Tackling complexities of cyclic inventory routing under conditions of limited modelling and computing capacity

Abstract

This paper presents a tractable approximation method for reliably finding near-optimal solutions to the cyclic inventory routing problem (IRP) without requiring the specialised modelling skills and computing technology to deploy more exact solution approaches. Together with simultaneously attaining high solution quality and efficiency, the method's contribution to the IRP literature stems from modelling transportation cost in a way that parsimoniously captures (a) complexities of estimating vehicle requirements and (b) the heretofore ignored phenomenon of possible correlation between customer locations and demands. An extensive set of computational settings to test the method's performance indicated that vis-à-vis a well-founded benchmark from the literature, the average gap in cost is just slightly above 1%, with a worst-case gap of 4.8% across all settings.

Keywords:

• inventory routing
• vehicle routing
• heuristics
• transportation cost modelling
• computational experiments