Agent-based experimental investigations of the robustness of tactical planning and control policies in a softwood lumber supply chain

Abstract

This article investigates the robustness of different tactical planning and control policies for a softwood supply chain using an agent-based environment that simulates a distributed advanced planning and scheduling system and its corresponding supply chain operations. Simulations were modelled using a novel agent-based methodology combined with a robust experimental design approach and an industrial data set obtained from two companies in Eastern Canada. Experimental results provide insights about the dynamic relations among factors related to control levels, planning methods and planning horizon lengths. Results indicate that supply chain control levels play a relevant role in defining robust service levels, while the planning horizon and the planning method have lower impact in this context. In addition, from the supply chain inventory level point of view, we verified that the three investigated tactical rules have to be configured together if one desires to maximise their contribution for a robust supply chain system capable of coping with uncertainties from the business environment. When these rules are evaluated individually, it is not possible to make the most of their potential due to interactions between them. The article concludes by proposing an optimum robust configuration of the tactical rules to minimise the impact of supply chain uncertainties.

Keywords:

• supply chain management
• tactical planning
• softwood lumber industry
• agent-based simulation
• robust design

Acknowledgements

The authors thank National Science and Engineering Research Council of Canada (NSERC) and the FORAC Research Consortium (www.forac.ulaval.ca) for their financial support, as well as the referees for their valuable suggestions which have helped us improve this article.

Notes

1. ‘Board feet measurement’ (bfm) is a common unit of measurement used in the lumber or timber domain, which represents 144 cubic inches of wood (1 foot × 1 foot × 1 inch).

2. The ‘smaller-is-better’ approach is selected because we need to minimise inventories and backorders. When, for example, we need to maximise profits, the indicated approach is ‘larger-is-better’. In both situations, the larger SN ratio gives the best setting because it is a robustness indicator. The difference is that the SN ratio can be calculated in three different ways depending on the situation: function to minimise, to maximise or even to reach a target.