Reliability Boundaries under Synthetic Post-Earthquake Road Disruption

A Reproducible Graph-Based Monte Carlo Study of Emergency-Response Accessibility in Beykoz, Istanbul

Abstract

This study investigates how responder availability and resource-allocation strategy affect incident accessibility when portions of a road network become unavailable following severe disruption. The road network of Beykoz, Istanbul, was represented as a computational model, and road disruptions were simulated across multiple stochastic scenarios. The principal reliability measure was the 80/80 boundary, defined as the disruption level at which the probability of maintaining access to at least 80% of incidents reaches 80%.

The results showed that the estimated 80/80 boundary increased from 0.1425 with 12 responders to 0.1625 with 32 responders. The study also compared a sequential Greedy assignment strategy with a Global Minimum-Cost assignment approach. Across all 102 tested cases, the Global Minimum-Cost method produced lower total response time than the Greedy method. However, the two approaches yielded identical accessibility-boundary estimates at all six tested resource levels.

These findings suggest that improved allocation can make more efficient use of responders and road connections that remain available, but optimization alone cannot restore accessibility once the underlying road network loses connectivity. The disruption-severity parameter used in this study is a synthetic network-stress variable and should not be interpreted as earthquake magnitude, peak ground acceleration (PGA), a physical damage measure, or an operational safety threshold.