Maintenance plays a crucial role in the management of rail infrastructure systems as it ensures that infrastructure assets (e.g., tracks, signals, and rail crossings) are in a condition that allows safe, reliable, and efficient transport. An important and challenging problem facing planners is the scheduling of maintenance activities which must consider the movement and availability of the maintenance resources (e.g., equipment and crews). The problem can be viewed as an inventory routing problem (IRP) in which vehicles deliver product to customers so as to ensure that the customers have sufficient inventory to meet future demand. In the case of rail maintenance, the customers are the infrastructure assets, the vehicles correspond to the resources used to perform the maintenance, and the product that is in demand, the inventory of which is replenished by the vehicle, is the asset condition. To the best of our knowledge, such a viewpoint of rail maintenance has not been previously considered.

In this thesis we will study the IRP in the rail maintenance scheduling context. There are several important differences between the classical IRP and our version of the problem. Firstly, we need to differentiate between stationary and moving maintenance. Stationary maintenance can be thought of having demand for product at a specific location, or point, while moving maintenance is more like the demand for product being distributed along a line between two points. Secondly, when performing maintenance, trains may be subject to speed restrictions, be delayed, or be rerouted, all of which affect the infrastructure assets and their condition differently. Finally, the long-term maintenance schedules that are of interest are developed annually. IRPs with such a long planning horizon are intractable to direct solution approaches and therefore require the development of customised solution methodologies.