Mechanical properties and normal-incidence sound absorption performance of crumb-rubber-modified porous asphalt mixtures

1 Introduction

Each year, more than one billion tires reach the end of their service life worldwide, creating a mounting environmental concern (Alsheyab et al., 2025). Landfills face capacity issues, while the risk of fires and toxic leachate contamination grows. Because landfilling and incineration cause lasting ecological harm, many regions have tightened regulations on these disposal methods (Li et al., 2025). This tightening increases the demand for alternative, sustainable ways to manage end-of-life tires. One promising option involves using crumb rubber in asphalt mixtures (He et al., 2023). This recycled rubber waste may also improve pavement durability.

Porous asphalt mixtures (PAMs)—also called open-graded friction courses (OGFC) or permeable friction courses—have found broad use in highway engineering (Sharma et al., 2024; Manrique-Sanchez et al., 2022). Their key feature lies in a macroporous structure that allows water to drain quickly from the surface and reduces noise caused by tire-pavement contact. The voids, making up roughly 18%–25% of the mixture’s volume, form an array of Helmholtz resonators (Alber et al., 2018). These resonators dissipate sound energy as heat through viscous effects inside the pores (Knabben et al., 2016). Such characteristics make PAMs especially appealing for busy urban roads and neighborhoods sensitive to noise, where wet-weather safety and quiet surroundings matter (Wu et al., 2023). However, compared to dense-graded asphalt, PAMs often show weaker mechanical performance (Mallick et al., 2000). These weaknesses limit PAM’s use in areas with harsh temperature swings and heavy rainfall, where early pavement deterioration can occur well before the expected service life.

Incorporating crumb rubber (CR) into asphalt binders offers a strategy that tackles the end-of-life tire waste problem while potentially improving the durability issues found in porous asphalt mixtures (Tian et al., 2025). Studies have found that binders modified with CR tend to have higher softening points and lower sensitivity to temperature changes (Xu et al., 2023). They also show better elastic recovery, which together may enhance resistance to rutting and extend fatigue life. When incorporated into asphalt through the wet process, CR swells through the absorption of lighter binder fractions and modifies the rheological and damping characteristics of the binder phase. The resulting rubberized binder may consequently influence both the mechanical response and the acoustic behavior of porous asphalt mixtures (Chai et al., 2020; Chen et al., 2025). Unlike rigid minerals, CR’s viscoelastic behavior helps dissipate both mechanical vibrations and acoustic energy through hysteresis. This interaction supports the two main goals of PAM: maintaining structural performance and enhancing acoustic energy absorption. Such a combination could help overcome the usual compromise between performance and durability that limits traditional porous pavements (Qian and Fan, 2020).

The literature remains divided on the dosage of crumb rubber (CR) that best suits PAM, and that uncertainty has slowed any move toward standard engineering practice. Shirini and Imaninasab (Shirini and Imaninasab, 2016) found that 10% CR, measured by binder mass, gave the most balanced results in porous asphalt, although higher contents reduced resilient modulus and permeability. Quan et al. (2022) proposed 4% CR by aggregate volume for optimal damping, and they stressed the role of particle-size distribution. Notably, most studies treat mechanical and acoustic properties separately, so they miss the trade-offs that arise when sound absorption and structural integrity are optimized at the same time. Few studies have examined the long-term aging of rubberized PAM in a systematic way (Zhang et al., 2025). That gap matters. Oxidative aging is the main degradation mechanism in surface layers, and it can alter both mechanical and acoustic behavior over the pavement’s service life (Kang et al., 2026). Without a clearer picture of these coupled effects, engineers still lack dependable guidance for designing rubberized PAM that will perform consistently over its intended lifespan.

Previous studies have generally examined the mechanical, moisture-related, or acoustic properties of rubberized porous asphalt separately, and the stability of the recommended CR dosage after laboratory aging has received comparatively limited attention. The contribution of the present study does not lie in introducing new laboratory test methods. Instead, it lies in integrating volumetric and drainage characteristics, mechanical durability, moisture resistance, aggregate-loss resistance, and normal-incidence sound absorption within a consistent wet-process CR dosage series. In addition, the performance ranking was evaluated before and after accelerated laboratory aging, and a multi-dimensional performance index with weight-sensitivity analysis was used to identify a balanced dosage range rather than the maximum of a single indicator.

Accordingly, this study evaluated the effects of CR dosage on the mechanical, volumetric, moisture-related, and normal-incidence sound absorption performance of OGFC-13 mixtures. Paired specimens were tested before and after accelerated laboratory aging at 85 °C for 120 h to compare their relative aging susceptibility. The objectives were to: (1) quantify the effects of CR dosage on high-temperature rutting resistance and low-temperature cracking resistance; (2) evaluate moisture resistance, pore connectivity, permeability, binder retention, and aggregate-loss resistance; (3) characterize frequency-dependent normal-incidence sound absorption; and (4) identify a balanced CR dosage range using a multi-dimensional CPI and weight-sensitivity analysis. The CPI weights represent an engineering evaluation scenario rather than universal theoretical constants.

