This study systematically compares the predictive performance of single machine learning (ML) models (KNN, Bayesian ridge regression, decision tree) and ensemble learning methods (bagging, boosting, stacking) for quantifying the rheological properties of mineral powder-modified asphalt, specifically the complex shear modulus (G*) and the phase angle (δ). We used two emulsifiers and three mineral powders for fabricating modified emulsified asphalt and conducting rheological property tests, respectively. Dynamic shear rheometer (DSR) test data were preprocessed using the local outlier factor (LOF) algorithm, followed by K-fold cross-validation (K = 5) and Bayesian optimization to tune model hyperparameters. This framework uniquely employs cross-validated predictions from base models as input features for the meta-learner, reducing information leakage and enhancing generalization. Traditional single ML models struggle to characterize accurately as a result, and an innovative stacking model was developed, integrating predictions from four heterogeneous base learners—KNN, decision tree (DT), random forest (RF), and XGBoost—with a Bayesian ridge regression meta-learner. Results demonstrate that ensemble models outperform single models significantly, with the stacking model achieving the highest accuracy (R² = 0.9727 for G* and R² = 0.9990 for δ). Shapley additive explanations (SHAP) analysis reveals temperature and mineral powder type as key factors, addressing the “black box” limitation of ML in materials science. This study validates the stacking model as a robust framework for optimizing asphalt mixture design, offering insights into material selection and pavement performance improvement.

The utilization of oil-rich reclaimed asphalt pavement fine aggregate (O-RAP), characterized by its high asphalt content, which has inherent compatibility with the high asphalt demand of stress-absorbing layer (SAL) mixtures, enables significant recycling rates, thereby promoting resource efficiency and a promising pathway for sustainable infrastructure development. This study provides a comprehensive evaluation of the rheological properties of recycled asphalt binder through dynamic shear rheometer and bending beam rheometer tests. Furthermore, the pavement performance of SAL mixtures was systematically assessed via fatigue testing, moisture susceptibility evaluation, rutting resistance analysis, and overlay testing. The results indicate that increasing the O-RAP content enhances the complex shear modulus while reducing the phase angle, suggesting improved stiffness but reduced flexibility of the binder. In SAL mixtures, higher O-RAP content was associated with decreased fatigue life, moisture stability, and low-temperature cracking resistance, yet it contributed to improved resistance to reflective cracking and high-temperature rutting. Pearson correlation analysis further revealed that the fatigue life of the binder exhibits a strong positive correlation with creep rate and significant negative correlations with creep stiffness modulus, high-temperature stability, and reflective cracking resistance. These findings underscore the viability of high-content O-RAP incorporation in SAL mixtures as a technically sound and environmentally sustainable strategy for low-carbon pavement construction, offering significant reductions in virgin material consumption and associated carbon emissions.

Emulsified asphalt has become a popular green material in modern road construction due to its environmentally friendly and energy-saving characteristics. The method of obtaining residues is crucial for accurately evaluating the performance of emulsified asphalt and ensuring the quality of road construction. The evaporated residues of 3% and 6% SBR modified emulsified asphalt and 6% SBS modified emulsified asphalt were obtained by the low-temperature evaporation (LTE) method and the direct heating (DH) method, respectively. The rheological properties of the modified asphalt residues were tested by the temperature frequency sweeping (TFS), multi-stress creep recovery (MSCR), and linear amplitude sweeping (LAS) tests using the dynamic shear rheometer (DSR), and their aging resistance in the form of carbonyl index was measured by the Fourier infrared spectroscopy (FTIR). The test results showed that, as compared to the DH method, the SBR modified emulsified asphalt residue obtained by the LTE method had higher elastic recovery rate, lower irrecoverable creep compliance, smaller accumulated performance deformation, longer fatigue life, and higher aging resistance, indicating that the LTE method using the low-temperature curing conditions (25°C, 24 h—60°C, 24 h) is more suitable for curing the SBR modified emulsified asphalt. However, the opposite trends of the performance indices except the aging resistance were observed for the SBS modified emulsified asphalt residue obtained from the LTE method, showing that the high-temperature curing condition (120°C, 30 min—163°C, 1 min) can give the SBS modified asphalt residue better rheological properties and higher fatigue resistance performance, but it also has the disadvantage of weakening the aging resistance. Therefore, in practical application, an appropriate curing temperature should be selected to condition the SBS modified emulsified asphalt for balancing its rheological performance and aging resistance.

