Verification and Validation of Agent-Based Models for Resilience Analysis and Simulation

Abstract

Natural hazards can inflict significant damage on communities. It may take a long time for hazard-stricken communities to recover from the losses induced by natural hazards. The concept of community resilience has received increased attention in recent decades as hazards have occurred at an increasing rate and magnitude. To model the postdisaster recovery process, an agent-based modeling approach was developed to simulate the behavior of different systems of the community as well as their interactions. For a model to simulate a certain process or solve a specific problem, a verification and validation (V&V) process is crucial. This paper presents the V&V process of the agent-based model (ABM) for resilience analysis of hazard-stricken communities. Tornado hazards, which occur at an ever-increasing rate in the US, were selected as the hazard of interest. Centerville, a virtual community located in the midwestern US, was chosen as the numerical testbed to exemplify the V&V process for the simulation models. A detailed V&V process indicated that the ABM developed for the community resilience analysis is able to characterize the postdisaster recovery process of a community, and therefore, this simulation model provides validated results of resilience assessment.