Predictive modeling of mechanical properties in cold recycled asphalt mixtures enhanced with industrial byproducts

Emulsified cold recycled mixtures are primarily utilized as base layers within a pavement systems. Emulsion-recycled asphalt layers often encounter challenges such as raveling, stripping, weak initial strength, and extended curing periods. This study investigates the mechanical performance of emulsified cold recycled mixtures enhanced with cement kiln dust (CKD) and steel slag, compared to baseline mixtures and those incorporating 1 % and 2 % cement. The modified Marshall method was used to determine the optimal asphalt binder and water content, followed by ambient curing. Mechanical tests, including Marshall stability, indirect tensile strength, moisture susceptibility were conducted at 25°C, 5°C, and −5°C, with fatigue performance assessed at 25°C after 7 and 42 days. Machine learning models specifically Random Forest and XGBoost, were employed to predicte tensile strength and fatigue life, with XGBoost achieved an R² of approximately 0.95 and RMSE of approximately 0.02 for the tensile strength ratio, and R² of approximately 0.95 with MSE of approximately 15,000 for fatigue life. The results indicate that a blend of 3 % CKD and 3 % steel slag significantly improved mechanical properties, often outperforming mixtures with added cement, particularly after extended curing periods. Fatigue performance was superior at strains below 200 microstrain, aligning with typical base layer conditions, thus confirming the potential of CKD-steel slag as a sustainable additive.