2 Materials and methods

2.1 Materials

The base binder used in this study was PG 58–22 base asphalt produced by a local refinery in Shandong Province, China. Its comprehensive physical properties are presented in Table 1. This binder grade was selected for its wide applicability in temperate climate zones of China, balancing high-temperature rutting resistance and low-temperature cracking performance.

TABLE 1. Physical properties of PG 58–22 base asphalt.

We produced crumb rubber by mechanically grinding waste passenger car tires at ambient temperatures. After grinding, we sieved the rubber particles. The resulting gradation primarily featured the 0.6–2.36 mm fraction, as shown in Table 2. This particle-size range was selected to facilitate CR swelling while maintaining workable binder production and mixture preparation. Table 3 provides details on the key properties of the crumb rubber.

TABLE 2. Crumb rubber gradation.

TABLE 3. Physical properties of crumb rubber.

We used crushed basalt for both the coarse and fine aggregates. Its abrasion resistance and angularity make it well suited to porous asphalt, where skeleton stability matters. Table 4 summarizes the aggregate properties. Limestone mineral filler served to improve binder–aggregate adhesion.

TABLE 4. Physical properties of aggregates and mineral filler.

The mixture followed the OGFC-13 grading specified in JTG D50-2017, with a nominal maximum aggregate size of 13.2 mm. We chose this gradation to form a stable stone-on-stone skeleton and, at the same time, preserve the interconnected voids needed for drainage and noise reduction. Figure 1 presents the full gradation.

2.2 Mixture design and specimen preparation

2.2.1 Binder modification protocol

All CR-modified binders were prepared using the wet process. In this study, CR was used solely as a binder modifier and did not replace mineral aggregate. The CR dosage was defined using Equation 1:

C R d o s a g e % = m C R / m b a s e a s p h a l t × 100 % (1)

Where m C R is the mass of crumb rubber and m b a s e a s p h a l t is the mass of the base asphalt before CR addition. In the present study, CR was used solely as a wet-process binder modifier and did not replace mineral aggregate.

To prepare the rubberized binder, we first preheated the base asphalt to 135 °C in a stainless steel container for 2 h to remove moisture. Then, we gradually raised the temperature to 180 °C while continuously stirring at 500 rpm. Crumb rubber, preheated to 100 °C to avoid thermal shock and clumping, was added in three equal portions over 10 min. This approach ensured uniform dispersion. We sheared the mixture at 3,500 rpm for 30 min, keeping the temperature between 175 and 185 °C. This condition allowed the crumb rubber to swell properly without causing significant devulcanization or thermal degradation of the asphalt. After shearing, we continued stirring at 500 rpm for another 30 min at 170 °C to eliminate air bubbles and ensure the rubber particles fully swelled.

2.2.2 Marshall mix design

Five CR dosages, namely, 0%, 10%, 15%, 20%, and 25% by mass of the base asphalt, were investigated. The corresponding mixtures were designated R-0, R-10, R-15, R-20, and R-25, respectively. For example, R-15 denotes a mixture prepared using a wet-process binder containing 15% CR by mass of the base asphalt. Standard Marshall specimens, measuring φ101.6 mm × 63.5 mm, were prepared using a Marshall compactor with 50 blows per side. The final OBC values for mixtures R-0, R-10, R-15, R-20, and R-25 were established at 4.8%, 5.0%, 5.2%, 5.4%, and 5.6%, respectively, indicating the increased binder demand associated with rubberized mixtures.

2.2.3 Specimen preparation for performance testing

For performance testing, standard Marshall specimens were prepared as previously described for water stability and indirect tensile strength tests. Additionally, rutting slabs measuring 300 mm × 300 mm × 50 mm were fabricated using a wheel rolling compactor, with 12 passes per side to achieve the target air void content of 18%–22%. Bending beams sized 250 mm × 30 mm × 35 mm were cut from the rutting slabs using a diamond saw for low-temperature bending tests. Furthermore, impedance tube specimens with a diameter of 100 mm and a height of 50 mm were cored from the rutting slabs for sound absorption testing. All specimens were allowed to cool at room temperature for 24 h before demolding and subsequent testing.

2.2.4 Accelerated laboratory aging procedure

Compacted specimens were subjected to accelerated laboratory aging to compare the relative oxidative-aging susceptibility of the mixtures. After demolding, the specimens were placed horizontally on wire-mesh shelves in a forced-draft oven and conditioned at 85 ± 1^∘C for 120 h. The specimens were positioned at least 5 cm apart and were inverted every 24 h to promote relatively uniform exposure. After conditioning, the oven was switched off, and the specimens were allowed to cool gradually to room temperature before testing.

The adopted condition was used as a comparative laboratory exposure and was not regarded as a direct or universal equivalent of a specific field-service period. Field aging is additionally influenced by climatic temperature history, ultraviolet radiation, oxygen transport, rainfall, traffic loading, pavement depth, and construction quality. Therefore, the results describe relative performance retention under the adopted laboratory condition rather than actual service-life equivalence (Abouelsaad et al., 2024).

2.3 Testing methods

2.3.1 Volumetric and functional characterization

Total air void content (V a), connected air void content (V c), voids in mineral aggregate (VMA), permeability coefficient (k), calculated average binder-film thickness (δ), draindown loss (D), and Cantabro loss (CL) were determined for each mixture. Three independently prepared specimens were used for the volumetric, permeability, and draindown measurements, and six specimens were used for the Cantabro test. The results were expressed as the mean ± standard deviation.

