Influence of combined aging on the mechanical and physical properties of reclaimed asphalt pavement mixtures blended with rejuvenator agents

Recently, reclaimed asphalt pavement (RAP) with high RAP contents with rejuvenator agents have been implemented in Japan. However, a unified view of the durability of RAP mixtures with rejuvenators has not been established. Therefore, this paper aims to assess the durability of RAP mixtures blended with different types of rejuvenators through combined aging, compared with that of hot mixed asphalt (HMA) mixture. Mechanical and physical properties of RAP mixtures with rejuvenators subjected to a combination of moisture, heat, and oxidation were investigated. Saturation aging tensile stiffness conditioning was used to simulate the combined aging. Four types of RAP mixtures were prepared by employing two types of oil-based rejuvenators with two RAP contents. HMA mixtures were also prepared. The test results demonstrated that the RAP mixtures did not exhibit a reduction in the retained stiffness, whereas the HMA mixtures exhibited a significant drop in stiffness with conditioning. In addition, the RAP mixtures exhibited a longer residual service life than the HMA mixtures after conditioning. Detailed analyses revealed differences in the stiffness and fatigue behaviors of the RAP mixtures with different types of rejuvenators. Finally, certain aspects of the deterioration mechanisms of RAP mixtures with rejuvenators were discussed.


Introduction
In many countries, especially developed countries, assessing the durability of aging infrastructures has become an urgent issue, as the number of aging infrastructures that fail during their service has been increasing [1]. Meanwhile, increases in torrential rainfall and heat waves have become apparent worldwide and might be related to global warming [2]. Therefore, many materials related to aging infrastructures are exposed to more severe environments; accordingly, the possibility of further failures during their service is expected to increase. In particular, the durability of road infrastructure is a concern because they are directly subjected to heavy rainfall and heat waves while withstanding traffic loads. Thus, waste from road sector might be risen through those maintenance cycles.
In the field of road pavements, the use of reclaimed asphalt pavement (RAP) has increased for asphalt pavements owing to the aim of preserving natural resources, resulting in increased demand for RAP mixture production [3]. In Japan, asphalt pavement recycling has been started since 1970s as a countermeasure against the country's oil crisis. Among several types of asphalt recycling technologies, plant recycling has been a major method in Japan [4]. Rejuvenators have been implemented in the plant recycling to recover penetrations of aged binders to target levels [5]. Therefore, the influences of rejuvenators on the physico-chemical properties of recovered binders have been intensively investigated [6][7][8][9]. In addition, mechanical properties and performances of RAP mixtures have been examined, related to rejuvenated agents [10][11][12][13]. As the results, saturate-rich rejuvenators have been widely used in Japan, because they can recover aged binders with smaller dosage amounts compared with aromatic and resin-rich rejuvenators [14,15]. Owing to these technical development efforts, RAP production has increased dramatically since 2000s.
Meanwhile, recently, there has been a trend toward the use of RAP mixtures with high RAP contents blended with rejuvenator agents [16]. In addition, repetitive recycling with high RAP contents has been attracting attention [17]. Recent studies related to repetitive recycling in Japan suggested that saturate-rich rejuvenators might give negative effects on the properties of binders and mixtures, while aromaticrich rejuvenators did not show any significant changes in the properties [15,[17][18][19][20][21]. Not only in Japan but also in other countries, many research studies regarding the effects of rejuvenators have been widely conducted looking at both binder and mixtures [22][23][24][25][26][27][28][29][30][31][32][33]. However, these studies deal with binder and mixtures with high and intermediate temperatures, respectively, and the combination of these factor was not investigated. Therefore, it is especially important to investigate durability for such RAP mixtures in more detail. However, although the assessment and advantages of RAP mixtures have been highlighted in previous studies [34][35][36][37][38][39], recent studies have suggested that there is a contradiction in the durability of RAP mixtures blended with rejuvenators.
For example, Hamburg wheel tracking (HWT) test results in a previous study showed that RAP mixtures with rejuvenators might contribute to permanent deformation and moisture sensitivity [40]. The performance is related to the blended state of the virgin binder and rejuvenators [41]. In contrast, the tensile strength ratio (TSR) conducted to evaluate moisture susceptibility suggested that the use of RAP with rejuvenators might increase the moisture damage of the mixtures with an increase in the RAP contents [42,43]. Thus, the moisture susceptibility of RAP mixtures with rejuvenators remains unclear.
Focusing on the effects of aging on the durability of RAP mixtures, it is observed that the use of rejuvenators may influence the fatigue behavior of aged RAP mixtures. A recent study concluded that aging has a negative impact on the fatigue performance of RAP mixtures with rejuvenators compared to that of the control mix [44]. Meanwhile, another study indicated that long-term aging did not have a significant effect on the fatigue characteristics of RAP mixtures with high RAP contents with and without rejuvenators [41]. Therefore, a universal agreement has not been established regarding the effect of aging on the fatigue behavior of RAP mixtures with rejuvenators. Consequently, further efforts are required to understand how the performance of RAP mixtures with rejuvenators changes when the mixtures are subjected to the effects of moisture, heat, and oxidation. This paper aims to assess the durability of RAP mixtures blended with different types of rejuvenators through combined aging, compared with that of hot mixed asphalt (HMA) mixture. In this study, the mechanical and physical properties of RAP mixtures with 30 and 60% RAP content blended with two types of oil-based rejuvenator agents were experimentally investigated. RAP mixtures with and without the combined effects of moisture, heat, and oxidation, referred to as "combined aging," were investigated. Saturation aging tensile stiffness (SATS) conditioning was used to simulate the combined aging. Note that although SATS conditioning has been intensively conducted on various types of asphalt mixtures in previous studies [45][46][47][48][49][50][51][52][53], the number of applications of RAP mixtures with rejuvenators is limited. Furthermore, the properties of binders extracted from mixtures with and without combined aging were examined in this study. The properties of the binder obtained through a two-stage extraction process were also examined. The details of the experimental program and test results are described in the subsequent sections.

