Quantifying Directional Dependencies from Infrastructure Restoration Data

Abstract

Lifeline utilities and critical infrastructures are becoming increasingly interactive and dependent on one another for normal operation. With a natural disaster or disruptive event, these dependencies can be studied under stressed conditions. To replicate events and inform future simulations, such dependencies can be quantified in both magnitude and direction. This paper builds on recent efforts by proposing a new dependency index methodology that gives importance to the direction of dependency between coupled infrastructures and equally weighting the multiple dependencies that may be realized across a variety of lag times. The effectiveness of this methodology is presented as a case study for the 22 February 2011 earthquake experienced in Christchurch, New Zealand. Dependencies are quantified for a range of critical infrastructure couplings, which provide insight into the future application of these results and the requirement for integration with qualitative studies to accurately inform interdependency models.

A Generalized Natural Hazard Risk Modelling Framework for Infrastructure Failure Cascades

“…This offers the versatility to calibrate and adjust parameters in the model based on evidence, such as tailoring impact functions to match print media coverage on structural damages, or amending dependency heuristics to fit utility provider's outage reports. To the best of our knowledge, only few quantitative modelling studies [71] incorporate such feedback possibility. Obtaining results on direct and cascading infrastructure failures further allows to quantify the role of infrastructure dependencies in causing wide-spread impacts:…”

Post-disaster infrastructure restoration: A comparison of events for future planning

“…Such functions tend to suit those datasets where recovery is modeled back to the initial pre-disruption conditions due to observed datasets commonly showing restoration rates slowing over time [20,56,80]. However, as discussed, the proposed methodology seeks to only address the response phase meaning that functions that asymptotically tail-off are undesired.…”

“…For telecommunications networks, an increase in the fraction of inoperability beyond the initial event is frequently observed when redundant sources of electricity are drained soon following a disaster event [24,50,80]. For this reason, the time of positive restoration is taken as the maximum outage -typically observed at t 0 for the water and electricity infrastructures.…”

Congestion and observability across interdependent power and telecommunication networks under seismic hazard

“…It represents the magnitude of interdependency between two network components. Previous studies have quantified coupling strengths between two infrastructure networks from time-dependent correlations between restoration trajectories (Dueñas-Osorio and Kwasinski, 2012; Zorn and Shamseldin, 2016). However, available system-level coupling strengths lack component-level details, which we estimate here as the ratio of the number of outages of a central office due to substation outage to the total number of substation outages in a given time for a given event.…”