Total air void content and VMA were calculated using the measured bulk specific gravity and theoretical maximum specific gravity of the mixtures. Connected air void content was determined using a vacuum-saturation method, and the connected-air-void ratio was calculated using Equation 2:where R c is the connected-air-void ratio (%), V c is the connected air void content (%), and V a is the total air void content (%).

R c = V a V c × 100 % (2)

The permeability coefficient was measured using a constant-head permeameter. Binder-film thickness was estimated from the effective binder volume and the calculated aggregate surface area. Draindown loss was determined from the mass of binder and fine material collected after high-temperature conditioning of the loose mixture. Cantabro loss was calculated from the mass loss of compacted specimens after abrasion in a Los Angeles machine without steel balls.

2.3.2 High-temperature rutting resistance

Dynamic stability (DS) was evaluated using the wheel tracking test in accordance with JTG E20 T0719. Prior to testing, the rutting slabs were conditioned at 60 °C ± 1 °C for 5 h. A solid rubber wheel, applied at a contact pressure of 0.7 ± 0.05 MPa, was reciprocated across the specimen at a frequency of 42 cycles per minute for 60 min. Rut depth was recorded at 1-min intervals using a linear variable differential transformer (LVDT).

2.3.3 Low-temperature cracking resistance

Three-point bending tests were performed at −10 °C ± 0.5 °C in accordance with JTG E20 T0715. Prior to testing, beam specimens were conditioned in a low-temperature chamber for a minimum of 4 h. The test was conducted with a span length of 200 mm and a loading rate of 50 mm/min. Load and mid-span deflection were recorded continuously until failure of the specimen occurred.

2.3.4 Water stability

Water stability was assessed through two complementary tests: residual Marshall stability and freeze-thaw splitting strength ratio (TSR).

For the residual Marshall stability test, Marshall specimens were immersed in a water bath maintained at 60 °C for 48 h, after which their Marshall stability was evaluated. The residual stability was determined by calculating the ratio of the stability measured after immersion to the initial stability.

In the freeze-thaw splitting test, two sets of Marshall specimens were prepared. One set served as the control and was tested for indirect tensile strength (ITS) at 25 °C. The second set underwent a freeze-thaw cycle, which involved vacuum saturation at a pressure of 98.3–98.7 kPa for 15 min, followed by freezing at −18 °C ± 2 °C for 16 h. This was followed by thawing in a 60 °C water bath for 24 h. The TSR was then calculated as the ratio of ITS after freeze-thaw conditioning to the control ITS.

2.3.5 Sound absorption performance

Sound absorption coefficients were measured using a transfer function impedance tube (BSWA SW422) in accordance with GB/T 18696.1. The tube had an internal diameter of 100 mm and a test frequency range of 100–2000 Hz, which encompasses the dominant frequency range of tire-pavement interaction noise (400–1,600 Hz). Cylindrical specimens (φ100 mm × 50 mm) were carefully prepared to ensure a tight fit within the impedance tube without any air gaps. Each specimen was tested three times, and the frequency-dependent absorption coefficient was averaged. The impedance-tube results characterize the normal-incidence sound absorption of laboratory specimens. They are used in this study as comparative acoustic indicators and should not be interpreted as direct measurements of field tire–pavement noise reduction.

Two key indicators were extracted for comparative analysis: the peak absorption coefficient, defined as the maximum absorption coefficient over the entire frequency range, and the average sound absorption coefficient (SAC), defined as the mean absorption coefficient within the 400–1,600 Hz range. It should be emphasized that the measured absorption coefficients were obtained under normal-incidence laboratory conditions. Although these indicators are relevant to the acoustic energy dissipation capacity of porous mixtures, they cannot be directly converted into field noise-reduction values in decibels. Field validation using controlled pass-by, close-proximity, or on-board sound-intensity measurements is required before drawing conclusions regarding actual tire–pavement noise reduction.

2.4 Statistical analysis and comprehensive performance evaluation

All experimental results are presented as the mean ± standard deviation. Unless otherwise stated, three independently prepared specimens were tested for each mixture and aging condition. The effects of crumb rubber (CR) dosage and laboratory aging on each performance indicator were evaluated using two-way analysis of variance, with CR dosage and aging condition treated as fixed factors. When a statistically significant effect was identified, Tukey’s honestly significant difference test was used for pairwise comparisons. A difference was considered statistically significant when the probability value was less than 0.05 (p < 0.05).

Because the investigated performance indicators had different units and numerical ranges, min–max normalization was used to transform all indicators into dimensionless values ranging from 0 to 1. The normalized value of a benefit-type indicator, for which a higher measured value represents better performance, was calculated as:

Z i , c , j = X i , c , j ‐ X j , ⁡ min X j , ⁡ max ‐ X j , ⁡ min (3)

The normalized value of a cost-type indicator, for which a lower measured value represents better performance, was calculated as:

Z i , c , j = X j , ⁡ max ‐ X i , c , j X j , ⁡ max ‐ X j , ⁡ min (4)

In Equations 3, 4, Z i , c , j is the normalized value of performance indicator j for mixture i under condition c; X i , c , j is the corresponding measured value; X j , ⁡ max and X j , ⁡ min are, respectively, the maximum and minimum measured values of indicator j among all five mixtures under both unaged and laboratory-aged conditions. The subscript i denotes the mixture type, namely, R-0, R-10, R-15, R-20, or R-25; the subscript c denotes the aging condition, where u represents the unaged condition and a represents the laboratory-aged condition; and the subscript j denotes the performance indicator. A superscript asterisk (^∗) is used hereafter to denote a normalized value.

