of the rheological type. Table 3-28 summarizes the Christensen-Anderson Model parameters for tank, RTFOT, SAFT, and MGRF condition- ing. The effect of conditioning on the rheology of the binder is best represented by changes in the model parameters from tank condition to short-term aged condition. Figures 3-35 through 3-37 compare changes for SAFT and MGRF condi- 45 y = 0.3978x R2 = 0.6971 -1.20 -1.00 -0.80 -0.60 -0.40 -0.20 0.00 0.20 -1.20 -0.80 -0.60 -0.40 -0.20 0.00 0.20 M G R F M as s Ch an ge , % RTFOT Mass Change, % Line of Equality -1.00 Figure 3-33. Comparison of mass change for MGRF and RTFOT. 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07 1.0E+08 1.0E+09 1.0E-06 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 1.0E+06 1.0E+08 1.0E+10 Reduced Frequency at 25 °C, rad/sec G *, Pa -28 C BBR -22 C BBR -16 C BBR -10 C BBR 10 C DSR 22 C DSR 34 C DSR 46 C DSR 58 C DSR 70 C DSR FIT Glassy Modulus ≈ 1 GPa Viscous Asymptote Rheological Index, R Cross-Over Frequency, ωc Figure 3-34. Christensen-Anderson Model master curve.

tioning to RTFOT conditioning for R, ωc, and Td, respectively. The general trends shown in these figures are reasonable. The rheological index increases with short-term conditioning, indi- cating that the master curve is becoming flatter. The crossover frequency decreases, indicating that the binder is becoming harder with short-term conditioning. Finally, the defining temperature increases on short-term aging, indicating greater temperature dependency. If the changes in model parameters are the same for SAFT- and MGRF-conditioned binders compared to RTFOT- conditioned binders, the data should plot along the line of equality. As shown, there is significant scatter in the data because the master curve parameters are somewhat inter- related and depend on the quality of the data. Trend lines are shown for the SAFT and the MGRF data in Figures 3-35 and 3-36 to make it easier to interpret these plots. The trend lines for the MGRF data are much closer to the line of equality than those for the SAFT data. The results of regression analyses for the change in the Christensen-Anderson Model parameters are summarized in Table 3-29. Details of this statistical analysis are presented in Appendix E (see the project webpage on the TRB website). 46 Table 3-28. Christensen-Anderson Model parameters. Rheological Parameter, R, log10 Pa log10 Crossover Frequency, , rad/s Defining Temperature, Td, °C Binder Source Log10 Glassy Modulus, Pa Tank SAFT RTFOT MGRF Tank SAFT RTFOT MGRF Tank SAFT RTFOT MGRF AAC-1 8.6 1.21 1.62 1.50 1.55 3.42 2.41 2.68 2.42 -9.8 -0.7 -3.6 -1.5 AAD-2 9.2 1.92 2.17 2.16 2.14 4.11 3.18 3.28 3.31 -18.1 -15.6 -15.1 -14.7 AAF-1 8.6 1.39 1.63 1.65 1.85 2.39 1.56 1.41 1.24 -1.3 3.7 3.8 4.0 AAM-1 8.6 1.49 1.67 1.86 1.91 2.39 1.92 1.41 1.40 -3.1 0.3 2.7 3.7 ABL-1 8.7 1.55 1.65 1.77 1.75 2.97 2.62 2.15 2.27 -14.9 -13.1 -10.7 -11.0 ABM-2 8.7 0.90 0.91 0.90 1.02 2.97 2.74 2.69 2.59 -4.1 -4.1 -3.1 -2.9 Airblown 8.8 2.31 2.42 2.41 2.29 0.88 0.35 0.12 0.38 3.7 5.0 4.2 3.8 ALF 10.9 4.69 5.13 5.36 5.47 2.71 1.48 0.81 0.70 -15.2 -12.6 -12.8 -10.4 Citgoflex 9.8 2.95 3.06 3.33 3.18 1.82 1.35 0.63 0.92 -7.5 -6.2 -2.6 -2.1 Elvaloy 9.2 2.06 2.18 2.44 2.41 3.29 2.77 1.99 2.32 -13.4 -11.1 -9.5 -10.8 Novophalt 8.8 1.60 1.63 1.83 1.87 2.20 1.80 1.14 1.26 -9.7 -9.4 -5.1 -4.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Ch an ge in R fo r M G RF A gi ng Ch an ge in R fo r S AF T Ag in g Change in R for RTFOT Aging SAFT MGRF SAFT MGRF Figure 3-35. Comparison of change in rheological index for RTFOT, SAFT, and MGRF aging.

