Accurate prediction of tunnel face stability in complex geological formations remains a critical challenge in underground engineering, necessitating innovative computational approaches. This research proposes a Comprehensive Hybrid Gaussian (CHG) algorithm for predicting tunnel face stability in complex rock formations. The algorithm introduces a failure penetration probability index (FPPI) as an intermediate variable that establishes a probabilistic mapping between Tunnel Boring Machine parameters and rock mass stability, overcoming limitations of traditional linear mapping approaches. The CHG algorithm integrates feedforward layer normalization from Transformer architectures to enable dynamic kernel function optimization, resolving the manual hyperparameter tuning constraints in conventional Gaussian processes. A multi-scale regularization framework, from Tikhonov regularization to dropout, provides effective complexity control while maintaining expressive capacity. The posterior process incorporates Chebyshev’s inequality to enhance confidence interval estimation and prediction robustness. Validation across three geological points in the Yinsong Project demonstrates an average prediction deviation of 20.453%, with the CHG algorithm (R² = 0.982) significantly outperforming support vector regression (R² = 0.846) and random forest (R² = 0.923). While slightly underperforming the Transformer model (R² = 0.992) statistically in cross-project validation on the Yinchao data set, the CHG algorithm (R² = 0.869) exhibits superior adaptability to geological uncertainties. The synergy between FPPI and dynamic kernel functions establishes an innovative framework for predicting mechanical behavior in heterogeneous geological conditions, particularly in lithological transition zones, providing a theoretically sound and practically applicable decision support system for tunnel stability assessment.

Natural and human-made hazards pose significant risks to bridges, disrupting transportation systems and causing severe economic and social impacts. Earthquakes and blasts are particularly critical in evaluating long-term bridge resilience. Current resilience assessment methods, however, focus primarily on single and deterministic hazards, neglecting the uncertainty associated with hazard randomness, hazard interrelationship, structural robustness, and variability in restoration. This can underestimate risks and lead to structural failures, highlighting a critical knowledge gap. This paper proposes a novel approach to assess bridge resilience under multi-hazards, specifically earthquakes and blasts. The approach incorporates underexplored uncertainties, accounts for damage accumulation through state-dependent fragility, and introduces the resilience quantification probabilistically. An illustrative case study demonstrates its application, showing that hazard randomness, particularly the sequence and timing of sequential hazards during restoration, significantly influences bridge resilience. The findings emphasize the importance of detailed and probabilistic consideration of hazard randomness and interrelationship in the multi-hazard context. The proposed approach has the potential for broader application to other hazard types and structural systems, addressing an urgent need for resilience assessment in infrastructure systems subjected to multiple hazards.

In this paper a framework of quantile-based sequential optimization and reliability assessment (SORA) is extended to consider the global stability constraint in the optimization of plane frames. Uncertainty is considered on the structural side without any prior assumptions on their distribution information, and two novel stopping criteria with a reduction coefficient are employed to smoothly shift the constraint boundary for the next iteration. Force density method is introduced for the shape optimization of plane frames to avoid the existence of the melting nodes, and the geometrical stiffness matrix is also penalized to exclude pseudo local buckling modes. The numerical examples illustrate that with the help of reduction coefficients, the shifts of different constraint boundaries in SORA become smoother and the convergence of sequential optimization is improved, and due to shape optimization, the reliability of structural stability can be satisfied with limited increase of structural volume of the one without considering stability. Moreover, it is also shown in the cantilever beam and bridge examples that global structural stability can be enhanced by applying nodal displacement constraints with higher reliability.

Due to the growing needs of strengthening steel tubular truss, a new method for enhancing tubular joints by partially filling the chord with grout is proposed. However, the strengthening design of a whole truss is a challenging task, mainly because of multiple design objectives and various fabrication uncertainties. Current practice based on empirical or simple rule-based strategies is not able to handle the task. To address this challenge, a design framework for tubular truss strengthening is developed. The proposed framework can reduce the maximum deflection, improve the load capacities of the truss, and minimize the usage of grout. Furthermore, it considers geometric and modeling uncertainties through Monte Carlo simulation and predict intervals, thereby preventing over-idealization during practical optimization. To demonstrate the proposed design framework, a comparative structural analysis was conducted on a typical Warren truss between pre- and post- optimal strengthen. The results show that, by building upon the Machine Learning models, the proposed framework can produce an effective strengthening scheme. After considering uncertainties in optimization, some idealized samples are filtered out, resulting in a more practical strengthening scheme. The proposed framework is versatile and can be applied to other similar optimal strengthening designs with minimal additional effort.