Preparation of mixtures
To prepare RAP mixtures, the optimum ratio of rejuvenator to RAP binder extracted from waste reclaimed asphalt (RA) sources was investigated. Subsequently, RAP mixtures with 30 and 60% RAP contents blended with the rejuvenators were produced. For comparison, HMA mixtures were prepared using a virgin binder and virgin aggregates.

Binder blending design
In this study, virgin and RA binders were prepared to design HMA and RAP mixtures. The virgin binder was 60/80 pen bitumen, whereas the RA binder was extracted from RA sources provided by a plant through a fractioning column with trichloroethylene as the solvent, according to the American Society for Testing and Materials (ASTM) D 1856-95a [54]. The properties of the virgin and RA binders were examined through penetration tests, softening tests, and rotational viscosity tests from 150 °C to 120 °C according to ASTM D5-97, ASTM D 36, and ASTM D 4402-02, respectively [55][56][57]. The test results are listed in Table 1.
As shown in Table 1, the penetration of the RA binder is significantly lower than that of the virgin binder. The table also demonstrates that the softening point of the RA binder is much higher than that of the virgin binder. This trend is consistent with the viscosities observed for the virgin and RA binders. These results indicate that the RA binder was heavily aged under in-service conditions. Therefore, it was necessary to recover the RA binder by dosing it with rejuvenating agents for the production of RAP mixtures. In this study, two types of oil-based rejuvenators (Rejuvenators 1 and 2) were selected. The properties of these two rejuvenators are listed in Table 2. Note that Rejuvenator 1 exhibited a slightly higher density than Rejuvenator 2. However, it can be observed in the table that both rejuvenators exhibit similar viscosity characteristics. In terms of the saturation, aromatic, resin, and asphaltene (SARA) fraction in the rejuvenators, Rejuvenator 1 had high aromaticity, whereas Rejuvenator 2 was rich in saturates. Because the two rejuvenators had different compositions, the SARA fraction of the RA binder also changed after blending. Consequently, a binder blending design was conducted to obtain RAP binders that could meet the grade of 70 penetration, such that the penetration of the RAP binders became close to that of the virgin binder. First, the RA binder was mixed with each rejuvenator at rejuvenator ratios of 0, 15, and 20% (percentage by mass of the RA binder). Subsequently, a linear relationship between the penetration and rejuvenator ratio was obtained for the mixed binders. Based on these relationships, the optimum rejuvenator ratios for achieving a 70 pen grade were determined to be 18.3 and 21.5% for Rejuvenators 1 and 2, respectively. Finally, two types of blended binders were prepared with the optimum rejuvenator ratios of Rejuvenators 1 and 2.
Each blended binder with the optimum rejuvenator ratio was analyzed using a chromatographic method (ASTM-D2007) [58]. The results are shown in Fig. 1, along with those obtained for the rejuvenators. The figure also shows the results obtained for the virgin and RA binders. As can be observed in the figure, the two blended binders, that is, the RA binders blended with Rejuvenators 1 and 2, showed differences, especially with respect to the aromatic and resin components. The blended binder with Rejuvenator 1 showed a higher percentage of resin relative to Rejuvenator 2, whereas the latter binder showed a higher percentage of aromatics. However, both binders showed a higher percentage of aromatics and a lower percentage of resin components than the RA binder. Therefore, it can be said that the RA binder was recovered to some extent by blending the rejuvenators.

