Mechanistic insights into the structure–void–interface–performance coupling of High-RAP hot-recycled asphalt mixtures

ABSTRACT

To elucidate the multiscale mechanisms governing performance formation in high–reclaimed asphalt pavement (RAP) mixtures, this study integrates macro-, meso-, and microscale analyses into a unified framework. Based on large-sample statistical analysis, the concept of RAP “inherent gradation” is proposed, demonstrating consistent convergence toward the upper limit of the AC-13 gradation. A quantitative clustering degree index (C, average ≈44%) is introduced, identifying clustering as the key structural origin of systematic gradation coarsening and void-structure destabilization. An internal-porosity–based discriminant model, combined with a water-to-asphalt film-thickness conversion (h_a = 1.77h_w), enables accurate prediction of the optimum asphalt content (OAC), with deviations within ±5%. Microscale chemical and morphological analyses indicate that aging-induced increases in interfacial tension act as a dominant barrier to binder compatibility, whereas moderate incorporation of Trinidad Lake Asphalt (TLA, ~40%) effectively reduces interfacial energy and improves wetting behavior. Building on these findings, a “Structure–Void–Interface–Performance” multiscale interaction model is established and experimentally validated, revealing a dual-control mechanism: high-temperature performance can be enhanced through meso–micro synergistic regulation, while low-temperature and fatigue performance remain constrained by aged-binder rheology. The results provide a theoretical basis and practical guidance for refined high-RAP mixture design.

Reclaimed asphalt pavement (RAP); Inherent gradation; Clustering effect; Interfacial compatibility; Multiscale mechanism

1. INTRODUCTION

With the rapid expansion of transportation infrastructure in China, the consumption of natural aggregates and asphalt binders in road construction has risen sharply, resulting in severe resource depletion and escalating environmental concerns [¹, ²]. In this context, the high-value recycling and reuse of reclaimed asphalt pavement (RAP) have become an essential pathway for achieving sustainable highway development [³, ⁴]. Specifically, hot recycling technologies incorporating high RAP contents (typically exceeding 30%) are widely regarded as a crucial direction for future asphalt pavement development, owing to their substantial potential to conserve resources, reduce costs, and mitigate carbon emissions [⁵, ⁶].

Nevertheless, the large-scale implementation of high-RAP-content mixtures in practical engineering continues to face formidable challenges, primarily due to the limited understanding of the intrinsic complexity and performance evolution of RAP materials [⁷, ⁸]. The gradation characteristics of RAP fundamentally determine the skeletal structure of recycled asphalt mixtures. Traditionally, RAP gradation has been considered highly variable [⁹, ¹⁰], necessitating individual extraction and testing during mix design. However, recent studies and field observations have revealed a contrasting trend: RAP materials from diverse sources, after long-term service and repeated maintenance, tend to exhibit a “convergent” gradation distribution within a specific range [¹¹, ¹²]. Repeated traffic loading causes coarse aggregates to fracture and become finer, increasing the proportion of fine aggregates and resulting in a stabilized overall gradation curve [¹³]. However, this potentially “inherent gradation” pattern and its statistical characteristics have not yet been systematically validated through large-scale datasets, thereby limiting its reliability as a design reference.

At the microscale, RAP particles often exhibit a “clustering effect” resulting from the cohesive action of residual aged asphalt binders [¹⁴, ¹⁵]. This phenomenon produces an apparently coarser gradation in the RAP and markedly alters its effective specific surface area as well as the distribution of asphalt film thickness. Consequently, this effect induces deviations in the predicted optimum asphalt content (OAC) and destabilizes the volumetric properties of recycled mixtures [¹⁶]. Although the clustering phenomenon has been widely recognized, most existing studies have remained qualitative in nature. Quantitative characterization—such as defining a measurable degree of clustering—and systematic understanding of its impact on the macroscopic performance of recycled mixtures remain lacking. Consequently, universally applicable evaluation methods and mechanistic explanations for this effect have yet to be established.

The interfacial compatibility between aged and virgin asphalt binders forms the microscopic foundation governing the performance of recycled asphalt mixtures, especially their resistance to low-temperature cracking and fatigue. During long-term service, aged asphalt undergoes oxidation and polymerization, leading to a substantial increase in polar functional groups, including carbonyl and sulfoxide moieties. This process promotes asphaltene aggregation [¹⁷], coarsens the colloidal structure, increases viscosity [¹⁸], and reduces molecular diffusion capacity. These microscopic mismatches at the binder interface represent the fundamental cause of performance degradation in recycled mixtures [¹⁹]. Although conventional rejuvenators can partially restore the rheological properties of aged asphalt, their ability to reduce interfacial tension and promote molecular-level blending between aged and virgin binders remains limited [²⁰, ²¹].

Trinidad Lake Asphalt (TLA), a naturally occurring asphalt with high modulus and strong polarity, provides a promising approach to addressing the aforementioned challenges. Studies have demonstrated that incorporating TLA can significantly improve the high-temperature stability of asphalt mixtures [²²], while its abundant natural asphaltenes and highly polar components can effectively enhance interfacial compatibility between aged and virgin binders [²³], existing studies on the use of TLA in high-RAP systems have predominantly focused on macroscopic performance verification [²⁴]. Comprehensive, multiscale investigations linking gradation convergence, clustering evolution, and binder interfacial compatibility remain scarce. Consequently, an integrated mechanistic framework connecting material characteristics, interfacial interactions, and macroscopic performance has yet to be established.

