Predicting road blockage due to building damage following earthquakes

Transportation infrastructure supports the social and economic activities of communities. One of the impacts of roads’ disruption is the obstruction of emergency services (e.g., ambulance, firefighting, evacuation). Furthermore, the recovery process of a community following an extreme event (e.g., a natural hazard) depends on the functionality of the transportation infrastructure. Therefore, conducting a risk and resilience analysis of transportation infrastructure is critical to help communities minimize the initial impact of hazards and promote a rapid recovery. Current approaches model the probability of road blockage due to building damage using high-resolution optical satellite images and aerial photographs collected after past events. However, the data used by these methods are limited, and few data have been collected before 2010. Besides, data may not be available for specific regions that have not experienced recent earthquakes. Thus, a probabilistic predictive method is needed for risk and resilience analysis of roads. This paper proposes a probabilistic model using the data from the 2010 Haiti Earthquake and calibrated by Bayesian approach to predict the debris distance from undamaged buildings (e.g., the distance debris can reach from the footprint of the undamaged building). The model is then used to construct fragility curves that give the conditional probability of road blockage at a given road section for a given seismic intensity. The proposed model considers the relevant factors affecting the road blockage probability, including building types, damage level, and road characteristics. The probability of road blockage at a given road section is estimated for the four general road section types, considering buildings on only one side of the road or both sides, and with or without a raised traffic median. The probability of road blockage for an entire road is then calculated by system and parallel reliability analysis. The proposed models apply to any general urban area without the dependence on historical data from past earthquakes.