Mitigating supply chain disruptions through interconnected logistics services in the Physical Internet

TLDR

Traditional supply chain design creates independent heterogeneous networks, whereas the Physical Internet seeks to integrate them into a global, open system, and prior work shows PI‑enabled inventory models can reduce inventory levels through increased flexibility. The study examines how PI‑based inventory models handle disruptions at hubs and plants and proposes a simulation‑based optimisation model to determine inventory control decisions. The authors model a single‑product inventory problem with uncertain demand and stochastic supply disruptions and use a simulation‑based optimisation model to determine inventory control decisions. The PI inventory model outperforms classic models in resilience, with performance gains growing as product value, penalty costs, and disruption frequency rise, indicating a novel approach to building resilient supply networks.

Abstract

This paper investigates the resilience of inventory models using interconnected logistics services in the Physical Internet (PI). With traditional supply chain network design, companies define and optimise their own logistics networks, resulting in current logistics systems being a set of independent heterogeneous logistics networks. The concept of PI aims to integrate independent logistics networks into a global, open, interconnected system. Prior research has shown that new inventory models enabled by and applied to PI could help reduce inventory levels thanks to its high flexibility. Continuing along these lines, this paper examines how inventory models applying PI deal with disruptions at hubs and plants. To attain this, a single product inventory problem with uncertain demands and stochastic supply disruptions is studied. A simulation-based optimisation model is proposed to determine inventory control decisions. The results suggest that the PI inventory model, with greater agility and flexibility, outperforms the current classic inventory models in terms of resilience. Moreover, the difference in performance increases when the product value, penalty costs and disruption frequency increases. This paper indicates a novel approach to build a resilient supply network.