In summary, advancing high-RAP-content recycling technologies urgently requires a comprehensive design framework that integrates macro-, meso-, and microscale perspectives. To address this need, this study has four primary objectives. First, it aims to systematically verify and quantify the existence and statistical characteristics of RAP’s “inherent gradation” through large-sample statistical analysis. Second, it seeks to propose a novel parameter, the clustering degree (C), to quantitatively characterize the clustering effect and elucidate its impact on gradation and volumetric parameters. Third, the study will develop a void ratio discrimination model and a water-film–asphalt-film conversion method, based on internal porosity correction, to accurately predict the optimum asphalt content (OAC). Finally, it will employ FTIR, AFM, SEM, and Fluent numerical simulations to comprehensively investigate the interfacial diffusion behavior and compatibility modulation mechanisms between aged and virgin binders in the TLA-modified system. By establishing a multiscale interaction framework linking structure, void, interface, and performance, this study seeks to provide a robust theoretical foundation and an innovative technical pathway for enhancing the performance and refining the design of high-RAP hot-recycled asphalt mixtures.

2. MATERIALS AND METHODS

2.1. Materials

In this study, over one hundred groups of reclaimed asphalt pavement (RAP) samples were collected from diverse geographic regions, covering service lives from 3 to 15 years and originating from different pavement structural layers (surface, intermediate, and base courses), to ensure comprehensive representativeness of the dataset. The virgin asphalt employed was a 70# base binder. To adjust the performance of the composite binder, a portion of the samples was modified with Trinidad Lake Asphalt (TLA). A commercially available rejuvenator served as the regeneration agent. The properties of asphalts are shown in Table 1.

Table 1
Properties of the base and TLA-Modified asphalt.

PROPERTY	70# BASE BINDER	40% TLA-MODIFIED BINDER
Penetration (25°C, 5s, 100g, 0.1 mm)	64	31
Softening Point (Ring & Ball, °C)	49	59
Ductility (cm)	151 (15°C)	63.6 (25°C)
Density (25°C, g/cm3)	1.03	1.383
Dynamic viscosity (Pa·s)	180 (60°C)	1.06 (135°C)
Mass Loss after RTFOT (%)	0.21	0.24

• Note: TLA content is 40% by mass of the total binder.

2.2. Inherent gradation and clustering degree

The gradation of RAP samples was determined by extraction and sieving, followed by statistical analysis to quantify variability. Based on the large-sample dataset, the concept of “inherent gradation” (see Appendix A) is proposed to describe the stabilized particle-size distribution that RAP develops after long-term service.

To quantify RAP particle clustering, two indices are defined: the residual mass ratio (n) and the clustering degree (C). The residual mass ratio n is calculated using Equation 1, representing the clustering level at each sieve size:

(1) $$n\text{=}\frac{m_1}{m_2}\times \text{100\%}$$

where m₁ is the mass of aggregates retained on the sieve after extraction (g), and m₂ is the total mass of aggregates subjected to extraction (g). The C was calculated using Equation 2.

(2) $$C\text{=}1-{\displaystyle \sum P_b}\times n$$

where P_b is the proportion of retained aggregates on each sieve prior to extraction (%), and n is the residual mass ratio of RAP at each sieve size.

2.3. Dispersion characteristics and void ratio testing

To evaluate the dispersion behavior of clustered RAP particles during mixing, laboratory dispersion tests were performed at 140 °C. Sodium chloride powder was employed as an anti-adhesion agent to simulate field mixing conditions.

The bulk and apparent specific gravities of RAP aggregates were measured using the saturated surface-dry (SSD) method, and the initial void ratio (VV) was calculated using Equation 3.

(3) $$VV\text{=}\left(1-\frac{{\gamma }_f}{{\gamma }_t}\right)\times 100$$

where γ_f is the bulk specific gravity of RAP aggregates, and γ_t is the apparent specific gravity.

Since open surface pores on RAP particles are likely to be filled with virgin asphalt during recycling and thus no longer contribute to the final void structure of the recycled mixture, an internal void ratio (VV_in) was introduced to more accurately represent the volumetric influence of RAP on mixture design, as expressed by Equation 4.

(4) $${\text{VV}}_{\text{in}}\text{=}\left(1-\frac{{\gamma }_f}{{\gamma }_t}-\frac{m_f-m_a}{\raisebox{1ex}{$m_a$}\!\left/ \!\raisebox{-1ex}{${\gamma }_f$}\right.}\right)\times 100$$

where m_f and m_a denote the sample masses before and after immersion, respectively.

2.4. Asphalt film thickness and OAC prediction

Because water and asphalt exhibit different adhesion/affinity to aggregate surfaces, the thicknesses of the corresponding films formed on the aggregate surface are not identical. In this study, an empirical, dimensionless conversion factor t is introduced to relate the asphalt-film thickness to the experimentally measured water-film thickness, i.e., asphalt-film thickness = water-film thickness × t. The factor t is determined by coupling the binder–aggregate ratio of each size fraction obtained from extraction with the corresponding measured water-film thickness. Based on the water-film thickness measurement principle, a conversion relationship between aggregate water-film thickness and asphalt-film thickness is established, where the water-film thickness is (h_w) calculated according to Equation 5.

