The areal fracture intensity (P21) of the surrounding rock is a critical indicator for evaluating tunnel excavation stability. However, conventional methods often fail to capture geological conditions ahead of the tunnel face. With the development of measurement-while-drilling (MWD) technology, real-time acquisition of surrounding rock information has become feasible. Therefore, establishing a relationship between MWD data and the P21 is of significant importance. This study proposes a semi-supervised deep learning framework that combines a multi-autoencoder unit training module (MAUT) with a multi-branch feature extraction network (MBEN). By integrating MWD, construction, and blasting data as dataset input, the proposed model is applied to the Yangjiawopu tunnel. The predicted P21 values exhibited a relative error of less than 10%, confirming the model’s superior predictive performance and engineering applicability. Ablation studies demonstrate that both the MAUT module and feature fusion enhance model accuracy. The shapley additive explanations (SHAP) analysis highlights excavation length, water inflow, and mechanical specific energy (MSE) as key predictors. The model effectively leverages unlabeled data and provides robust geological insights, offering a data-driven approach for early fracture detection and risk assessment ahead of the tunnel face.

Currently, research on waterproofing at shield tunnel joints remains largely focused on pure EPDM (Ethylene Propylene Diene Monomer) rubber gaskets, while there is still a lack of in-depth understanding regarding the underlying physical mechanisms of waterproofing performance and the leakage failure mechanisms in EPDM-WSR (Water-Swelling Rubber) composite gaskets, which are increasingly adopted in engineering applications. To address this, this study employs a refined Coupled Eulerian-Lagrangian (CEL) numerical method, validated through full-scale mechanical and waterproofing tests, to reproduce the entire dynamic process of joint failure, encompassing gasket deformation, WSR swelling, and high-pressure water breakthrough. Subsequently, this model was used to systematically compare the performance differences between conventional EPDM and three typical EPDM-WSR composite gaskets, providing an in-depth analysis of their mechanical evolution characteristics, leakage failure mechanisms, and the influence of WSR arrangement schemes on sealing performance. Based on these mechanistic insights, the cross-sectional configuration of the traditional EPDM-WSR composite gasket was optimized, and multi-dimensional indicators such as the swelling enhancement coefficient (γse) and effective high-pressure coverage ratio (ηesc) were introduced to enable a quantitative and comprehensive evaluation of the composite gasket’s overall performance. The results indicate that the optimized gasket significantly reduces assembly loads while exhibiting the highest water-induced expansion efficiency. Furthermore, it maintains the highest interfacial contact stress during the critical leakage stage, effectively enhancing the waterproofing redundancy of the joints. This research provides universal engineering guidance for the failure prevention and design optimization of joint waterproofing in deep-sea, high-pressure shield tunnels.

In earth pressure balance (EPB) shield tunneling, chamber pressure control is essential for maintaining excavation-face stability and limiting surface settlement. However, the influence mechanisms of key tunneling parameters on chamber pressure and ground response remain difficult to clarify when soil conditioning, cutterhead squeezing, muck discharge, and ground deformation are considered together. To address this issue, this study establishes a theoretical mechanical model for EPB shield tunneling and develops a calibrated three-dimensional FDM–DEM coupled numerical model based on the Jinan Metro Line 6 project. Triaxial and slump tests were used to calibrate the macro–micro parameters of unconditioned and foam-conditioned soil. The effects of tunneling speed, cutterhead speed, and screw conveyor speed on chamber pressure distribution and surface settlement were then analyzed. The results show that tunneling speed mainly affects the overall chamber pressure and face-support condition, screw conveyor speed controls muck discharge and pressure release, and cutterhead speed influences the spatial distribution of chamber pressure and settlement response. This study provides a theoretical and numerical basis for understanding chamber pressure evolution and coordinating tunneling parameters in EPB shield construction.

