Cascading Failure Propagation and Perfect Storms in Interdependent Infrastructures

Abstract

The increasingly complex conditions that are reshaping environments demand novel analysis of infrastructure weaknesses and behavior. Of critical concern are cascading failures and how small disruptions can spiral into large-scale outages. Significant evidence indicates infrastructures are increasingly stressed given a combination of disruptions including extreme climate events, disrepair, cyberattacks, and emerging and disruptive technology integration. Small disruptions appear increasingly likely to cascade to larger failures within and beyond infrastructures, and there is limited insight into how to protect systems that are increasingly integrated. As novel capabilities emerge to expedite the analysis of cascades (namely, synthetic infrastructure models and open-source network solvers), new opportunities exist to provide critical insights into cascading failures. Using the City of Phoenix as a case study, synthetic power and water networks are constructed and coupled, and disturbances are simulated to capture cascading failure behaviors within and across power and water distribution systems. Network solvers [PyPSA (version 0.24.0) for power and EPANET (version 2.2) for water] are used to capture network rebalancing. Failures are simulated starting with transmission line outages and 120,000 simulations used to capture stochasticity in the rebalancing of power and water systems and resulting differences in failure dynamics. In 89% of the simulations initial transmission line outages did not cause outages at substations or in water systems. Power failures did not lead to water outages in 96% of simulations. Despite significant variability in the networks, emergent failure patterns are observed when substation and resulting pump outages occur—a critical insight for resilience planning. Approximately 3.69% of the simulations lead to large cascading failures across both power and water systems. Furthermore, low-likelihood but high-consequence perfect storm outcomes were observed in many of the simulations, often resulting in widespread power outages. Combined, the results provide important insights for resilience planning across increasingly vulnerable and interdependent infrastructures.