-2.2 -2.0 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 -2 .2 -2 .0 -1 .8 -1 .6 -1 .4 -1 .2 -1 .0 -0 .8 -0 .6 -0 .4 -0 .2 0. 0 -2.2 -2.0 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 Ch an ge in lo g 1 0 ( c) fo r M GR F A gin g Ch an ge in lo g 1 0 ( c) fo r S AF T A gin g Change in log ( c) for RTFOT Aging SAFT MGRF Figure 3-36. Comparison of change in crossover frequency for RTFOT, SAFT, and MGRF aging. 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 Ch an ge in T d fo r M G RF A gi ng , ° C Ch an ge in T d fo r S AF T Ag in g, °C Change in Td for RTFOT Aging, °C SAFT MGRF MGRF SAFT Figure 3-37. Comparison of change in defining temperature for RTFOT, SAFT, and MGRF aging. Table 3-29. Regression analysis of change in Christensen- Anderson Model parameters. Neat Modified Hypothesis Test of Equality of Average Aging Index for Neat and Modified Binders Method Average StandardDeviation Average Standard Deviation Pooled s t tcritical Conclusion R30 3.30 1.08 3.05 1.32 1.19 0.43 2.26 No difference RTFOT 2.44 0.46 2.26 0.58 0.52 0.55 2.26 No difference MGRF 2.44 0.47 2.30 0.35 0.42 0.54 2.26 No difference SAFT 1.97 0.48 1.34 0.27 0.40 2.60 2.26 Neat > Modified

This analysis shows that there is a significant relationship in the change in the model parameters between the SAFT and RTFOT and the MGRF and RTFOT. The strength of the rela- tionships, as indicated by the R2 values, is better for the MGRF compared to the SAFT. Additionally, the slopes for the MGRF data are close to 1, and the 95 percent confidence intervals for the slope capture 1 for all three parameters. The slopes for the SAFT data range from 0.6 to 0.7, and only the 95 percent con- fidence interval for the slope of the change in Td relationship captures 1. This analysis shows that MGRF aging produces changes in the Christensen-Anderson master curve parameters that are not significantly different from those for the RTFOT, while the SAFT aging produces different changes in the master curve parameters. 3.6.1.4 Aging Indices Another technique for judging the conditioning that occurs in the SAFT and MGRF relative to the RTFOT is to calculate and compare aging indices for the procedures. This was done by calculating aging indices at a common temperature for each aging condition (but different for each binder) and at a series of moduli. This approach par- allels the aging that occurs in an actual pavement—the change that occurs in stiffness at temperatures correspon- ding to different unaged moduli. The temperatures where G* for the unaged binders at 10 rad/s is equal to 1, 10, 100, 1,000, 10,000, and 100,000 kPa were calculated using the fitted Christensen-Anderson Model for each binder. These temperatures are shown in Table 3-30. The fitted Chris- tensen-Anderson Model was then used to calculate the complex moduli for the binders for the three aging condi- tions at these temperatures and 10 rad/s. The moduli for the aged binders were then divided by the moduli for the unaged binders. The resulting aging indices are shown in Table 3-30. Figure 3-38 compares aging indices from the SAFT and MGRF with those from the RTFOT for the binders used in the study. If the aging indices for the SAFT and MGRF are the same as the RTFOT, the data should plot along the line of equality. As shown, there is significant scatter in the data. Trend lines are shown for the SAFT and MGRF data in Fig- ure 3-38. The trend line for the MGRF data falls nearly on the line of equality while the trend line for the SAFT data is much lower. The results of regression analyses for the aging indices are summarized in Table 3-31. Details of this statis- tical analysis are presented in Appendix E. This analysis shows that there are significant relationships between the aging indices from the SAFT and RTFOT, and the MGRF and RTFOT. The strength of the relationship, as indicated by the R2 value, is approximately the same for the MGRF compared to the SAFT. However, the slope for the MGRF data is close to 1, and the 95 percent confidence interval for the slope captures 1. The slope for the SAFT data is 0.74 and the 95 percent confidence interval for the slope does not capture 1. The conclusion from this analysis is that the MGRF produces aging indices that are approximately the same as the RTFOT. Aging indices from the SAFT are less than those from the RTFOT. 3.6.2 Oven-Aged Mixture Experiment In the oven-aged mixture experiment, binder properties back-calculated from mixture dynamic modulus test data were used to assess how well the binder aging procedures simulate the aging that occurs in mixtures during short- term oven conditioning. Dynamic modulus master curve tests were performed on samples prepared from uncondi- tioned mixture and from mixture conditioned for 4 hours at 135°C as specified in AASHTO R30. From the mixture dynamic modulus master curves, the binder shear modulus master curves were estimated using the Hirsh Model (6). The back-calculated binder modulus data were then ana- lyzed to assess how well the binder aging procedures simu- late the aging that occurs in mixtures during short-term oven conditioning. The EVA-modified binder was not included in the comparisons due to the difficulties in testing this binder that were discussed earlier. The sections that follow describe the major findings from the oven-aged mixture experiment. The complete analysis is included in the verification study report in Appendix E (see the project webpage on the TRB website). 3.6.2.1 Back-Calculated Binder Properties Binder properties for the mixtures were obtained by preparing mixtures, developing a dynamic modulus master curve for each mixture, and then back-calculating binder properties from the mixture modulus data. The dynamic modulus master curve testing was performed in accordance with AASHTO TP79, “Determining the Dynamic Modulus and Flow Number for Hot Mix Asphalt (HMA) Using the Asphalt Mixture Performance Tester (AMPT),” and PP61, “Developing Dynamic Modulus Master Curves for Hot Mix Asphalt (HMA) using the Asphalt Mixture Performance Tester (AMPT),” which were developed in NCHRP Project 9-29 (34). The testing was conducted at three temperatures and four frequencies as summarized in Table 3-32. The low and middle temperatures were set at 4°C and 20°C, respec- tively. The upper temperature was selected based on the binder grade: 34°C for the softer binders (AAC-1, AAD-2, ABL-1, ABM-2, ALF, and Elvaloy) and 40°C for AAF-1, 48

