Planning and coordination for decentralised business units in a lumber production company

Ludwig Dumetz (1) , Jonathan Gaudreault (1) , André Thomas (2) , Hind Bril El-Haouzi (2) , Nadia Lehoux (1) , Philippe Marier (1)

1 Forac Research Consortium
2 CRAN - Centre de Recherche en Automatique de Nancy

Mathematical models are frequently used in various industries to help decision makers to plan their activities. Many researchers develop planning models for a specific business unit. As an example, the forestproducts industry has access to specialised mathematical models for each production centre of thesawmill (sawing, drying, finishing units) which calls for decentralised planning. In the literature, these models are most of the time evaluated/tested separately, or connected using heuristics and testedin a " static " context (" one shot " demand datasets are provided).In this paper, we want to simulate theuse of these tools by companies for a long period of time (new orders are arriving dynamically, planning need to be updated periodically, etc.).Different coordination mechanisms are compared. We show what the performance of the company would be in terms of accepted orders and average inventory using various coordination mechanisms and order acceptance policies such as available-to-promise (ATP) and capable-to-promise (CTP). Our main finding is that previously published coordination mechanisms for decentralised planning leads to bad CTP implementations which open very interesting research avenues.

Mots clés: co-production; coordination mechanisms; simulation; planning; divergent process

Domaines: Automatique / Robotique

Ludwig Dumetz, Jonathan Gaudreault, André Thomas, Hind Bril El-Haouzi, Nadia Lehoux, et al.. Planning and coordination for decentralised business units in a lumber production company. 2017. ⟨hal-01425811⟩