Functionality Assessment of the Seismic-Damaged Lifeline Systems under Cascading Failures

Abstract

Lifeline system is a highly complex network consisting of diverse components that are spatially distributed and interconnected each other. As such, during an earthquake, it is common that the system encountered problems in maintaining reliable operation. Moreover, damage at a single-site component readily propagates to other interdependent components in same and different lifeline systems. In this context, many researchers have continued their efforts to offer useful indices to measure the degraded performance and to ensure the constant service supply of the lifeline systems. The essential research perspective, thus, shifts to understanding the secondary disruptions in the lifeline systems and how malfunctions arise. However, complex inter-dependency is still made challenges in estimating the lifeline system performance under abnormal conditions.

Therefore, this research develops a comprehensive framework for functionality assessment of the seismic-damaged lifeline systems to solve the problems: (a) destruction due to ground shaking, (b) reduction of inflow due to internal/external dependency, and (c) demand fluctuation due to environment changes. In detail, target of estimation is divided into ground motion at particular site, common-cause failure, cascading failure (in terms of internal and external dependency), and escalating failure. In particular, this research use inoperability input-output model incorporating Bayesian network (BN) and System dynamics (SD). To be specific, BN can facilitate prediction of the probability of the unknown event base on the input information or spatial path analysis in situations of data scarcity. On the other hand, SD can be handled demand fluctuation during an earthquake. Due to the inherent uncertainty in earthquake occurrences, this research conducts scenario-based performance assessment using the data from the 2011 Tohoku earthquake and the 2016 Gyeongju earthquake. The analysis results show that the operational state of a component is even dependent through the availability of input inflow from adjacent components rather than its physical damage. Moreover, since the actions taken immediately following an earthquake can play a significant role on the extent of cascading failures, this research provides useful information for those with a concern in the community resilience maintaining.