Dynamic stability (DS), flexural strength (R B), maximum flexural strain (ε B), residual Marshall stability (M S 0), freeze–thaw splitting strength ratio (TSR), peak sound absorption coefficient (α max), and average sound absorption coefficient (SAC) were treated as benefit-type indicators. The low-temperature stiffness modulus (S B) was treated as a cost-type indicator because a lower stiffness modulus generally indicates greater low-temperature deformation and stress-relaxation capacity.

To reduce the influence of differences in the number of indicators assigned to each performance category, the normalized indicators were grouped into four performance dimensions: high-temperature performance, low-temperature performance, water stability, and normal-incidence sound absorption performance.

The high-temperature performance score was calculated using Equation 5:where H i is the high-temperature performance score of mixture i; D S i , u * is the normalized dynamic stability of mixture i under the unaged condition; and D S i , a * is the normalized dynamic stability of the same mixture after laboratory aging.

H i = D S i , u * + D S i , a * 2 (5)

The low-temperature performance score was calculated using Equation 6:where Li is the low-temperature performance score of mixture i; R B , i , u * and R B , i , a * are the normalized flexural strengths before and after laboratory aging, respectively; ε B , i , u * and ε B , i , a * are the corresponding normalized maximum flexural strains; and S B , i , u * and S B , i , a * are the normalized low-temperature stiffness moduli. Because S B was normalized using Equation 4, a higher value of S B * represents better low-temperature cracking resistance.

L i = R B , i , u * + ε B , i , u * + S B , i , u * + R B , i , a * + ε B , i , a * + S B , i , a * 3 (6)

The water-stability score was calculated using Equation 7:where W i is the water-stability score of mixture i; M S 0 , i , u * and M S 0 , i , a * are the normalized residual Marshall stability values before and after laboratory aging, respectively; and T S R i , u * and T S R i , a * are the corresponding normalized freeze-thaw splitting strength ratios.

W i = M S 0 , i , u * + T S R i , u * + M S 0 , i , a * + T S R i , a * 4 (7)

The normal-incidence sound-absorption score was calculated using Equation 8:where A i is the normal-incidence sound absorption score of mixture i; α max ⁡ , i , u * and α max ⁡ , i , a * are the normalized peak sound absorption coefficients before and after laboratory aging, respectively; and S A C i , u * and S A C i , u , a * are the corresponding normalized average sound absorption coefficients within the frequency range of 400–1,600 Hz.

A i = α max ⁡ , i , u * + S A C i , u * + α max ⁡ , i , a * + S A C i , u , a * 4 (8)

In addition to the normalized performance dimensions, an aging-durability index was introduced to quantify the proportion of each property retained after laboratory aging. For the seven benefit-type indicators, the retention ratio was calculated by dividing the aged value by the corresponding unaged value. For the low-temperature stiffness modulus, the unaged value was divided by the aged value because an aging-induced increase in stiffness represents deterioration rather than improvement. The aging-durability index was calculated using Equation 9:where D i is the aging-durability index of mixture i. The subscripts u and a denote the unaged and laboratory-aged values, respectively. A higher D i value indicates that a greater proportion of the initial performance was retained after laboratory aging.

D i = 1 8 D S i , a D S i , u + R B , i , a R B , i , u + ε B , i , a * ε B , i , u * + S B , i , u * S B , i , a * + M S 0 , i , a * M S 0 , i , u * + T S R i , a * T S R i , u * + α max ⁡ , i , a * α max ⁡ , i , u * + S A C i , u , a * S A C i , u * (9)

The comprehensive performance index of mixture i was calculated using Equation 10:where C P I i is the comprehensive performance index of mixture i; H i, L i, W i, A i, and D i are the high-temperature, low-temperature, water-stability, sound-absorption, and aging-durability scores, respectively; and w H, w L, w W, w A, and w D are their corresponding weights, which satisfy the normalization constraint given in Equation 11:

C P I i = 100 · w H H i + w L L i + w W W i + w A A i + w D D i (10)

w H + w L + w W + w A + w D = 1 (11)

For the baseline engineering scenario, w H = 0.45, w L = 0.15, w W = 0.15, w A = 0.10, and w D = 0.15. Therefore, the comprehensive performance index under the baseline engineering scenario was calculated using Equation 12:

C P I i = 100 · 0.45 H i + 0.15 L i + 0.15 W i + 0.10 A i + 0.15 D i (12)

The weights represent an engineering decision scenario for a heavy-traffic OGFC surface layer and are not unique theoretical constants. Project-specific traffic, climate, moisture, and acoustic requirements may justify alternative weighting schemes. Low-temperature performance, water stability, and aging durability were each assigned a weight of 0.15, while normal-incidence sound absorption was assigned a weight of 0.10. These weights represent an engineering decision scenario rather than unique or universal theoretical constants. For projects with different traffic levels, climates, moisture conditions, or acoustic requirements, the weights may be adjusted accordingly.