(5) $$h_w\text{=}\frac{m_1-m_2}{SA\rho m_2}\times {10}^3$$

2.5. Interfacial compatibility and diffusion behavior

A three-layer micro-diffusion model was developed using Fluent software to simulate the effects of viscosity and interfacial tension on binder diffusion. The model comprised an air layer, a virgin asphalt layer, and an aged asphalt layer (each 1 mm thick and 10 mm wide), with the air layer treated as semi-infinite. A spread-area test was also conducted to assess the spreading capacity of rejuvenated asphalt with varying TLA contents, serving as an auxiliary indicator of interfacial compatibility. The basic parameters of the asphalt model are shown in Table 2.

Table 2
Thermophysical properties of virgin and aged asphalt.

MATERIAL	DENSITY (kg/m3)	SPECIFIC HEAT (J/kg·K)	THERMAL CONDUCTIVITY (W/m·K)
Virgin asphalt	1030	1340	0.17
Aged asphalt	1050	1550	0.2

2.6. Pavement performance testing and microstructural characterization

This study’s experimental program, centered on the binder scale, evaluated the compatibility and rheological behavior of aged and virgin asphalt through a multi-technique approach. Two sample configurations were designed: homogeneous blends (50:50 mass ratio of hard virgin to aged asphalt) and interface-isolated models simulating partial blending. A comprehensive rheological assessment was conducted using a Dynamic Shear Rheometer (DSR) to measure high-temperature performance (G/sin δ) and fatigue resistance (G·sin δ), and a Bending Beam Rheometer (BBR) to determine low-temperature properties. Complementary microstructural analysis of interfacial morphology, phase distribution, and surface topography was performed via Scanning Electron Microscopy (SEM) and Atomic Force Microscopy (AFM).

3. RESULTS AND DISCUSSION

3.1. Statistical characteristics of inherent gradation

To clarify the long-term gradation evolution of reclaimed asphalt pavement (RAP), systematic extraction and sieving analyses were first conducted on RAP collected from multiple lanes and structural layers of the Nanjing Airport Expressway. As shown in Figure 1a, despite variations in original mix design and material composition across layers, the extracted gradation curves were highly concentrated, predominantly near the upper limit of the AC-13 gradation band (nominal maximum aggregate size: 13.2 mm), which is a dense-graded mixture specified by Chinese standards. Notably, RAP from the surface layer—constructed with hard basalt aggregates—exhibited a slightly coarser gradation than the intermediate and base layers. Aggregates larger than 16 mm fractured extensively during service and milling, indicating the coupled influence of traffic loading and aggregate mechanical properties on gradation evolution.

Figure 1
Comparison of RAP gradations from different sources and pavement layers. (a) Extracted gradation curves from different layers on the Nanjing Airport Expressway; (b) Gradation curves from Nanjing Gaojia Road & Bridge Co. and representative literature sources.

To further verify the generality of this pattern, 28 RAP gradation datasets from Nanjing Gaojia Road & Bridge Co., together with five representative datasets from the literature, were integrated for comparative analysis. As illustrated in Figure 1b, despite differences in source and service history, all RAP gradation curves fluctuated narrowly within the upper boundary of the AC-13 specification. This finding suggests that, after prolonged service and subsequent milling, the gradation of dense-graded asphalt mixtures does not diverge randomly but converges toward a relatively stable state.

Building on these observations, large-sample statistical analyses were conducted on over one hundred RAP samples collected from multiple provinces across China. Figure 2 summarizes the grouped statistical characteristics and convergence behavior of the full dataset under different conditions.

Figure 2
Statistical comparison of RAP gradations based on grouped analysis of over 100 samples under varying conditions. (a) Provinces; (b) Service lives; (c) Pavement layers.

Analysis of Figure 2 yields three insights. (i) Minimal regional variation: small standard deviations among provinces indicate limited geographic influence (Figure 2a). (ii) Pronounced temporal evolution: with increasing service life, coarse fractions decrease due to fragmentation, whereas fines increase, producing a general fining trend (Figure 2b). (iii) Layer-specific regularity: gradation coarseness follows surface < intermediate < base, consistent with original design gradations (Figure 2c).

Integrating statistical evidence across regions, service ages, and structural layers indicates that RAP gradation exhibits marked concentration and stability. Accordingly, this study formally proposes the “inherent gradation” of RAP, with recommended target values listed in Table 3.

Table 3
Recommended target gradation values for RAP (“inherent gradation”).

Sieve size (mm)	19	16	13.2	9.5	4.75	2.36	1.18	0.6	0.3	0.15	0.075
Target passing (%)	98.82	96.34	91.58	80.59	56.23	38.37	29.23	22.02	15.32	10.79	7.79

Comparative analysis with typical AC-20 and EME-20 gradations (Figure 3) further shows that the inherent RAP gradation is generally finer, with a lower proportion of coarse aggregates. This fine-grained tendency can partially compensate for the reduced low-temperature and fatigue performance associated with aged binders. Therefore, adopting the “inherent gradation” as a design baseline for high-RAP mixtures can simplify design, reduce variability, and provide an intrinsic performance-compensation mechanism, offering substantial practical value.

Figure 3
Comparison of gradation curves for AC-20, EME-20, and RAP “inherent gradation.”