Preparation of mixtures
Densely graded asphalt mixtures (AC 13) were prepared using rejuvenators, RAP aggregates, virgin aggregates, and virgin binder to assess the durability of the RAP mixtures. In this study, the mixing ratio of RAP aggregates to the total amount of aggregates was set to 30 or 60%. Four types of RAP mixtures were prepared by employing two types of rejuvenators and two types of RAP contents. Hereafter, the mixtures with 30 and 60% RAP content using Rejuvenator 1 are denoted as 30% RAP1 and 60% RAP1, respectively. Similarly, the mixtures with 30 and 60% RAP contents using Rejuvenator 2 are denoted as 30% RAP2 and 60% RAP2. AC 13 was also prepared using a virgin binder and aggregates to assess the durability of the HMA mixtures. Both HMA and RAP mixtures were designed to meet the standard grading range for AC 13, and the binder content was determined using the Marshall design method. In this study, the target penetration was set to 70 pens, and the optimum rejuvenator amount was determined by changing it to 0, 15, and 20% by weight of the RA binder. The ratio indicated in the 70 pen in the binder was selected from linear regression analysis. To manufacture each type of RAP mixture, the RAP aggregates were first mixed with each rejuvenator considering the optimum rejuvenator ratio. The blended aggregates were then placed in an oven at 130 °C for 3 h. After curing, the aggregates were mixed with a virgin binder and virgin aggregates, which had been heated at 160 °C. As aforementioned, the RAP aggregate content ratio was set to 30 or 60%. After mixing, the mixed aggregates were compacted at 140 °C to obtain the laboratory test specimens. The details of the laboratory tests are presented below.
The densities and air void contents of the five mixtures (HMA, 30% RAP1, 60% RAP1, 30% RAP2, and 60% RAP2) were measured after manufacturing the Marshall specimens. The densities and void contents of the RAP mixtures ranged from 2356 to 2380 kg/m 3 and 5.5 to 5.6%, respectively, which were close to those of the HMA mixture. This suggests that all RAP mixtures were successfully designed and ready for the laboratory experiment program, as presented in the next section.

Experimental program and combined aging with saturation aging tensile stiffness conditioning
As shown in Fig. 2, laboratory specimens (diameter: ϕ = 100 mm, height: h = 50 mm) were prepared from five mixtures (HMA, 30% RAP1, 60% RAP1, 30% RAP2, and 60% RAP2). The physical and mechanical properties of the mixtures, as well as those of the binders extracted from the mixtures, were assessed through various laboratory experiments, as shown in Fig. 2. To simulate combined aging, a saturation aging tensile stiffness test was conducted on the targeted laboratory specimens based on EN 12697-45 [59]. Hereafter, the combined aging through SATS tests will be referred to as SATS conditioning.

Overview of the experimental program
As shown in Fig. 2, after the specimens were prepared, indirect tensile stiffness modulus (ITSM) tests were performed on the specimens in accordance with British Standard (BS) EN 12697-26 [60]. The specimens were then divided into two groups. The specimens in each group were subjected to SATS conditioning. The detailed procedure of the SATS conditioning is presented later. Following SATS conditioning, ITSM and indirect tensile fatigue tests (ITFTs) were performed on the specimens. The ITFTs were conducted based on BS EN 12697-24 [61]. After the ITFTs were conducted, the stripping status of each specimen was quantitatively evaluated by image analysis of the cross-sectional profile of the specimen. Subsequently, the binders, that is, aged binders, were extracted from the specimens, followed by dynamic shear rheometer (DSR) tests. The binders were extracted from the specimens using a fractioning column with a chromatographic solvent, following EN 12697-4 [62].
In contrast, the specimens in the other groups were not subjected to SATS conditioning. After the ITSM tests were conducted, ITFTs were conducted on the specimens in the same manner as those for the SATS conditioning. Subsequently, the binders, that is, the unaged binders, were extracted from the specimens. Finally, DSR tests were conducted on the extracted binders.