A weight-sensitivity analysis was conducted to evaluate whether the recommended CR dosage depended strongly on the selected weighting scheme. Each of the five baseline weights was independently increased and decreased by 20%. When one weight was adjusted, the other four weights were proportionally rescaled so that the sum of all weights remained equal to 1.0. The resulting C P I i values and mixture rankings were compared with those obtained using the baseline weights. A dosage recommendation was considered relatively robust when the corresponding mixture remained first or second in all alternative weighting scenarios.

Pearson correlation coefficients were also calculated to describe the linear relationships among CR dosage and the investigated performance indicators. Because only five CR dosage levels were included, the correlation results were interpreted as exploratory relationships and were not considered evidence of causality. Polynomial regression was used only to describe the general dosage-dependent trend, and the fitted peak was treated as an approximate engineering range rather than an exact universally applicable optimum.

3 Results and discussion

3.1 Volumetric and functional characteristics

Table 5 summarizes the volumetric and functional characteristics of the investigated mixtures. As the CR dosage increased from 0% to 25%, the total air void content decreased from 21.4% to 19.3%, while the connected air void content decreased from 18.8% to 16.3%. All mixtures remained within the target air-void range of 18%–22%. The permeability coefficient decreased correspondingly from 0.36 to 0.25 cm/s, indicating a gradual reduction in pore connectivity at higher CR dosages.

TABLE 5. Volumetric and functional characteristics of the investigated mixtures.

V a = total air void content; V c = connected air void content; Rc = connected-air-void ratio; VMA = voids in mineral aggregate; k = permeability coefficient; δ = calculated average binder-film thickness; D = draindown loss; and CL = Cantabro loss. Each measured value should be reported as the mean ± standard deviation based on the original replicate results.

VMA increased slightly from 28.6% to 29.8%, and the calculated average binder-film thickness increased from 9.0 to 14.5 μ m. Draindown loss decreased from 0.22% for R-0–0.06% for R-25, indicating improved binder retention. Cantabro loss decreased from 18.7% for R-0 to a minimum of 9.8% for R-15 and subsequently increased to 13.1% for R-25.

Overall, R-15 maintained adequate pore connectivity and permeability while exhibiting low draindown and the minimum Cantabro loss. R-20 provided slightly higher sound absorption but showed lower permeability and higher Cantabro loss than R-15, indicating a different balance between acoustic and structural functions.

3.2 High-temperature rutting resistance

Figure 2 presents the dynamic stability (DS) of the unaged and aged specimens at different CR dosages. The unaged control mixture (R-0) exhibited a DS of 1,180 cycles/mm, which increased markedly to 3,350 cycles/mm at 10% CR and reached a maximum of 5,450 cycles/mm at 15% CR, corresponding to an improvement of 362%. This enhancement cannot be attributed solely to binder stiffening; rather, it is primarily associated with the formation of a three-dimensional elastomeric network within the asphalt mastic (Khodaii and Mehrara, 2009). As CR particles absorb aromatic oils from the binder, they swell to approximately 2–3 times their original volume, generating interlocking rubber domains that function as elastic springs within the viscous asphalt matrix.

When the CR content exceeded 15%, DS declined to 4,980 cycles/mm at 20% CR and further decreased to 3,680 cycles/mm at 25% CR, indicating a transition from beneficial modification to adverse overdosage. This reduction is attributed to the limited capacity of the asphalt binder to accommodate excessive swollen rubber particles. At CR contents above 15%, the binder phase becomes progressively diluted, and insufficient asphalt remains to adequately coat both the aggregate surfaces and the rubber particles. As a result, weak rubber-rubber interfaces and localized binder starvation develop, thereby compromising the cohesive strength of the mastic. Nevertheless, even at 25% CR, the DS remained 212% higher than that of the control mixture, confirming the substantial potential of CR modification for improving high-temperature performance.

Aging caused a consistent reduction in DS across all mixtures, with the decay rate increasing monotonically from 30.5% for R-0–38.0% for R-25. The aged R-15 mixture retained a DS of 3,520 cycles/mm, which was still 329% higher than that of the aged control and exceeded the 3,000 cycles/mm threshold specified for heavy-traffic highways in China. Figure 3 shows the polynomial trend lines for the unaged and aged DS results, indicating optimal CR dosages of 16.2% and 15.4%, respectively. The narrowing gap between the unaged and aged curves at higher CR contents suggests that rubberized mixtures become increasingly susceptible to oxidative hardening. Oxidative aging has competing effects on the high-temperature response of asphalt mixtures. At the binder scale, oxidation increases viscosity and stiffness, which would generally be expected to reduce permanent deformation. At the mixture scale, however, prolonged oxidative exposure may reduce binder ductility and weaken the ability of the mastic bridges and binder–aggregate interfaces to accommodate repeated wheel loading, particularly in a high-void OGFC skeleton. The observed decrease in DS therefore represents the net mixture-level response rather than the effect of binder hardening alone. In the present mixtures, aging-induced embrittlement and local structural deterioration may have outweighed the beneficial effect of increased binder viscosity. Because binder rheology and post-aging microstructure were not directly measured, this explanation should be regarded as a plausible interpretation rather than direct mechanistic verification.

