Open-pit haul roads are increasingly exposed to extreme climate conditions, while life-cycle carbon emissions are rarely incorporated into material selection in engineering practice. This study presents a fuzzy AHP-TOPSIS-based evaluation approach for mine-derived materials, in which extreme climate factors are integrated into the hierarchical weighting process and life-cycle carbon emissions are incorporated into the ranking procedure. The approach was applied to an open-pit haul road project at the Zijinshan Gold-Copper Mine in China to identify suitable pavement materials. Results under the SSP5-8.5 scenario indicated that mechanical strength and heavy metal control dominated decision priorities, while climate-related performance criteria, particularly stability and durability, also exhibit considerable importance. Life-cycle assessment results shown that polyvinyl alcohol-granulated blast-furnace slag stabilized blasting slag powder (PGBP) achieved the lowest carbon emissions due to reduced maintenance demand. The ranking consistently identified PGBP as the optimal alternative, and sensitivity analysis confirmed the robustness of the decision results under varying climate and carbon-weight scenarios. The proposed approach provided a practical and transferable decision-support tool for material selection in open-pit haul roads under complex environmental conditions.
Expansive soils pose significant challenges to infrastructure due to their high swelling potential, often requiring chemical stabilization with lime-based binders. However, conventional assessments rarely integrate short-term swell-control and environmental performance within a unified evaluation framework. At the same time, the environmental implications of nano eggshell lime for soil stabilization remain largely unexplored. This study evaluates the techno-environmental performance of four binders — nano eggshell lime, eggshell lime, calcitic lime, and dolomitic lime — using an integrated methodology combining laboratory free-swell tests and life cycle assessment. Environmental impacts were quantified using the ReCiPe 2016 (H) method at midpoint and endpoint levels, while technical efficiency was defined by the binder dosage required to reach the target free swell. To integrate both dimensions, a Techno-Environmental Efficiency Score (TEES) was developed based on the relative ranking of binder dosage and endpoint single-score performance. Results show that nano eggshell lime requires the lowest dosage (1.40%), representing reductions of up to 91% compared with the other binders, but exhibits moderate environmental performance due to chemical reagent production and energy-intensive processing. Eggshell lime exhibits the best environmental performance (1.30Pt) and the highest TEES value (5), followed by nano eggshell lime (TEES = 4), calcitic lime (TEES = 3), and dolomitic lime (TEES = 0). The findings show that waste origin and low binder dosage alone do not guarantee superior environmental performance, highlighting the need to evaluate material efficiency and production-related impacts together. The proposed framework demonstrates how integrating short-term swell-control performance with life cycle assessment can support more sustainable decision-making in soil stabilization practice
During the construction of cast-in-place piles in permafrost regions, the hydration heat of cement inevitably causes significant thermal disturbances, which adversely affect the early stability of pile foundations. Based on a bridge pile foundation project on the Qinghai-Tibet Plateau, this study investigates the thermal disturbance characteristics and refreezing law of pile foundations in warm permafrost regions, based on a combination of field experiments and numerical simulations. An active cooling system using pre-embedded acoustic pipes combined with circulating coolant was proposed, and its cooling effect was evaluated. Field results indicate that after concrete casting, the pile temperature peaks within approximately one day, while permafrost warming exhibits a pronounced lag, with a maximum thawing radius of about 1 m. The latent heat effect makes the refreezing time of the pile more than twice that of the soil 0.1 m outside it at the same depth during natural refreezing. The active cooling system significantly accelerates the refreezing process. The average temperature of the pile is reduced by 0.73 °C, with a maximum decrease of 1.1 °C. The numerical simulation further shows that reducing the inlet temperature of the circulating coolant, increasing the flow rate, or increasing the pipe diameter can shorten the refreezing time, and increasing the pipe diameter has the most significant effect on improving the refreezing uniformity. Overall, active cooling enhances cold energy storage, shortens the refreezing time, mitigates hydration heat disturbance, and ensures early thermal stability of pile foundations in permafrost regions.
The resilient modulus (Mr) of unbound granular base materials is a fundamental input parameter in mechanistic-empirical pavement design, yet its laboratory determination through repeated load triaxial (RLT) testing remains resource-intensive and inaccessible to many transportation agencies. This study presents an interpretable machine learning framework for Mr estimation using 82,682 laboratory measurements obtained from 603 Long-Term Pavement Performance (LTPP) test sections distributed across 40 U.S. states, representing one of the largest and most geographically diverse datasets employed for this purpose to date. Four machine learning algorithms, Random Forest, Extreme Gradient Boosting (XGBoost), Gradient Boosting, and Artificial Neural Network (ANN), were trained and evaluated under a section-based train-test split protocol specifically designed to prevent data leakage and produce methodologically sound generalization estimates. The Gradient Boosting model achieved the best generalization performance, yielding R2 = 0.775 and RMSE = 32.92 MPa on the held-out test set comprising 16,862 observations from 121 independent LTPP sections. SHapley Additive exPlanations (SHAP) analysis revealed that confining pressure (σ3) accounted for 50.3% of the total mean absolute prediction contribution, followed by bulk stress (θ, 16.4%) and climate zone (12.0%), providing mechanistically interpretable insights consistent with established constitutive geotechnical theory. Climate zone stratification demonstrated that models performed best in Wet-Freeze regions (R2 = 0.823, MAPE = 21.4%) and exhibited substantially higher errors in Dry-No-Freeze climates (MAPE = 36.0%), underscoring the need for region-specific calibration. An ablation study confirmed that structural and regional features contribute an independent R2 improvement of 0.027 beyond stress-state variables alone. This study is deliberately scoped to stress-state and structural features available directly from the LTPP RLT test protocol; integration of gradation and compaction state variables is reserved for future work. The proposed framework offers a practical, data-driven, and physically interpretable tool for preliminary Mr estimation that complements, rather than replaces, laboratory RLT testing, and establishes a methodological benchmark for large-scale, leakage-free ML applications in transportation geotechnics.
