Modelling cascading failures in lifelines using temporal networks

Lifelines are critical infrastructure systems with high interdependency. During a disaster, the interdependency between the lifelines can lead to cascading failures. In the literature, the approaches used to analyze infrastructure interdependencies within the social, political, and economic domains do not properly describe the infrastructures’ emergency management. During an emergency, the response phase is very condensed in time, and the failures that occur are usually amplified through cascading effects in the long-term period. Because of these peculiarities, interdependencies need to be modeled considering the time dimension. The methodology proposed in this paper is based on a modified version of the Input-output Inoperability Model. The lifelines are modeled using graph theory, and perturbations are applied to the elements of the graph, simulating natural or man-made disasters. The cascading effect among the interdependent networks has been simulated using a spatial multilayer approach. The adjancency tensor has been used to for the temporal dimension and its effects. Finally, the numerical results of the simulations with the proposed model are represented by probabilities of failure for each node of the system. As a case study, the methodology has been applied to a nuclear power plant. The model can be adopted to run analysis at different scales, from the regional to the local scales.