LAUSR

Abstract When a disaster strikes many roads are blocked and the affected network may break up into a number of isolated parts. The reconnection of the network is therefore necessary for both relief distribution and planning of construction work. Shortening the time during which the road network is separated into isolated parts helps decrease indirect losses from disasters. The obstacles usually faced during the process of reconstruction include both the large number of blocked links and extensive affected areas (road networks). A reduction of the network into a much smaller complete graph and metaheuristic based on an ant colony optimization has been introduced to overcome this issue. We demonstrate that, for small networks, the metaheuristic produces the same results as other deterministic algorithms. We further show that the method is still a viable approach for large networks (723 nodes and 974 links, where we artificially blocked 46 links) when the NP-hard nature of this problem began to affect the computational time of the deterministic algorithms. We demonstrate how the various scenarios can be included into the algorithm. We finally introduce a new ranking of feasible solutions which enables the algorithm to minimize the time of reconstructions for all repair units. Reasonable results were obtained after five minutes of computation. There is nevertheless an up-to-38% improvement of the initial solution. The algorithm can also be used for both relief distribution, when no roads were damaged, and for planning of construction work when damaged roads occur.