AAM-1, Air Blown, Citgoflex, Novophalt, and EVA. This testing protocol produces 10 dynamic modulus and phase angle measurements for each specimen. Three replicate specimens were tested in this project. Dynamic modulus master curves were fitted to the meas- ured data using the procedure in AASHTO PP61. Equations 7 and 8 present the form of the master curve that was fitted to the data. Equation 7 is the form of the dynamic modulus mas- ter curve recommended in the Mechanistic-Empirical Pave- ment Design Guide (MEPDG) for use in pavement structural design (35). This sigmoidal function describes the frequency dependency of the modulus at the reference temperature, 49 Table 3-30. Complex moduli and temperatures used to calculate aging indices. Aging Index Aging Index Binder G*,kPa T, °C SAFT RTFOT MGRF Binder G*,kPa T, °C SAFT RTFOT MGRF 1 56.9 2.35 2.11 2.70 1 77.1 1.86 3.17 2.56 10 43.0 2.56 2.21 2.92 10 59.5 1.76 2.81 2.40 100 30.8 2.59 2.19 2.94 100 44.0 1.64 2.42 2.20 1,000 19.7 2.35 2.00 2.66 1,000 29.8 1.49 2.00 1.95 10,000 8.6 1.86 1.63 2.06 10,000 15.5 1.31 1.59 1.65 AAC-1 100,000 -5.2 1.16 1.16 1.27 Airblown 100,000 -2.5 1.09 1.20 1.30 1 57.1 2.51 3.08 2.87 1 70.1 1.52 2.51 2.58 10 40.6 2.32 2.85 2.73 10 46.6 1.41 2.12 2.25 100 26.2 2.08 2.55 2.51 100 27.2 1.32 1.78 1.96 1,000 13.1 1.80 2.18 2.20 1,000 10.7 1.23 1.49 1.72 10,000 0.5 1.49 1.77 1.82 10,000 -4.0 1.15 1.26 1.51 AAD-2 100,000 -13.2 1.20 1.36 1.42 ALF 64-40 100,000 -17.6 1.03 1.03 1.18 1 64.5 2.29 2.99 3.18 1 87.1 1.05 1.43 1.68 10 50.4 2.32 2.97 2.90 10 64.4 1.04 1.39 1.73 100 38.0 2.24 2.78 2.46 100 45.3 1.03 1.35 1.76 1,000 26.5 2.01 2.39 1.93 1,000 28.6 1.02 1.30 1.76 10,000 14.9 1.65 1.85 1.40 10,000 13.2 1.01 1.24 1.71 AAF-1 100,000 0.3 1.14 1.19 0.92 Citgoflex 100,000 -2.0 1.00 1.17 1.59 1 65.6 1.42 2.29 1.92 1 66.0 1.21 3.10 3.05 10 50.8 1.44 2.23 1.90 10 48.3 1.22 2.78 2.67 100 37.7 1.42 2.05 1.79 100 32.9 1.22 2.40 2.25 1,000 25.6 1.35 1.77 1.58 1,000 19.1 1.20 1.98 1.82 10,000 13.5 1.22 1.41 1.30 10,000 5.7 1.16 1.57 1.43 AAM-1 100,000 -1.9 1.03 0.98 0.92 Elvaloy 100,000 -8.7 1.10 1.22 1.11 1 67.7 1.55 2.71 2.27 1 78.0 1.32 2.23 2.49 10 50.3 1.53 2.63 2.23 10 59.6 1.30 2.20 2.46 100 35.2 1.48 2.43 2.10 100 43.7 1.26 2.09 2.33 1,000 21.4 1.39 2.12 1.88 1,000 29.3 1.22 1.88 2.07 10,000 7.8 1.26 1.70 1.57 10,000 15.2 1.15 1.59 1.71 ABL-1 100,000 -8.5 1.10 1.25 1.21 Novophalt 100,000 -1.1 1.06 1.24 1.29 1 61.5 1.70 1.69 1.96 10 48.3 1.68 1.73 1.96 100 36.8 1.65 1.76 1.89 1,000 26.6 1.58 1.74 1.74 10,000 16.8 1.44 1.64 1.48 ABM-2 100,000 5.4 1.21 1.36 1.14