This study presents a comprehensive investigation into the deformation mechanisms of existing metro stations subjected to the simultaneous construction of adjacent foundation pits and underground tunnels. A refined three-dimensional numerical modeling framework is developed to simulate the entire construction process, capturing the complex interactions between excavation activities and station structures. The modeling encompasses deep excavation, side-crossing, and overcrossing passage construction, and the staged installation of support systems. Six construction schemes, varying in excavation sequence, interlayer thickness (clear distance), and passageway layout, are systematically analyzed. Field monitoring data are incorporated to validate the numerical models, enhancing the reliability of the results. The analysis identifies the construction sequence as the primary factor influencing station deformation. Specifically, the strategy of constructing passageways first, followed by excavation of the interchange hall, effectively reduces both vertical and horizontal displacements by leveraging the early-stage portal-frame reinforcement effect. Increasing the clear distance between new structures and the existing station helps mitigate construction-induced deformation, although the benefits plateau beyond a certain threshold. Sensitivity analysis shows that overcrossing passages are most sensitive to variations in clear distance, followed by foundation pits and side-crossing tunnels. Additionally, the spatial positioning of passageways significantly impacts deformation magnitude and propagation. Passageways near expansion joints cause the greatest uplift, while those placed at mid-span experience minimal disturbance due to enhanced structural stiffness. This research provides a quantitative understanding of metro station deformation under concurrent construction activities and offers practical insights for optimizing excavation sequences, structural layouts, and interlayer spacing. The findings contribute to ensuring structural safety and minimizing risks in densely built urban metro environments.

Water loads on subsea tunnel linings are governed by the combined effects of seepage, structural deformation, and the grouting zone. However, these factors are often treated separately in existing analytical studies. In this study, an analytical model is developed for steady-state saturated seepage around a deep-buried subsea tunnel. The surrounding rock, grouting zone, primary lining, and secondary lining are represented as distinct hydraulic zones, with a constant permeability coefficient assigned to each zone. The analytical results are compared with numerical simulations and show good agreement. Parametric analyses indicate that increasing the grouting-zone thickness or decreasing its permeability reduces the water pressure behind the secondary lining and the lining stress. This effect becomes more pronounced at higher allowable drainage rates but gradually diminishes beyond a certain parameter range. Field monitoring in the Zhujiangkou Tunnel further shows a maximum relative error of 6.89% for the water pressure behind the lining, supporting the applicability of the proposed analytical method.

In the context of tunnel infrastructure maintenance, ensuring structural integrity remains a critical concern, with particular attention given to the early detection of lining cracks—among the most prevalent and hazardous forms of tunnel deterioration. However, despite numerous efforts, two major challenges continue to impede detection accuracy: the elusive nature of micro-scale cracks and the difficulty in distinguishing true cracks from construction seams. To address the former, we enhanced the classical U-Net segmentation framework by designing a multi-view fusion module that incorporates the dynamic snake convolution and classical convolution, thereby strengthening its ability to capture fine, elongated features. The training process further adopted a compound loss function (Lc), which harmonizes Dice and centerlineDice (CLDice) components to better preserve crack topology. For the latter challenge-false identification caused by seam interference-we introduced a two-step solution. First, a fracture-reconstruction strategy was developed for separating simple skeletons. Building on this, separation of complex skeletons was achieved via a disassembly-merging strategy. Secondly, leveraging the structural patterns of these separated skeletons, a graph convolutional network (GCN) was applied to classify and eliminate seam-like structures effectively. Evaluation results demonstrated that this approach significantly enhanced micro-crack detection, yielding a 0.042 improvement in intersection over union (IoU) over baseline methods. Moreover, the proposed skeleton processing algorithm not only achieved more accurate separation of complex skeletons but also reduced computational time by a factor of 2 to 3. After incorporating the seam elimination model, the precision and recall rates of crack detection increased to 0.682 and 0.915, respectively—highlighting the method’s robustness in suppressing linear seam noise while maintaining crack precision. Field validations conducted on two distinct categories of engineering imagery further confirmed the system’s adaptability and effectiveness under realistic conditions.