3.3 Low-temperature cracking resistance

The results of the low-temperature bending tests are summarized in Figure 4. Flexural strength (R _*B*) exhibited a steady increase from 4.52 MPa for the control mixture (R-0) to 8.42 MPa at 20% CR, before experiencing a slight decline to 7.68 MPa at 25% CR. Maximum flexural strain (ε _*B*) followed a similar trend, peaking at 3,950 με for the R-20 mixture. In contrast, the stiffness modulus (S _*B*) decreased from 592 MPa for R-0–465 MPa for R-20, indicating that CR contributes ductility and stress relaxation capacity to the binder film.

This simultaneous enhancement of strength and ductility is characteristic of elastomeric modifiers and sets CR apart from rigid fillers. At low temperatures, asphalt binder behaves as a brittle, glassy material, with cracking initiating at stress concentrations around aggregate particles. CR particles disrupt this brittle failure mechanism by acting as flexible inclusions capable of undergoing significant elastic deformations without fracturing. As a crack approaches a rubber particle, the particle absorbs energy at the crack tip through elastic stretching, thereby blunting the crack and preventing catastrophic failure. Furthermore, the rubber-binder interface functions as a debonding zone that dissipates energy through controlled interfacial sliding, further enhancing fracture toughness. The peak performance observed at 20% CR corresponds to the optimal dispersion of rubber particles, potentially reducing the severity of local stress concentrations within the mixture with a nearby rubber inclusion.

The decline in performance at 25% CR aligns with the trend observed in dynamic stability, suggesting that overdosage compromises the structural integrity of the mastic-aggregate interface. At this elevated content, CR particles may become less uniformly dispersed, forming large weak zones within the mixture that serve as preferential sites for crack initiation.

Aging resulted in a reduction of R _*B* and ε _*B* by 15–22% across all mixtures, while S _*B* increased by 25–35% due to oxidation and stiffening of the binder. Nevertheless, the relative performance advantage of CR-modified mixtures was maintained after aging. The aged R-15 mixture exhibited a favorable balance with R B at 6.88 MPa, ε B at 2,950 με, and S B at 638 MPa, satisfying typical low-temperature design thresholds for cold-climate regions (ε B ≥ 2,500 μ ε). Notably, the aged R-20 mixture demonstrated an ε B of 3,080 μ ε, which was higher than the unaged control value of 2,880 μ ε, indicating that CR provides durable low-temperature cracking resistance even following oxidative aging.

3.4 Water stability

Figure 5 presents the residual Marshall stability (MS₀) and TSR results. The unaged MS₀ increased from 82.5% for the control mixture (R-0) to 93.2% for R-20, indicating enhanced binder-aggregate adhesion mediated by the polar components released during CR digestion. During high-temperature mixing, mechanical shear and thermal energy induce partial desulfurization of the CR surface, generating thiol and sulfoxide functional groups. These polar groups can form strong hydrogen bonds with hydroxyl groups on the basalt aggregate surface, producing interfacial interactions that are substantially more resistant to water-induced displacement than the physical adhesion of unmodified asphalt.

The TSR increased to 90.5% at 15% CR and reached its highest level at 20% CR. The TSR remained above 88% at 25% CR, indicating that the moisture-damage resistance of the CR-modified mixtures remained favorable over the investigated dosage range. The difference in the optimal dosages for MS₀ and TSR reflects an important distinction: MS₀ measures the residual strength of the mixture after prolonged water immersion and is therefore governed primarily by the cohesive strength of the binder mastic. In contrast, TSR is a more sensitive indicator of interfacial adhesion because the freeze-thaw cycle generates hydraulic pressure that preferentially acts on the binder-aggregate interface. At CR contents above 15%, although the cohesive strength of the mastic continues to improve, excessive rubber particles begin to compete with the aggregate for binder coverage, resulting in thinner asphalt films on some aggregate surfaces and a slight reduction in interfacial adhesion.

The similar trends of M S 0 and TSR indicate that moderate CR modification improved both mastic cohesion and resistance to moisture-induced interfacial damage. The slight decline observed at 25% CR may be associated with less uniform binder distribution and reduced pore connectivity. The superior water stability of CR-modified mixtures is further supported by the hydrophobic nature of rubber particles, which form a water-repellent barrier around aggregate surfaces and impede water ingress into the binder-aggregate interface (Ren et al., 2021). The improved moisture resistance of the CR-modified mixtures was also associated with the increase in calculated average binder-film thickness. As shown in Table 5, the binder-film thickness increased from 8.8 μ m for R-0 to 11.5 μ m for R-15. A thicker and more continuous binder film may delay water penetration and reduce the direct exposure of the aggregate surface to moisture.

However, the binder-film thickness continued to increase to 13.1 μ m for R-20 and 14.3 μ m for R-25, whereas the moisture and Cantabro performance did not improve proportionally. Therefore, binder-film thickness alone cannot explain the observed water stability. Binder–aggregate adhesion, binder distribution, pore connectivity, and the homogeneity of the rubberized mastic also contributed to the moisture-damage response.