Procedure of the SATS conditioning
As mentioned before, SATS conditioning, which represented the combined aging through the SATS tests, was applied to the targeted specimens. The SATS test was originally developed at the University of Nottingham [46][47][48][49]. The test method was standardized by the UK Highway Agency, and is currently standardized as EN 12697-45. As shown in Figs. 3a, b, the vessel in the SATS tests can provide specimens with combined aging of moisture, temperature, and pressure. In this study, SATS conditioning was conducted in accordance with EN 12697-45, but as conditioning was particularly important for applying combined aging to the specimens in this study, the detailed procedure is described below.
First, the dry mass of the targeted specimens was measured, and then the specimens were soaked in distilled water at 20 °C and saturated by vacuum pressure of − 65 kPa for 30 min. Subsequently, the wet mass of each specimen was measured to calculate the percentage saturation as the "initial saturation." Meanwhile, while maintaining the top lid closed, the vessel with water was maintained at 85 °C for at least 2 h before inserting the specimens. Subsequently, the saturated specimens were positioned in the vessel on the trays, the lid of the vessel was closed again, and the air pressure inside was gradually raised to 0.5 MPa. One of the reasons for setting the pressure to 0.5 MPa instead of 2.1 MPa was to minimize the damage to each specimen by referring to previous studies [63][64][65]. In fact, preliminary experiments were conducted in this study with setting the air pressure to 2.1 MPa. As the results, microcracks were seen in bitumen mastics between aggregates in the specimens. Meanwhile, with setting the air pressure to 0.5 MPa instead of 2.1 MPa, no cracks were observed in the specimens. Therefore, it was decided to set the air pressure to 0.5 MPa inside the vessel for SATS conditioning, which seemed to give a more realistic degradation. Thereafter, the conditions were maintained at 85 °C for 24 h in accordance with previous studies. After 24 h, the vessel temperature was reduced to 30 °C, at which point the vessel was cooled for another 24 h. Finally, the air pressure was gradually decreased to atmospheric pressure, making the surface of the specimens dry, and each specimen was weighed. Percentage saturation was calculated as the "retained saturation."  Immediately after measuring the mass of the specimens, each specimen was wrapped in a plastic cling film to retain the moisture content. As previously described, the indirect tensile stiffness modulus of the specimens was measured using ITSM tests. Consequently, the fatigue properties of the specimens were examined using ITFTs. In practice, SATS conditioning has been conducted to assess the durability performance of asphalt mixtures, particularly focusing on stiffness [50,63]. Therefore, conditioning could be beneficial for the evaluation of the moisture sensitivity of aggregates because the reduction in stiffness would be related to stripping. It is also vital to assess the resiliency of RAP mixtures based on their fatigue properties. Therefore, in this study, attempts were made to evaluate the fatigue performance of RAP mixtures subjected to conditioning. It was expected that the use of high-quality aggregates in the mixtures and the low pressure (0.5 MPa) in the SATS conditioning might make it possible to investigate the fatigue performance of RAP mixtures even after they experienced conditioning. As a result, the residual service lives of the HMA and RAP mixtures experiencing conditioning were successfully compared based on the ITFT results. The details are presented later in this study.

Changes in binder properties with and without SATS conditioning
As shown in Fig. 2, the asphalt binders were extracted from the laboratory specimens of the mixtures to investigate the binder properties with and without SATS conditioning. The binders were extracted from specimens of five types of mixtures (HMA, 30% RAP1, 60% RAP1, 30% RAP2, and 60% RAP2). The properties of the extracted binders were evaluated via frequency sweep and fatigue tests using DSR. In this section, first, the results obtained from the frequency sweep tests are presented; the fatigue test results are then presented in the subsequent section.