3.2. Clustering effect and gradation deviation

During milling, transportation, and storage, RAP particles often occur in clustered form due to the cohesive action of residual aged binders. This clustering behavior is a fundamental cause of the discrepancy between measured and intrinsic RAP gradations. The clustering morphology is influenced by the properties of the original pavement materials and the milling process, resulting in considerable variability. Accordingly, the clustering evaluation approach used in this section is based on the method proposed by XU et al. [²⁵]. and is employed here to support subsequent multiscale analysis in this study.

Based on morphological observations, three representative clustering types were identified (Figure 4).

Figure 4
Representative RAP clustering types: (a) mild; (b) moderate; (c) severe.

To quantify the macroscopic influence of clustering on gradation, comparative sieving and extraction–sieving tests were conducted. As illustrated in Figure 5, the post-extraction gradation curve shifts upward and becomes noticeably finer, indicating that the apparent “coarse” gradation observed before extraction arises primarily from clustering rather than from the intrinsic characteristics of RAP.

Figure 5
Comparison of RAP gradation curves before extraction and after extraction.

To further characterize clustering across particle sizes, the residual mass ratio (n) was introduced. As shown in Figure 6, pronounced differences in n were observed among sieve sizes. The lowest n values—indicating the most severe clustering—occurred within the 9.5–16 mm and <0.075 mm fractions, with the 16 mm group exhibiting clustering ratios exceeding 70%. Using the residual proportions (P_b) and corresponding n for each sieve, the overall clustering degree (C) calculated from Equation 2 was approximately 44%. This result indicates that nearly half of the RAP mass enters the mixing process not as individual particles but as aggregated clusters.

Figure 6
Residual mass ratio (n) of RAP across particle-size ranges.

A critical engineering implication of clustering is its resistance to complete elimination under standard mixing conditions. To assess this, an extreme-dispersion test was conducted at 140 °C. As shown in Figure 7, although large clusters (>4.75 mm) were partially disintegrated, the overall gradation remained notably coarser than that of the fully extracted material. The fraction of particles within 0.3–2.36 mm increased after dispersion, whereas fines <0.15 mm decreased markedly due to re-agglomeration. These findings confirm that, under practical production conditions, RAP clusters cannot be fully dispersed and instead persist as partially clustered entities that participate in forming the aggregate skeleton of recycled mixtures.

Figure 7
Gradation comparison of RAP before extraction, after extraction, and after 140 °C extreme-dispersion testing.

In summary, clustering is an intrinsic characteristic of RAP, and the gradation deviation it induces is a primary cause of systematic discrepancies between the designed and in-place gradations of recycled mixtures. The clustering degree (C) proposed herein provides a reliable quantitative index for evaluating this effect. The findings highlight the necessity of explicitly considering clustering in both mix design and process control for high-RAP mixtures. Targeted measures—such as gradation adjustment, optimization of the binder–aggregate ratio, and enhanced dispersion techniques (e.g., staged preheating, high-shear mixing)—should be incorporated to mitigate clustering effects and improve mixture uniformity.

3.3. Void Ratio and Optimum Asphalt Content (OAC) prediction

In this section, clustering degree is incorporated as an intermediate volumetric variable to establish a discriminant model for void-ratio evolution and optimum asphalt content (OAC) prediction, rather than being treated as an independent performance indicator. Recent studies have examined the influence of RAP clustering degree and design parameters on the volumetric and performance characteristics of recycled asphalt mixtures [²⁶].

The void ratio is a key volumetric parameter governing asphalt-mixture performance. To enable precise volumetric design for high-RAP mixtures, it is essential to clarify how RAP clustering and internal porosity jointly influence the void structure and the prediction of the optimum asphalt content (OAC).

In this section, clustering degree is incorporated as an intermediate volumetric variable to establish a discrExperimental results reveal a distinct particle-size dependence of RAP internal porosity (Figure 8). For particles <2.36 mm, internal porosity is negligible, whereas for those >4.75 mm it increases markedly—reaching up to 2.5%. This indicates that large, clustered RAP particles are the dominant contributors to the internal void structure of recycled mixtures.

Figure 8
Variation of RAP internal porosity with particle size.

To clarify this influence, void-ratio measurements were conducted at two binder contents (5.5% and 7.5%) while varying RAP content (Figure 9). At 5.5% binder, the void ratio decreases with increasing RAP because the RAP internal porosity is lower than the design target and thus acts as a filler. Conversely, at 7.5% binder, the void ratio increases with RAP content, implying that internal RAP pores are not fully filled by the virgin binder and instead act as defects. Based on this dual behavior, a discriminant model was established that compares the design void ratio with RAP internal porosity as the core variable, enabling accurate prediction of void-ratio trends across binder levels.

Figure 9
Variation of void ratio with RAP content at different binder contents.

To validate the effect of clustering on volumetric properties, specimens with different clustering degrees were prepared (Table 4). As clustering decreased (from “clustered” to “fully opened”), a larger fraction of RAP surface area became available for coating, which increased the effective specific surface area and consequently increased the measured air-void content. Consistent with this volumetric change, the corresponding performance results showed improved high-temperature stability and reduced low-temperature cracking susceptibility with decreasing clustering. These results indicate that promoting RAP dispersion can beneficially affect both volumetric properties and mixture performance.

Table 4
Volumetric properties under different clustering states.

