Balancing traffic flow in the congested mass self-evacuation dynamic network under tight preparation budget: An Australian bushfire practice

Abstract

Disaster preparation budget plays a key role in the success of an emergency response operation. This research presents a novel bi-level model for mass self-evacuation discrete network design problems to find the best balance between the preparation budget and evacuation congestion. In this model, the importance of including non-compliance behavior in evacuation is considered. The proposed model aims to design an evacuation network with the smoothest traffic flow by using the maximum capacity of a dynamic transportation network where a number of roads become inaccessible as the bushfire spreads. The output of this research aids the emergency authorities to decide the best roads and shelters to equip in the evacuation process considering the access orders of the roads. The proposed approach covers the last four key phases of the entire evacuation process. Benders Decomposition and heuristics accelerators have been employed to help generate the outputs in less computational time for large-scale instances. The model and the solution approach have been validated through numerical experiments in different sizes. In addition, the model has been tested on the real case of the Churchill bushfire, Victoria. The results show that altering a number of factors such as budget and the number of allowable shelters will significantly reduce the overall evacuation time and makes the roads less congested. A sensitivity analysis is also conducted which demonstrates that it is possible to reduce traffic congestion without raising the budget.