Stiffness modulus
From the frequency sweep tests, the complex modulus G* and phase angle δ of each binder were evaluated for a range of frequencies of 0.1-10 Hz at eight temperatures from 34 to 78 °C in 6 °C increments. These results were assessed using the master curves of G* and black diagrams. Figures 4a, b present the master curves of G* for the binders extracted from the specimens without and with SATS conditioning, respectively. The master curves were obtained at a reference temperature of 34 °C. In general, old binders tend to increase G* compared to virgin binders. However, as shown in Fig. 4a, the master curves obtained for each binder were almost similar. This suggests that the rejuvenators successfully worked when mixed with the RAP aggregates and that the master curves of the binders of the RAP mixtures could be similar to that of the HMA mixture. Comparing the results in Figs. 4a, b, it can be observed that there is no significant change in the properties of each binder with SATS conditioning. This is probably because a low pressure (0.5 MPa) and a curing period of 24 h were set in the SATS conditioning in this study.
To understand the detailed rheological properties of the binders, the relationship between G* and δ is represented by a black diagram. Figures 5a, b show the black diagrams for the five extracted binders with and without the SATS conditioning, respectively, using the results measured at a temperature of 34 °C. The results shown in the figures reveal that there is no significant difference in the G*-δ relationship of each binder with or without SATS conditioning. It was also found that the G*-δ relationships shifted toward a lower phase angle as the RAP content increased from 0 (i.e., HMA) to 60%. This trend can be observed in both RAP 1 and RAP 2 binders, with or without SATS conditioning.
As shown in Fig. 6, detailed analyses were conducted on the changes in the binder properties with and without SATS conditioning, based on the ratio of G* with SATS conditioning (G* conditioned ) to G* without SATS (G* unconditioned ), as obtained at 10 Hz for each type of mixture. In the figure, the ratios are plotted against the temperature. It can be observed that the ratio of each binder is nearly 1.0, at all temperatures. It also appears that there is little change in binder stiffness with and without SATS conditioning. However, carefully examining the details, the binder extracted from the 30% RAP1 mixture showed ratios lower than 1.0, whereas the binder from the 60% RAP1 mixture demonstrated higher ratios than those from the HMA and RAP 2 mixtures. Different trends in the effects of SATS conditioning on the RAP1 and RAP2 binders were also observed for the results obtained at other frequencies. Although details of the reason(s) for the different trends need to be studied in the future, the difference may be attributed to the difference in the chemical compositions of the binders. As shown in Fig. 1, both the virgin binder and the RA binder with Rejuvenator 2 show similar components in the SARA fractions. However, the RA binder with Rejuvenator 1 had a different chemical composition in the SARA fraction compared with those of the virgin binder and the RA binder with Rejuvenator 2.

Fatigue behavior
To understand the fatigue characteristics of the binders extracted from the specimens of five types of mixtures with and without combined aging, fatigue tests were also performed on the binders using DSR. As shown in Fig. 2, the tests were conducted under stress-control conditions.
The failure criterion shown in Fig. 7 is used to determine the number of cycles at fatigue failure. Figure 7 shows an example of the relationship between the dissipated energy ratio (DER) and the number of cycles obtained from the test. The DER is defined as follows: where W N is the dissipated energy in cycle N, W i is the dissipated energy in cycle i, and ∑ N 1 W i is the total sum of the dissipated energy up to cycle N. W i can be obtained as follows: where σ i is the stress level in cycle i, ε i is the strain level in cycle i, and φ i is the phase angle in cycle i. In general, cracks propagate when the actual DER decreases by 20% relative to the DER in the non-destructive state, as represented by the straight line in Fig. 7 [66]. Thus, the number of cycles at failure (N f ) was determined when the actual DER decreased by 20%. Figure 8 shows the test results obtained for the specimens without SATS conditioning and those with SATS conditioning. In Fig. 8a, b, the relationships between the initial DER and the number of loading cycles at failure of the RAP binders without SATS conditioning are located on the upper left side of those of the HMA binder. This indicates that RAP binders are less durable against fatigue than HMA binders. In addition, there is a tendency for the fatigue life of the RAP binders without SATS conditioning to decrease with an increase in the RAP content, irrespective of the rejuvenator type. These observations are consistent with those shown in Fig. 5a.
Comparing the results in Figs. 8, it can be observed that the changes in the fatigue behaviors of the binders from SATS conditioning are insignificant. However, a more detailed analysis revealed differences in the effects of SATS conditioning on the fatigue behaviors of RAP1 and RAP2 binders. The RAP 2 binders did not show any significant differences in fatigue behaviors with and without SATS conditioning. Notably, the HMA binder also showed little difference in fatigue behaviors with and without SATS conditioning. However, the binder from the 30% RAP1 mixture showed a slight reduction in fatigue life, whereas that from the 60% RAP1 mixtures showed a slight increase in failure life when experiencing SATS conditioning. These trends are consistent with those observed in Fig. 6.

