The Siting Of Multi-User Inland Intermodal Container Terminals In Transport Networks

Abstract

Almost without exception, cargo movements by sea have their origins and destinations in the hinterlands and efficient land transport systems are required to support the transport of these cargo to and from the port. Furthermore, not all goods produced are exported or all goods consumed are imported. Those produced and consumed domestically also require efficient transport to move them from their production areas to areas of consumption. The use of trucks for these transport tasks and their disproportionate contribution to urban congestion and harmful emissions has led governments, transport and port authorities and other policy-makers to seek for more efficient and sustainable means of transport. A promising solution to these problems lies in the implementation of intermodal container terminals (IMTs) that interface with both road and rail or possibly inland waterway networks to promote the use of intermodal transport. This raises two important linked questions; where should IMTs be located and what will be their likely usage by individual shippers, each having a choice of whether or not to use the intermodal option. The multi-shipper feature of the problem and the existence of competing alternative modes means the demand for IMTs are outcome of many individual mode choice decisions and the prevailing cargo production and distribution patterns in the study area. This thesis introduces a novel framework underpinned by the principle of entropy maximisation to link mode choice decisions and variable cargo production and distribution problems with facility location problems. The overall model allows both decisions on facility location and usage to be driven by shipper preferences, following from the random utility interpretation of the discrete choice model. Several important properties of the proposed model are presented as propositions including the demonstration of the link between entropy maximisation and welfare maximisation. Exact and heuristic algorithms have been also developed to solve the overall problem. The computational efficiency, solution quality and properties of the heuristic algorithm are presented along with extensive numerical examples. Finally, the implementation of the model, illustration of key model features and use in practice are demonstrated through a case study.