which for this testing was selected to be 22°C to coincide with that used in the binder testing. The shift factors describe the temperature dependency of the modulus. Equation 8 provides the form of the Arrhenius equation used for the shift factors. where E = dynamic modulus Emax = limiting maximum modulus Emin = limiting minimum modulus ωr = frequency of loading at the reference temperature β, γ = parameters describing the shape of the sigmoidal function where a(T) = shift factor as a function of temperature T = temperature, Kelvin Tr = reference temperature, Kelvin ΔEa = activation energy log ( )a T Tr ( ) = −⎛⎝⎜ ⎞⎠⎟ ΔE 19.14714 T a 1 1 8 log log log log min max min E E E E e    ( ) = + −( ) + +1 β γωr ( )7 The final form of the dynamic modulus master curve equa- tion is obtained by substituting Equation 8 into Equation 7. The limiting maximum modulus of the mixture E*min was estimated using the Hirsch Model (6) and a limiting maxi- mum binder shear modulus of 1 GPa (145,000 psi) (32). Equation 10 presents the Hirsch Model. As shown, the limit- ing maximum modulus of the mixture is a function of VMA and VFA. It ranges from about 3,100 to 3,800 ksi for typical ranges of VMA and VFA. E P VMA G VF mix c binder  = − ⎛⎝⎜ ⎞⎠⎟ +4 200 000 1 100 3, , A xVMA P VMA c 10 000 1 1 100 , ⎛⎝⎜ ⎞⎠⎟⎡⎣⎢ ⎤ ⎦⎥ + − − ⎛⎝⎜ ⎞⎠⎟ 4 200 000 3 10 , , ( ) + ⎡ ⎣ ⎢⎢ ⎤ ⎦ ⎥⎥VMA VFA G binder  log log log log min max min log E E E E e     = + −( ) + + 1 β γ ω+ ⎛⎝⎜ ⎞⎠⎟ −⎛⎝⎜ ⎞⎠⎟ ⎡ ⎣⎢ ⎤ ⎦⎥ ⎧⎨⎩ ⎫⎬ΔEa T Tr19 14714 1 1 . ⎭ ( )9 50 Table 3-31. Regression analysis of aging indices. Measure SAFT MGRF Slope 0.64 1.00 Lower 95% CI 0.70 0.96 Upper 95% CI 0.79 1.03 R2 0.93 0.96 Table 3-32. Temperatures and frequencies used in the dynamic modulus master curve testing. Temperature, °C Frequency, Hz 4.0 10, 1, and 0.1 20.0 10, 1, and 0.1 34.0 or 40.0 10, 1, 0.1, and 0.01 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0. 5 1 .0 1. 5 2 .0 2. 5 3 .0 3. 5 SA FT o r M G RF A gi ng In de x RTFOT Aging Index SAFT MGRF Equality SAFT MGRF Figure 3-38. Comparison of aging indices for RTFOT, SAFT, and MGRF.

