System-reliability-based disaster resilience analysis: Framework and applications to structural systems

As part of the recent effort to understand the performance capabilities of various engineering systems from their initial disruptions to the recovery phase, interest in the concept of resilience has been growing. In particular, to assess the disaster resilience of civil infrastructures subjected to natural or man-made hazards, various resilience criteria have been proposed. Given that infrastructures are complex systems consisting of components whose post-disaster performance capabilities are uncertain and interdependent, a system-reliability-based perspective is needed for a comprehensive evaluation of their disaster resilience. To this end, this paper characterizes disaster resilience from a system-reliability-based perspective in terms of three criteria: reliability, redundancy, and recoverability. These criteria are then discussed at each of the three scales of infrastructure systems, i.e., individual structures, infrastructure networks, and urban communities. Among the research needs and opportunities identified for the nine combinations of the resilience criteria and application scales (termed a "3x3 resilience matrix "), this paper focuses on a comprehensive assessment of the reliability and redundancy of an individual structure and proposes what is termed a "reliability-redundancy (beta-n) analysis " method along with a consideration of recoverability. For each of the initial disruption scenarios of component failures, the proposed analysis method computes the reliability index (beta) and redundancy index (n) based on the probabilities of the scenario and the corresponding system-level failure, respectively. Using a beta-n diagram that shows the pairs of the calculated indices for a given type of hazard, one can compute the system-level failure probability per hazard occurrence and identify critical initial disruption scenarios requiring further investigations and actions to assure proper disaster resilience. By incorporating a recoverability index into the beta-n diagram, decision-makers can identify top-priority initial disruption scenarios from a disaster resilience viewpoint. Numerical examples illustrate the proposed beta-n analysis method and demonstrate its general applicability and effectiveness during the effort to evaluate and manage the disaster resilience of structural systems. The source codes of the paper are available for download at https://github.com/Seonghyun-Lim/beta-pi_analysis.