To study different lateral restraints, different constituents of expansion agents, the influence of different steel ratios, and concrete creep properties, we carried out experiments with lateral restraint and without lateral restraint conditions separately on 12 specimens with the expansion agent content accounting for 4%, 8%, and 12% respectively. In addition, the creep tests were performed on specimens with different steel ratios of 0.0%, 3.8%, 6.6%, and 9.2%. The test results show that the lateral restraint improves the strength of the system (concrete-filled steel tubes) which resists further load after the concrete ultimate strength is surpassed and reduces the creep. The creep degree of the concrete-filled steel tube with lateral restraint is about 0.09–0.30 times smaller than that of the tube without lateral restraints. The creep degree of the concrete-filled steel tube increases as the steel ratio decreases. Creep tests with different amounts of expansion agent indicate that the creep degree of the concrete structure increases as expansion agent content decreases. To study the internal mechanism of the creep of concrete-filled steel tubes with different lateral restraints and different expansion agent concentrations, a microscopic pore structure test on the steel core concrete was conducted using the RapidAir457 pore structure instrument. Microscopic studies show that the air content and the length of the bubble chord of the laterally restrained core concrete are lower than those without lateral restraint core concrete. The amount of air content and the length of the bubble chord of core concrete specimens increase as the expansion agent content in the core concrete specimens decreases from 12% to 4%. Under the same external loading conditions, as steel ratio increases, the lateral restraint causes a further reduction of creep. The results of this study suggest that the creep of concrete can be reduced by selecting appropriate lateral restraint conditions and an optimal amount of expansion agent in the mix design of concrete for concrete-filled steel tubes.

In the built loess tunnel, a large number of diseases have been found during the operation period, especially when the tunnel basement is after water inflow. At present, a large number of high pressure jet grouting pile are used to reinforce the tunnel basement, but there are few studies on the reinforcement effect of high pressure jet grouting pile when the tunnel basement is after water inflow during the operation period. Due to the operation control of the expressway, it is difficult to carry out field research during the operation period. In this paper, by changing the physical parameters of the water inflow layer of the surrounding rock of the tunnel basement, numerical simulation is used to analyze the reinforcement effect of high pressure jet grouting pile on the tunnel basement under the loads, and field monitoring means are used to confirm the correctness of the three-dimensional model. The results show that the reinforcement of high pressure jet grouting pile can prevent the uplift caused by the softening of tunnel basement surrounding rock. The error of monitoring data and simulation results is also within a reasonable range.

Accurate classification of manufactured and natural sand, a key natural sand substitute, is vital for controlling concrete performance. This study proposes a machine - learning - based sand classification method. By extracting sand particles’ morphological features (length /width ratio × roundness), a multi - dimensional feature engineering system is built. An ensemble learning model (combining Gradient Boosting, XGBoost, LightGBM) is used to distinguish manufactured (class 0) and natural (class 1) sand. It achieves an F1 score of 0.6211 and an ROC - AUC of 0.7105. SHAP analysis shows the “Length/width ratio×Roundness” interaction feature most impacts results, aiding sand classification mechanism study. This method helps rapid construction sand identification and concrete quality control.