CLUSTERING STATE	DRY MASS (g)	MASS IN WATER (g)	SSD MASS(g)	BULK DENSITY (g/cm3)	MEAN(g/cm3)	THEORETICAL MAX DENSITY (g/cm3)	VOID RATIO (%)
Fully opened	1209.9	715.9	1214.5	2.43	2.44	2.52	3.00
	1218.5	725.5	1221.1	2.46
	1210.2	719.4	1213.7	2.45
Semi-opened	1208.7	718.3	1214.9	2.43	2.45	2.52	2.68
	1210.5	721.8	1215	2.45
	1216.1	725.3	1217.8	2.47
Clustered	1205.1	717	1208.3	2.45	2.46	2.52	2.30
	1207.1	721.1	1211.4	2.46

To enable accurate OAC prediction, an empirical conversion between the water-film thickness (h_w) and the asphalt-film thickness (h_a) was established based on the water-film test principle. Centrifugation tests on aggregates of different size fractions yielded an empirical relationship indicating that h_a is approximately 1.77 times h_w (Figure 10). This relationship can be expressed as Equation 6.

Figure 10
Relationship between asphalt-film thickness and water-film thickness.

(6) $$h_a=1.77h_w$$

To verify the microscale distribution of asphalt films, SEM observations of asphalt-mixture cross-sections were conducted (Figure 11). The results reveal highly non-uniform films with wavy morphology (Figure 11a), thicker films in surface depressions, and thinner—or locally absent—films at protrusions and angular edges (Figure 11b). Comparison across particle sizes (Figure 11c, d) shows that coarse aggregates (588 μm) are typically coated by thicker films, whereas fine aggregates (36 μm) exhibit sharper edges and significantly thinner films. This morphological evidence confirms that particle size is the primary determinant of asphalt-film thickness, consistent with thickness distributions predicted from specific-surface-area calculations. Integrating RAP gradation with the film-thickness model yielded OAC predictions within 5% of experimental optima, validating the accuracy and engineering applicability of the proposed approach. These observations are intended to provide microstructural support for the film-thickness-based OAC prediction model, rather than to report isolated new morphological features.

Figure 11
SEM images of asphalt-mixture cross-sections at different magnifications: (a) wavy film morphology; (b) thin/absent films at protrusions; (c) coarse aggregate; (d) fine aggregate.

3.4. Interfacial compatibility and performance regulation mechanism

Interfacial compatibility between aged and virgin binders forms the microscopic foundation governing the durability of recycled mixtures. To elucidate the governing mechanisms and identify effective modulation strategies, comprehensive analyses were conducted from chemical, morphological, kinetic, and wetting perspectives.

FTIR results (Table 5) show that, with increasing laboratory aging, the absorption areas corresponding to carbonyl (C=O) and sulfoxide (S=O) groups increase significantly. The calculated carbonyl index (CI) and sulfoxide index (SI) increase monotonically, confirming that aging enhances the chemical polarity of asphalt. Consequently, intermolecular forces increase, viscosity rises, and molecular mobility and diffusion capacity decrease—collectively impeding binder blending and rejuvenation.

Table 5
Functional-group indices under different aging durations.

AGING TIME(h)	AC=O	AS=O	AC-H	CI	SI
0	0.083	0.973	3.949	0.021	0.246
12	0.258	1.233	4.220	0.061	0.292
26	0.589	1.585	4.401	0.134	0.360

AFM and SEM observations provide morphological evidence supporting this chemical evolution. AFM phase images (Figure 12) show that, with increased aging, the characteristic “bee-structure” domains become fewer and larger, indicating colloidal coarsening driven by asphaltene aggregation and depletion of light components. These AFM observations corroborate the aging-induced colloidal coarsening and polarity enhancement assumed in the interfacial compatibility analysis.

Figure 12
AFM phase images at different aging levels: (a) 2D—virgin; (b) 3D—virgin; (c) 2D—12 h; (d) 3D—12 h; (e) 2D—26 h; (f) 3D—26 h.

SEM images (Figure 13) reveal pronounced phase separation and blurred transition zones at aged/virgin interfaces, indicating that interfacial microscale heterogeneity is a primary cause of macroscopic performance deterioration. These observations provide microstructural support for the interfacial compatibility mechanism discussed herein, rather than serving as isolated morphological findings.

Figure 13
SEM images of interface regions in asphalt-mixture cross-sections.

To quantitatively isolate key parameters affecting binder blending, a micro-diffusion model was constructed in Fluent. Simulations (Figure 14) reveal a critical distinction: increasing virgin-binder viscosity gradually decreases diffusion volume, whereas elevated interfacial tension causes a drastic reduction. Under the investigated binder system, aging level, and temperature range, interfacial tension shows a stronger influence than viscosity in our simulations/observations, and can be interpreted as a major thermodynamic contributor that limits interfacial diffusion.

Figure 14
Diffusion model results: (a) diffusion volume vs. binder viscosity; (b) diffusion volume vs. interfacial-tension coefficient.

From a mechanistic standpoint, interfacial tension controls the energetic cost of phase contact and wetting, while viscosity primarily governs the kinetic rate of molecular transport. The pronounced sensitivity of diffusion volume to interfacial tension observed in the simulations therefore reflects a thermodynamic limitation rather than a purely rheological constraint.

Building on the finding that interfacial tension is the governing barrier, we introduced highly polar Trinidad Lake Asphalt (TLA) as an interfacial modifier. Spreading-area tests (Table 6) identified a condition-specific optimal range. The specific spreading area per unit mass reached a maximum at a TLA content of approximately 40% by mass of the total binder, indicating a favorable balance between wetting enhancement and rheological compatibility under the tested conditions. This suggests that TLA lowers interfacial energy through its polar components, while an excessive dosage increases system viscosity and impairs spreading, thus defining a clear optimum window.