Changes in properties of mixtures with and without SATS conditioning
The properties of the five types of mixtures (HMA, 30% RAP1, 60% RAP1, 30% RAP2, and 60% RAP2) were evaluated through ITSM tests and ITFTs, as shown in Note that the fatigue behaviors of the HMA and RAP mixtures subjected to SATS conditioning were investigated to assess their residual service life after the damage from the conditioning.

Stiffness modulus
To investigate the changes in the stiffness of the mixtures with and without SATS conditioning, ITSM tests were conducted on specimens of five types of mixtures (HMA, 30% RAP1, 60% RAP1, 30% RAP2, and 60% RAP2). The test conditions are shown in Fig. 2. Table 3 summarizes the obtained stiffness and air void content values of the mixtures with and without SATS conditioning. The stiffness shown in the table is the average stiffness obtained from the three specimens for each mixture. The standard deviation and coefficient of variation values are also listed in the table. As shown, without SATS conditioning, the average stiffness of the RAP mixtures was lower than that of the HMA mixture, although the RAP 1 and RAP 2 binders were adjusted to 70 pen grade with the rejuvenators. This indicates that the stiffness of the RAP mixtures can differ from that of the HMA mixture, even if the penetration of the binder and grading of the RAP mixtures are adjusted to be close to those of the HMA mixture. The RAP mixtures showed a stiffness lower than that of the HMA mixture probably because of the rheological properties of the RAP binders, as shown in Fig. 5a. As discussed previously, the G* of the RAP binders becomes lower than that of the virgin binder as the phase angle decreases.   Figure 9a shows the retained stiffness ratio, that is, the ratio of stiffness with SATS conditioning to that without SATS, as calculated from the stiffness of an identical specimen with and without SATS conditioning. The ratio of the HMA mixture was significantly lower than 1.0, whereas the ratios of the RAP mixtures were higher than those of the HMA mixtures. Moreover, the ratios obtained for the 60% RAP1 and 60% RAP2 mixtures were higher than those for the 30% RAP1 and 30% RAP2 mixtures, respectively.
In Fig. 9b, the retained stiffness ratio is plotted against the retained saturation of each specimen. As shown, the ratio decreases with an increase in the retained saturation, suggesting that the decrease in the mixture stiffness caused by SATS conditioning is strongly related to the increase in the retained saturation. In the figure, the RAP mixtures show higher retained stiffness ratios with lower retained saturations compared to the HMA mixtures, indicating that the RAP mixtures will degrade less against combined aging than the HMA mixture in terms of stiffness. This trend became more significant with an increase in the RAP content, irrespective of the rejuvenator type. Among the RAP mixtures, the 30% RAP2 mixture had a higher retained saturation than the other RAP mixtures, suggesting that it was more affected by SATS conditioning. As shown in Fig. 10a, b, both the RAP 1 and RAP 2 mixtures show longer fatigue cycles than the HMA mixture, despite their initial strains being nearly the same as that of the HMA mixture. This observation is different from that observed for the fatigue properties of binders (refer to Fig. 8). This suggests that the adhesion state of the binder and aggregates significantly influences the fatigue properties of the mixtures. In addition, the fatigue cycles of the RAP mixtures increased when the RAP content increased from 30 to 60% for both the RAP1 and RAP2 mixtures. This suggests that the service life of the RAP mixtures increased as  . 9 Retained stiffness ratio from ITSM tests a retained stiffness ratio for each mixture, b relationships between the retained stiffness ratio and retained saturation for each mixture the chances of adhesion between the RAP aggregates and binders increased. Comparing the results in Fig. 10a-d, notably, the RAP mixtures with SATS conditioning demonstrate a longer residual service life compared to the HMA mixture with SATS conditioning. Similar to the case without SATS conditioning, this tendency was more pronounced when the RAP aggregate content was increased from 30 to 60% for both RAP1 and RAP2 mixtures. It can also be observed that the number of cycles at failure of the RAP mixtures with SATS conditioning is reduced relative to that without SATS conditioning. However, the initial strain of the RAP mixtures did not change significantly with or without SATS conditioning. This agrees with the fact that the retained stiffness ratios of the RAP mixtures were relatively close to 1.0, as shown in Fig. 9a. Figure 10 also demonstrates that the RAP1 mixtures showed differences in the initial strains between 30 and 60% RAP contents, whereas the RAP 2 mixtures showed similar strains even when the RAP contents were different. This difference may be attributed to the difference in the air void content between the RAP1 and RAP2 mixtures, as shown in Table 3. The RAP1 mixtures showed a decrease in the air void content as the RAP aggregate content increased, whereas the RAP2 mixtures showed the same air void content between 30 and 60% RAP contents. Because the RAP1 mixtures included higher resin components, their compaction performance became worse than that of the RAP2 mixtures, and higher air voids were attained in the RAP1 mixtures. This suggests that the difference in the type of rejuvenator affects the compaction performance of the RAP mixtures, resulting in different fatigue behaviors.

