Improved Lagrangian-relaxation based approaches for multi-period multi-stage fixed charge transportation problem

Abstract

This paper considers the Fixed-Charge Transportation Problem (FCTP), which spans multiple time periods and includes multiple stages, termed as Multi-Period, Multi-Stage Fixed-Charge Transportation Problem (MPMS-FCTP) in a supply chain. A generalised mathematical model which minimises the fixed and variable transportation, storage and backlog costs is proposed for the problem under study. Subsequently, two improved Lagrangian-Relaxation (LR)-based approaches are developed to efficiently solve the MPMS-FCTP. In addition, heuristics based on the improved LR approaches are proposed to obtain good upper bounds. The experimental results show that the proposed LR approaches perform better than the traditional LR approach with the sub-gradient optimisation. The lower bounds obtained using the proposed approaches are always non-decreasing in successive iterations of LR, providing tighter lower bounds. In addition, the proposed LR-based heuristics yield better solutions than the existing heuristics in the literature. Finally, the convergence to optimality of the proposed approaches is also discussed.