During the construction of underwater shield tunnels (excavated using a slurry pressure balance shield machine), whether seawater (Sw) can be used to replace freshwater (Fw) in the preparation of cement–sodium silicate grout (CSG) has become a major concern in the engineering community. CSG is formed by mixing components A and B, where component A is a liquid prepared by mixing bentonite, cement, and water, and component B is a sodium silicate solution. In this paper, the CSG was prepared using Sw instead of part of Fw. The properties, including bleeding rate, initial and final setting time, gel time, compressive strength, and microscopic characteristics, were tested to investigate the influence of Sw on the performance of CSG and explore its impact mechanism. The results showed that when expanding bentonite with Sw, the bleeding rate of Component A exceeded 50%, failing to meet the engineering requirement of 10%. However, expanding bentonite with Fw, the seawater replacement ratio has almost no effect on Component A, with all values remaining below 10%. As the seawater replacement ratio increases, the setting time of CSG is significantly shortened. Although the inclusion of seawater results in a marginally lower 1-day strength for CSG, it notably boosts the strength at later ages. Specifically, at a 45% seawater replacement ratio, the 28-day strength showed a marked increase of 52% relative to the CSG without seawater. In the later stage of hydration, the positive effect of Cl⁻ in seawater, promoting the hydrolysis of C₃S and C₂S on strength, is significantly higher than the negative effect of sulfate ion erosion in seawater on strength. Therefore, seawater significantly increases the 28-day compressive strength of CSG. This study can provide reference and guidance for the application of seawater in the preparation of two-component grout for submarine shield tunnels.

Artificial ground freezing (AGF) is widely used for ground reinforcement and groundwater control during underground shield docking, but freezing-pipe failures can alter the thermal response of the frozen wall. This study examines an ultra-large-diameter shield-docking project in water-rich sandy strata using a three-dimensional transient heat-transfer model incorporating ice-water phase change and comparison with field temperature measurements. At 11 representative monitoring points, R² ranged from 0.923 to 0.993, and the overall relative error was 6.872%. Under normal freezing, frozen-wall closure occurred at approximately 45 d; the average-temperature criterion (≤−13 °C) and effective-thickness criterion (≥3.9 m) were reached at approximately 54 d and 64 d, respectively. Missing pipes delayed attainment of these criteria to 55–85 d and 84–129 d, depending on their number and spatial arrangement. During shutdown, the average-temperature rise remained below 0.88 °C within 48 h. In the 360 h sensitivity analysis, the rise reached 1.315–2.517 °C for local shutdowns and 4.089 °C for system-wide shutdown. Formation-stage failures were governed by local cooling deficits and neighboring-pipe compensation, whereas shutdown responses were dominated by progressive dissipation of stored cold energy. The results support stage-specific thermal assessment of AGF-assisted shield docking; mechanical stability and excavation safety require separate verification.