In a railway system, empty freight cars have to be repositioned to enable transportation demand to be fulfllled. These empty car movements are planned and implemented in the empty freight car distribution process. Shortcomings in the planning instrument is a contributory cause to large distribution costs, delayed deliveries of freight cars, and au underutilized car fleet. In this thesis, we study optirni2ation models and methods that could constitute the kernel of a distribution planning system, resulting in a more efficient distribution process.

The first model, as well as all the following, includes a heterogeneous car fleet and transportation capacities on trains. The results show that the model is tractable and that it is important to take the train capacities into account in the planning. The modeling ideas are further developed in the second model, which is used for operational empty distribution planning at Swedish Railways. The problem conditions are changed continuously and the proposed distribution plan is fixed in small portions. The model is decomposed both geographically and by freight car type. The subproblems are solved in an on-line setting to calculate if empty freight cars can be delivered to demands. A heuristic is used to coordinate and reallocate capacity assigned to the subproblems.

The third model captures the economies of scale in the empty distribution process by including a fixed cost for each transported car-cluster. The model is solved with a tabu search heuristic and the results show that the nmnber of car-clusters transported in the empty distribution can be significantly reduced without large increase in the total transportation distance. Finally, the fourth model is designed to study the correlation between the empty freight car distribution and the dimensioning of the transportation capacities in the tactical planning. The results illustrate the possibilities and advantages to consider the empty distribution already in the tactical planning.