An agent-based approach for supply chain retrofitting under uncertainty

Abstract

In this work, decisions that have a long lasting effect on the supply chain (SC) such as the design and retrofit of a production/distribution network are considered. The retrofitting tasks are accomplished by using a SC agent-oriented simulation system, which model each entity belonging to the SC as an independent agent. The starting point is a set of possible design options for the existing SC. For each design alternative a performance index is obtained through the agent-based framework by looking for the best value of the operational variables associated to the resulting network. The proposed methodology allows to address the design of complex SCs which are hard to be modelled otherwise, for example by means of standard mathematical programming tools. Specifically, the multi-agent system is suitable for SCs that are either driven by pull strategies or operate under uncertain environments, in which the mathematical programming approaches are likely to be inferior due to the high computational effort required. The advantages of our approach are highlighted through a case study comprising several plants, warehouses and retailers.