Logic-Based Benders for intermodal operations with delay penalties

TL;DR

This paper introduces a novel Logic-Based Benders Decomposition for intermodal logistics, effectively reducing costs and penalties by exploiting the problem's staged structure, with demonstrated improvements on real and random instances.

Contribution

The paper develops and validates a new Logic-Based Benders Decomposition tailored for intermodal logistics with delay penalties, including extensions for real-world applications.

Findings

Significant reduction in solution gaps and computation time compared to monolithic MILP.

Effective application of the method to large-scale random instances.

Versatile extension supporting different planning approaches in real logistics cases.

Abstract

Intermodal logistics typically include the successive stages of intermodal shipment and last-mile delivery. We investigate this problem under a novel Logic-Based Benders Decomposition, which exploits the staged nature of the problem to minimise the sum of transport costs and delivery penalties. We establish the validity of our decomposition and apply effective optimality cuts. Apart from models and formal proofs, we provide extensive experimentation on random instances of considerable scale that shows the improvement achieved in terms of small gaps and shorter time compared to a monolithic MILP approach. Last, we propose a major extension of our generic method for a real logistics case. The implementation of the extension on real instances show the versatility of our method in terms of supporting different planning approaches thus leading to actual cost improvements.