A vulnerability-weighted stochastic framework for urban stormwater drainage resilience: Enhancing estimation accuracy and unveiling failure mechanisms.

Structural failures are widespread in stormwater drainage networks during long-term operation, making quantitative structural resilience assessment essential. However, current structural resilience assessments typically simplify the threat to the number of failed pipes, implicitly assuming that all pipes exert equal influence on system performance. This coarse representation weakens the explanatory power of the threat metric and requires extensive simulations for stable assessment; meanwhile, simplified failure scenarios can distort results. To address these issues, this study develops an efficient framework that introduces two key improvements: (i) a vulnerability-weighted threat metric whose statistical efficiency is benchmarked against the traditional equal-weight approach, and (ii) enhanced scenario realism through expanded pipe-failure states, complemented by targeted scenario sampling and efficient regression to improve computational performance. A case study demonstrates that, even when using only 1% of the simulation scenarios required by the traditional method, the proposed approach still yields more accurate fitted threat-performance curves. Error decomposition further confirms that the new metric yields a smoother threat-performance relationship and lower inherent variability, fundamentally improving assessment efficiency. Furthermore, attribution analysis is applied to derive intrinsic structural resilience by isolating the contribution of structural failures.