Data-mechanism hybrid-driven digital twin for spatiotemporal prediction of multiple evolving risk in deep excavation

A spatiotemporal settlement estimation framework integrating augmented clustering and state-space modeling during subway deep foundation excavation

Wall deformation of a zone-divided deep excavation with concrete servo struts in Hangzhou soft clay

Fracture evolution and failure mechanism of coal-rock composite under deep excavation disturbance: experimental and DEM investigation

Adjacent ultra-long deep excavation induced deformation of metro station–tunnel system: mechanism and intelligent prediction

Surrogate-assisted optimization of dewatering well layouts for deep excavations

Predicting foundation pit deformation is a significant challenge for foundation pit engineering. The inaccuracy of deformation prediction is increased by the intricacy of subterranean space and the variety of construction conditions. Thus, this study develops a deformation prediction model that combines the attention mechanism and bidirectional long short term memory network (BiLSTM) in order to increase the accuracy of deep foundation pit deformation prediction. Meanwhile, to enhance the generalization ability of the model, this study introduces combined regularization in the loss function and adds a Dropout mechanism in the network structure. This study takes a deep foundation pit excavation project in Guangzhou as an example. Experiments shows that the model proposed in the study can complete convergence in about 30 training rounds, and the training loss is maintained at the 0.03 level. Meanwhile, the maximum absolute error of the model in the prediction of verification data is 1.44mm, and the minimum error is 0.001mm. The mean absolute error of the model is 0.311mm, the root mean square error is 0.433mm, and the R2 is 0.906, which is better than the comparison model. The attention mechanism and BiLSTM model suggested in this study provides good generalization performance and high prediction accuracy in deep foundation pit deformation prediction, according to experimental results. Its potential for use in engineering safety management is promising.

ABSTRACT The study provides a generalised approach for the estimation of earth pressure forces on the excavation support system using a diaphragm wall, taking into consideration the presence of a cavity in the retained ground as well as load transferred to the supporting wall due to the self-weight of the buried pipe or utility tunnel, as well as the overburden load of the soil. The theoretical perspective of the proposed solution is based on Rankine’s earth pressure theory and Spangler’s solution, based on Modified Boussinesq’s approach. A parametric study is then performed, taking into account different buried pipe or utility tunnel diameters located very near the 15 m deep excavation support system in the upper, middle, or lower zone. Numerical analysis using STAADPro is involved to estimate the deformation as well as the shear and bending moment in the diaphragm wall by assuming its cantilever action, and the results are presented to analyse and discuss the influence of the presence of a buried pipe or utility tunnel very near the deep excavation support system using a diaphragm wall.

The surrounding rock around deep underground excavations generally contains fractures and joints that significantly reduce mechanical integrity, particularly in zones where excavation-induced unloading, stress rotation, and anisotropic deformation generate tensile stress concentrations. These tensile regimes govern crack initiation, fracture propagation, spalling, and progressive instability around tunnels, caverns, mines, and wellbores. Although rock grouting is widely used as a reinforcement strategy, its effectiveness in restoring the inherent strength of tension-dominated surrounding rock is often assumed, and the tensile fracture behavior of fracture-grouted rock remains insufficiently understood. This study investigated the evolution of tensile strength (TS) and tensile-induced fracture behavior in fractured surrounding rock before and after fracture grouting, providing mechanistic evidence for assessing strength recovery. In addition to TS, failure modes and total fracture length (TFL) were quantified to assess fracture propagation and grout-reinforcement performance in tension-prone surrounding rock zones. Brazilian disc tests (BDT) were conducted on natural (limestone and dolomite) and synthetic (3D-printed) rock samples, allowing direct comparison of the same samples in ungrouted and fracture-grouted states. To mechanistically interpret and validate the experimental findings, a finite element model employing the cohesive zone method (FEM-CZM) implemented in ABAQUS was developed to simulate the evolution of tensile strength and the initiation and propagation of micro-fracture in the specimens before and after grouting. Results revealed that while fracture grouting does not fully restore the inherent tensile strength of fractured rocks, it significantly altered the tensile failure process. Grout-rock interfaces in the fracture-grouted rocks constrained and redirected crack propagation, reduced total fracture length, and shifted failure modes toward more localized and controlled fracture patterns. A positively correlated TS-TFL relationship observed in ungrouted samples reversed in fracture-grouted samples, indicating that higher tensile resistance in grouted rock mass corresponds to more limited fracture development. The FEM-CZM simulation confirmed the experimentally observed post-grouting delay in rock damage onset with reduced fracture path and an increase in Mode-II energy dissipation, and provided direct visualization of stress concentration, damage evolution, and fracture-path control. These findings demonstrate that surrounding rock control in tensile regimes depends not only on tensile strength recovery but also on the ability of grouting to suppress fracture propagation and damage evolution. The results provide new mechanistic insight into fracture-grouting performance in tension-prone underground environments and demonstrate that TFL, when used alongside tensile strength and other mechanical parameters, is a valuable metric for assessing reinforcement effectiveness. This work advances the understanding of grouting as a surrounding rock control strategy and informs the design of reinforcement systems aimed at stabilizing underground excavations subjected to tensile stress concentrations.