Table 6
Spreading-area results of binders with different TLA contents

TLA CONTENT (%)	70%	55%	55%	40%	0%
Specific spreading area (cm2)	2.86	3.54	3.73	3.72	2.86
	2.89	2.88	3.4	3.25	2.89
	3.11	3.23	3.75	3.46	3.11
Average spreading area (cm2)	2.95	3.22	3.63	3.48	2.95

It should be noted that this optimal range is not intended to represent a universal threshold, but rather reflects a trade-off between interfacial-energy reduction and viscosity increase for the specific binder system, aging level, and temperature investigated in this study.

In summary, interfacial tension is identified as a key controlling factor influencing binder compatibility under the investigated conditions. Aging increases chemical polarity and structural coarseness, elevating interfacial tension and suppressing diffusion; in contrast, moderate TLA incorporation (approximately 40%) reduces interfacial tension and helps overcome this barrier. This establishes a theoretical foundation for tuning recycled-mixture performance through targeted material design and interfacial-energy regulation.

It should be noted that the absolute values of clustering degree, OAC, and optimal TLA content are influenced by RAP source characteristics, aging level, and climatic conditions. The framework proposed in this study is therefore intended to provide a mechanistic and transferable basis for mixture design, rather than fixed numerical thresholds. Variations in material properties or environmental conditions are expected to shift the optimal ranges but not alter the underlying interfacial-regulation principles.

3.5. Multiscale mechanism and macroscopic performance verification

Building on the preceding systematic investigation of RAP—from macroscopic gradation to microscopic interfacial behavior—this section establishes a unified theoretical framework to elucidate the cross-scale mechanisms governing the performance of high-RAP mixtures. Accordingly, a “Structure–Void–Interface–Performance” multiscale interaction model is proposed and subsequently validated through macroscopic pavement-performance testing.

The model shows that the performance of recycled mixtures arises from hierarchical transmission and coupling of factors across the macro-, meso-, and microscale levels. The core findings are summarized as follows.

Macroscopic structural scale—Inherent gradation and clustering as the foundation. Large-sample statistics revealed an “inherent gradation” pattern in RAP (Section 3.1). However, a clustering degree C of up to 44% (Section 3.2) induces systematic deviations between designed and as-formed gradations, constituting a primary source of performance variability.

Mesoscopic volumetric scale—Voids and film-thickness distribution governing stability. Particle clustering and internal porosity (Section 3.3) jointly determine the mixture void structure. The discriminant model and film-thickness prediction developed herein enable precise control of the optimum asphalt content (OAC).

To directly assess pavement performance and verify the impacts of clustering and interfacial mechanisms, comprehensive tests of high-temperature, low-temperature, and fatigue properties were conducted. The experimental focus concerned binder blending state (two idealized models: homogeneous mixing and interface isolation) and asphalt aging level. The results are summarized in Figure 15.

Figure 15
Macroscopic performance validation: (a) rutting factor; (b) creep stiffness (homogeneous state); (c) fatigue factor (homogeneous state).(Error bars represent mean ± SD (n = 3)).

High-temperature performance. DSR results (Figure 15a) show that recycled binders in the interface-isolated state exhibit markedly higher rutting factors (G/sin δ) than those in the homogeneous-mixing state. This validates the beneficial effect of high-modulus interfacial layers and incomplete blending, consistent with simulations in Section 3.4 that identified interfacial effects as critical to high-temperature performance.

Low-temperature and fatigue performance. BBR and DSR-fatigue results (Figure 15b, c) indicate that, in the homogeneous-mixing state, creep stiffness (S) and fatigue factor (G·sin δ) of recycled binders follow the degradation trend of aged asphalt. The presence of virgin binder exerts negligible improvement. This demonstrates that, once blending is complete, the recycled system inevitably inherits the intrinsic deterioration of aged asphalt, imposing a clear macroscopic performance ceiling.

Integrating macroscopic evidence with the multiscale framework yields the following mechanistic chain: at the mesoscopic scale, clustering and porosity govern the optimization potential of high-temperature performance; at the microscopic scale, interfacial compatibility controls blending efficiency, whereas the intrinsic rheology of aged asphalt fundamentally constrains low-temperature and fatigue performance.

This understanding provides a pathway for engineering practice: high-temperature stability should be enhanced through coordinated meso- and microscale interventions—clustering dispersion (e.g., staged preheating, high-shear mixing) and material design (e.g., TLA modification to lower interfacial tension). By contrast, improvements in low-temperature and fatigue resistance are inherently limited by RAP-binder aging, necessitating strict control of RAP content and service history as hard constraints in mix design.

Through construction and validation of the “Structure–Void–Interface–Performance” multiscale model, this study identifies dual controlling factors for recycled-mixture performance—clustering/porosity at the meso-scale and interfacial tension/aging at the micro-scale. The framework clarifies the mechanistic origin of performance variability and provides a theoretical basis for accurate prediction and targeted regulation of high-RAP hot-recycled asphalt mixtures.

4. CONCLUSIONS

Through cross-scale experimental characterization and theoretical modeling, this study systematically elucidates the material regularities and performance-formation mechanisms of high-RAP hot-recycled asphalt mixtures. The key findings are as follows.

