Skip to Main content Skip to Navigation
Theses

Design, simulation and analysis of collaborative strategies in multiagent systems: The case of supply chain management

Abstract : A supply chain is a set of companies that manufacture and distribute products to end-customers. We model each company as an intelligent agent to study the ``bullwhip effect'', that is, the amplifcation of the variability in orders placed by companies. This phenomenon can also be seen as a particular case of stream fluctuations in a distributed system. Such fluctuations incur costs due to higher inventory levels and supply chain agility reduction. This cost has been estimated at 40-60 million USD for a 300 kton paper mill.

Since the bullwhip effect is due to a lack of coordination, we propose two principles that any coordination mechanism should respect: (i) ordering what is ordered from us eliminates the bullwhip effect but do not manage inventory, and (ii) companies should react only once to each market consumption change. In order to validate these two principles, we simulate a forest supply chain called the Québec Wood Supply Game. This game is used to teach the bullwhip effect. Each company-player is modelled as an intelligent agent applying a given ordering rule. For this simulation, we have designed two rules that respect our two principles.

Firstly, we compare experimentally the efficiency of these two rules with five other rules. We assume here that the supply chain is homogeneous, that is, every company uses the same ordering rule. In this way we check that our two coordination mechanisms, i.e., our two ordering rules, are efficient for the supply chain as a whole.

Secondly, we stop assuming the supply chain is homogeneous to carry out additional simulations in order to build a game. The analysis of this game according to Game Theory shows that companies do not have incentive to cease unilaterally from using our two coordination mechanisms (Nash equilibrium).
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00091295
Contributor : Thierry Moyaux <>
Submitted on : Tuesday, September 5, 2006 - 4:51:39 PM
Last modification on : Monday, July 20, 2020 - 1:06:04 PM
Long-term archiving on: : Tuesday, April 6, 2010 - 12:47:03 AM

Identifiers

  • HAL Id : tel-00091295, version 1

Citation

Thierry Moyaux. Design, simulation and analysis of collaborative strategies in multiagent systems: The case of supply chain management. Networking and Internet Architecture [cs.NI]. Université Laval, 2004. English. ⟨tel-00091295⟩

Share

Metrics

Record views

794

Files downloads

1453