The design of internal bracing is critical for controlling deformations in soft-soil foundation pits, especially in congested urban areas where ground anchors are prohibited. This study investigates a novel combined support system consisting of 45° inclined H-beam strength composite piles (SCPs) and triaxial deep-mixing columns (TDCs). A full-scale field investigation was conducted in a 6.15-7.10m deep excavation in Nanjing mucky silty clay, adjacent to sensitive metro tunnels and dense underground utilities. Field load tests on three inclined SCPs demonstrated ultimate vertical capacities of 1,350-2,090 kN, with total settlements limited to 14.07-15.17mm. Regression analysis using a power-function model (s = aQb) predicted extrapolated ultimate capacities of 2,299-3,682 kN, indicating that current technical specifications (JGJ/T 327) provide reliable but conservative estimates (approximately 62%-84%). Continuous field monitoring revealed that the maximum lateral wall deflection was confined to 6.36mm (less than 0.1%H), significantly superior to conventional cantilever systems. These results validate a structural system transformation from a quasi-cantilever to a quasi-simply supported configuration enabled by the inclined bracing. This study provides a practical benchmark and quantified performance data for the application of inclined SCPs as a low-carbon, high-efficiency alternative to traditional internal bracing in complex soft-soil environments.

Deep excavations in densely built urban areas frequently employ jet grouting to enhance base stiffness and limit diaphragm wall deformation. However, most design approaches treat the improved soil mass as homogeneous and rarely consider the spatial variability inherent in jet-grouted materials, which may lead to biased deformation predictions and incomplete risk evaluation. This study investigates the influence of spatial variability in the secant stiffness modulus (E₅₀) of jet-grouted soil on diaphragm wall displacement using a probabilistic numerical framework. A two-dimensional lognormal random field of E₅₀ was generated using the spectral representation method and incorporated into a finite element model within the Random Finite Element Method (RFEM)framework. The mean stiffness of the improved soil was taken as 1336.85MPa based on laboratory testing, while variability was characterized by coefficients of variation (COV) ranging from 0.34 to 0.8 and spatial correlation lengths defined by scales of fluctuation of SOFₓ = 3-5m and SOFγ = 0.5-2m. Monte Carlo simulations were conducted with a converged sample size of 40 realizations. The results indicate that spatial variability significantly affects both the magnitude and dispersion of wall displacement, with predicted maximum values generally ranging from 11.0 to 11.8mm and following a lognormal distribution. For an allowable displacement of 11.44mm, the exceedance probability decreases from approximately 58% at COV = 0.34 to about 35% at COV = 0.8. Changes in spatial correlation length have comparatively smaller effects, although vertical correlation shows a slightly stronger influence than horizontal correlation. These findings demonstrate that deterministic analyses assuming uniform ground improvement may underestimate the range of possible wall movements, while incorporating spatial variability provides a more realistic basis for reliability-based assessment and risk-informed design of jet-grouting-reinforced excavations.