Accurately predicting surrounding-rock deformation is vital for ensuring tunnel safety. However, three critical gaps persist: first, conventional numerical methods are unsuitable for real-time feedback owing to their low speed; second, geological-parameter uncertainty limits simulation accuracy; and third, existing data-driven models cannot forecast unexcavated sections based on data monitored from prior sections. Hence, this study proposes an efficient surrogate analysis method that integrates mechanical-parameter inversion with surrounding-rock deformation prediction. First, an improved sine-cosine and Cauchy sparrow search algorithm (SCSSA) is developed by incorporating refraction reverse learning and Cauchy mutation to optimize randomforest hyperparameters. Subsequently, this established SCSSA-RF surrogate model is employed to invert rock mass parameters using field monitoring data from the Qingdao Jiaozhou Bay Second Undersea Tunnel. Finally, a stacking ensemble strategy is adopted to construct an efficient surrogate model that integrates inverted parameters, tunnel burial depth, and excavation advance for deformation prediction. The results show that the SCSSA-RF model achieves an average coefficient of determination (R2) of 0.96 for mechanical-parameter inversion. On the test set, the stacking surrogate model yields mean (R2), root mean square error, and mean absolute error values of 0.986, 0.23, and 0.11, respectively. The approach enables high-precision and highefficiency deformation prediction. Future investigations shall focus on incorporating datasets from more diverse and complex geological conditions to enhance the model's generalizability.

We develop a time-domain analysis framework that couples the indirect boundary element method(IBEM)with the discrete element method(DEM)to investigate the dynamic response patterns and failure mechanisms of rock slopes subjected to near-fault ground motions.This framework enables a nonlinear dynamic simulation approach for near-fault slope systems,capturing the discontinuous deformation characteristics of rock and soil masses.Firstly,we construct a high-precision numerical model of the kilometer-scale,semi-infinite near-fault seismic wavefield using IBEM.Subsequently,based on Green's function theory and the IBEM solution of the wavefield,we derive an explicit formulation of the equivalent seismic loads on the boundaries of the DEM computational domain.This enables cross-scale energy transfer within the IBEM-DEM coupled system.Finally,the DEM resolves the nonlinear dynamic response of meter-scale rock slopes,yielding a multi-scale nonlinear seismic simulation framework that spans from kilometer-scale faults to meter-scale slopes.Numerical simulations combined with dynamic monitoring results demonstrate that the IBEM-DEM coupling algorithm can accurately capture the dispersion characteristics and energy attenuation patterns of near-field seismic wave propagation.Under near-fault seismic loading,progressive shear failure first occurs within weak interlayers,leading to strength degradation,the formation of through-going rupture surfaces,and subsequent accelerated instability of the sliding mass along the shear plane.This process induces significant displacement and velocity responses,ultimately forming a typical debris accumulation at the slope toe.The surface velocity of the sliding mass is markedly greater than that at the base,with the mean surface velocity reaching 3.6 times that of the base,and peak X-and Z-direction velocity components of 4.98 m/s and 5.92 m/s,respectively,exhibiting a pronounced surface-acceleration effect.The monitoring points of the sliding mass exhibit maximum displacements of up to 41 m in the X-direction and 35 m in the Z-direction from the initial slope surface to the final accumulation position,with the displacement-time history showing a distinct step-like growth pattern,indicative of abrupt sliding behavior during the alternating transformation of kinetic and potential energy.The IBEM-DEM coupled method developed in this study reconstructs the full evolutionary sequence from rock mass rupture to landslide formation,providing an innovative analytical framework for the dynamic failure analysis of landslides induced by near-fault ground motions,as well as theoretical and technical support for identifying landslide mechanisms and mitigating seismic hazards in complex geological settings.

This paper introduces a novel two-step multi-scale coupled method for simulating the nonlinear dynamic behavior of a mountain tunnel subjected to fault movement. In the first step, the broadband seismic responses within a large-scale mountain-fault model can be accurately solved by the indirect boundary element method, converting them into effective input forces around the specified region of interest within the mountain. The second step involves finely simulating the nonlinear dynamic response of the tunnel cross-section in the designated region using the finite element method, with the implementation of a viscoelastic artificial boundary to absorb the reflection of scattered waves at truncated boundaries. Two verification processes are employed to validate the accuracy of the multi-scale coupled method. Furthermore, we illustrate the applicability and efficacy of the new method with an example involving the elastoplastic dynamic analysis of a mountain tunnel under the influence of normal fault movement. The presented example highlights the impact of fault motion parameters, including fault dislocation value and dip angle, on the responses of the mountain tunnel. The results demonstrate that the proposed multi-scale coupled method can achieve full-process seismic simulation, ranging from kilometer-scale fault rupture to centimeter-scale mountain tunnel section damage, with a considerably reduced computational expense.