Stripping characteristics of the mixtures with the SATS conditioning
As shown in Fig. 2, the stripping status of the HMA and RAP mixtures with SATS conditioning was examined after the ITFTs. Each mixture specimen was divided into two cross-sections, and the stripping areas were measured through digital image analysis following the Japanese standard of the immersed wheel tracking test, B004  [67]. The stripping ratio of each specimen was defined as the area of the stripping aggregate divided by the crosssectional area. Consequently, an average stripping ratio was obtained for each mixture. Figure 11 shows the average stripping ratio of the mixtures with the SATS conditioning as well as the images of the cross-sections of HMA and 60% RAP1 mixtures. As a result, the average ratio of the HMA mixture showed approximately 8% whereas those of RAP mixtures indicated less than 4%. Moreover, the stripping resistance increased as the RAP content increased. The 30 and 60% RAP1 mixtures demonstrated 1.1 and 0.8% while those of RAP2 mixtures illustrated 3.4 and 0.6%, respectively. These results indicate that among the RAP mixtures, the damage by SATS conditioning was more pronounced in 30% RAP2 mixture. This was in harmony with the high residual saturation after the conditioning shown in Fig. 9. Although it seems that the differences in rejuvenator might influence the stripping resistance of mixtures, the obtained results demonstrate that the RAP mixtures have better stripping resistance than the HMA mixture. Similar to the findings obtained for the fatigue characteristics, the results indicate that the stripping resistance of the mixtures increases as the chances of adhesion between the RAP aggregates and RAP binders increase.

Discussion on the deterioration mechanism
In the previous sections, different trends were observed in the changes in the properties of the HMA and RAP mixtures with SATS conditioning. For example, the RAP mixtures did not show a reduction in the retained stiffness, whereas the HMA mixture showed a significant drop in stiffness with SATS conditioning. It was also found that the RAP mixtures exhibited a longer residual service life than the HMA mixture after SATS conditioning. In addition, the RAP mixtures exhibited higher stripping resistance than the HMA mixture. These trends were more pronounced for higher RAP content.
However, no significant changes were observed in the properties of the binders extracted from the HMA and RAP mixtures with SATS conditioning. Therefore, the fatigue test results indicated different trends between the mixture and binder, despite the general expectation that the trend should be the same. To examine this contradiction, the RAP binder without SATS conditioning was investigated in more detail using two-stage extraction, with reference to a previous  [68]. Two-stage extraction was performed on the 60% RAP1 mixture without SATS conditioning. First, the RAP mixtures were soaked in dichloromethane for 3 min, after which the outer binder layer was extracted. The same procedure was followed for the rest of the RAP mixture to obtain the inner-layer binder. Both the outer-and inner-layer binders were examined through frequency sweep tests using DSR. Figure 12 shows the master curves of the inner-and outer-layer binders extracted from the 60% RAP1 mixture without SATS conditioning. The master curve of the binder extracted from the HMA mixture without SATS conditioning in Fig. 4a is also shown. The difference in the master curve between the outer-and inner-layer binders can be observed for the 60% RAP1 mixture. The inner-layer binder demonstrates a higher complex modulus G* than the outerlayer binder at all frequencies. Meanwhile, the outer-layer binder showed a modulus G* similar to that of the binder extracted from the HMA mixture. These facts suggest that even though the RA binder was recovered by the rejuvenators, the inner-layer binder was still older than the outer-layer binder and the HMA binder, as demonstrated in a previous study [69,70]. Thus, it can be assumed that the inner-layer binder adheres to the aggregate better than the HMA binder [71][72][73].
As mentioned above, the properties of the binder extracted from the HMA mixture did not change significantly with the SATS conditioning. However, in the case of the HMA mixture, the retained saturation was high with the SATS conditioning. Owing to this situation, as shown in Fig. 13a, the adhesion between the aggregate and binder may be lost after SATS conditioning. Water might then infiltrate into the voids that had not been continuous before being subjected to SATS conditioning. The loss of adhesion between the aggregate and binder allows the aggregate to move more easily during loading. This could be in accordance with the lower stiffness and higher initial strain of the mixtures with SATS conditioning compared to those without conditioning, as observed in the ITSM tests and ITFTs, respectively.
In contrast, in the case of the RAP mixture, although the properties of the binder did not change significantly when subjected to SATS conditioning, as in the case of the HMA mixture, the retained saturation remained relatively low. This indicates that, as shown in Fig. 13b, the RAP mixture is expected to have little water infiltration into the discontinuous voids, even after SATS conditioning. Therefore, it is natural to assume that the adhesion between the binder and aggregate is not lost. In fact, the results shown in Fig. 12 support the idea that the inner-layer binder adheres to the aggregate better than the HMA binder. This mechanism is consistent with the fact that the stiffness in the ITSM tests and the initial strain in the ITFTs did not change significantly for the RAP mixture with SATS conditioning. Notably, because the properties of the binders did not change significantly after SATS conditioning, it was assumed that the bonds in the binders were not lost in either the HMA or RAP mixtures, as shown in Fig. 13.