SUPPLEMENTARY MATERIAL

The following online material is available for this article.

Appendix A

Glossary of key terms and parameters.

5. ACKNOWLEDGMENTS

The authors are grateful for the Transportation Department of Jiangxi Province (grant number: 2023Z0001, 2024YB047) and Science Technology Development Program of Jilin Province (grant number: 20220203159SF).

6. BIBLIOGRAPHY

• ^[1] TSAKOUMAKI, M., PLATI, C., “A critical overview of using Reclaimed Asphalt Pavement (RAP) in Road pavement construction”, Infrastructures, v. 9, n. 8, pp. 128, 2024. doi: https://doi.org/10.3390/infrastructures9080128.
  » https://doi.org/10.3390/infrastructures9080128
• ^[2] ZHANG, F., ZHU, J., SUN, Y., et al, “Effect analysis of using Tall Oil Pitch (TOP) to partially extend bitumen in asphalt pavements: comparison of different tops”, Road Materials and Pavement Design, v. 26, n. sup1, pp. 361, 2025. doi: https://doi.org/10.1080/14680629.2025.2483478.
  » https://doi.org/10.1080/14680629.2025.2483478
• ^[3] MAIDI, M., SHERZER, G.L., GAL, E., “Multiscale numerical study of enhanced ductility ratios and capacity in carbon fiber-reinforced polymer concrete beams for safety design”, Polymers, v. 17, n. 2, pp. 234, 2025. doi: https://doi.org/10.3390/polym17020234. PubMed PMID: 39861305.
  » https://doi.org/10.3390/polym17020234
• ^[4] ATHIAPPAN, K., MURUGASAN, R., MAKENDRAN, C., et al, “Mechanical properties of asphalt mixtures containing reclaimed asphalt incorporating acrylonitrile butadiene styrene (ABS)”, Matéria, v. 28, n. 4, e20230201, 2023. doi: https://doi.org/10.1590/1517-7076-rmat-2023-0201.
  » https://doi.org/10.1590/1517-7076-rmat-2023-0201
• ^[5] ELNIHUM, A., LU, Q., ALHARTHAI, M., et al, “Evaluation of an asphalt mixture containing a high content of Reclaimed Asphalt Pavement (RAP) materials with epoxy asphalt”, Sustainability, v. 16, n. 12, pp. 4988, 2024. doi: https://doi.org/10.3390/su16124988.
  » https://doi.org/10.3390/su16124988
• ^[6] ZHANG, F., FALCHETTO, A.C., WANG, D., et al, “Prediction of asphalt rheological properties for paving and maintenance assistance using explainable machine learning”, Fuel, v. 396, pp. 135319, 2025. doi: https://doi.org/10.1016/j.fuel.2025.135319.
  » https://doi.org/10.1016/j.fuel.2025.135319
• ^[7] WANG, L., SHEN, A., MOU, G., et al, “Effect of RAP gradation subdivision and addition of a rejuvenator on recycled asphalt mixture engineering performance”, Case Studies in Construction Materials, v. 18, e02136, 2023. doi: https://doi.org/10.1016/j.cscm.2023.e02136.
  » https://doi.org/10.1016/j.cscm.2023.e02136
• ^[8] FENG, D., CAO, J., GAO, L., et al, “Recent developments in asphalt-aggregate separation technology for reclaimed asphalt pavement”, Journal of Road Engineering, v. 2, n. 4, pp. 332–347, 2022. doi: https://doi.org/10.1016/j.jreng.2022.07.002.
  » https://doi.org/10.1016/j.jreng.2022.07.002
• ^[9] AI, X., PEI, Z., XU, K., et al, “Rap agglomeration and partial blending of recycled hot mix asphalt: a literature review”, Journal of Road Engineering, v. 5, n. 2, pp. 230–243, 2025. doi: https://doi.org/10.1016/j.jreng.2024.12.006.
  » https://doi.org/10.1016/j.jreng.2024.12.006
• ^[10] XU, G., SHEN, Z., HAO, S., et al, “Mitigating rap variability through detailed particle size classification: applications in high rap content asphalt mixtures”, Case Studies in Construction Materials, v. 23, e04993, 2025. doi: https://doi.org/10.1016/j.cscm.2025.e04993.
  » https://doi.org/10.1016/j.cscm.2025.e04993
• ^[11] LI, M., HAN, Z., CHENG, H., et al, “Low-temperature performance improvement strategies for high rap content recycled asphalt mixtures: focus on rap gradation variability and mixing process”, Fuel, v. 387, pp. 134362, 2025. doi: https://doi.org/10.1016/j.fuel.2025.134362.
  » https://doi.org/10.1016/j.fuel.2025.134362
• ^[12] OLIVEIRA, M.S., FARIAS, M.M., “Hot recycled asphalt mixtures with the incorporation of milled material: analysis of the mechanical performance and economic presumption in the applicability of the technique”, Matéria (Rio de Janeiro), v. 27, n. 4, e20220217, 2022. doi: https://doi.org/10.1590/1517-7076-rmat-2022-0217.
  » https://doi.org/10.1590/1517-7076-rmat-2022-0217
• ^[13] WU, Z.Y., CAI, H., WU, X.S., et al, “Study on the factors affecting temperature and workability during the mixing process of hot in-place Recycled Asphalt Mixtures (RHMA)”, Matéria, v. 30, e20240816, 2025. doi: https://doi.org/10.1590/1517-7076-rmat-2024-0816.
