PrioritEvac: an Agent-Based Model (ABM) for Examining Social Factors of Building Fire Evacuation
- PDF / 1,268,754 Bytes
- 14 Pages / 595.276 x 790.866 pts Page_size
- 32 Downloads / 144 Views
PrioritEvac: an Agent-Based Model (ABM) for Examining Social Factors of Building Fire Evacuation Eileen Young 1
&
Benigno Aguirre 1
Accepted: 5 October 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Fire evacuation modeling benefits from the application of social science both in terms of accuracy and external validation. This paper describes PrioritEvac, a novel agent-based model which incorporates the social dimension of group loyalty into fire evacuation and responses to fire and smoke. It uses individual priorities, making for a dynamic approach that allows greater agency and nuance. PrioritEvac is programmed in NetLogo and validated using extensive data collected from the Station nightclub fire. The statistical analysis of the results of the model indicate that, compared to historical patterns, it reproduces along multiple metrics including a mean of 114 deaths (std. dev. = 38) over 50 runs, which puts the actual result of the fire within one standard deviation of the mean results of the simulation. Overall, the mean differential along all the metrics is 79, significantly outperforming all published ABM models of the Station nightclub fire that did not incorporate social relationships. Keywords Egress . Agent-based model . ABM . Social relationships . The station building fire
1 Introduction
1.1 Problem Statement
In 2016, there were 475,500 structure fires in the United States (Haynes 2017). Combined, they caused 2950 civilian deaths, 12,775 civilian injuries, and $7.9 billion in property damage (US Fire Administration 2018). The number of people who died due to fires per million people in the population has decreased 15% in the last 10 years (US Fire Administration 2018), perhaps indicating that there have been improvements in the design and engineering of buildings as well as in the response to fire by fire personnel and in the training of civilians to respond to fires. Nevertheless, there is still room for improvement, particularly in the understanding of evacuations of people during fires and the best way to save additional lives. This research seeks, in part, to assist in this effort by presenting a novel evacuation simulation framework, PrioritEvac (Young 2019), designed to improve the use of this methodology by making open source software with a visualization component available.
Sociological as well as other personal and environmental factors influence evacuation behavior and have been studied using agent-based modeling (ABM), a popular approach which allows for the input of social rules that impact behavior in a stochastic rather than a deterministic way. In ABM it is possible to input an initial state and then watch social behavioral changes occur rather than inputting the expected outcome and then attempting to reverse-engineer the social factors which led to it. It is a modeling framework that can adequately accommodate the social forces in an initial state and incorporate individual priorities. Additionally, PrioritEvac can accommodate different floorplan
Data Loading...
