The rapid development of coastal cities has intensified land resource constraints and is leading to an increasing number of foundation pit projects near existing operational tunnels. This necessitates careful consideration of coastal excavation impacts on adjacent tunnels. Taking a foundation pit project in Qingdao as a case study, this paper investigates tunnel deformation through statistical analysis, numerical simulation, and field monitoring. By adjusting numerical model parameters, the research examines the influence of horizontal clearance distances, existing structure burial depths, and different retaining structure configurations on tunnel deformation, providing guidance for deformation control. Key findings include the following: (1) Statistical analysis reveals that tunnels in silty clay strata experience more significant excavation-induced deformation compared to those in silt strata, with relative positional relationships between pits and tunnels playing a critical role. (2) Numerical and monitoring results demonstrate that pit excavation induces tunnel displacement towards the excavation zone. Maximum lateral displacement reached 3.57 mm (simulated) and 4.79 mm (measured), while maximum vertical displacement was 3.11 mm (simulated) and 3.85 mm (measured), all within safety thresholds. (3) Sensitivity analysis shows that shallower tunnels exhibit more pronounced deformations. Increasing horizontal separation distance from 10 m to 25 m reduces deformation by one-third. However, adjusting diaphragm wall thickness and retaining structure embedment depth proves limited in deformation control, necessitating reinforcement measures on the tunnel side. These findings provide valuable references for protecting coastal silty clay stratum tunnels.

Water and mud inrush hazards pose significant threats to the safety, environmental stability, and resource efficiency of tunnel construction, representing a critical barrier to the development of sustainable transportation infrastructure. Misjudgment—especially missed detections of severe hazards—can lead to extensive geological disturbance, excessive energy consumption, and severe socio-environmental impacts. However, pre-trained large-scale models still face two major challenges when applied to tunnel hazard classification: limited labeled samples and the high cost associated with misclassifying severe hazards. This study proposes a sustainability-oriented intelligent classification framework that integrates a large-scale pre-trained model with multi-strategy data augmentation to accurately identify hazard levels during tunnel excavation. First, a Synthetic Minority Over-Sampling Technique (SMOTE)-based multi-strategy augmentation method is introduced to expand the training set, mitigate class imbalance, and enhance the model’s ability to recognize rare but critical hazard categories. Second, a deep feature extraction architecture built on the robustly optimized BERT pretraining approach (RoBERTa) is designed to strengthen semantic representation under small-sample conditions. Moreover, a hierarchical weighting mechanism is incorporated into the weighted cross-entropy loss to emphasize the identification of severe hazard levels, thereby ensuring zero missed detections. Experimental results demonstrate that the proposed method achieves an accuracy of 99.26%, representing a 27.96% improvement over the traditional SVM baseline. Importantly, the recall for severe hazards (Levels III and IV) reaches 100%, ensuring zero misjudgment of major hazards. By effectively reducing safety risks, minimizing environmental disruptions, and promoting resilient tunnel construction, this method provides strong support for sustainable and low-impact underground engineering practices.

In highly permeable strata, the rapid formation of a dense filter cake is critical to the safety of slurry-shield tunnelling beneath rivers and seas. However, conventional bentonite slurries often struggle to develop an effective cake under these conditions. To investigate the permeation behavior of bentonite slurries modified with graded sand in highly permeable strata, based on the Nanjing Heyan Road river-crossing tunnel project, a field-reflective column infiltration experiment was established to systematically examine the permeation–cake-formation behavior of five graded sand gradations and three dosages (5–15 g/L) in 2–5 mm formations. Results indicate two end states: permeation zone, or a coexisting filter cake with a permeation zone. The cake consists of a graded sand skeleton interwoven with bentonite fines, with a small fraction of sand entering pore throats to produce localized internal clogging. Surface retention shows a pronounced negative relationship with permeation flux and exhibits saturation; in the 4–5 mm formation, the maximum surface retention at 15 g L⁻¹ is comparable to that at 10 g/L (0.55 kg/m²), both corresponding to the lowest flux. Filter cake formation is jointly governed by gradation–pore-throat matching and dosage, and low dosages struggle to establish a continuous filter cake. For formations with a permeability coefficient below 5.29 × 10⁻² m /s, and with particle-size matching between the sand and the formation’s pore diameter, forming a dense filter cake requires a minimum retention of 0.51 kg/m ² on the excavation surface graded sand surface. Based on these results, an empirical criterion centered on particle-size matching and surface retention, together with mix-design recommendations, provides a verifiable design basis for face-pressure stabilization and slurry selection in river-crossing shield tunneling.