Conclusion
To assess the durability of RAP mixtures, the mechanical and physical properties of RAP mixtures with 30 and 60% RAP contents blended with two types of oil-based rejuvenator agents (denoted as 30% RAP1, 60% RAP1, 30% RAP2, and 60% RAP2 mixtures, respectively), both with and without combined aging, were experimentally investigated. For comparison, the properties of the HMA mixtures were investigated in the same way. In the experiments, SATS conditioning was used to simulate combined aging. The following conclusions were obtained from the test results.
1. Asphalt binders were extracted from five mixtures (HMA, 30% RAP1, 60% RAP1, 30% RAP2, and 60% RAP2) to investigate the binder properties with and without SATS conditioning. The DSR test results demonstrated that there was no significant change in the properties of each binder after SATS conditioning. This was because a low pressure (0.5 MPa) and curing period of 24 h were used in the SATS conditioning. Detailed analyses showed that there were differences in the stiffness and fatigue results of the RAP binders with different rejuvenators. The different trends in the effects of SATS conditioning on the RAP1 and RAP2 binders might be attributed to differences in the chemical composition of the binders, as caused by the different rejuvenators. 2. Different trends were observed in the changes in the properties of the HMA and RAP mixtures with SATS conditioning. The ITSM test results demonstrated that the RAP mixtures with rejuvenators did not show a reduction in retained stiffness, whereas the HMA mixtures showed a significant drop in stiffness with SATS conditioning. Among the RAP mixtures, the 30% RAP2 mixture had a higher retained saturation than the other RAP mixtures, suggesting that it was more affected by SATS conditioning. In addition, in the ITFTs, the RAP mixtures with rejuvenators showed a longer residual service life than the HMA mixture after the damage from SATS conditioning. These trends were more pronounced for higher RAP content. In addition, detailed analyses showed that there were differences in the fatigue behaviors of RAP mixtures with different rejuvenators. Meanwhile, stripping results demonstrated that among the RAP mixtures, the damage by SATS conditioning was more pronounced in 30% RAP2 mixture. This was in harmony with the high residual saturation after the conditioning. These differences might be attributed to differences in the compaction performances of the RAP mixtures with different rejuvenators, which could (in turn) be due to the different components of the chemical properties of the RAP binders. 3. The RAP mixtures with rejuvenators showed a lower stripping ratio with SATS conditioning than that of the HMA mixture. This trend became more pronounced as RAP content increased. All the experimental results suggested that the adhesion states of the binder and aggregates caused a difference in durability between the HMA mixture and RAP mixtures with SATS conditioning. To assess the deterioration mechanism, DSR tests were conducted on the binders using two-stage extraction of the RAP mixture with rejuvenators. The results indicated that even though the RA binder was recovered by the rejuvenators, the inner-layer binder was still older than the outer-layer binder and the HMA binder. Based on these observations, it is proposed that the RAP mixture with rejuvenators did not lose adhesion between the binder and aggregate, whereas the adhesion could be lost in the case of the HMA mixture with SATS condition- ing. The proposed mechanism can reasonably explain the trends in the changes in the properties of binders and mixtures with SATS conditioning, as obtained from the experiments in this study.
Funding Open access funding provided by Yokohama National University.

Conflict of interest
The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.