where ⏐E⏐mix = mixture dynamic modulus, psi VMA = voids in mineral aggregates, % P VFAx G VMA VFAx G c binder = + ⎛ ⎝⎜ ⎞ ⎠⎟ + 20 3 650 3 0 58   . binder VMA ⎛ ⎝⎜ ⎞ ⎠⎟ 0 58. VFA = voids filled with asphalt, % ⏐G⏐binder = shear complex modulus of binder, psi Table 3-33 summarizes the dynamic modulus master curve parameters obtained by numerical optimization of Equation 9 after substitution of the limiting maximum modulus obtained from Equation 10. Figure 3-39 shows a typical comparison of unconditioned and conditioned mixture master curves and temperature shift factors as well as the measured data for ABL-1. Similar plots for all of the 12 binders are included in Appendix E. The error bars shown are 95 percent confidence 51 Table 3-33. Mixture modulus master curve parameters. Parameter Binder Unconditioned Conditioned Binder Unconditioned Conditioned Air Voids, % 4.6 4.4 4.3 4.0 VMA 16.4 16.0 16.4 15.9 VFA 72.3 72.7 73.6 74.6 E*max, ksi 3342.2 3366.2 3349.1 3381.6 E*min, ksi 7.5 4.1 3.6 2.2 β -0.1910 -0.5708 -1.1127 -1.3366 γ 0.6503 0.4924 0.4485 0.3721 Ea, kJ/mole AAC-1 213344 211294 Airblown 217602 208852 Air Voids, % 4.2 4.2 4.3 4.3 VMA 16.0 15.5 16.4 15.7 VFA 73.7 73.1 73.6 72.7 E*max, ksi 3371.5 3396.0 3363.9 3382.8 E*min, ksi 5.1 5.3 21.7 21.1 β -0.0012 -0.3389 0.5936 0.2370 γ 0.5674 0.4767 0.5903 0.5194 Ea, kJ/mole AAD-2 180577 182085 ALF 169894 173512 Air Voids, % 3.9 4.0 3.7 3.9 VMA 15.6 15.6 15.5 15.5 VFA 75.2 74.2 76.1 76.1 E*max, ksi 3401.2 3396.1 3411.3 3417.9 E*min, ksi 5.3 2.6 18.0 40.0822 β -0.8889 -1.3679 -0.6852 -0.8515 γ 0.5905 0.5199 0.5239 0.6811 Ea, kJ/mole AAF-1 212784 211229 Citgoflex 187462 194955 Air Voids, % 4.4 4.2 4.2 4.0 VMA 16.4 16.1 15.9 15.5 VFA 73.1 74.1 73.6 74.0 E*max, ksi 3346.5 3368.1 3376.5 3400.7 E*min, ksi 3.6 2.4 12.5 7.6 β -0.6677 -0.9797 -0.1281 -0.6076 γ 0.5018 0.4312 0.5684 0.4448 Ea, kJ/mole AAM-1 225645 234911 Elvaloy 188958 193486 Air Voids, % 4.2 4.0 4.0 4.0 VMA 15.8 15.4 15.9 15.6 VFA 73.3 74.2 75.0 74.1 E*max, ksi 3380.4 3407.3 3383.7 3395.6 E*min, ksi 6.2 8.1 29.0 7.1 β -0.4518 -0.6041 -0.7557 -1.1755 γ 0.5407 0.5080 0.5999 0.4138 Ea, kJ/mole ABL-1 187271 188768 EVA 190346 191384 Air Voids, % 4.1 4.2 3.8 3.8 VMA 15.9 15.8 15.3 15.5 VFA 73.9 73.0 75.2 75.4 E*max, ksi 3378.0 3378.8 3385.7 3409.8 E*min, ksi 7.4 8.7 9.0 13.8 β -0.7397 -0.9640 -0.9371 -1.1633 γ 0.8694 0.8167 0.5481 0.5569 Ea, kJ/mole ABM-2 195125 200619 Novophalt 183832 190785

intervals based on the pooled standard deviation of the dynamic modulus data for all binders. As shown in this figure, short-term oven conditioning in accordance with AASHTO R30 results in significant stiffening of the mixture. The mixture dynamic modulus is highly sensitive to the stiffness of the binder in the mixture. The shear modulus of the binder can be estimated from the mixture dynamic mod- ulus using the Hirsch Model (6) previously presented as Equation 10. Knowing the dynamic modulus of the mixture and the VMA and VFA of the test specimens, Equation 10 can be solved for the binder shear modulus. This is best done by trial and error. Figure 3-40 presents an example of the back- calculated binder shear moduli for the testing conditions used in the dynamic modulus testing. Back-calculated binder shear modulus data for all of the binders is included in Appendix E. Figure 3-40 presents data for both the uncondi- tioned mixture and the mixture conditioned for 4 hours at 135°C in accordance with AASHTO R30. The effect of the short-term oven conditioning is clearly evident. It results in significant stiffening, particularly at the higher temperatures. Master curves were developed for the back-calculated binder modulus data by fitting the Christensen-Anderson Model to the data as described previously for the binder test data. The glassy modulus from the binder testing was used in fitting the back- 52 1 10 100 1000 10000 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 Reduced Frequency, Hz E* , k si UNAGED STOA -3 -2 -1 0 1 2 3 0 20 30 40 Temperature, C Lo g Sh ift F ac to r 10 Figure 3-39. Mixture dynamic modulus master curves for ABL-1. 1 10 100 1000 10000 100000 1.0E-03 1.0E-02 1.0E-01 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 Reduced Frequency at 22 °C, rad/s B in de r G *, kP a Mix STOA Mix Unaged Figure 3-40. Back-calculated binder modulus master curves for ABL-1.

