Multicriteria evacuation plan for natural disasters
Résumé
Since the 90s, the problems of evacuation of persons have been extensively studied in the literature. The proposed models are classifed intotwo main categories: macroscopic and microscopic models. The DSS EVAC LOGISTIQUE project is interested in the evacuation of people in the context of flooding, burning or seismic event for which insecurity, capacity and time of crossing roads vary over time. Considering as solved the shelter locations problem, we develop discrete macroscopic models and methods incorporating a notion of risk inherent of the context studied for the evacuation of persons. The problem will be to determine a minimum overall evacuation time while minimizing the risk incurred by each evacuee. In this context, we propose two methods based on Shortest augmenting Paths without using a time-expended network. A e cient exact pseudopolynomial algorithm to solve the Earliest Arrival Flow and the Quickest Flows problems with time dependent data in graphs with one strongly connected component. An epsilonconstrains method for solving Quickest/Safest ows problem with time-dependent data. This research has been supported by the French National Agency of Research ANR-11-SECU-002-01- project DSS EVAC LOGISTIQUE.