This paper establishes a novel full-process numerical simulation framework for analyzing the 3D seismic response of mountain tunnels induced by active faults. The framework employs a two-step approach to achieve wavefield transmission through equivalent seismic load: first, a highly efficient and accurate FMIBEM (Fast multipole indirect boundary element method) is used for large-scale 3D numerical simulations at the regional scale to generate broadband ground motions (1-5 Hz) for specific sites; subsequently, using the FEM (Finite element method), a refined simulation of the plastic deformation of surrounding rock and the elastoplastic behavior of the tunnel structure was conducted at the engineering scale. The accuracy of the framework has been validated. To further demonstrate its effectiveness, the framework is applied to analyze the impact of different fault movement mechanisms on the damage to mountain tunnels based on a scenario earthquake (Mw 6.7). By introducing tunnel structure damage classification and corresponding damage indicators, the structural damage levels of tunnels subjected to active fault movements are quantitatively evaluated. The findings demonstrate that the framework successfully simulates the entire process, from fault rupture and terrain amplification to the seismic response of tunnel structures. Furthermore, the severity of tunnel damage caused by different fault types is ranked as follows: reverse fault > normal fault > strike-slip fault.

To investigate the bending performance and damage characteristics of segmental joints with double sealing gaskets in large-diameter shield tunnels under high water pressure, this study established a three-dimensional high-fidelity numerical model of the segment-joint system based on the Pearl River Estuary Tunnel project. A comprehensive analysis was conducted on the mechanical and deformation behavior of large-diameter shield tunnel segmental joints under combined compressive/flexural loading. The research systematically examined the evolving relationships between bending moments, vertical displacements, and joint opening at the double-sealed gasketed joints under varying axial compression conditions, thereby elucidating the phased characteristics of joint deformation. The results indicate that the deformation patterns of double-sealed gasketed segmental joints under compressive/flexural loading exhibit pronounced nonlinearity and stage-dependent features. Both positive and negative bending moment scenarios demonstrate four distinct failure phases. Under high-water-pressure conditions, structural damage initiation consistently occurs at waterproof sealing grooves and bolt holes, regardless of bending moment direction. As loading intensifies, cracks propagate symmetrically at 45° angles from the joint interface, generating extended fracture networks, which creates additional water infiltration pathways, significantly compromising the joint’s waterproofing integrity.

Composite construction methods may be adopted in some underwater tunnels due to complex geological conditions. However, the seepage field at the junction between parts constructed by different methods has been rarely reported. This study used the Pearl River Estuary Tunnel as a case study to conduct model experiments on the connection part and its adjacent mining and shield parts, which have different drainage systems. Through model experiments, the study explored the variation laws of water inflow and external water pressure on the tunnel lining. The experimental results were validated against numerical simulations. The results indicate that, longitudinally, the junction experiences pressure fluctuations due to variations in construction methods and tunnel cross-sectional dimensions. A logistic fit of the experimental data shows that fluctuations range from 12 to 21 m in the mining part and from 27 to 42 m in the shield part. The longitudinal distributions indicate that water pressure remains stable in the shield part, while in both the connection and mining parts it is influenced by the drainage pipes. Transversely, water pressure increases from the arch crown to the arch bottom, forming an approximately circular distribution. Both water pressure and inflow increase linearly with rising water levels. A comparison between experimental and simulation results shows a relative error of less than 9 %, confirming the accuracy of the experimental model. The study's findings provide deeper insight into the seepage field at the junction of underwater tunnel parts, with implications for the design and construction of similar projects.