A FEM-MARS hybrid framework for predicting wall deflection and seepage flow in deep excavations

Load transfer mechanism and spatio-temporal control of deep underground excavations under repeated mining disturbances: A case study

Three-dimensional HSS finite-element analysis of a shield shaft-station deep excavation with bored pile–strut support

Use of Steel Sheet Pile Walls for Deep Excavations

Towards reliable deep excavation monitoring through graph recurrent neural network-based spatio-temporal imputation

To evaluate the effectiveness of a prestressed anchor cable combined with soil nailing technology in supporting deep foundation pits in loess regions, comprehensive field monitoring was conducted to measure horizontal and vertical displacements at the pit crest, as well as deep-seated horizontal deformations of the support system. A two-dimensional numerical model was developed using MIDAS GTS NX to facilitate comparative analysis. The results demonstrate that the integrated support system effectively controls deformation during excavation. The maximum displacement at the top of the support structure was recorded as 3.91mm, while the maximum horizontal displacement below a depth of 0.5m did not exceed 1.6mm-both values are substantially below the predefined alarm thresholds. The vertical profile of horizontal displacement exhibits a fluctuating pattern, with relatively smaller displacements observed at the elevations corresponding to soil nails and anchor cables, indicating localized reinforcement effects. Finite element analysis reveals that horizontal displacement within the excavation zone and surface settlement at the pit edge increase progressively with excavation depth, exhibiting an approximately linear trend. In the unsupported scenario, numerical instability occurred at an excavation depth of 3.6m, beyond which shear failure propagated through the soil mass. Although the simulated displacements prior to failure were higher than those measured in the field, this discrepancy is attributed to the minimal external loading under actual site conditions and the simplified representation of full-length grouting effects in the simulation model, which may underestimate the confinement provided by reinforced soil. In conclusion, the integration of prestressed anchor cables with soil nailing walls demonstrates high performance in deep foundation pit engineering in loess regions, offering effective deformation control and enhanced structural stability.

This study investigates the deformation mechanisms of an existing tunnel adjacent to a deep basement excavation supported by circular beams in soft clay deposits. Initially, a well-instrumented basement in soft clay is back-analyzed using three-dimensional numerical modeling with the Hardening Soil model with small-strain stiffness (HSS). The computed lateral wall movements and ground surface settlements show good agreements with field measurements, validating the numerical model and model parameters. Subsequently, a comprehensive parametric study is conducted to examine the effects of tunnel cover depth (C/He), tunnel-basement clearance (S/He), and basement geometry (G/He) on tunnel responses. The results indicate that the lateral tunnel movements are more significant than settlements, with the maximum values typically occurring at the basement centerline. Tunnel movements increase with decreasing S/He to peak values when C/He is approximately 0.6. A linear relationship is observed between tunnel movements and basement geometry. A simplified design chart is developed to categorize tunnel deformation influence zones. This chart provides a practical tool for engineers to assess excavation-induced tunnel deformations and implement necessary protective measures during the design of deep basements adjacent to existing tunnels.

One Health-relevant ground-movement risks from chemically weakened urban surface soils during deep excavations: a 3D finite-element study.

Abstract Deep subway station excavation in dense urban areas can induce rapid municipal pipeline deformation under complex, time-varying conditions. Existing statistical models often struggle to capture nonlinear spatiotemporal deformation patterns, while numerical simulation approaches remain computationally demanding for ultra-short-term, high-frequency forecasting. In addition, current Building Information Modeling (BIM) applications are still used mainly for static visualization and are not well integrated with real-time monitoring data. To address these limitations, this study proposes a BIM-integrated intelligent prediction framework driven by an improved spatiotemporal Swin-U-Net (ST-SwinUnet) model for ultra-short-term pipeline deformation forecasting. A four-dimensional BIM/GIS-based digital twin is constructed by integrating geometric semantics, geological information, construction schedules, and hourly monitoring data into a unified spatiotemporal data cube. On this basis, prediction-oriented feature engineering is designed to represent spatial relationships, construction sequences, and time-dependent deformation effects. The ST-SwinUnet model further incorporates hierarchical working-condition embedding and a memory-enhanced attention mechanism to improve adaptability under complex operating scenarios. The framework was evaluated using an eight-month monitoring dataset from a deep subway station excavation project under a chronological train/validation/test design. On the 1–6 hour forecasting task, ST-SwinUnet achieved an average RMSE of 3.14 mm, corresponding to reductions of 46.9%, 37.1%, and 18.7% relative to LSTM, ConvLSTM, and the original Swin-U-Net, respectively. Additional multi-seed experiments further showed that the proposed model maintained stable predictive performance across repeated runs. These findings support the proposed framework as a reproducible project-scale proof-of-concept for ultra-short-term deformation assessment and rolling early warning, while broader cross-project validation remains necessary in future work.