Intermodal service network design with stochastic demand and short-term schedule modifications. (September 2021)

• Record Type: Journal Article
• Authors: Müller, Jan Philipp; Elbert, Ralf; Emde, Simon
• Is Part Of: Computers & industrial engineering. Volume 159(2021)
• Publication Date: 2021-09
• Subjects: Stochastic service network design -- Intermodal transportation -- Matheuristics -- Stochastic demand; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2021.107514
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: We introduce a new variant of the intermodal service network design problem. Ad-hoc modifications of the schedule are possible to cope with stochastic demand. A new matheuristic solution approach clearly outperforms a commercial solver. Results of a real-world case study for a large intermodal operator are presented. Substantial cost savings can be achieved compared to an expected value model. Abstract: Service network design is a crucial planning task for intermodal operators on the tactical level as it defines the transportation schedule and the routing of commodities through the intermodal network. In many cases, the transportation demand over the planning horizon cannot be known with certainty, resulting in a stochastic service network design problem. Whereas current literature focuses on outsourcing as a recourse action when reacting to fluctuating demand patterns, we introduce a new variant of this problem in which an ad-hoc modification of the planned schedule for the transportation assets is possible. Although this recourse action is commonly applied in practice, a detailed investigation including high-performance solution procedures for this specific service network design problem has been missing until now. We present a two-phase matheuristic and respective computational experiments that demonstrate its superior performance compared to commercial solvers, including the capability of solving large stochastic instances. Moreover, by presenting the results of a real-world case study for a large intermodal rail operator, we illustrate substantial cost reductions compared to an expected value model and a stochastic model without consideration of schedule modifications. Our detailed analysis further reveals a different solution structure when schedule modifications are considered which explains the cost savings achieved.