A three-stage service network design model for intermodal transport under uncertainty

Abstract

Intermodal transport is the transport of freight with multiple modes whereby freight remains in the same loading unit. The combination of road transport with other high-capacity transport modes such as trains and barges is potentially more sustainable than unimodal road transport for long-haul freight shipping. A drawback of those high-capacity transport modes is their lower flexibility compared to trucks, hindering the modal shift. Logistics service providers typically book capacity in advance on those services, at which point complete demand information is missing. This work aims to assist logistics service providers in their implementation of intermodal transport, which is done with a capacity decision support model. The proposed model tackles a service network design problem with stochastic demand. In this type of problem, the aim is to select transport services based on the estimated demand and route freight through those services. In academic literature, these types of problems are studied with two-stage models. The first stage takes place before demand has materialised and only the demand distribution is known. Capacity on each service is determined at this stage. The second stage occurs in the short term when complete demand information is available. Routing decisions are taken at this stage, as well as recourse actions in case the capacity selected in the first stage is insufficient. A shortcoming of these two-stage models is that recourse actions are only performed once complete demand information is available. In reality, transport orders arrive up to a few days in advance when it might be too late to perform large updates to the transport plan. Logistics service providers revise their capacity gradually over time as new orders arrive. To more accurately represent reality, we propose a three-stage model that includes an additional intermediate stage. A logistics service provider is the decision maker in our proposed stochastic optimisation model. It considers a rail-road intermodal network in which train services are offered by third parties. Since different logistics service providers can book those same services, the remaining amount of capacity on the market declines over time. Included uncertainties are stochastic demand and a stochastic decrease in available capacity. Regarding demand, it is assumed that only the distribution of total demand is known in the first stage. In the second stage, additional information on the total demand in the transport market leads to more accurate forecasts of demand volume. Demand materialises in the third stage. The model includes capacity decisions taken in each of the three stages and routing decisions in the last stage. Stochasticity is captured with a scenario tree and the objective function minimises the average cost over all scenarios. The model is solved with an exact algorithm with a time limit.

This work is supported by VLAIO (cSBO project DISpATch, HBC.2016.0412)