Efficient and accurate numerical methods are essential for analyzing seismic wave propagation and amplification in near-fault complex sites. Understanding the complete process of ground motion simulation, including fault rupture, path propagation, and near-surface complex site response, is crucial for studying earthquake damage mechanisms, seismic zoning, and designing large-scale engineering structures. In this study, we propose a fast multipole indirect boundary element method (FMIBEM) to achieve broadband and high-efficiency simulation, enabling a comprehensive analysis of the complete process involving seismic ground motion. The FMIBEM significantly reduces the computational and storage costs associated with 3D near-fault complex site seismic wave scattering problems to O(N). We validate the accuracy of the method by comparing it with the analytical solution. Compared to the conventional indirect boundary element method (IBEM), our proposed method reduces computational time by over 95 % and storage costs by nearly 90 %. FMIBEM greatly enhances the efficiency of the boundary element method for simulating seismic wave scattering in near-fault complex sites. To demonstrate the effectiveness of our approach, we apply the FMIBEM method to two typical 3D near-fault complex site examples of seismic wave scattering. The simulation results successfully capture both the local site amplification effect and the characteristic near-fault seismic features, such as the hanging wall effect, permanent displacement effect, and large velocity pulse.

Concealment of filling constructions poses significant challenges for quality assurance in filling engineering. Direct surveillance of fill dispersal currently remains infeasible, while conventional detection techniques suffer deficiencies in efficiency. This research proposes a framework integrating elastic wave monitoring and hybrid deep learning for predictive modelling of filling state transitions and diffusion range. During the sand filling of the immersed tunnel, elastic wave data is collected via elastic wave testing, and the response energy characteristic is derived through time-domain analysis. The trends in elastic wave response energy are correlated with three filling states: free diffusion, accumulation, and filled state, using Seasonal and Trend decomposition using Loess (STL) for seasonal trend analysis. Convolutional Neural Networks (CNN) and Long Short-Term Memory Networks (LSTM) are utilized to extract spatiotemporal features from the response energy trends, facilitating accurate prediction of the trends' development and the sand filling state over time. The performances of the proposed strategy are illustrated through an application to the case study of the sand filling construction of the Chebeilu immersed tunnel. The CNN + LSTM model with the proposed strategy gave excellent results (MAE 0.0663, MSE 0.0071, RMSE 0.0845). The model can predict fill state changes and quantify diffusion radii to optimize and guide the construction process.

A tunnel-group metro station built in rock site is composed of a group of tunnels. Different tunnels and their interconnections can show inconsistent responses during an earthquake. This study investigates the dynamic responses of such a metro station in a rock site, by shaking table tests. The lining structures of each tunnel and surrounding rock are modeled based on the similitude law; foam concrete and gypsum are used to model the ground-structure system, keeping relative stiffness consistent with that of the prototype. A series of harmonic waves are employed as excitations, input along the transverse and longitudinal direction of the shaking table. The discrepant responses caused by the structural irregularities are revealed by measurement of acceleration and strain of the model. Site characteristics are identified by the transfer function method in white noise cases. The test results show that the acceleration response and strain response of the structure are controlled by the ground. In particular, the acceleration amplification effect at the opening section of the station hall is more significant than that at the standard section under transverse excitation; the amplification effect of the structural opening is insignificant under longitudinal excitation.

This study proposes a rapid seismic resilience assessment framework of tunnels in mountain regions considering the topography amplification effect and tunnel-soil dynamic interaction based on the indirect boundary element method (IBEM) coupled with the finite element method (FEM). The high efficiency is achieved by using a surrogate model to determine the tunnel fragility curves. This model reflects the relationship between the geometric and material variables of mountains and tunnels, as well as the tunnel damage index. To obtain the surrogate model, the identification of model variables is first explored quantitatively based on the random forest algorithm due to the high variable quantity. The dataset for training and testing the random forest is constructed from 600 numerical simulations. The IBEM-FEM coupling scheme is employed to describe the large-scale site response for tunnel damage analysis and significantly reduce the number of finite element grids for each sample. This scheme solves the nonlinear dynamic response of mountain tunnels under near-fault earthquakes. The surrogate model is then used to obtain the tunnel functionality and resilience. Based on the proposed framework, the influence of the mountain material, mountain height-span ratio, and tunnel position on the seismic fragility, functionality, and resilience are investigated. The results reveal that a surrogate model can be employed to replace a series of nonlinear time-history analyses of tunnels, with a high accuracy and efficiency. The shear modulus of the surrounding rock, the height-to-span ratio of the mountain, and tunnel position have a significant impact on tunnel fragility and resilience. This impact is correlated with the tunnel height. The mountain topography can cause a difference of approximately 20 % in the tunnel resilience.