calculated master curves. The resulting Christensen-Anderson master curve parameters are summarized in Table 3-34 for the 11 binders included in the binder evaluation. 3.6.2.2 Comparison of Master Curve Parameters The purpose of the oven-aged mixture experiment was to compare the degree of aging from the short-term binder aging procedures with that from mixtures conditioned at 135°C for 4 hours in accordance with AASHTO R30. The first approach for comparing the short-term binder and mixture aging procedures was to compare the Christensen-Anderson master curve parameters obtained from the short-term binder aging procedures with those obtained from the back- calculated binder moduli from the mixture dynamic modu- lus testing. The parameters include the rheological index, R, the crossover frequency, ωc, and the defining temperature, Td. Recall, the physical significance of the Christensen-Anderson master curve parameters are as follows: • The rheological index, R, is the difference between the log of the glassy modulus and the log of the dynamic modu- lus at the crossover frequency. It is an indicator of the rheological type. • The crossover frequency, ωc, is the frequency where the phase angle is close to 45 degrees and is an indicator of the hardness of the binder. • The defining temperature, Td, is an indicator of the glass transition of the binder. Figures 3-41 through 3-43 compare the Christensen- Anderson master curve parameters back-calculated from the unaged mixture dynamic modulus tests with those measured for the tank binder. These figures show that there is reason- able agreement in the master curve parameters for unaged conditions. Table 3-35 summarizes the results of regression analyses for each of the parameters. Details of this statistical analysis are presented in Appendix E (see the project webpage on the TRB website). The 95 percent confidence intervals for the slope of the best-fit regression line for each of the three parameters captures 1, the line of equality, indicating that the back-calculated master curves vary in a similar manner as the tank master curves for the range of binders tested. Figures 3-45 through 3-46 compare changes in the mas- ter curve parameters for AASHTO R30 mixture aging with RTFOT and SAFT aging for the binders. Comparisons were not made for MGRF aging because analysis of the binder aging data showed RTFOT and MGRF aging were essen- tially the same. Figures 3-44 and 3-46 show that there is no relationship for the change in the rheological index and the change in the defining temperature between short-term mixture and binder aging. The trend lines in Figure 3-45 show that there is a weak relationship for the crossover frequency between the short-term mixture and binder aging, with the mixture aging producing somewhat harder binders. 3.6.2.3 Aging Indices The second approach for comparing the short-term mixture and binder aging procedures was to compare aging indices computed from the fitted binder master curves. Aging indices were computed from the back-calculated binder modulus data using the method previously described for the binder testing. Table 3-36 presents aging indices computed in this manner for each of the binders for various unaged binder modulus values ranging from 1 to 100,000 kPa. The 1 kPa values are based on unaged binder modulus values that were below the range measured in the dynamic modulus test and 53 Table 3-34. Christensen-Anderson Model parameters for back-calculated binder moduli. Rheological Parameter, R, Log10 Pa Log10 Crossover Frequency, , rad/s Defining Temperature, Td, °C Binder Source Log10 Glassy Modulus, Pa Unaged R30 Unaged R30 Unaged R30 AAC-1 8.6 1.16 1.85 3.69 2.57 -6.4 -5.5 AAD-2 9.2 1.90 2.47 4.33 2.98 -14.4 -15.6 AAF-1 8.6 1.32 1.46 2.64 1.85 -3.5 -4.2 AAM-1 8.6 1.83 2.07 2.51 1.66 0.4 2.3 ABL-1 8.7 1.72 2.01 2.92 2.07 -13.3 -12.4 ABM-2 8.7 0.59 0.68 3.43 3.09 -9.5 -7.3 Airblown 8.8 2.27 2.94 1.14 -0.57 -2.2 -3.1 ALF 10.9 4.49 5.18 3.18 1.32 -18.4 -15.8 Citgoflex 9.8 3.82 3.33 0.15 0.29 -4.2 2.0 Elvaloy 9.2 2.32 3.03 3.02 1.23 -12.1 -10.4 Novophalt 8.8 1.73 2.10 2.03 0.73 -13.5 -4.5

