Network design to anticipate selfish evacuation routing

Abstract

When a disaster occurs the population of the endangered zone must sometimes be evacuated as fast as possible. In this case, a large number of vehicles move through a street network to reach safe areas. In such a situation it might be impossible to communicate the routes to the evacuees they have to choose to optimize the traffic flow, moreover it is difficult to ensure that the evacuees take the communicated routes. With our approach we optimize the traffic routing without determining optimal routes for every evacuee. In the developed method, the street network for a given traffic flow is optimized. With the blockage of street segments we reach an improvement of traffic distribution what leads to a better traffic flow and results in a faster evacuation. To integrate human behaviour every evacuee is modelled as an independent acting agent that chooses a route dependent on her preferences. So the individual behaviour of the evacuees and also the structure of the street network are integrated in the solution. In the computational study, the results from the unmodified network, the network modified with our method and a solution where the optimal routes for every evacuee are given are compared. We also compare different implementation variants of our approach to investigate which works best. The results of the computational study indicate that our approach reduces the negative influence of selfish routing on the evacuation.