EVACUATION SIMULATION WITH LIMITED CAPACITY SINKS An evolutionary approach to solve the shelter allocation and capacity problem in a multi-agent evacuation simulation

Evacuation modeling is an important problem that has been intensively studied in the last 40 years. In a general evacuation situation, there might be more then one safe place. A common approach to reduce this multidestination problem to a single destination problem is to connect all safe places via zero cost links to a super sink. However, this approach works only as long as the safe places have no capacity constraints. In this paper, we present an approach to solve an evacuation problem where the safe places have limited capacities. This is, for example, the case if the safe places are shelter buildings for tsunami evacuation. The problem is solved through an evolutionary learning algorithm for the combined routeand shelter-assignment problem combined with a heuristic approach for the fair minimization of shelter capacities. Different behavioral assumptions (”fair” vs. ”globally optimal”) are investigated. The proposed approaches are discussed in the context of a real-world tsunami evacuation problem.