Capacity planning under uncertainty in synchromodal transport

Abstract

The shift towards more sustainable modalities such as rail transport is effective in reducing the negative externalities caused by unimodal road transport. However, despite the more sustainable alternatives being potentially cheaper, the share of road transport in continental freight transport in the EU remains high and has even increased in the last decade. The low adoption rate of intermodal transport, the transport of goods with multiple modalities in which freight remains in the same loading unit, is due to it being perceived as slower, less reliable and less flexible compared to road transport. The need for flexibility arises from the various uncertainties encountered in transport networks. Synchromodal transport is an extension of intermodal transport which leverages the advantages of multiple transport modalities with an integrated approach. It addresses the lower flexibility of intermodal transport, which forms a major barrier to the modal shift. The main features of synchromodal transport are synchronised operations between actors in the supply chain, mode-free booking, and real-time planning. This dissertation focuses on rail capacity planning under uncertainty in a synchromodal context from the perspective of logistics service providers (LSPs). LSPs are tasked with transporting customer orders through transport services. LSPs typically do not own the vehicles with which their orders are shipped, instead resorting to buying transport capacity from carriers such as rail operators. In contrast to road transport, rail services generally follow fixed schedules and transport capacity is bought for several months at a time. This causes greater demand uncertainty for LSPs compared to road transport, since their own customer orders only arrive on short notice. Due to this demand uncertainty, capacity plans are updated in the short term when demand is known. However, since rail capacity is bought externally, this introduces uncertainty on the remaining amount of capacity at the time of replanning. Two LSPs that make use of both rail and road transport were consulted to discover their main challenges related to intermodal transport. Both LSPs found the demand uncertainty when making long-term rail capacity commitments and the capacity uncertainty for short-term adjustments the most challenging aspects of intermodal planning. A literature review on uncertainty in intermodal and synchromodal transport is performed first to identify researched uncertainties and existing measures to mitigate their impact. The considered capacity planning decisions can be modelled as a service network design problem (SND). Findings from the literature review indicate that short-term replanning of intermodal services is not considered in existing SND models. Consequently, the impact of capacity uncertainty when replanning is not researched either. Moreover, in reality, the consulted LSPs do not wait for complete information before replanning their capacity. They update their capacity decisions twice: once based on partial information and once when complete demand information is available. In the second part of this thesis, we investigated the impact of intermodal capacity replanning with stochastic demand and capacity. To this end, we developed a three-stage SND model. Initial capacity decisions are made in the first stage, which are updated in the second stage based on partial demand information, and updated with complete information in the third stage. Capacity updates take on the form of additional bookings and cancellations. The model with two replanning stages is compared against a model with a single replanning stage, which only updates capacity when complete information is available, and a model without replanning. Additional comparisons against the case with perfect information indicates the cost of uncertainty. A full factorial experiment is performed to compare the performance across various market conditions. Our results reveal that replanning leads to improved planning under uncertainty, which is reflected by lower costs and higher rail shares. As such, it is effective in stimulating the modal shift. Factors which positively influence the impact of replanning are cheaper and more rail capacity on the market, more direct rail connections, more transhipment options, and increased demand uncertainty. In addition, replanning is more advantageous at lower costs of cancelling capacity and if the individual demand is negatively correlated with the market demand. Introducing replanning with complete information yields the largest improvements. Further cost savings and rail share increases of a second replanning stage, with partial information, are smaller. As such, there are diminishing returns with multiple replanning stages. The advantages of replanning with partial information increase with the amount of uncertainty that is resolved at the time of replanning. Coordinated planning is a core feature of synchromodality. However, the majority of literature on synchromodal transport focuses on the real-time aspect, whereas research on cooperation as a means to mitigate uncertainty is scarce. Therefore, in the third part of this thesis, cooperative capacity planning is researched, with and without replanning. A centralised cooperative approach between two LSPs is compared against the case with no cooperation. In the considered cooperative approach, the demands are pooled, reducing the demand volume fluctuations over the planning horizon and enabling a more efficient allocation of capacity to customer orders. Results of a sensitivity analysis indicate that cooperation is more advantageous in cases with sufficient capacity on the market for both participants. Cooperation yields the largest improvements for LSPs that experience large fluctuations in demand volume over their planning horizon, for instance if their demand structure consists of few large orders with uncertain time windows or due to low demand volumes. As such, smaller LSPs benefit more from cooperation. In addition, collaborative savings and rail share increases are larger with negatively correlated demand volumes between cooperants compared to uncorrelated and positively correlated demand. Our findings reveal that cooperation is effective at mitigating the impact of uncertainty, even in combination with replanning. However, as part of the uncertainty is mitigated through cooperation, the impact of replanning is diminished. The impact of a single replanning stage with complete information remains significant. In contrast, under cooperation, replanning with partial information only leads notable improvements if a large amount of uncertainty is resolved at the time of replanning.