54 0 1 2 3 4 5 0 1 2 3 4 5 R fo r T an k Bi nd er R for Backcalculated from Unaged Mixture Figure 3-41. Comparison of rheological index from unaged mixture and tank binder tests. 0 1 2 3 4 5 0 1 2 3 4 5 lo g 1 0 c fo r T an k Bi nd er log10 c for Backcalculated from Unaged Mixture Figure 3-42. Comparison of crossover frequency from unaged mixture and tank binder tests.

the 100,000 kPa values were above the range measured in the dynamic modulus test. The remaining values were within the range of the measured data. Figure 3-47 compares aging indices from the mixture testing with those from the RTFOT, SAFT, and MGRF for an unaged binder modulus of 10 kPa, the lowest stiffness included in the mixture testing conditions. Figure 3-47 shows a reasonable correlation between the AASHTO R30 mixture aging and the RTFOT and MGRF. The correlation for the SAFT is much poorer. The regression analysis is summarized in Table 3-37. Details of this statistical analysis are presented in Appendix E (see the project webpage on the TRB website). The slope and intercept from the RTFOT and the MGRF regression models are both significant at the 99 percent level, while those for the SAFT are not significant at the 95 percent level. The slopes of the relationships indicate that the aging index from the short-term binder aging tests is less than that obtained from short-term aging of mixtures in accordance with AASHTO R30. Rankings for the aging indices can be used to compare the short-term binder aging procedures to AASHTO R30. Table 3-38 summarizes the rankings for each test, with the rank of 1 given to the binder with the highest aging index. Table 3-38 also presents the Spearman’s rank correlation coefficient and its significance level (p-value) for the three short-term binder aging procedures. Higher values of the Spearman’s rank correlation coefficient indicate greater sim- ilarity in the rankings. The p-values for the Spearman’s rank correlation coefficient indicate the level of statistical signifi- cance for the rankings. The ranking analysis in Table 3-38 shows that the RTFOT provides the closest rankings compared to AASHTO R30 with a Spearman’s rank correlation coefficient of 0.91 and a significance level exceeding 99.9 percent. The MGRF also provides similar rankings to AASHTO R30 with a Spearman’s rank correlation coefficient of 0.80 and a sig- nificance level of 99.7 percent. The SAFT provides rankings having the greatest difference compared to AASHTO R30. 55 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4 T d fo r Ta n k B in de r Td for Backcalculated from Unaged Mixture Figure 3-43. Comparison of defining temperature from mixture and binder tests. Table 3-35. Regression analysis of unaged mixture and tank, Christensen- Anderson Model parameters. Parameter Measure Value Slope 0.94 Lower 95 % CI 0.85 Upper 95 % CI 1.02 R R2 0.88 Slope 0.95 Lower 95 % CI 0.81 Upper 95 % CI 1.08 c R2 0.86 Slope 0.95 Lower 95 % CI 0.71 Upper 95 % CI 1.19 Td R2 0.78

56 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Ch an ge in R fo r R TF O T A gi ng Ch an ge in R fo r SA FT A gi n g Change in R for R30 Aging SAFT RTFOT Figure 3-44. Comparison of change in rheological index for short-term mixture and binder aging. -2.2 -2.0 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 -2.2 -2.0 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 -2.2 -2.0 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 Ch an ge in lo g 1 0 ( c) fo r R TF O T A gi ng Ch a n ge in lo g 1 0 ( c) fo r SA FT A gi n g Change in log ( c) for R30 Aging SAFT RTFOT RTFOT SAFT Figure 3-45. Comparison of change in crossover frequency for short-term mixture and binder aging.