Currently, research on the modification mechanisms of activated rubber/SBS (styrene–butadiene–styrene) composites and the microscopic processes involved remains limited. To investigate the impact of the rubber activation treatment combined with SBS modifier on asphalt modification, this study employs composite-modified asphalt formulations using either a conventional mix or activated rubber in conjunction with SBS. Infrared spectroscopy (IR) and scanning electron microscopy (SEM) were utilized to analyze the chemical components and microscopic morphology of the composite-modified asphalt following activation treatment. Microscopic analysis revealed that the asphalt stirred for 20 min has a characteristic peak with a wave number of 966 cm⁻¹, while the characteristic peak with a wave number of 700 cm⁻¹ is not obvious. That is, the asphalt sample contains the polybutadiene component and a reduced amount of the polystyrene component. Therefore, it can be inferred that the asphalt sample only contains activated rubber, along with less SBS modifier content. Traditional rubber undergoes significant expansion reactions during the mixing stage, but there are difficulties in degradation, which leave large particles and reduce the proportions of the lightweight asphalt components. However, active rubber and SBS mainly expand and degrade more completely during the shear stage, forming many micro-volume particles in asphalt. Additionally, frequency scanning and multiple creep recovery tests were conducted to evaluate the high-temperature rheological properties of the asphalt. The results indicate that activated rubber, doped at 20%, and SBS, doped at 2%, significantly enhance the high-temperature rheological properties of the composite-modified asphalt compared to base asphalt, exhibiting a 417.16% increase in the complex modulus at 64 °C and 1 Hz. Furthermore, these modifiers interact synergistically to improve modification efficiency.

High-volume utilization of steel slag in cementitious materials requires a proportioning method that accounts for coupled effects among multiple components. This study investigated a four-component system comprising steel slag (SS), blast furnace slag (BFS), ordinary Portland cement (OPC), and desulfurized gypsum (DG), using 17 formulations within a constrained four-component mixture design. Unlike previous studies focused mainly on ternary binders or empirical proportioning, this work combined four-component formulation screening with age-dependent hydration characterization in one SS-BFS-OPC-DG framework. The 7 d compressive strength was selected as the modeled response because it provided a consistent early-age basis for comparing all 17 formulations within a practical testing period, whereas flexural strength was retained as a complementary experimental response. The compressive-strength mixture model was significant (p = 0.0044), with R² = 0.692, adjusted R² = 0.589, predicted R² = 0.435, and adequate precision = 8.545. Its performance was evaluated using prediction indices, residual diagnostics, and an independent confirmation batch. The selected mass fractions of SS, BFS, OPC, and DG were 0.322, 0.429, 0.131, and 0.116, respectively. The confirmation batch achieved a 7 d compressive strength of 24.80 MPa, compared with the predicted value of 24.93 MPa. The model was therefore used only for formulation screening and interpolation within the investigated component ranges. The selected formulation reached compressive and flexural strengths of 32.6 and 7.6 MPa at 28 d, respectively. XRD, SEM, TG, and FTIR results collectively supported continued formation of hydration products and progressive matrix densification from 3 d to 28 d. The C-S-H gel index increased from 2.07% to 6.15%, while the number of large visible voids decreased and the continuity of the matrix improved. These results provide a constrained mixture-design route for screening high-solid-waste SS-BFS-OPC-DG binders.