3.5 Sound absorption performance

Figure 6 illustrates the frequency-dependent sound absorption coefficient (α) for unaged specimens. All mixtures displayed a characteristic broadband absorption profile, with α increasing from 0.08–0.11 at 125 Hz to 0.71–0.79 at 1,600 Hz, followed by a decline at 2000 Hz. This pattern corresponds to the resonant frequency of the Helmholtz resonators formed by the interconnected voids within the porous structure. Notably, the peak frequency shifted slightly from 1,600 Hz for the control mixture (R-0) to 1,500 Hz for R-25, reflecting changes in the effective pore throat diameter due to partial occlusion by CR particles.

Figure 7 presents quantitative data on the peak absorption coefficient (α max) and the average absorption coefficient (SAC) across the 400–1,600 Hz range, which is the dominant frequency range for tire-pavement interaction noise. Both metrics peaked at 20% CR, with αmax reaching 0.79 and SAC at 0.50, while the R-15 mixture exhibited close values of α max = 0.77 and SAC = 0.48. The control mixture (R-0) achieved α max = 0.71 and SAC = 0.44.

The sound absorption behavior was governed by the combined effects of pore structure and material damping. Although the total and connected air void contents decreased moderately with increasing CR dosage, R-15 and R-20 exhibited higher sound absorption coefficients than R-0. This result indicates that sound absorption was not controlled by porosity alone.

The viscoelastic rubberized binder may have enhanced acoustic-energy dissipation through hysteretic deformation and internal friction (Faßbender and Oeser, 2020). Meanwhile, the moderate narrowing of the connected pore channels may have increased airflow-path tortuosity and viscous interaction between the oscillating air and the pore walls. These effects may have compensated for the moderate reduction in pore volume at CR dosages up to 20%.

At 25% CR, the more pronounced reductions in connected air void content and permeability may have limited further improvement in sound absorption. Because pore-size distribution, airflow resistivity, and three-dimensional pore morphology were not directly measured, these explanations should be regarded as mechanism-based interpretations rather than direct microstructural verification.

Aging reduced α max by approximately 9%–11% and SAC by approximately 10%–12%. This reduction may be associated with changes in the effective pore structure and a decrease in the viscoelastic damping capacity of the aged binder. Oxidative aging increased binder stiffness and may have reduced the capacity of the pore walls to dissipate acoustic energy through cyclic deformation. However, because the post-aging pore morphology and airflow resistivity were not directly measured, the relative contributions of pore-structure change and binder stiffening could not be independently quantified. Nevertheless, the aged R-15 mixture retained an SAC of 0.43, which was close to the unaged R-0 value of 0.44, indicating relatively stable normal-incidence sound absorption under the adopted laboratory aging condition.

3.6 Aging decay and performance degradation

Figure 8 compares the aging decay rates across six performance indicators. Dynamic stability (DS) exhibited the greatest absolute decay, ranging from 30.5% to 38.0%, followed by maximum flexural strain (ε _*B*), which decreased by 15.3–24.3%, and the average sound absorption coefficient (SAC), which declined by 10.9–12.5%. In contrast, the freeze-thaw splitting strength ratio (TSR) and residual Marshall stability (MS₀) showed the lowest decay rates, at 4.7–5.5% and 4.8–5.9%, respectively, suggesting that water stability is relatively insensitive to thermal oxidation.

This hierarchy of decay rates reflects the distinct mechanisms through which aging influences each performance attribute. High-temperature rutting resistance is governed primarily by the viscoelastic properties of the binder, which are most strongly altered by oxidative aging. The formation of carbonyl and sulfoxide groups increases the molecular weight of the asphalt, making it more viscous and less elastic. Low-temperature ductility depends on the ability of the binder to flow and relax stress, both of which are substantially reduced by oxidative hardening. Sound absorption is also affected by aging through two pathways: degradation of the pore structure and reduction in the damping capacity of the material, leading to moderate performance loss. By comparison, although aging reduces the cohesive strength of the binder, it also increases binder viscosity and thereby improves adhesion to the aggregate surface. These two opposing effects largely offset one another, resulting in only minor net changes in water stability.

Notably, the decay rate of DS increased monotonically with CR content, indicating that while CR improves initial rutting resistance, it also makes the mixture more susceptible to oxidative embrittlement. This trade-off should be carefully considered in life-cycle design, especially for pavements exposed to high temperatures and elevated ozone levels, where aging is accelerated. For such applications, a slightly lower CR dosage, approximately 12–14%, may be more appropriate to achieve a better balance between initial performance and long-term durability.

3.7 Correlation analysis

Figure 9 presents the Pearson correlation matrix for unaged performance indicators. Rubber content demonstrated strong positive correlations with flexural strength (R _*B*) (r = 0.93), maximum flexural strain (ε _*B*) (r = 0.92), and peak absorption coefficient (αmax) (r = 0.89), thereby confirming the beneficial role of crumb rubber (CR) in enhancing both low-temperature and acoustic performance. These high correlation coefficients indicate that these three attributes share a common mechanistic origin: the elastomeric nature of CR particles, which improves energy dissipation for low-temperature cracking resistance and noise reduction, as well as stress transfer for flexural strength.