The Spearman’s rank correlation coefficient for the SAFT rankings is only 0.60 and the correlation is not significant at the 95 percent level. 3.7 Short-Term Aging of Modified Binders One of the purported issues with the RTFOT is that the aging of modified binders is less because the higher viscos- ity modified binders do not maintain a thin film during the test. Table 3-39 summarizes average aging indices for the neat and modified binders for 10 kPa unaged binder stiffness. Data are presented for four aging procedures (AASHTO R30 mixture aging and binder aging using the RTFOT, MGRF, and SAFT). In computing the averages and standard deviations given in Table 3-38, the data for the EVA binder were eliminated due to the testing dif- ficulties caused by phase separation that were discussed previously. The average aging indices for neat and modified binders are compared in Figure 3-48 for the four short-term aging procedures. The error bars shown in Figure 3-48 are 95 per- cent confidence intervals for the average. These comparisons show that there is little difference in the average aging index for the neat and modified binders included in NCHRP Pro- ject 9-36. Table 3-39 also summarizes the results of hypothesis test- ing for the average of the mean 10-kPa aging index between neat and modified binders for the four short-term aging procedures. This testing shows that the average aging index is the same for neat and modified binders for short-term mixture aging in accordance with AASHTO R30 and short- term binder aging in the RTFOT and the MGRF. For the SAFT, the hypothesis testing shows that the average aging index for the neat binder is greater than that for the modi- fied binder. 57 -2 0 2 4 6 8 10 -2 0 2 4 6 8 10 -2 -1 0 1 2 3 4 5 6 7 8 9 10 Ch an ge in T d fo r R TF O T A gi ng , °° C Ch an ge in T d fo r SA FT A gi ng , °° C Change in Td for R30 Aging, °C SAFT RTFOT Figure 3-46. Comparison of change in defining temperature for short-term mixture and binder aging. Table 3-36. Aging indices for AASHTO R30 conditioning based on back-calculated binder modulus values. Aging Indices from AASHTO R30 for Various Unaged Binder Modulus Values Binder 1 kPa 10 kPa 100 kPa 1,000 kPa 10,000 kPa 100,000 kPa AAC-1 4.78 3.50 2.34 1.43 0.84 0.55 AAD-2 6.46 4.53 3.01 1.91 1.19 0.77 AAF-1 5.26 4.49 3.61 2.68 1.83 1.17 AAM-1 2.95 2.62 2.23 1.80 1.40 1.00 ABL-1 3.43 2.89 2.32 1.77 1.30 0.97 ABM-2 1.70 1.79 1.87 1.89 1.73 1.32 Airblown 6.66 4.34 2.71 1.65 1.02 0.72 ALF 3.33 2.74 2.24 1.82 1.48 1.19 Citgoflex 0.85 1.20 1.72 2.40 3.09 2.79 Elvaloy 6.13 4.34 2.94 1.93 1.26 0.86 Novophalt 2.61 2.66 2.58 2.34 1.94 1.43

58 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 RT FO T, S A FT , o r M G RF Ag in g In de x R30 Aging Index SAFT RTFOT MGRF Equality SAFT R2 = 0.26 RTFOT R2 = 0.87 MGRF R2 = 0.67 Figure 3-47. Comparison of 10 kPa aging indices for mixture and binder testing. Table 3-37. Regression analysis of aging indices from AASHTO R30, RTFOT, SAFT, and MGRF. Model Measure Value R2 0.87 Slope 0.41 p-value for slope 0.0002 Intercept 1.04 RTFOT vs R30 p-value for intercept >0.0001 R2 0.26 Slope 0.25 p-value for slope 0.0639 Intercept 0.88 SAFT vs R30 p-value for intercept 0.0563 R2 0.67 Slope 0.30 p-value for slope 0.0012 Intercept 1.42 MGRF vs R30 p-value for intercept 0.0010 Table 3-38. Comparison of ranking of binder aging indices. Rank Binder R30 RTFOT SAFT MGRF AAC-1 5 7 1 1 AAD-2 1 2 3 3 AAF-1 2 1 2 2 AAM-1 9 6 7 10 ABL-1 6 5 6 8 ABM-2 10 10 5 9 Airblown 3 3 4 6 ALF 7 9 8 7 Citgoflex 11 11 11 11 Elvaloy 4 4 10 4 Novophalt 8 8 9 5 0.91 0.60 0.80 Spearman Rank Correlation Coefficient p-value 0.0001 0.0510 0.0031 Table 3-39. Average 10 kPa aging indices for neat and modified binders. Neat Modified Hypothesis Test of Equality of Average Aging Index for Neat and Modified Binders Method Average StandardDeviation Average Standard Deviation Pooled s t tcritical Conclusion R30 3.30 1.08 3.05 1.32 1.19 0.43 2.26 No difference RTFOT 2.44 0.46 2.26 0.58 0.52 0.55 2.26 No difference MGRF 2.44 0.47 2.30 0.35 0.42 0.54 2.26 No difference SAFT 1.97 0.48 1.34 0.27 0.40 2.60 2.26 Neat > Modified

59 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 R30 RTFOT MGRF SAFT A gi ng In de x Aging Procedure Neat Modified Figure 3-48. Comparison of average 10-kPa aging indices for neat and modified binders.

Next: Chapter 4 - Conclusions and Recommendations »
Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Report 709: Investigation of Short-Term Laboratory Aging of Neat and Modified Asphalt Binders provides a proposed method of testing for short-term laboratory aging of neat and modified asphalt binders using the modified German rotating flask as an alternative to the rolling thin film oven test.

The following appendixes A-E to NCHRP Report 709 are only available in electronic format:

Appendix A: Binder Aging Bibliography

Appendix B: Selection Study Report

Appendix C: Volatile Collection System Study Report

Appendix D: SAFT Optimization Study Report

Appendix E: Verification Study Report

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!