Contributions of social science to agent-based models of building evacuation

Abstract

This paper describes the general characteristics of the agent-based models (ABM) approach to the simulation of building evacuations, and offers details of some popular computer codes for building evacuation. Contemporary ABM computer programming emphasising the characteristics of individuals and their propensities to act should be supplemented with codes informed by group-level dimensions such as norms, values, commitments to lines of action, leadership, and a sense of identification and membership in meaningful groups. The paper discusses social science considerations that need to be borne in mind when developing an ABM simulation of building evacuation in a disaster. It includes information on an ABM simulation of a disastrous incident: the fire in a US nightclub in Rhode Island in 2003. The model incorporates aspects of the physical environment in which the evacuation took place, the nature and spread of the fire and the smoke, information on individual and groups in the fire, and the impact of the density of human collectivities on the ability of individual and groups to evacuate. It concludes with seven suggestions for computer programmers developing ABM of human behaviour in building evacuation.

Acknowledgements

This research was supported in part by the Disaster Research Center, University of Delaware; the Department of Civil and Environmental Engineering, University of Michigan; and the National Science Foundation through Grant Numbers SES-0824737 and 0825182. The authors are solely responsible for the contents of the paper.