Predicting the punching shear strength (PSS) of flat slabs is crucial for ensuring the safety and efficiency of reinforced concrete structures. This study presents novel hybrid approaches combining support vector regression (SVR) with advanced optimization algorithms to enhance the accuracy of PSS predictions. Four optimization algorithms, krill herd algorithm, biogeography-based optimization, equilibrium optimizer, and genetic algorithm (GA), were employed to optimize SVR parameters for improved PSS estimation. A data set of 264 samples with seven design parameters was used as input to model PSS. Sensitivity analysis and comparison to standard equations were conducted to evaluate the significance of input variables and the reliability of proposed models in predicting PSS. The results demonstrated that integrating optimization algorithms significantly improved the predictive performance of SVR models. Among the proposed approaches, the SVR-GA model achieved the highest accuracy, with a correlation coefficient of 0.95 and a mean absolute error of 132.28 kN in the testing phase. Sensitivity analysis revealed that slab thickness and depth, followed by concrete strength, were the most influential parameters for predicting PSS. The proposed SVR-GA model was found more accurate than American, European, and Canadian concrete code standards in modeling PSS. These findings underscore the effectiveness of hybrid SVR models in accurately modeling PSS and highlight the importance of optimizing input features to ensure robust predictions.

Point load test is an effective method for obtaining the mechanical parameters of the rock. However, the theoretical basis of the test is not clear at present, which greatly limits its acceptability and application. Therefore, it is essential to find the theoretical solution for point load test and its testing results. To calculate the uniaxial compressive strength, tensile strength and elastic modulus of rocks, this paper establishes the quantitative relationship between point load strength and rock mechanical parameters based on the theory of elasticity mechanics. The point load tester was used to test 120 cylindrical basalt specimens, the thickness range of basalt is from 18 to 60 mm. Comparing the calculated results with the experimental results, it is found that the prediction error range of the proposed method is from 4% to 8%. The approach provides a new way for predicting the mechanical parameters of the rock.

Adhering fiber reinforced plastics (FRP) is a typical method for reinforcing cracked tunnel linings. Due to the influence of the asymmetric effect, the mechanical response of FRP-strengthened lining with unsymmetric cracks is not known. Investigating the mechanical performance and strengthening mechanisms of FRP-strengthened linings with unilateral cracks is critical in guiding the reinforcing strategy. In this study, a series of model tests were conducted, two unilaterally cracked linings strengthened with FRP for different ranges (unilateral and full-span strengthening), and one intact lining was tested. The deformation field and damage behavior during the test process were monitored by digital image correlation analysis and acoustic emission technology. Results show that the unilaterally strengthened lining exhibits 30% higher bearing capacity and energy-dissipation capability than the full-span strengthened lining. During the damage process, a comparable percentage (50%) of tensile and shear cracks developed in all the lining specimens. However, compared to the unilateral strengthening, the extra FRP strengthening resulted in a higher severity of cracking. In addition, the strengthening mechanism of FRP-strengthened lining was analyzed based on the section-moment and section-stiffness of the linings. The full-span strengthening caused a greater stress concentration near the pre-crack, resulting in greater section moments and stiffness decay rate, contributing to structural failure. For tunnel linings with unilateral cracking, extra FRP strengthening may compromise reinforcement efficiency.

