
Abstract The bearing capacity and settlement behavior of foundation piles are subject to considerable uncertainty due to the inherent variability in soil properties and structural loads. A global-scale, site-specific reliability analysis is presented based on 1,172 static load tests from 60 sites across 19 countries. Load–settlement curves were modeled using both power-law and hyperbolic regression functions, and uncertainty in the regression parameters was quantified. A geometric reliability index was computed directly in the physical space through a limit-state equation associated with serviceability criteria. At a standard settlement threshold of 40 mm, the power-law model consistently yields higher estimated bearing capacities and larger reliability indices than the hyperbolic model. For each site, an evaluation index circle was constructed to visualize the reliability index, factor of safety, and correlation coefficients. The power-law model is shown to provide a more robust and interpretable basis due to its ease of fitting and physical interpretability. Shapley additive explanations–based interpretability analysis indicates that the coefficient of variation of the regression parameters is the dominant factor governing the computed reliability index. The findings of this study aim to support the alignment of pile foundation design codes with those used in the structural engineering domain.
Abstract This study proposes a novel probabilistic framework for assessing cascading failure risk in transmission systems subjected to hurricane hazards. The framework integrates the probability of initial component outages with cascading failure analysis to identify critical scenarios characterized by both high likelihood and severe system-level consequences. Transmission system impacts are evaluated using a direct-current power flow model, while outage probabilities are derived from transmission-line fragility functions combined with joint probability distributions of maximum wind speeds constructed from historical hurricane records. The proposed methodology is demonstrated on the Institute of Electrical and Electronics Engineers 39-bus New England power system using data from 119 historical hurricanes. A vine-copula model is employed to capture complex spatial dependencies of wind extremes across the network, enabling realistic representation of correlated wind loading and simultaneous line failures. Results indicate that transmission lines located in a close proximity or along dominant hurricane trajectories are particularly vulnerable to concurrent extreme wind events. The most critical components are identified in southern coastal New England, where high load demand, limited network redundancy, and frequent intense hurricanes collectively amplify cascading failure risk. The findings, supported by sensitivity and uncertainty analyses, provide insights for transmission system planning and risk-informed mitigation strategies against hurricane-induced disruptions.
Abstract Despite ongoing efforts and advancements in safety management, metro construction in China continues to experience frequent accidents due to complex, dynamic, and nonlinear risk factors. Existing studies often focus on isolated risk factors and overlook the complex, multidimensional coupling mechanisms among various hazards, thereby limiting the practical impact of current prevention measures. To address this critical gap, this study proposes a novel topology analysis method based on knowledge graphs to systematically uncover the hidden patterns and structural logic of metro construction accident risks. A comprehensive metro construction accident knowledge graph (MCAKG) is constructed by extracting data from 214 official accident investigation reports spanning from 2003 to 2025. The MCAKG is specifically designed to explore and quantify hazard interactions across six distinct dimensions: hazards, hazard types, hazard occurrence frequencies, hazard occurrence times, accidents, and accident consequences. Furthermore, targeted topology indicators, including the direct and indirect casualty indices, are developed to assess the severity and mediating roles of different hazards within the global network. The topology analysis reveals that the harmful effects of accidents can be significantly mitigated by systematically identifying and eliminating key intermediary hazards. Notably, the results demonstrate that removing eight key hazards can reduce the total direct casualty index by 80% and the indirect casualty index by 90.5%. Ultimately, this study provides a robust, data-driven approach to metro construction safety management, offering targeted risk mitigation strategies that enhance overall construction safety.
Abstract In structural reliability analysis, the estimation of the failure probability function (FPF) is often computationally prohibitive and numerically unstable. To address this issue, this paper proposes an augmented-space integral–based Hamiltonian Monte Carlo subset simulation (ASI-HMC-SS) method for efficient FPF estimation. Based on the Bayesian formulation, the distribution parameters are treated as auxiliary random variables with instrumental prior distributions. Accordingly, the FPF is formulated based on the failure probability in the augmented space and the parameter posterior density conditioned on failure, which supports efficient global FPF estimation through a single-run calculation. In the conditional sampling stage of subset simulation, a Hamiltonian Monte Carlo (HMC) scheme with an adaptive trajectory-length adjustment strategy is incorporated to significantly improve sampling efficiency and numerical stability in failure domains. Several static and dynamic engineering examples are presented to validate the proposed method. The results suggest that, in terms of efficiency and accuracy, the proposed method outperforms conventional augmented-space subset simulation methods. This superiority highlights its significant potential for engineering applications.
Abstract Service contracts in logistics and transportation define acceptance conditions that must be verified against timestamped obligation records. However, resilience and reliability metrics usually describe degradation and recovery without producing a third-party auditable compliance verdict. This paper defines a reproducible two-level limit-state protocol to obtain such a verdict under finite evidence and uncertain recovery dynamics. The framework is not tied to a specific network topology and applies to services where obligations are set ex ante and classified as compliant, noncompliant, or failed under declared rules. The aggregated service process X Δ ( t ) is computed over a contractual denominator that cannot be revised ex post. Two limit states are defined: the service limit state, triggered by threshold crossing, and the recovery limit state (RLS), triggered when cumulative time below threshold within a recovery window exceeds D max . Level 1 yields a deterministic verdict from the accepted record, including conservative treatment of low-evidence days through one-sided Clopper–Pearson bounds. Level 2 estimates P ( RLS | M , θ , R ) by Monte Carlo under a declared occurrence–recovery generator, with one-at-a-time sensitivity to epistemic assumptions. The protocol is illustrated on a Chile–Europe reefer maritime transit through two reproducible normative examples. In Level 1, two underthreshold episodes are detected, and the global verdict shifts from compliant to noncompliant when D max is reduced from 22 to 19 days. In Level 2, the Monte Carlo estimate is 0.638; sensitivity ranges from 0.55 to 0.71 for frequency and duration perturbations and falls to 0.00 when the degraded-regime mean crosses the threshold. The result is a checking rule for auditors and contract managers.