Studying dynamic response and damage assessment of mountain tunnels is paramount in earthquake engineering. This paper proposed a novel method, the indirect boundary element method-finite element method (IBEM-FEM) coupled method, aimed to concurrently simulate the amplification effect of kilometer-scale mountain topography and the damage evolution in a centimeter-scale lining tunnel section. The proposed method involves two fundamental steps. Initially, the input motion on the truncation boundary within the mountain is accurately determined using the indirect boundary element method (IBEM). Subsequently, the nonlinear dynamic behavior of the tunnel structure and its adjacent area is analyzed by the finite element method (FEM) based on the results from the previous step. This study numerically simulates the seismic damage and failure mechanisms of a mountain tunnel considering three influence factors: incident wave types, incident wave intensities, and surrounding rock mass grades. The results indicate that mountain topography amplifies the seismic response, thereby affecting the dynamic behavior of the tunnel. Under the incidence of SV-waves and P-waves, the dynamic damage distribution patterns of the tunnel lining are significantly different. Moreover, as seismic wave intensity increases, the peak values of tunnel stress exhibit a non-monotonic increasing trend. The surrounding rock mass grade also significantly affects the dynamic response and damage distribution of the mountain tunnel. Overall, the proposed IBEM-FEM coupled method can effectively consider complex mountain topography and is applicable for evaluating the nonlinear seismic response of mountain tunnels.

As for the running tunnel section between Wuyuanwan satation-Liuwudian station of Xiamen metro line 3, the construction method of "dismantling shield in the subsea first, and continually excavating the rest of section by the drilling-blasting method" is adopted. It is required to overcome such technical difficulties as shield dismantling in a confined space, long-distance single-line transportation, long-distance blind ventilation, protection of formed segments, waterproofing of joint structures, etc. The paper systematically expounds the transfer technology of various construction methods in Wuyuanwan station-Liuwudian station section and mainly introduces the solutions to the above construction difficulties. It has been proved in practice that this technology can greatly improve the construction progress under the premise of guaranteeing construction safety and structural function completeness, having reference significance for the subsea tunnel constructed by multiple combined method.

为解决 20～70 m超大跨洞室围岩分级难题,提出在初步分级阶段采用面积加权平均法确定围岩基本质量指标、在详细分级阶段增加跨度影响修正系数的超大跨扁平洞室围岩分级方法,跨度影响修正系数值根据洞室稳定系数与围岩基本质量指标和洞室跨度之间的关系来确定.采用三维数值模拟和强度折减法,通过模拟洞室的开挖全过程,计算Ⅰ—Ⅳ级围岩中 20～70 m跨度洞室的稳定系数,分析得出洞室稳定系数与围岩基本质量指标和跨度之间的经验公式.结果表明:1)当跨度由20 m增大至70 m后,洞室稳定系数减小了 46.54%～50.67%;2)当洞室跨度相同时,围岩基本质量指标越小,洞室稳定性越差.定义跨度影响修正系数为洞室稳定系数偏离基准跨度(20 m)洞室稳定系数的相对偏离度与基准跨度洞室稳定系数级差的比值.跨度影响修正系数取值范围为 0～1.0,跨度增大对围岩稳定性的最不利影响相当于围岩级别降低 1 级.

为有效提高矿井排风热湿能量提取效率,在上喷式矿井排风流热回收装置内,采用马尔文激光粒度仪,开展喷嘴雾化压力、排风流风速对上喷式液滴群特征粒径与分散相分数影响试验,完成21个工况的沿程数据实测;应用临界韦伯数和韧带直径计算式,明确喷嘴操作参数与排风流动力学参数对液滴群分散相分数的影响;引入分散相分数,推导出具有液滴群特征粒径的单一液滴运移关联式.结果表明:液膜的初次破碎类型为长波长线性不稳定性破碎,当喷嘴雾化压力增大时,液滴粒径减小,分散相分数增大;引入分散相分数后,阻力系数增大,分散相分数对排风流热回收装置内的液滴群运移表现出类润滑作用.