Diss. Lyngby : DTU Transport, 2010

The central problem treated in this thesis is that of designing and maintaining a service network in a liner container shipping context. Based on a unified description of the individual planning processes involved in the definition of the liner container service network design (SND) problem, a series of integrated models for the SND problem are developed in aniterative fashion. The first models that are proposed represent simple abstractions of the SND problem but gradually, additional dimensions are added to obtain rich models capable of capturing many of the requirements imposed on the liner container SND. Two concrete problems, the liner container feeder service network design problem and the network transition problem are presented and analyzed in further detail with the purpose of providing tools to support the planning processes related to the design of the service network as well as the realization of a new design. The first problem, the liner container feeder service network design problem, addresses the tactical planning of the service network. The problem is inspired by a real-world case and problem specific structural properties are exploited to develop a new decomposition strategy. Essentially, the problem is decomposed into two types of sub-problems; a route generation problem and a route packing problem. The key to the success of this strategy is the introduction of the concept of a route pool which is iteratively augmented with route candidates using a dual based heuristic. Furthermore, dual estimation is used to select attractive routes from the route pool. Selected routes are managed in a master problem which is dynamically expanded as new routes and route packings are generated. Through the proposed decomposition strategy, a very rich representationof the liner service network design problem can be achieved modeling complex aspects such as service level dependent demand.