Impact of timing in post-warning prepositioning decisions on performance measures of disaster management: a real-life application

This research analyzes the impact of timing of post-warning and pre-disaster stock prepositioning decisions in disasters with an advance warning, such as hurricanes. From the warning time to the landfall time, disaster planners receive updated information regarding the hurricane's trajectory and the potential target region iteratively. The planners should decide when the best trigger time (TT) is to start the preparedness activities (prepositioning stocks of emergency goods). The quality of a given TT is evaluated concerning two performance measures: (i) the total logistics cost in the preparedness and response phases; and (ii) the minimum response time (RT) needed to transfer goods from the response facilities to affected areas in the response phase. A stochastic optimization model is designed to determine the best TT, preparedness decisions, and response operations. The computational complexity of the model is reduced using a graph-theoretic conversion. We test the converted model on experimental data based on major hurricanes from 2001 to 2017 on the United States southeastern coast. Sensitivity analysis of the model shows that delaying TT makes a non-linear reduction in the total logistics cost and non-linear increment in the shortest possible RT. The optimal TT makes the best compromise between the total logistics cost and the speed of the response operations, according to the preference of disaster planners.