A Bayesian Network Modeling Approach for Time-Varying Flood Resilience Assessment of Housing Infrastructure System

Abstract

Natural hazard causes severe types of damage to infrastructure systems at regular intervals, and the occurrence of such events is inevitable. The concept of resilience is adopted to make infrastructure systems more reliable, adaptable, and recoverable against natural disasters. Resilience is defined as the ability of an infrastructure to resist the impact of a disaster and bounce back to its desirable performance level after the disaster. Recovery of infrastructure, which forms a part of resiliency, is a time-dependent process in nature. In this work, a dynamic Bayesian network (BN) model is developed for the resilience assessment of housing infrastructure against flood hazards. The proposed resilience model is then implemented in the testbed of Barak Valley in North-East India. An extensive field survey is performed to collect relevant indicator data for resilience assessment and validation. To assess the recovery process, the housing infrastructure resiliency of the Barak Valley testbed is evaluated and compared between multiple flood event time periods. Lastly, the most critical indicators of the proposed dynamic BN model are identified by performing sensitivity analysis. This study will help the planner, designers, policymakers, and stakeholders to provide resilience-based decisions on flood resiliency of housing infrastructure systems.