  » https://doi.org/10.1590/1517-7076-rmat-2024-0816
• ^[14] CAI, F., XIAO, T., SHE, H., et al, “Performance optimization of asphalt blending behavior considering the effect of rap agglomerates in hot in-place recycled asphalt mixture”, Construction & Building Materials, v. 470, pp. 140633, 2025. doi: https://doi.org/10.1016/j.conbuildmat.2025.140633.
  » https://doi.org/10.1016/j.conbuildmat.2025.140633
• ^[15] LIU, Q., LIU, J., WU, J., et al, “Investigating agglomeration behavior of recycled asphalt mixtures in mixing stage using discrete element method”, Construction & Building Materials, v. 440, pp. 137504, 2024. doi: https://doi.org/10.1016/j.conbuildmat.2024.137504.
  » https://doi.org/10.1016/j.conbuildmat.2024.137504
• ^[16] YE, W., CHEN, M.J., GAO, C., et al, “Sustainable asphalt concrete containing RAP and RCA: volumetrics, mechanical properties, and economic analysis”, Matéria v. 30, e20250161, 2025. doi: https://doi.org/10.1590/1517-7076-rmat-2025-0161.
  » https://doi.org/10.1590/1517-7076-rmat-2025-0161
• ^[17] WANG, C., TIAN, X.G., WANG, Y., et al, “Asphalt aging and its anti-aging mechanism based on quantum chemistry”, Case Studies in Construction Materials, v. 20, e02802, 2024. doi: https://doi.org/10.1016/j.cscm.2023.e02802.
  » https://doi.org/10.1016/j.cscm.2023.e02802
• ^[18] CUI, B., WANG, H., “Oxidative aging mechanism of asphalt binder using experiment-derived average molecular model and reaxff molecular dynamics simulation”, Fuel, v. 345, pp. 128192, 2023. doi: https://doi.org/10.1016/j.fuel.2023.128192.
  » https://doi.org/10.1016/j.fuel.2023.128192
• ^[19] HU, D., GU, X., YANG, F., et al, “Atomic mechanisms of separation failure at the asphalt–aggregate interface and its dependence on aging and rejuvenation: insights from molecular dynamics simulations and dft calculations”, Applied Surface Science, v. 598, pp. 153775, 2022. doi: https://doi.org/10.1016/j.apsusc.2022.153775.
  » https://doi.org/10.1016/j.apsusc.2022.153775
• ^[20] FANG, Y., ZHANG, Z., YANG, J., et al, “Research on interfacial fusion characteristics of rejuvenator-aged asphalt system: wettability, permeability and molecular simulation”, Construction & Building Materials, v. 443, pp. 137646, 2024. doi: https://doi.org/10.1016/j.conbuildmat.2024.137646.
  » https://doi.org/10.1016/j.conbuildmat.2024.137646
• ^[21] CAVALLI, M.C., WU, W., POULIKAKOS, L., “Bio-based rejuvenators in asphalt pavements: a comprehensive review and analytical study”, Journal of Road Engineering, v. 4, n. 3, pp. 282–291, 2024. doi: https://doi.org/10.1016/j.jreng.2024.04.007.
  » https://doi.org/10.1016/j.jreng.2024.04.007
• ^[22] VIGNESWARAN, S., YUN, J., KIM, H., et al, “Enhancing asphalt binder performance and storage stability with Trinidad Lake Asphalt (TLA)”, Applied Sciences, v. 14, n. 14, pp. 6023, 2024. doi: https://doi.org/10.3390/app14146023.
  » https://doi.org/10.3390/app14146023
• ^[23] ZHANG, M., YUAN, D., JIANG, W., “Effects of TLA on rheological, aging, and chemical performance of SBS modified asphalt”, Construction & Building Materials, v. 438, pp. 137283, 2024. doi: https://doi.org/10.1016/j.conbuildmat.2024.137283.
  » https://doi.org/10.1016/j.conbuildmat.2024.137283
• ^[24] LI, C., ZENG, Y., FANG, Y., et al, “On the use of recycled asphalt and trinidad lake asphalt (tla) for the preparation of high modulus asphalt concrete”, Fluid Dynamics & Materials Processing, v. 19, n. 7, pp. 1939–1950, 2023. doi: https://doi.org/10.32604/fdmp.2023.026103.
  » https://doi.org/10.32604/fdmp.2023.026103
• ^[25] XU, G., MA, T., FANG, Z., et al, “The evaluation method of particle clustering phenomena in RAP”, Applied Sciences, v. 9, n. 3, pp. 424, 2019. doi: https://doi.org/10.3390/app9030424.
  » https://doi.org/10.3390/app9030424
• ^[26] ZHANG, Y., CHU, X., ZHU, W., et al, “Effect of the original clustering degree of reclaimed asphalt pavement and design parameters on properties of recycled asphalt mixtures”, Journal of Materials in Civil Engineering, v. 37, n. 2, pp. 04024498, 2025. doi: https://doi.org/10.1061/JMCEE7.MTENG-18910.
  » https://doi.org/10.1061/JMCEE7.MTENG-18910

Publication Dates

• Publication in this collection
  30 Mar 2026
• Date of issue
  2026

History

• Received
  19 Oct 2025
• Accepted
  10 Feb 2026