Conversely, dynamic stability (DS) exhibited a moderate negative correlation with SAC (r = −0.42) and αmax (r = −0.38), revealing an inherent conflict between structural stiffness and acoustic porosity. This trade-off arises because high-temperature rutting resistance necessitates a dense, stiff mastic with high cohesive strength, while effective sound absorption requires an open, porous structure with significant damping capacity. As CR content exceeds 15%, the continued enhancement in acoustic performance comes at the expense of reduced rutting resistance, underscoring the necessity for multi-objective optimization rather than the maximization of a single parameter.

The strong correlation between TSR and MS₀ (r = 0.96) validates their mutual representation of water stability. Both indicators are critically dependent on the quality of the binder-aggregate interface and the cohesive strength of the mastic, leading them to vary in tandem. This high collinearity suggests that future studies may consider omitting one of these indicators without significant loss of information, thereby reducing testing time and costs.

These correlations inform the weighting scheme in multi-criteria optimization. Indicators with high collinearity, such as MS₀ and TSR, can be consolidated, while orthogonal objectives, such as DS and SAC, necessitate a balanced trade-off analysis to ensure that no critical performance attribute is compromised.

3.8 Comprehensive performance evaluation

Figures 10, 11 illustrate the normalized comprehensive performance profiles for both unaged and aged mixtures. For the unaged specimens, R-15 and R-20 exhibited the strongest combined rutting-resistance and normal-incidence sound-absorption performance, while R-0 and R-10 lagged behind in these areas. Additionally, water stability and freeze-thaw resistance improved progressively with increasing CR content, peaking at 20%. In the case of aged specimens, the overall performance envelope contracted uniformly; however, R-15 maintained the most balanced profile, with all axes remaining above 60% of the maximum values observed in unaged mixtures.

The calculated dimension scores and CPI values are summarized in Table 6. R-15 achieved the highest baseline CPI of 76.42, followed closely by R-20 with a CPI of 76.22. R-25, R-10, and R-0 exhibited substantially lower CPI values of 58.72, 50.64, and 21.40, respectively.

TABLE 6. Normalized performance dimensions, aging-durability index, CPI values, and rankings of the investigated mixtures.

The CPI values were calculated using the mean values of the experimental indicators.

The difference between R-15 and R-20 was small because the two mixtures exhibited different performance advantages. R-15 showed the highest dynamic stability and slightly better overall aging retention, whereas R-20 exhibited superior low-temperature deformation capacity, water stability, and normal-incidence sound absorption. Therefore, the CPI result should not be interpreted as evidence that 15% CR was universally superior to 20% CR for every application.

The weight-sensitivity analysis further showed that the first-ranked mixture alternated between R-15 and R-20 when the engineering priorities were changed. R-15 ranked first in six of the ten alternative weighting scenarios, whereas R-20 ranked first in the remaining four scenarios. No other mixture achieved the highest rank. These results identify 15%–20% CR as the stable high-performance range rather than a single universally optimal dosage.

The small CPI difference between R-15 and R-20 reflects their different performance advantages. R-15 exhibited the highest dynamic stability and the lowest Cantabro loss, whereas R-20 provided the highest low-temperature deformation capacity, water stability, and normal-incidence sound absorption. Therefore, the CPI result should not be interpreted as evidence that 15% CR is universally superior to 20% CR. The stable high-performance range was 15%–20% CR. Under the baseline heavy-traffic weighting scenario, approximately 15% CR is recommended; approximately 20% may be more appropriate where low-temperature, moisture, or acoustic performance is assigned greater importance.

4 Conclusion

This study evaluated the mechanical, volumetric, moisture-related, normal-incidence sound absorption, and laboratory-aging performance of wet-process CR-modified OGFC-13 mixtures. The following conclusions were obtained.

• CR modification substantially improved the performance of OGFC-13 mixtures, but the optimum dosage depended on the evaluated indicator. R-15 exhibited the highest unaged dynamic stability and the lowest Cantabro loss, whereas R-20 exhibited the highest low-temperature deformation capacity, water stability, and normal-incidence sound absorption.
• Increasing CR dosage reduced total and connected air void contents and permeability while increasing calculated binder-film thickness. Moderate CR modification improved binder retention and aggregate-loss resistance, but the Cantabro loss increased again at dosages above 15%.
• Under the baseline heavy-traffic weighting scenario, R-15 and R-20 achieved CPI values of 76.42 and 76.22, respectively. Weight-sensitivity analysis showed that the first-ranked mixture alternated between R-15 and R-20. Therefore, 15%–20% CR should be regarded as the stable high-performance range rather than a single universally optimal dosage. Approximately 15% is recommended when rutting resistance and aging durability are prioritized, whereas approximately 20% may be preferred for cold-region or acoustically sensitive applications.
• Under the adopted accelerated laboratory aging condition of 85 °C for 120 h, dynamic stability decreased by approximately 30%–38%, and average sound absorption decreased by approximately 10%–12%. These results describe relative performance retention under controlled laboratory exposure and should not be interpreted as a direct conversion to a specific field-service period.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Wensheng Wang, Jilin University, China

Reviewed by: Mingjun Hu, Dalian University of Technology, China

Qinglin Guo, Hebei University of Engineering, China