Leakage at tunnel structural joints represents a critical vulnerability in tunnel waterproofing systems. The interaction between secondary lining cavity development and waterproofing system failure creates a vicious cycle that accelerates tunnel structural joint leakage instability risks. Forty physical model tests were conducted to investigate and quantify the impacts of cavity depth, dimension, and spatial distribution on joint leakage under varying hydraulic heads. The study employed a self-developed tunnel lining leakage defect simulation apparatus. Experimental investigation elucidates the coupled hydro-mechanical mechanisms governing seepage evolution and stress redistribution in defective lining structures. Key findings indicate that water pressure progressively accumulates at the lining with increasing hydraulic head height, while cavity zones maintain consistently lower pressure due to pressure relief effects. This pressure distribution causes the water pressure sharing ratio η to increase with higher head height, while gradually decreasing as cavity breakage becomes more severe. Cavity extent expansion demonstrates the most significant influence, decreasing vault earth pressure by 23.8% compared to 15.8% and 19.5% reductions from depth extension and spatial redistribution. Cavity depth expansion resulted in a 25.4% increase in the axial force at the vault, along with a 3-fold increase in the bending moment. Expansion of the cavity extent mainly led to a 40.3% increase in axial force and a 1.8-fold increase in bending moment. These findings provide quantitative insights for optimizing waterproofing design and developing targeted maintenance strategies for cavity-affected tunnel structures.

This study extended the average strain energy density theory, traditionally applied for notch analysis in isotropic materials, to predict mixed mode I/II fracture in functionally graded materials. The effectiveness of existing experimental criteria designed for orthotropic materials is examined for their capacity to predict crack initiation in graded materials. A toughening mechanism based on fiber bypassing during crack propagation is introduced and investigated through experimental methods. Scanning electron microscopy images are employed to elucidate the role of this mechanism in enhancing the fracture resistance of chopped/short fiber-reinforced composites. The findings reveal that the average strain energy density method demonstrates strong correlation and agreement with experimental–numerical data, affirming its robustness for predicting crack initiation in mixed mode fracture scenarios.

Recycled aggregate concrete (RAC) undergoes irreversible freeze–thaw damage, with more complex and stochastic crack propagation than natural aggregate (NA). Consequently, steel stirrups are commonly used to enhance its durability and mechanical performance. 36 specimens with four recycled aggregate (RA) replacement ratios (0%, 30%, 70%, and 100%) and three stirrup types (no stirrup restraint (PCC), rectangular stirrup restraint (RSC), spiral stirrup restraint (SSC)) were therefore tested for uniaxial compression after 100 freeze–thaw cycles. Afterward, the stress–strain behavior of specimens was analyzed. The results demonstrate that RA exhibits superior resistance to freeze–thaw cycles compared to NA under similar strength, the peak stress increases with higher RA replacement ratios. SSC exhibits a stronger strengthening effect than RSC due to the uniform confinement of spiral stirrups. Lateral strain increases with axial strain, following a cubic nonlinear relationship. Subsequently, by employing the cohesive elastic ring model and the modified elastic beam theory, the lateral effective stress of RSC and SSC can be accurately predicted, with a standard deviation of 0.113 and an average absolute error of 0.186. A novel compressive damage constitutive model for RAC, incorporating lateral stress effects, shows strong agreement with test stress–strain curves, with R² > 0.91. Finally, a comparison of bearing capacity calculations shows the proposed method aligns best with experimental results.

This study proposes an optimized fine-grained concrete incorporating marine quartz sand (FGCMQS) and blended mineral admixtures to enhance mechanical performance while promoting sustainability. Marine quartz sand (MQS) was combined with crushed sand at a 30:70 ratio, resulting in an optimal gradation that meets the requirements of ASTM C33. Using response surface methodology (RSM), this study developed four statistical models to optimize the proportions of fly ash (22%), silica fume (9%), ground granulated blast-furnace slag (32%), superplasticizer (0.85%), and a water-to-cement (W/C) ratio of 0.32. The optimized FGCMQS achieved a slump of 4 cm, a compressive strength of 65.1 MPa, a chloride ion permeability of 812 C (Coulombs), and a sulfate-induced length change of 0.04%. Compared to conventional fine-grained concrete using river sand, the FGCMQS exhibited a 10.5% improvement in compressive strength but slightly higher chloride permeability and sulfate expansion. SEM analysis confirmed a denser microstructure with well-developed C-S-H and C-(A)-S-H gels. Despite durability trade-offs, the optimized FGCMQS presents a viable, eco-friendly alternative to traditional concrete, reducing cement consumption while offering enhanced strength. This study provides a foundational approach for developing high-performance, low-carbon fine-grained concrete, with potential applications in sustainable constructions. Future research should focus on durability enhancement under aggressive environmental conditions.