Normative behaviour-based first-best analysis for disaster evacuation

The most uncertain aspect in designing an evacuation plan is the evacuation behaviours of individuals in the immediate aftermath of a natural disaster. Hence, this study attempts to reduce behavioural uncertainty and proposes a first-best analysis assuming that all evacuees perfectly follow the best evacuation strategy that is termed as normative behaviour. Although the first-best performance may not be practical or feasible, it clearly shows the benchmark performance of evacuation planning. This type of a benchmark is quite useful to understand the relative performances of several more practical scenarios. The results indicate that the first-best strategy is obtained by solving a Linear Programming in a time-dependent manner. Additionally, a practical solution algorithm is proposed based on the Lagrangian relaxation method. The method is then applied to the city of Ishinomaki, Japan, to confirm its computational feasibility.